Citation
The mass mediated sports hero as a role model for adolescent males

Material Information

Title:
The mass mediated sports hero as a role model for adolescent males
Creator:
Strudler, Keith Andrew
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Adolescents ( jstor )
Behavior modeling ( jstor )
Heroes ( jstor )
Modeling ( jstor )
Regression analysis ( jstor )
Role models ( jstor )
Self concept ( jstor )
Self image ( jstor )
Statistical models ( jstor )
Statistics ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
23010395 ( ALEPH )
45080028 ( OCLC )

Downloads

This item has the following downloads:


Full Text










THE MASS MEDIATED SPORTS HERO AS A ROLE MODEL FOR ADOLESCENT MALES: MEASURING SELF-CONCEPT CONGRUITY WITH THE PERCEIVED IMAGE OF THE HERO TO PREDICT ROLE MODELING BEHAVIORS

















By

KEITH ANDREW STRUDLER












A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2000





























For Seth, CJ, and X Butlers everywhere.














ACKNOWLEDGMENTS

Thanks to my parents and family for always believing that I could make it and to my friends for not holding it against me if I did not. Thanks to all the professors who made me think and all the staff who made me remember. Finally, a special thanks to Dr. John Sutherland and Dr. Kim Walsh-Childers who, both in their own way, provided the guidance, knowledge, and patience necessary to write this dissertation.



































i1














TABLE OF CONTENTS



ACKNOWLEDGMENTS ..................................................................... ........................ 1

A B STRA CT ................................................ ...................................................... ... .... vi

CHAPTERS

1 IN T R O D U C T IO N .................................................................... ................................

M ass M edia R esearch .............................................................. .................................. 3
Self-Concept Theory ....................... ............ ............................................. 11
Problem Statement .............................................................................................. 13

2 REVIEW OF LITERATURE ..........................................................16

Role Models and Behavioral Modeling .................................................16
The Sports Hero as a Role Model Figure.............................................................24
Self-C oncept ................................................................................ ............................ 38
C ongruity Theory...................................................... ...............................................54
H ypotheses .................................................................................. ............................. 58

3 M E T H O D S ................................................................................ ............................. 62

Subjects for Study ..................................................... ...............................................62
Tests and Test Administration ........................................................ 65
Issues of Validity for this Study....................................................75
D ata A nalysis ......................................................... ..................................................80

4 RESULTS AND DISCUSSION ...................................................... 96

D escriptive Statistics........................................................ .......................................97
Self Concept, Image of Sports Hero, and Distance Scores .......................................... 110
Results for Hypothesis I and Hypothesis II ........................................ ......124
Summary Chart of Significant Results of Regression Analyses for H 1 and H2.......... 135 Summary of Results of H1 and H2 ..............................................................................38
Domain Importance and Distance Worth.....................................................................40
Results for Hypothesis III and Hypothesis IV ...................................... ......143
Summary Chart of Significant Results of Regression Analyses for H3 and H4.......... 155

iv








Summary of Results of H3 and H4 .............................................................................58
Sum m ary of R esults ..................................................................................................... 59

5 C O N C LU SIO N S..................................................... ..............................................161

D escriptive Statistics.................................................. ............................................161
Hypothesis I and Hypothesis II................................................... ...................168
Hypothesis III and Hypothesis IV...................................................175
Lim itations of the Study .............................................................................................. 78
Recommendations for Future Research .......................................................................81

APPENDICES

A SELF-CONCEPT/IMAGE OF THE SPORTS HERO QUESTIONNAIRE ............188

B ANSWER GUIDE FOR SELF-CONCEPT/IMAGE OF THE SPORTS HERO......193 C ROLE MODELING QUESTIONNAIRE ..................................................................97

R EFER EN C ES .......................................................... .................................................200

BIOGRAPHICAL SKETCH .........................................................218






























V













Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

THE MASS MEDIATED SPORTS HERO AS A ROLE MODEL FOR
ADOLESCENT MALES: MEASURING SELF-CONCEPT
CONGRUITY WITH THE PERCEIVED IMAGE OF THE HERO TO PREDICT ROLE MODELING BEHAVIORS By

Keith Andrew Strudler

May 2000

Chair: John Sutherland
Major Department: Journalism and Communications


The sports hero has become a popular and influential figure in modern American society. The growth of this social phenomenon can be attributed largely to heavy mass media coverage of sports, with a particular emphasis placed on the star athlete. This emphasis by a variety of media outlets, most notably television, allows fans to learn more about their favorite athletes, both positive and negative, than possible for previous generations of fans. Such coverage has created concern over the potential role modeling influence of star athletes, particularly such behavioral influence on adolescents. Two primary areas of interest are the actual behaviors that are most likely to be role modeled and the conditions under which such behavioral role modeling will occur. A theoretical foundation to understanding human behaviors is provided by congruity theory. According to congruity theory, the closer one's self-concept is to his or her image of a consumer product, the more likely this person is to purchase this product.




vi








This study adapts congruity theory to examine behavioral role modeling of mass mediated sports heroes by American adolescent males. Surveys were given to 172 9th and 10th grade male student athletes. These subjects first identified their individual sports hero. Subjects then used Harter's Self-Perception Profile for Adolescents to rate their real and ideal self-concepts and the image of the their sports heroes. Finally, subjects rated role modeling of their sports heroes for a variety of behaviors. Regression analyses were completed to examine whether the proximity between self-concept and the image of a sports hero, or congruity, correlated with role modeling behaviors of the sports hero. According to the statistical analysis, behavioral role modeling correlated with congruity between real self-concept and the image of the sports hero for a limited number of behaviors. This correlation came primarily with sports-related behaviors. Almost no correlation was found for the congruity between ideal self-concept and the image of the sports hero. Additionally, the increasing importance of an area of self-concept did not correlate with increased role modeling as predicted.

























vii













CHAPTER 1
INTRODUCTION



Star athletes for years have assumed a heroic position in American society.

Historically, these star athletes were assumed to represent the best of American society. As written by Lewis Lord in U.S. News & World Report as an obituary to the recently deceased baseball hero Joe DiMaggio,

the recent death of baseball legend Joe DiMaggio recalls an era when a
sports hero embodied America's hopes and dreams.. Stories about and
pictures of DiMaggio are extremely evocative, representing as they do the
way America was -- or at least wished it was. (1999, p. 18)

DiMaggio played his career in an era when television and media did not heavily influence and potentially overexpose star athletes and when such athletes did not command the enormous salaries relative to the society at large. However, mass media, particularly television, has had a profound influence in recent years on popular sports and the fans' perspectives of star athletes. Certainly, fans now have greater opportunity to watch their favorite athletes perform. Fans also have greater insight into the personal lives of star athletes, an insight that often has exposed a less than heroic side of these stars. The money generated through media coverage has allowed star athletes to command yearly salaries of millions of dollars, not including the income they derive from advertising and endorsements. Obviously, the perspective of today's mass mediated sports stars differs greatly from that of the sports heroes of past generations.

For a variety of reasons, many journalists and social critics have argued that the sports hero no longer exists in modern American society. Popular journalist Mike Lupica wrote in 1994,



1





2


America's sports heroes were once larger than life, but today even the idea
of the sports hero is dead in America. A great number of sports heroes
have fallen off their pedestals in the public eye. .. Sports heroes have
destroyed themselves, but the fans and the media have helped them along.
The fans want to know too much about the players, and the media are only
too willing to give them what they want. (p. 62)

Whether the sports hero still exists in America or whether this figure has merely changed with changing social conditions is debatable. However, it is obvious that the star athlete assumes a visible and lofty position in American society.

Of greater importance than the labeling of star athletes is the potential influence of these highly conspicuous persons. It is often assumed that popular star athletes, such as basketball players Michael Jordan and Charles Barkley, are role models to many views, particularly children and adolescents (Peart, 1996). If young sports fans model positive behaviors of athletes, then this relationship has a positive influence on American society. Positive behaviors of star athletes might include athletic skill, good sportsmanship, and community involvement. However, if adolescents model the highly publicized negative activities and characteristics of many star athletes, including drug use, greed, and violent behavior, then this status is highly problematic for society at large (Benedict & Yaeger, 1998; Keteyian, 1998).

Therefore, the more important debate about sports heroes focuses not only on the characteristics of these athletes, but rather whether these star athletes serve as role models to America's youth. This debate is complicated by the increasing diversity of American society, the diversity of star athletes, and the huge number of sports stars accessible through various media outlets. Some journalists and social critics have written that athletes are no longer role models for American adolescents. Harry Edwards expressed that opinion in Sport magazine (1994), writing,

athletes have long been expected to be role models, but this
notion is becoming outdated. The role model expectation is based on the
assumption that star athletes and their fans share a uniform group of
values, but this is no longer true. Indeed, the racial, ethnic, gender, and





3


lifestyle diversity of the United States has shattered all but the pretense of
consensus upon which this notion is founded. Athletes are sports
entertainment celebrities with no more special calling, capability, or
obligation to be role models than anyone else in the mass-oriented
entertainment business. (p. 32)

According to this viewpoint, not only have star athletes lost their heroic credential, but due to the complexity of modern American society, these athletes are no longer role models.

The question of whether sports heroes still serve as role models for young

American sports fans is of great interest to parents, educators, and society in general. Further, it has become increasingly important to understand the specific behaviors that are most likely to be modeled and the motivations for individual fans to model their own favorite sports stars. This research project will address these topics to help gain a better understanding of the function of the modem star athlete in the lives of American adolescents. One area of study crucial to understanding the modem American sports hero is mass media theory and research. This topic will be reviewed extensively throughout this dissertation.



Mass Media Research


Years of academic research has focused on the vast impact mass media, and

specifically television, historically have had upon society. This school of research has produced theories predicting a range of effects, varying from theories examining general societal impact of media to those dealing with a narrow range of media types and populations.

Obvious reasons exist for the ongoing interest in this research agenda. By 1980, televisions were present in 98% of American households, a percentage that has held steady since that time (Harris, 1994b; Andreasen, 1990). Television viewing comprised an inordinate amount of American leisure time, more than four times more than any other





4


single activity (Cohen, 1993; Spring, 1993). Americans watch, on average, four hours of television a day, and the television remains on for seven hours a day, on average, in the American household (Harris, 1994b; Kubey & Csikszentmihalyi, 1990). Statistics on television in the lives of children are even more staggering. American youths, on average, watch approximately 1,500 hours of television a year, far greater than the 900 hours the American youth will spend in school annually (Harris, 1994b). With so much of America's time invested in television, it is natural and vital that researchers attempt to understand the various impacts of television content on society. Television and Society

America's interest in television has produced a society somewhat dependent upon television for its daily existence. Ball-Rokeach and DeFleur (1976) explained such a relationship with their "dependency theory," which suggests that, "particularly in a modern urban-industrial society, audiences have a high level of dependence on mass media information" (Severin & Tankard, 1992, p. 262). America, a highly industrial and increasingly urban society, exemplifies such a reliance on mass media information. This dependent relationship creates a need for continuous research on the ramifications of such dependence on American society.

One area of particular interest for this research project is the potential for

television viewers to copy the behaviors and actions of characters they see on television and to use television figures as role models. Much of the prior research has focused on violent behaviors and models, including such notable research works as the Surgeon General's Report on Television Violence, Television and Growing Up (1972), which established tentative relationships between viewing violent television content and modeling of those behaviors (Lowry & Defleur, 1983). Other examples of research concerning modeling of television content have shown that viewers may model anti-social behaviors such as smoking and drinking (Charlton & Blair, 1989; Botvin et al., 1991; Conrad et al., 1987), prosocial behaviors (Cohen, 1993; Bandura, 1994), styles





5


of dress and fashion (Muuss, 1975; Wilson & Sparks, 1996; Harris, 1994b), verbal expression (Muuss, 1975), and career selection (Muuss, 1975; Harris, 1994b).

Research has indicated that many television viewers look to television for role

models, and these televised role models have replaced traditional models such as family, peers, and teachers (Jung, 1986; Hendricks, 1992; Harris, 1994a; Pleiss & Feldhusen, 1995; "Role models", 1989; Duck, 1990). This may be particularly true during adolescence, an age when children often reject traditional role models, such as parents, and seek new role models to help establish their own identity (Jung, 1986; Orenstein, 1994; Krech et al., 1962; Hendricks, 1992; Felker, 1974). While one would hope television role models would display positive behaviors, this may not always be the case, and critics have suggested that much of television content portrays potential role models engaged in negative activity (Harris, 1994a). These potentially negative role models on television have drawn growing attention, and they are paramount in the study of the questionable behaviors of popular athletes as role models. Sports Television

Sports television, central to this research project, has received growing attention by networks and viewers alike in recent years. While the first televised sporting event was a college baseball game between Columbia and Princeton in 1939, sports did not gain much attention from television networks until the 1960s (McPherson et al., 1989; Davies, 1994; Coakley, 1994). Sports has been an increasingly vital part of the television industry since that time.

Growing interest in sports television developed from a variety of factors,

including a gradual exodus from downtown living areas, where stadiums are located, to the suburbs; realization by networks that sports offered an opportunity to boost ratings during traditionally slow times, such as Sunday afternoons; and the ability for advertisers to consistently attract young male viewers (Davies, 1994; Nixon, 1984; Harris, 1994a; Harris, 1994b; Katz, 1996; McPherson et al., 1989; Coakley, 1994; Clarkin, 1990).





6


Sports programming has increased through to the 1980s, when, on average, sports programming accounted for 15% of total programming on commercial television, a percentage that has held steady through the mid-1990s (Harris, 1994a; Coakley, 1994). Several television networks, including ESPN and Fox Sports, are now devoted solely to sports content, 24 hours a day, making sports extremely accessible to television viewers.

Along with the increase of sports media coverage has come a change in the

content of televised sports. Much of this can be attributed to the enormous amounts of money networks pay for the rights to televise amateur and professional sports. For example, in 1998 four networks (ABC, CBS, ESPN, and FOX) agreed to pay the National Football League (NFL) $17.6 billion over eight years for the right to broadcast professional football (Goldberg, 1998). This amount far surpasses the amount teams receive each year in ticket revenues. The National Basketball Association (NBA) has signed a similarly lucrative contract with two networks (NBC and Turner), receiving $2.4 billion for four years (Farhi, 1997). This does not include revenues received from lesser contracts signed with local networks for regional sports coverage. Given the dependence of major sports leagues on television for financial success, popular sports are influenced greatly by television networks and orchestrated for television viewing even more so than for live spectating. This coverage may cause the television audience to believe that they are watching the real game, not a media representation of the game (Temple, 1992; Coakley, 1994; Kinkema & Harris, 1992; Nixon, 1994; Harris, 1994b). In other words, viewers may be oblivious to the potential influence of networks to shape and influence the sporting events regularly seen on television. Additionally, the exorbitant prices of tickets to professional sporting contests (an average NBA ticket cost $36.32 in 1998; an average NFL ticket cost $40.93 in 1998) has made live attendance of such professional sporting contests virtually impossible for most Americans (Fehr, 1997; Heath & Solomon, 1997). Therefore, professional sports, an American institution rivaling religion in popularity and societal influence (Prebish, 1985; Novak. 1976; Sage, 1993), has





7


become a phenomenon Americans see almost exclusively on television (Temple, 1992; Chandler, 1985). Coakley (1994) explained,

the emphasis comes in a form that makes it seem as if the "real" game is
the one on television, not the one seen in the stadium. In other words,
television has constructed sports and spectator experiences in important ways. It's happened so smoothly that most people think that when they
watch a game on television they are experiencing sports in a "natural"
form! (p. 332)

What most Americans see and know about professional sports, they see and know from television. Considering the potential extent to which television can impact society, sports media should receive considerable attention from academics. Television Stars and Sports Heroes

A focal point of televised sports and the focal component for this research project is the star athletic performer. Through the process of mediation -- using narration, visuals, intimacy with particular athletes, and constant fixture on star performers -television has placed emphasis on the best individual performers and their greatness (Duncan & Brummett, 1987; Weisman, 1993; Katz, 1993; Harris, 1994b; Kinkema & Harris, 1992; Hargreaves, 1986; Figler & Whitaker, 1995; Shugar, 1988). Harris (1994b) wrote,

often the "star mentality" of television and the entertainment business in general affects the presentation of sports in the media. The superstars are exalted and glorified far more than, for example, a show of fine teamwork
on the field. Stories about sports tend to focus on individual performers,
much like stories about religion focus on the Pope or Jerry Falwell or
stories about government tend to focus on the president. (p. 129)

This emphasis on the star has been accomplished not only through the actual

coverage of regular sporting events, but also through other sports media, including sports highlights programs, pre- and post-game programs, and entertainment-based programs that highlight individual athletes over team performance (Kinkema & Harris, 1992; Figler & Whitaker, 1995; Harris, 1994b). Networks have promoted individuals in their





8


ongoing effort to increase the popularity of their programming, and the respective sports leagues hope to prosper from this coverage through the potential sale of merchandise from these star performers and overall increased interest in their sports product. This emphasis on the individual has created a society in which television sports viewers can learn more about popular athletes than in prior times, and the star sports figure has become one of the most prominent and well-known persons in American society (McPherson, 1989; Nixon, 1984; Harris, 1994a; Harris, 1994b; Coakley, 1994; Duncan & Brummett, 1987; Kinkema & Harris, 1992).

This focus on individual star athletes in sports television has promoted the

growing societal interest in the sports hero. As previously mentioned, the sports hero has been a part of American society for years before the existence of television, but television has contributed to changes in this construct (Porter, 1983; Smith, 1973; Voigt, 1978). Before the growth of sports television, sports heroes such as baseball legend Babe Ruth were accessible local figures, and many fans had the opportunity to meet their sports heroes in person (Walden, 1986; Katz, 1993). However, the lack of heavy coverage and scrutiny from mass media kept fans from knowing all the details, both good and bad, of the lives of their sports heroes. Due to television's current ubiquitous presence, while the sports fan today will not personally know his or her sports heroes or even have the opportunity to watch them play in person, these fans likely will know much more about their heroes (Walden, 1986; Nixon, 1984; Harris, 1994a; Harris, 1994b; McPherson et al., 1989; Malone, 1993; Paterno, 1994). Television has allowed greater access to sports heroes than ever before, as Caughley (1984) explained: "Even an average fan knows more than a baseball player's name. He also knows his team, uniform number, position, batting average, salary, and something about his appearance, personality, medical history, and off-the-field conduct" (p. 32).

Despite the increasingly visible information about the negative actions of popular sports stars, including rape, drug abuse, and a variety of criminal activities, which has





9


lead some theorists to question whether the sports hero still truly exists in American society (Rollin, 1983; Crepeau, 1985; Walden, 1986; Smith, 1973; Fairlie, 1978; Nixon, 1984; Harris, 1994a; Farrey, 1997; Ryan, 1995; Coakley, 1994; Nack & Munson, 1995; Goldberg, 1995), sports figures have frequently been chosen as heroes in surveys of America teens ("Heroes of"', 1990; Simons, 1997; Harris, 1994a; Harris, 1994b). Such survey results refute the idea of a recent extinction of the American sports hero and instead infer that the hero exists as a constantly evolving construct (Fishwick, 1983; Fishwick, 1985; Rollins, 1983; Harris, 1994a). In other words, the qualities once considered essential for the sports hero may be very different than those of the modem sports hero.

Sports fans now seem willing to overlook negative, or perhaps unheroic, activity in certain areas of the lives of their sports heroes. This is likely due to the impressive physical skills of the popular sports figure. A star athlete, while performing spectacular, seemingly impossible feats on the playing field or in the arena, transcends typical mortal athletic potential and appears, if only for that brief period of time, almost god-like (Nixon, 1984; Oriard, 1982; Goodman, 1993; Walden, 1986; Birrell & Cole, 1981; Loy & Hesketh, 1984). Sports may be one of the few remaining areas of American society in which one consistently can find greatness, and fans may be willing to overlook negative activity to embrace this greatness in their sports heroes. This, coupled with the astonishing growth of sports television, has made sports a very popular and frequent area from which American society finds its heroes. As Joan Ryan wrote (1995), "Every culture has its gods, and ours hit baseballs, make baskets, and score touchdowns" (p. 85). Regardless of the views one takes of the heroic condition in American society, the fact remains that the popular sports figure has continued to grow as a popular choice of hero for American society, and television has contributed greatly to this growth.





10


Societal Functions of Sports Heroes

A more clouded but perhaps more important topic is the popular belief that athletes inherently serve as role models, so much so that the terms "hero" and "role model" are often used interchangeably ("Role models", 1989; Malone, 1993; Harris, 1994b; Simons, 1997; Wilson & Sparks, 1996; Parker & Lord, 1993; Taylor, 1991; "Michael Jordan's", 1991; Goodman, 1993). In a column authored for Sports Illustrated magazine, NBA basketball star Karl Malone (1993) wrote, "We don't choose to be role models, we are chosen. Our only choice is whether to be a good role model or a bad one" (p. 84). A lesser held, contrary opinion is that while popular mass media figures such as sports heroes may be idolized, they often do not serve as role models because they are not perceived as similar enough to fans of the hero (Groller, 1995; Goodman, 1993). In other words, while fans may enjoy watching their sports heroes, these fans cannot actually view these heroes as role models due to the vast perceived differences between themselves and these athletes. Yet another suggestion is that sports heroes can serve as role models to certain people for certain areas of their life (Simons, 1997; Wilson & Sparks, 1996; Harris, 1994a; Harris, 1994b; Dononhew, Helm, & Haas, 1989; Nixon, 1984). Research about sports heroes should address whether the sports hero is a role model and the specific conditions under which a sports hero is most likely to serve as a role model.

Authors have addressed the potential for mass media figures to serve as role

models. Caughley (1984) explained the potential relationship between media figures and their fans, writing,

... relationships are those in which the media figure becomes the object
of intense admiration. Such relationships approximate the general
stereotypic conceptualization of the fan, but much more is involved than
esthetic appreciation. Characteristically, the admired figure comes to represent some combination of idol, hero, alter ego, mentor, and role
model. (p. 53)





11


As popular sports heroes have become mass mediated figures, they too are subject to the same types of relationships with their fans. Many athletes have become aware of their potential influence on the viewing public. Weisman (1993) wrote on this topic,

athletes' behavior, whether it involves drug use, the shoes they wear or their views on society, will be emulated. They are taught, very early in
their careers, to be the best example of American manhood, or at the very
least to escape with all their defects undetected. (p. 166)

In order for such a definitive statement to be made about the influence of televised star athletes, detailed research should address the specific impact of these mass mediated figures. Critical insight and research in this area would help society to further understand the impact of the positive and negative actions of sports heroes.



Self-Concept Theory


Importance of Self-Concept

One theoretical framework used to study motivations for human behavior is self-concept theory (Grubb & Grathwohl, 1967; Ross, 1971; Dolich, 1969; Hattie, 1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon & Bitta, 1979; Krech et al., 1962). Humans typically develop their wants and goals around maintaining and improving their self-concept, which can be defined as "myself as I see myself' (Kretch et al., 1962; Loundon & Bitta, 1979; Sirgy, 1983; Hattie, 1992). This desire to maintain and improve self-concept is a strong predictor for purchase intentions (Grubb & Grathwohl, 1967; Ross, 1971; Dolich, 1969; Hattie, 1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon & Bitta, 1979; Krech et al., 1962). Self-concept theory also has been linked to behaviors such as choosing friends and academic efforts in school (Felker, 1974; Hattie, 1992).

While most of the research on self-concept and behavior has focused on consumer behaviors, the link between self-concept and behavior has been strongly established and





12


is a valuable tool for future investigation into human behavior. Grubb and Grathwohl (1967) explained the importance of this topic, writing,

the self represents a totality which becomes a principal value around
which life revolves, something to be safeguarded and, if possible, to be
made still more valuable. An individual's evaluation of himself will
greatly influence his behavior. (p. 204)

Felker (1974) provided a more detailed explanation for the relationship between self-concept and behaviors, writing,

the role of self-concept is three-fold. The self-concept operates as a
mechanism for maintaining inner consistency; the self-concept
determines how experiences are interpreted; and the self-concept
provides a set of expectancies. Each of these three factors is a powerful
determiner of behavior. (p. 7)

Although self-concept theory has not previously been used to examine role modeling behaviors of sports heroes, use of this theory, which predicts that people continuously try to maintain and improve self-concept, provides a promising new framework to examine the motivations for such role modeling. Role Modeling Sports Heroes and Self-Concept Theory

This research project will look at the behaviors involved in role modeling sports heroes as related to self-concept. Similar types of studies have been done for consumer behaviors, determining that similarity between self-concept and product image correlates with purchase behaviors (Dolich, 1969; Grubb & Grathwohl, 1967; Ross, 1971; Hattie, 1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon & Bitta, 1979; Krech et al., 1962). Similarly, researchers have suggested that mass media figures can serve as role models, particularly for adolescents (French & Pina, 1991; Bandura, 1992; Harris, 1994b; Caughley, 1984). This research project incorporates these two concepts while examining a increasingly popular, criticized, and critical aspect of American society, the sports hero.





13


Problem Statement


Americans, and particularly American adolescents, historically have held up

popular sports figures as heroic figures. This is not surprising, as theorists have asserted that the need for heroes is inherent in human societies (Fishwick, 1983; Walden, 1986; Harris, 1994a; Schlesinger, 1968; Browne, 1983), and sports has become one of America's most celebrated pastimes and personifications of American dreams and values (Nixon, 1984; Porter, 1983; Goodman, 1993; Oriad, 1982; Ryan, 1995; Walden, 1986; Smith, 1973; Coakley, 1994; Crepeau, 1985; Harris, 1994a; "Role models", 1989; "Michael Jordan's", 1991).

Given the growing mass media interest in professional and major collegiate

sports, Americans now have more access to watch star athletes than American sports fans of generations past ("T.V. sports", 1996; McPherson, Curtis, & Loy, 1989; Coakley, 1994; Davies, 1994; Clarkin, 1990; Harris, 1994a; Harris, 1994b; Kellner, 1996; Caughley, 1984; Hargreaves, 1986; Kinkema & Harris, 1992; Denzin, 1996). Accordingly, the opportunity to find heroes in sports appears endless. In a society where popular athletes have become incredibly wealthy figures about which spectators know more than ever before, the issue of choosing a particular sports hero is a complex phenomenon. While theorists have suggested that heroes no longer truly exist in America (Boorstin, 1961; Klapp, 1962; Monaco, 1978; Lubin, 1968; Kelly, 1993; McLuhan, 1964), and many of today's popular athletes have displayed behaviors that certainly appear "unheroic," athletes are still popular choices as heroes among American youths and adolescents ("Heroes of", 1990; "Role models", 1989; Wilson & Sparks, 1996; Simons, 1997). Television has only aided this process, bringing heroes and hero worshippers together more easily (Higgs, 1978). However, one is left to wonder exactly what the impact of these sports heroes is on American adolescents.





14


Critics and fans of sports heroes alike have suggested that popular American sports heroes serve as role models for American adolescents (Harris, 1994a; Harris, 1994b; "Role models", 1989; Malone, 1993; Simons, 1997; Wilson & Sparks, 1996; Parker & Lord, 1993; Taylor, 1991; "Michael Jordan's", 1991; Goodman, 1993). Little scientific research has addressed the motivations for such role modeling or the conditions under which such role modeling is most likely to occur for the individual adolescent. Socioeconomic status, race, and parental and peer influences have been suggested as potential factors affecting such role modeling (Harris, 1994a; "Role models", 1989; Simons, 1997; Wilson & Sparks, 1996; "Michael Jordan's", 1991; Goodman, 1993). However, these factors do not address the individual role modeling decisions of each adolescent and the motivations for the individual to model the behaviors of his or her own particular sports hero. One area of research which has examined motivations for individual behavioral intentions is the study of individual self-concept and how this self-concept relates to the images of other objects (Dolich, 1969; Grubb & Grathwohl, 1967; Ross, 1971; Hattie, 1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon & Bitta, 1979; Krech et al., 1962). This research on self-concept has primarily been used to better understand consumer behaviors and individual purchase decisions. However, self-concept research also presents a potential means to better understand a variety of behaviors, including the emulation of sports heroes addressed in this research.

This theoretical approach to studying adolescent role modeling of sports heroes has lead to the following research questions:



1) Among adolescent males, how do the relationships between self-concept and perceived images of a sports hero relate to behavioral intentions to role model a

sports hero? In other words, how do the perceived similarities and differences between one's individual self-concept and the image of a sports hero relate to

one's intentions to emulate that sports hero?





15


2) What specific role modeling behaviors are most strongly related to these

relationships between self-concept and the image of a sports hero?

3) Which areas of self-concept are most influential in predicting adolescent role

modeling of a sports hero?


Chapter 2, the Literature Review, addresses prior research related to these

questions. Specifically, each construct used in this study is defined and discussed. Prior research on related subject matter is reviewed. Finally, Chapter 2 addresses areas of research that have not been studied adequately as well as the ways that this research project adds to the base of knowledge and contributes meaningful information to the academic community.

Three additional chapters are included in this dissertation. Chapter 3, Methods, addresses the specific methods and statistical analysis used in this dissertation. Chapter 4, Results and Discussion, details the results of the stated analysis. Finally, Chapter 5, Conclusions, addresses the results, discusses the importance of these results, and explains potential reasons these results were found.













CHAPTER 2
LITERATURE REVIEW



This chapter reviews relevant prior research. Additionally, important terms are defined for use in this research. Finally, the contribution of this research project is discussed. The topics reviewed in this chapter, in order, are as follows: role modeling, characteristics of the sports hero, the sports hero as a role model, self-concept, and self-concept congruity theory.



Role Models and Behavioral Modeling


Sports heroes are widely assumed to be role models, especially for children and adolescents, although such assumptions are often made based on personal opinion and lack concrete support (Malone, 1993; Harris, 1994a; Harris, 1994b; Wilson & Sparks, 1996; Nack & Munson, 1995; Taylor, 1991; Walden, 1986; "Michael Jordan's", 1991; Simons, 1997; "Role Models", 1989). One primary purpose of this research is to determine how the relationships between self-concept and the image of sports hero determine role modeling behaviors of this sports hero. Therefore, an adequate understanding of the role model and its function and influence is necessary. This section of the literature review defines role models, explains the process of imitating role models, and outlines the conditions under which such modeling is most likely to occur. Finally, issues specific to role models and adolescents are discussed.

The term "role model," like self-concept, has been used vaguely in reference to a variety of different constructs, so it is necessary to define this term specifically. A role



16





17


model has been described as "someone whose life and activities influenced the respondent in specific life decisions" (Bassoon & Owe, 1980, p. 559). This description does not address actual copying of behaviors or intentions to copy behaviors, but rather presents the role model as one influence in such intentions. Other definitions assert that role models are "adults who are worthy of imitation in some area of their life" (Pleiss & Feldhusen, 1995, p. 163) or "someone who demonstrates the appropriate behavior for a specific role or relationship with another person" (Jung, 1986, p. 528). Role models generally exert influence over one of three areas of life: educational, occupational, or personal (Bell, 1970; Pleiss & Feldhusen, 1995).

Many of the descriptions of the role model imply the role model is a positive influence whose actions should be copied and emulated. In other words, people who display negative behaviors, such as wife abuse, drug use, or poor performance in employment or school, are not role models (Malone, 1993; "Role models", 1989; Richards, 1998; Golden, 1986). However, this research adopts the perspective that role models can exhibit both positive and negative behaviors and characteristics. A description written by Hendricks (1992) gives insight into this position, noting, "We all have role models; some positive, some not so positive. They serve as archetypes, proxies we scrutinize as guides for our own actions and perhaps too as templates for possible selves" (p. 7).

Role modeling is a component of the more general term modeling. In fact, the term "model" is often interchanged with "role model." While role modeling certainly involves the processes of modeling, a clear distinction will be made to allow for sufficient definition of this term for this research. Role modeling refers to a more specific process than the more generic process of modeling of behaviors. Specifically, modeling is a measure of the influence of observed behavior on an observer (Bandura & Walters, 1963; Bandura, 1977; Bandura, 1994; Jung, 1986; Muuss, 1975). This process refers to a specific set of behaviors that are not necessarily tied to a specific role, and this process





18


can be replicated in laboratory settings. For example, learning the technical processes involved in juggling from a model in a laboratory setting would be an example of modeling. The person demonstrating the juggling behaviors would be the model. However, role modeling, while it does involve the technical processes of modeling, also involves an emotional attachment between the observer and role model. Jung (1986) explained distinct differences in the two constructs: "Modeling does not imply that there is any emotional tie between the model and observer. In contrast, many instances of role modeling .. seem to require some affectional bond for the model to be effective in influencing the observer" (p. 527). Role modeling involves a relationship between role model and observer, and such relationships can influence behavior through learned roles in specific situations (Hendricks, 1992; Brim, 1968; Bell, 1970; Jung, 1986; Miller, 1983). Therefore, the process of role modeling involves adopting and modeling behaviors based on an emotional tie between observer and model, and this process of modeling involves context and relationships with the role model. In order for role modeling to occur, the role model and the observer must interact and the observer must identify the role model as such (Bell, 1970; Basow & Howe, 1980).

One popular view of role model influence is that everyone has role models, and these figures serve as guides for our own actions and possible templates for our own selves (Hendricks, 1992; Gecas & Mortrimer, 1987). Such influence can be behavioral and attitudinal. However, behavioral influence has been studied and discussed more frequently than attitudinal influence by researchers and theorists (Basow & Howe, 1980; Almquist & Angrist, 1971; Hendricks, 1992; Jung, 1986), and these influences are the focal point of this research project.

Observers are not limited to one role model, and it is quite normal for observers to have several role models for different aspects of their lives (Jung, 1986). Additionally, some role models influence more aspects of observers' lives than other role models (Gecas & Mortimer, 1987; Basow & Howe, 1980). Hendricks (1992) explained this,





19


writing, "Some models are relevant in specific realms only; that of the older, wiser colleague at work, or the sports figure we seek to emulate. Others cut across specific roles and have a more profound influence on our lives" (p. 7). Most descriptions of role modeling refer to the process of copying the behaviors of the role model, although they also can, in some circumstances, refer to avoiding behaviors of the role model and actively seeking alternative behavioral techniques to those displayed by the role model (Pleiss & Feldhusen, 1995; Jung, 1986). While including such avoidance techniques broadens the definition of the role model, this research specifically examines the intention to model the behavior of a role model, not the intention to avoid certain behaviors.

Many available sources of role models exist. Persons who have frequent

interaction with and are well known by the observer, including parents, teachers, peers, co-workers, and siblings, typically are popular role models for adolescents. (Schaefer, 1978; Muuss, 1975; Basow & Howe, 1980; Malgady et al., 1990; Hackett et al., 1989; Almquist & Angrist, 1971; Betz & Fitzgerald, 1987). In addition to personal acquaintances, mass media figures also can serve as role models (Caughley, 1985; Schaefer, 1978; "Role models", 1989; Jung, 1986; Pleiss & Feldhusen, 1995; Duck, 1990). In contrast to personally known role models such as parents and peers, mass media figures, including famous athletes, actors, and politicians, have the potential to serve as role models for large numbers of people due to their mass exposure.

Researchers and critics have taken varying positions on the potential influence of mass-mediated figures such as sports hero. One popular belief is that popular, mass-mediated figures, particularly those portrayed as positive or heroic in their behaviors, inherently serve as role models due to their stature and position in society. In fact, the terms "public hero" and "role model" are often used interchangeably without regard to a distinction between the two (Pleiss & Feldhusen, 1995, p. 163; Malgady et al., 1990; Hunter, 1983). Conversely, some have concluded that a public hero cannot truly be a role model because there is almost no potential for interaction between the role





20


model and the observer (Singer, 1991; Hall & Post-Kammer, 1987). Both of these contradictory opinions are too inflexible to fully summarize the potential influence of the mass media figure, and the true influence of this figure lies somewhere inbetween these viewpoints. Research has indicated that mass-mediated figures, particularly those who are considered heroes, can serve as role models (Pleiss & Feldhusen, 1995; "Role models", 1989; Schaefer, 1978; Bandura, 1981; "Michael Jordan's", 1991; Malone, 1993; Kellner, 1996; Weisman, 1993; Wilson & Sparks, 1996). Parker and Lord (1993) have echoed this sentiment in research with African-American males, noting that "highly visible and successful individuals (such as sports personalities or entertainers) make excellent role models for young Black males" (p. 102). Therefore, while popular media figures are not inherently role models, they certainly have the potential to be role models.

Media users often develop intense imaginary relationships with their favorite

mass-media figures, and these relationships contribute to the role modeling potential of such figures (Caughley, 1985; McEvoy & Erikson, 1981). McEvoy and Erikson (1981) called these public heroes "reference idols." Reference idols, although not personally known by the observer, become part of the reference group of the observer, "a personification of values and behaviors perceived as important to the individual" (p. 114). The authors explained the potential influence of these mass media heroes, writing, "Reference idols, real or fictitious, can affect the formation of attitudes, beliefs, identities, and behaviors" (p. 114). In other words, mass media heroes provide example attitudes, beliefs, identities, and behaviors that individuals use in creating, modifying, and/or maintaining/reinforcing their self-concept. It is these potential influences of mass media figures that will be the focus of research for this dissertation.

In summary, it is necessary to define the term "role model" for use in this

dissertation. For this study, the role model is a mass mediated figure (in this instance, a sports figure). The processes of role modeling involve modeling of specific behaviors of an admired figure. Finally, an affectionate bond must exist between the observer and the





21


role model, unlike laboratory models. Therefore, role modeling will occur as part of daily life, making the study of role modeling one of naturalistic inquiry, not laboratory observation.

Conditions of Role Modeling

Many researchers have studied the conditions or factors which make observers

more likely to adopt the behaviors of a role model. These conditions are reviewed in this section. One primary factor is perceived personal similarity between observer and role model (Hackett et al., 1989; Jung, 1986; Decker & Nathan, 1985; Owens & Ascione, 1991; Wilson & Sparks, 1996; Muuss, 1975; Bandura, 1994; "Role models", 1989; Malgady et al., 1990; Burnstein et al., 1961; Stotland & Patchen, 1961; Stotland et al., 1962; Rosekrans, 1967). People are more likely to adopt the behaviors of a role model whom they perceive as having similar attributes to themselves. Specific attributes include sex, race, and cultural identification. Wilson and Sparks (1996) provided examples of this through interviews with African-American teens in reference to their sports heroes. They specifically studied African-American teenagers and their role modeling of African-American basketball stars. They wrote, in reference to potential modeling of clothing styles, "Many of the Black adolescents gave responses acknowledging the celebrity Black athletes' definitive influence on popular style through athletic apparel and the athletic-apparel commercial. This influence was manifested most often in the respondents' desire to look like the celebrity athlete" (p. 414). These adolescents were more likely to use African-American athletes as their role models due to perceptions of racial similarity, cultural identity, and masculine identity. Therefore, the perceived similarities between popular male, African-American athletes and male, African-American adolescent fans contributed greatly to this selection of role models.

Perceived self-efficacy of the observer also can affect the choice of a role model (Simons, 1997; Weisman, 1991; "Role models", 1989; Bandura, 1994). Self-efficacy refers to the a belief that one can successfully model an action (Bandura, 1994). As the





22


individual becomes more confident in his or her ability to perform a behavior, he or she will be more likely to model such behaviors.

Another factor influencing modeling is the perceived benefits the role model gains from his or her behaviors (Ostlund, 1974; Midgley, 1976). Role models who are rewarded for their actions are more likely to be imitated than those who are not rewarded. This has lead to the belief that popular, mass-mediated role models are often very appealing to observers because they are well compensated with fame and wealth (Simons, 1997; Wilson & Sparks, 1996). Viewers, therefore, will attempt to role model those behaviors perceived as instrumental in achieving such fame and wealth. Simons (1997) explained this appeal of wealthy, mass-mediated figures as potential role models to certain groups of adolescents, using basketball star Michael Jordan as an example. Simons wrote,

a lot of them (low-income, African-American adolescents) will tell you
they want to be like Mike.. In this context, being like Mike doesn't mean
becoming an entrepreneur, a corporate spokesman, or a college graduate.
It means being a high-flying, windmill slammer of a ballplayer. (p. 50) In other words, the most glamorous aspects of Michael Jordan (athletic fame, financial riches from athletic accomplishment) made Jordan appealing as a role model to these adolescents.

Self-Concept as a Condition for Role Modeling

Most research on the impact of role models, and specifically athletes as role

models, has focused on the characteristics of the potential role model; the societal role of the role model; and easily observable characteristics of the observer, such as race, socio-economic status, and perceived positive or negative value to society (Simons, 1997; Malone, 1993; "Role models", 1989; Wilson & Sparks, 1996; Chen, 1997; Golden, 1996; Walden, 1986; Pleiss & Feldhusen, 1995). This dissertation addresses the relationship between role model and observer from a slightly different perspective. Instead of measuring the aforementioned characteristics, this research examines





23


individual perceptions of the self, or self-concept, and the influence self-concept has in determining role modeling behaviors. Of primary importance for this research are the relationships between self-concept and the image one has of his role model (for this research, the sports hero). In other words, this research attempts to better understand whether one's behavioral intentions to model the behaviors of a role model are influenced by how similar or dissimilar one's self-concept is to his or her perceptions of his or her sports hero. This approach of examining self-concept has not been widely used to investigate behavioral intentions of role modeling. However, this research approach has been used quite frequently in marketing research, and this dissertation will borrow investigative strategies from such marketing studies. (See later section on "self-concept" for a detailed description of this construct and a review of research and theory in this area.)

Marketing researchers have provided a foundation for the use of self-concept in studies of behaviors and behavioral intentions (specifically, purchase intentions). First, they have established that self-concept maintenance and enhancement is a factor in determining behavior (Sirgy, 1983; Landon, 1974; Birdwell, 1968; Dolich, 1969; Grubb & Grathwohl, 1967; Ross, 1971; Delozier & Tillman, 1972; Loundon & Bitta, 1979; Krech et al., 1962; Wells & Marwell, 1976; Baughman & Welsh, 1962; Felker, 1974; Frey & Carlock, 1989). Several studies have established that consumers choose to purchase certain products based on the similarity between their reported self-concept and their perceived image of a product (Dolich, 1969; Ross, 1971; Delozier & Tillman, 1972; Grubb & Grathwohl, 1967; Hattie, 1992; Baughman & Welsh, 1962). These studies have shown that individual perceptions of self (self-concept) can influence behavioral intentions when viewed in reference to the individual perceptions of a consumer product. Therefore, each individual has perceptions that influence his or her behaviors, and these perceptions may be completely different than the perceptions of





24


another observer. This is the most important justification for the use of self-concept to examine behavioral intentions for this dissertation.

In the aforementioned consumer studies, subjects rated their perceptions of

consumer products, and these ratings were compared to the individual self-concepts of these subjects. From these comparisons, researchers examined the role of self-concept in influencing consumer behaviors toward selected products. However, this dissertation looks at perceptions and influence of sports heroes, not consumer products. Therefore, it is necessary to review the characteristics of popular, mass-mediated sports heroes and their potential to serve as role models for adolescents. This is the focus of the next section.



The Sports Hero as a Role Model Figure


The Hero

The term "hero" is a heavily cited yet frequently changing construct in American society. However, the hero has been a vital part of societies throughout history, and a brief review of this evolution will assist in better understanding the qualities of the modem hero and, more specifically, the modem sports hero.

The term "hero" dates back to ancient Greece and originally was defined as "a

superior man, embodiment of composite ideals" (Fishwick, 1979, p. 340). Qualities that the Greeks associated with the hero included an intimacy with gods; superiority to other humans; and an abundance of the best of human characteristics, such as strength and courage (Bowra, 1969; Brombert, 1969; Whitman, 1969; Fishwick, 1983). The ancient hero was a pre-destined character, and he was inherently greater than the normal man (Macmechan, 1840; Lehman, 1928). Another historical name for the heroic figure is the Great Man (Hook, 1957; Wecter, 1941).





25


According to Carlyle's Theory of the Hero, the hero is one who is destined for greatness from a divine source (Carlyle, 1841; Lehman, 1928; Smith, 1973; Wecter, 1941; Fishwick, 1985). Following this theoretical perspective, heroes were predestined by God as leaders to move the world forward (Carlyle, 1941; Lehman, 1928; Campbell, 1949; Wecter, 1941). Examples of Great Men in history include Moses, Frederick the Great, and Napoleon.

The evolution of the hero has been quite pronounced in the 20th century,

particularly in a relatively youthful American society. The youth of this society allowed Americans to define the characteristics of their own heroic types (Hook, 1957; Lubin, 1968; Wecter, 1941; Lerner, 1957). Theorists of this period asserted the hero was not pre-determined, but instead was a product of his society and those who followed him (Hook, 1957; Klapp, 1962). Additionally, an increasingly complex society has allowed for a variety of different heroic types and characteristics. In other words, 20th century America experienced a fragmentation of heroic types, allowing for greater heroic diversity and more heroic choices than in ancient Greek society.

Instead of possessing God-like qualities, the American hero of the mid-20th century had a closer relationship with the common man (Wecter, 1941; Hook, 1957). This presented a strong contrast with European Great Men of that time period, including Hitler and Mussolini. Americans, increasingly fearful of such powerful and destructive heroic figures, did not embrace Great Men as their heroic choices. They instead chose more familiar, less intimidating heroic types. (Hook, 1957; Fishwick, 1985).

The increasing influence of mass media in American society in the late 20th

century further changed the dynamic of the American hero. Theorists of the 1960s and 1970s described a new type of figure, the celebrity, which had become prominent in American society (Boorstin, 1961; Klapp, 1962; Lubin, 1968; Bowra, 1969; Monaco, 1978; Fishwick, 1979; Browne, 1983; Rollin, 1983; Hoagland, 1974; Harris, 1994). The celebrity, first described by Boorstin (1961), does not gain fame through greatness or





26


accomplishment, but instead due to mass exposure and visibility. Boorstin provided the following description of the American celebrity as:

a person who is known for his well-knownness. His qualities -- or lack of qualities -- illustrate our particular problems. He is neither good nor bad,
great nor petty. He is the human pseudo-event. He has been fabricated on
purpose to satisfy our exaggerated expectations of human greatness.
(1961, pp. 57-58)

Unlike heroes, celebrities do not display a model of greatness. They are a product of the media as much as they are a product of their accomplishments. Celebrities do not consistently display excellence, and they exhibit ordinary behaviors and accomplishments (Klapp, 1962; Kelly, 1993; Monaco, 1978).

Therefore, many theorists of the 1960s and 1970s believed, largely due to America's increased reliance on mass media and television, that the celebrity had replaced the hero in American society, and the heroic figure no longer existed in America (Schlesinger, 1968; Monaco, 1978; Hoagland, 1974; Kelly, 1993; Harris, 1994; "How to", 1995). This cynical view blamed television for initiating the extinction of the heroic figure. This viewpoint also expressed a static view of the hero. Accordingly, this heroic figure did evolve over time and was eventually replaced by the celebrity. However, many theorists of the 1980s and 1990s have refuted these arguments, writing instead about the ongoing evolution of the hero into what can be called the modem American hero (Fishwick, 1979; Fishwick, 1983; Fishwick, 1985; Browne, 1983; Walden, 1986; Harris, 1994). These authors argued that heroes have not disappeared, but instead heroic qualities have changed with the growing influence of mass media. In particular, because mass media have made heroes more available to the public, Americans are given a larger number of heroic choices, and these heroes cannot maintain the longevity of heroes of past generations.

Mass media exposure has become a critical component of modern heroic stature, and therefore the modem hero does share characteristics with the less substantial





27


celebrity. Rollin (1983) clarified the influence of mass media on the modem hero, writing, "Publicizing the heroic act and celebrating the performer, therefore, are crucial. All heroes then are celebrities; no hero is not a celebrity" (p. 14). Unlike the traditional hero, the modem hero must receive publicity by mass media in order to gain heroic consideration. Publicity is crucial in acceptance as a modem hero, regardless of the characteristics of the individual. However, unlike the celebrity figure, publicity in itself is not enough to create the modem hero, and this figure must also exhibit certain heroic qualities. This dissertation takes the perspective of these modem theorists, asserting that American heroes still do exist and that mass media have sped the evolutionary process of these heroic figures.

While the modem American hero can take several different forms, theorists have attempted to define some parameters of this modem heroic figure. These include attractiveness, winning, style, risk-taking behavior, individualism, and skill in a particular field (Browne, 1983; Rollin, 1983; Fishwick, 1985; Crepeau, 1985; Harris, 1994; "How to", 1995). Americans have become less likely to require moral greatness from their heroes than was required by previous generations. Additionally, due to an increasingly fragmented American society and an increasing number of heroic choices available through television, Americans of today are less likely than Americans of former generations to unite behind one particular hero (Walden, 1986; Fishwick, 1983; Rollin, 1983; Crepeau, 1985).

This research deals with one specific type of hero, the sports hero, and this is the focus of the following section.

The Sports Hero

Sports has become a popular and vital area in which Americans now find their heroes, and this trend has grown through the 20th century ("Heroes of', 1990; Ryan, 1995; Nixon, 1984; Oriard, 1982; Simons, 1997; "Role models", 1989). One potential reason for this is that sports remains one area where true greatness and superior beauty





28


still can be found in a complex society (Goodman, 1993). Regardless of the negative perceptions Americans may have of politicians, actors, and other frequently tarnished public figures, a star athlete, unlike other mass mediated figures, will have rare moments during which they appear to surpass typical mortal limitations through spectacular, seemingly impossible athletic feats. He will appear, if for only that brief period of time, almost "god-like" (Nixon, 1984; Oriard, 1982).

While the sports hero has been part of American culture since well before the turn of the 20th century, a fantastic growth and emphasis on the American sports hero occurred beginning in the 1960s with the rapid growth of sports television (McPherson, 1989; Davies, 1994; Harris, 1994; Nixon, 1984). By the mid-1980s, sports programming accounted for 15 percent of total programming on commercial television, including networks, such as ESPN, devoted solely to sports programming (Harris, 1994b; Davies, 1994; Nixon, 1984; Katz, 1996). This growth in sports media allowed for increased exposure and popularity of the sports hero to the general public. In addition to publicizing the modern sports hero, the current focus of the content of sports television has contributed also to the evolution of the American sports hero. To increase fan interest and provide more exciting television material, American sports programs are presented to emphasize heroic actions of the individual and individual emotions and personalities of the star athletes (Kinkema & Harris, 1992; Coakley, 1994; Hargreaves, 1986; Hilliard, 1984; Sabo & Jensen, 1992). This focus on the individual has helped to create a strong and unique relationship between the viewer and the individual star athletes. Duncan and Brummett (1987) explained media's ability to create such relationships, writing,

during sporting events, television tends to focus more on the individual
than the group, thus establishing a kind of intimacy between the audience and the player. This intimacy is created by close-ups of particular athletes
before, during, and after the competition and shots capturing the athletes' expressions and reactions at critical moments, by the announcers' creation
of imaginary conversations, but their use of the players' first names, and
by personal interviews with the athletes or coaches. Television, therefore





29


turns spectator sports into observance of intimates, albeit a controlled
intimacy within the confines of a 19-inch screen. (p. 171)

The individual has become so paramount in televised sports that television ratings often hinge upon whether a star athlete, such as basketball star Michael Jordan, plays in a particular contest (Fatsis, 1998; Shapiro, 1999; Starr, 1999). Professional basketball (NBA) in particular has employed a marketing strategy of promoting star players ahead of the actual game (Starr, 1999). For the 1997-98 basketball season, NBC's ratings for nationally televised Chicago Bulls games, featuring Michael Jordan, were 75% higher than for the network's other nationally televised NBA contests (Fatsis, 1998). With Jordan's recent retirement from basketball, both league and network executives have openly worried over the potential loss of league popularity and ratings of televised games. Peter Roby, vice president of marketing for Reebok Athletics, asserted that the league would never again reach the status attained through Jordan's star appeal, saying, "I don't think the NBA will ever again attain the glory it did with Michael" (Starr, 1999, p. 56).

To combat this potential loss, networks have embraced the opportunity to create and publicize new athletic stars. NBC Sports chairman Dick Eberson said in reference to the potential for decreased ratings upon Jordan's retirement, "We do have an opportunity to reintroduce a whole new generation of stars" (Shapiro, 1999, p. D5). In other words, the individual star remains the prime emphasis for television sports, and the retirement of one star leads to the creation of other stars. Nike corporation CEO Phil Knight, whose company airs numerous commercials featuring star athletes, has stated that even his television advertising campaigns alone are powerful enough to create popular sports figures. For instance, in reference to then-NBA newcomer Alonzo Mourning, Knight said, 'All we'd have to do is take the money we're using to advertise Charles Barkley and Michael Jordan, put it into Alonzo, and a lot more people would be talking about him' (Katz, 1994, p. 253). With the excessive television coverage of sports stars and the media influence on actually creating star appeal of accomplished athletes, the





30


distinction between sports hero and sports celebrity appears to be a confusing and blurred line. However, this research follows the perspective that mass media, specifically television, are now necessary to publicize sports heroes. Additionally, the fact that sports figures have primarily become mass-mediated figures does not mean all popular athletes must be celebrities instead of heroes, but instead reflects the evolution of the modem sports hero.

Due to the growth of media coverage of sports figures, Americans now know

more about popular sports figures than ever before, including both their on-the-field and off-the-field activities. As Caughley (1984) wrote, "Even an average fan knows more than a baseball player's name. He also knows his team, uniform number, position, batting average, salary, and something about his appearance, personality, medical history, and off-the-field conduct" (p. 32). The popular sports hero has been demystified, and fans now see the imperfections as well as the greatness (Hargreaves, 1986; Harris, 1994a; Hoagland, 1974; Coakley, 1994; McPherson et al., 1989). Such imperfections include wife/girlfriend abuse, drug and alcohol use, sports gambling, fighting, and fathering of illegitimate children (Harris, 1994; Paterno, 1992; Johnson, 1995; Messner & Solomon, 1994; Long, 1991; "Michael Jordan's", 1991; Nack & Munson, 1995; Starr & Samuels, 1997; Wilson & Sparks, 1996; Farhi, 1997; Franks, 1996; Howard, 1995). These imperfections must be considered in addressing the potential influence of popular sports figures, particularly their potential influence as role models for adolescents.

Despite these imperfections, many theorists still believe the modern,

mass-mediated sports figure can be a hero, and they have identified several characteristics that are commonly associated with this modern sports hero. These qualities include the following: supreme athleticism on the field or court, high winning percentages and the potential to win important games and championship matches, statistical records and greatness throughout a career, flair and charisma in style of play and personality, and





31


confidence in one's abilities (Nixon, 1984; Harris, 1994a; Harris, 1994b; Crepeau, 1985; Goodman, 1993; Smith, 1973; Porter, 1983; Starr & Samuels, 1997). According to surveys of American teenagers, particular sports-related behaviors, such as sportsmanship and fair play, are also important to some sports fans (Harris, 1994a). While not necessarily related to the athletic performance, financial success and lucrative commercial endorsement deals are commonly identified qualities of the sports hero, particularly to adolescent boys who aspire to reach similar financial heights through professional athletics (Katz, 1994; "Michael Jordan's", 1991; Weisman, 1993; Wilson & Sparks, 1996; Simons, 1997). Finally, theorists also have identified several non-performance-related characteristics, including civic and community involvement, academic accomplishment, strong family ties, and avoiding illegal and immoral behaviors, as components of the modem heroic qualification (Walden, 1986; Smith, 1973; Harris, 1994a; Harris, 1994b; Hoagland, 1974; Nixon, 1984; Coakley, 1994).

While the off-the-court actions of the sports star may have some impact on heroic classification, on-the-court excellence has been identified as more instrumental in determining heroic status. Nixon explained this, writing, "Wayward athletes may be excused by fans. in their lifestyle off the field as long as they work hard and produce on the field and. their behavior on or off the field does not depart too much from conventional standards" (1984, p. 174).

This dissertation accepts the perspective that the sports hero is a non-static,

evolving construct. Additionally, each individual may have his or her own views of the characteristics that are most important in being a sports hero. For this research, however, the sports hero will be defined as adhering to the following guidelines:



1) The sports hero must be well known. Therefore, he must be shown

prominently on television. This perspective acknowledges the impact of celebrity

on the heroic condition in America.





32


2) The sports hero must display athletic excellence on the field or court. This

can be displayed through record or notable performances (scoring, defense, etc.),

flamboyant or outstanding style of play, or other feats that set an athlete apart

from his or her peers on the field or court.

3) The sports hero must display the ability to be a winner on the field or court.

4) The sports hero may display the ability to overcome difficult odds against

success in sports, such as injury, physical limitations, or underprivileged family

life as a youth. The sports hero may also display extreme perseverance in

reaching athletic goals.

5) The sports hero may display exemplary behavior on and off the court, although this is less important than exemplary performance on the court.

Therefore, it is possible to have a sports hero who is quite flawed from a

behavioral perspective, as long as he or she performs well in athletics.

6) The sports hero will likely be extremely wealthy through high salaries

and sports marketing and endorsement deals.

7) While all the previously listed qualities are important in the definition of the sports hero, this research also will accept the idea that each individual may have his or her own priorities as to the most important characteristics of the American

sports hero.



The two concepts, the role model and the sports hero, described in the previous two sections have been synthesized in the following section, which describes the most critical concept for this research project, the sports hero as a role model. The Sports Hero as a Role Model

The popular sports figure is both praised and criticized as a potential role model, although limited research addresses these viewpoints. The function of star athletes as role models is often taken for granted and assumed to be true despite a lack of specific





33


evidence (Jung, 1986). The limited research that has been d9ne on this topic is addressed below.

The popular sports figure can serve as a role model in several different areas of adolescent life. Some of these roles are positive, while others are negative. One identified area of role modeling is style of play and athletic performance. Observers may pattern aspects of their athletic participation after the athletic accomplishments of their sports role models (Wilson & Sparks, 1996; Kellner, 1996; Simons, 1997; Harris, 1994b). For example, golfers reported a noticeable slowing of play at their local courses when golf tournaments were first televised (Harris, 1994b). This was because amateurs began lining up their putts as the pros did on television, despite the fact that many of them had no idea why this was done or the effectiveness of this procedure.

Similarly, observers model the athletic moves of NBA basketball stars (Harris,

1994; Wilson & Sparks, 1996; Simons, 1997; Nixon, 1984; "Michael Jordan's", 1991). Observers may attempt to adopt aspects of their athletic role model's play style, whether it be the flamboyant dribbling style of Julius Erving or the physical play of Karl Malone. The influence of professional athletes often extends from one generation of athlete to the next, as many current stars list former athletic stars as their role models in developing playing styles (Nixon, 1984; Wilson & Sparks, 1996).

Non-sports-related on-court behaviors also can be modeled by observers. These behaviors can be positive, such as exhibiting grace during defeat or helping players from other teams. They can also be negative, including throwing temper tantrums during competition, fighting with other players and officials, and abusing equipment (Weisman, 1993; Wilson & Sparks, 1996; Starr & Samuels, 1997; Harris, 1994). Several popular athletes, including Charles Barkley, Joe Namath, and Deon Sanders, have become well known for their brash and arrogant behavior, and adolescent observers may copy such behaviors (Harris, 1994b). Athletic role models also can display positive behaviors





34


during play, and observers can imitate these behaviors as well (Coakley, 1994; "Role models", 1989; Goodman, 1993; Harris, 1994b).

Theorists have suggested that athletes can serve as positive role models when observers imitate positive work behaviors of successful sports heroes. These behaviors include hard work toward reaching a goal, practice, desire to win, and prosocial behaviors such as charitable work and sportsmanship (Kellner, 1996; Wilson & Sparks, 1996; Coakley, 1994; Malone, 1993; Weisman, 1993; "Michael Jordan's", 1991; Nixon, 1984; "Role models", 1989; Harris, 1994a).

One prominent potential area of role modeling of popular mass-mediated athletes involves clothing style and athletic footwear, including athlete-endorsed shoes costing in excess of $100 per pair. While observers may not be able to perform the athletic feats of their role models, they can wear or hope to wear the clothing endorsed by role models (Weisman, 1993; Katz, 1994; "Michael Jordan's", 1991; Kellner, 1996; Simons, 1997). Mass media advertising campaigns have contributed to the marketing power of the athletic role model, with companies telling the consumer that shoes are a key component to being more like their athletic hero, whether it be Michael Jordan or Bo Jackson (Weisman, 1993; Wilson & Sparks, 1996; Simons, 1997). As Wilson and Sparks (1996) wrote about young African American males and their favorite NBA players, "Many of the Black adolescents gave responses acknowledging the celebrity Black athletes' definitive influence on popular style through athletic apparel and the athletic-apparel commercial. This influence was manifest most often in the respondents' desire to look like the celebrity athlete" (p. 414).

The coinciding occurrence of imitation of athletic play style and choice of athletic shoes points to the influence of the athletic role model. Katz (1994) described this in his chronicle of the Nike corporation, writing,





35


by 1991, a visitor to any playground in the world could find hefty young Charles Barkleys and towering David Robinsons playing Charles-style or
Mr. (David) Robinson-style in their Air Force shoes. Lean and athletic Scottie Pippens flew in Air Flights; little Bo Jacksons pumped iron in thousands of gyms, and, by then, miniature Andre Agassis lit up tennis
courts in their peacock-colored Nike clothes and shoes. (p. 148)

Fans could couple their role modeling of athletic behaviors of their sports hero with purchases of equipment of their sports hero, certainly an inviting concept to athletic companies marketing such endorsed sports products.

Popular athletes also may serve as role models of economic success. Professional athletes receive exorbitant salaries; NBA salaries average more than $2 million a season, not including endorsement income (Simons, 1997). Salaries of top athletes far exceed this figure, and such wealth is often accomplished at a very young age. Theorists have suggested that American youths use famous athletes as role models of economic success and mobility (Simons, 1997; Weisman, 1993; Wilson & Sparks 1996; Kellner, 1996; Harris, 1994; "Michael Jordan's", 1991; Nixon, 1984; Starr & Samuels, 1997). Therefore, such youths may desire to copy this financial success instead of imitating more realistic role models, such as teachers and local business persons. This occurs despite the fact that only one of every 50,000 high school basketball players actually reaches the NBA (Simons, 1997). In other words, an adolescent's chances of ever achieving fame and fortune through professional athletics are quite slim.

Additionally, many popular athletes have exhibited negative behaviors, both on and off-the-court, that possibly could be role modeled by fans. It is this type of role modeling that is most worrisome to critics of the modem sports hero. Although certainly not all popular sports stars display such behavioral tendencies, the list of athletes who have done so is long and covers a wide range of sports and teams. A few specific examples of such well publicized negative behaviors are described below.

Many of these negative behaviors occur during competition, referred to as on-court behaviors. These behaviors often involve arguments and confrontations





36


between players; perhaps the most notable of these occurred when boxer Mike Tyson bit Evander Holifield's ear during a title fight. These behaviors also include athlete confrontations with fans, coaches, and officials ("TKO: Tyson", 1999). For example, in a much publicized case, NBA star Latrell Sprewell, then of the Golden State Warriors, attacked his coach, P. J. Carlisimo, twice during a practice session (Starr and Samuels, 1997). In recent years, several players have been suspended in recent from league play for physically attacking a referee. These include Nick Van Exel and Dennis Rodman of the NBA, Orlando Brown of the NFL, and Roberto Alomar of Major League Baseball, who spit on an umpire after a disputed strike call (Collins, 1999). Rodman also has incurred fines and a suspension for kicking a TV cameraman during a game (Starr and Samuels, 1997).

Star athletes also have had conflicts with fans during games. Houston Rockets guard Vernon Maxwell punched a heckling fan in Portland during a 1995 game, costing Maxwell $20,000 and a ten-game suspension (Nance, 1995). Even professional coaches, who themselves have become highly paid, popular sports figures, have displayed extremely inappropriate behaviors during games. NFL coach Mike Ditka made an obscene gesture to home fans during a home New Orleans football game (Collins, 1999). Such negative on-court behaviors, including fighting, yelling at other players, shoving and arguing with officials, kicking cameramen, and, in the most extreme of cases, even physically harming a coach or fan, have become a somewhat regular occurrence for many popular sports figures (Wilson & Sparks, 1996; Weisman, 1993; Katz, 1994; Harris, 1994; Harris, 1994b; Starr & Samuels, 1997; Smith, 1973; Walden, 1986; Coakley, 1994).

Negative off-the-court behaviors of popular athletes have become even more troubling than such on-court behaviors, and many of these behaviors have resulted in much more serious consequences. One such example is the case of Ray Carruth, wide receiver for the NFL Carolina Panthers, who was indicted in December 1999 for





37


conspiring to murder his wife (Brady, 1999). Unlike many other athletes who have committed serious criminal offenses, Carruth did not have a prior record of negative and illegal behaviors. Other athletes, such as Mike Tyson and Dennis Rodman, became well known as regular law breakers. Among other offenses, Tyson has been sentenced to jail twice, once for rape and once for assault, and Rodman has been arrested for sexual assault, domestic abuse, and drug use ("TKO: Tyson", 1999; "Rodman sued", 1999). Although this paper cannot effectively address the many examples of the different varieties of negative off-the-court behaviors, some behaviors of popular athletes that are frequently reported by mass media include wife and girlfriend abuse, drug abuse, alcohol abuse, and assault (Howard, 1995; Taylor, 1991; Star & Samuels, 1997; Donohew, Helm, & Haas, 1989; Paterno, 1994; Messner & Solomon, 1994; Goldberg, 1995). Given the invasive media coverage of the off-the-field behaviors of athletes, these negative behaviors hold serious ramifications for any study of role modeling of popular athletes.

The previous section addressed the sports hero as a potential role model and some specific types of role modeling that could result from the actions of mass-mediated sports figures. This study, however, addresses not only different types of role modeling, but also the motivations for such role modeling and conditions under which role modeling might occur. Specifically, this study examines adolescent self-concept and the proximity of such self-concept to perceived images of sports heroes as a factor in determining a variety of role modeling behaviors of sports heroes. Therefore, the next section addresses self-concept and its relationship with behavior and role modeling.





38


Self-Concept


Introduction and Definitions

Much of the literature and debate dealing with the choice of sports heroes has

focused on the qualities of the sports figure, not the individual qualities of the sports fans. For example, theorists have debated whether the sports hero is determined solely through athletic feats, or whether he must also personify outstanding qualities in his life outside of sports (Harris, 1994; Walden, 1986; Smith, 1973). The limited discussion of the qualities of sports fans typically has focused on group affiliation and characterization of sports fans, including race, nationality, socio-economic status, and gender, and how these general characteristics might influence the choice of a sports hero. (Harris, 1994a; Wilson & Sparks, 1996; Lapchick & Stuckey, 1993; Klein, 1991). However, these perspectives ignore the fact that selection of a sports hero is an individual choice based on the individual characteristics and needs of each person. One potential means of addressing such individual characteristics in a standardized and measurable method is to examine self-concept in relation to the choice of a sports hero.

Self-concept is an extension of the term "self' that has been defined as "the individual's attitudes toward self' (Frey & Carlock, 1989, p. 2; Secord & Backman, 1964). The self serves as the reference point for each individual's view of the world. From "self," theorists have established self-concept. Self-concept has been defined with several slight variations, as explained below.

The simplest definition of self-concept is "myself as I see myself' (Loundon & Bitta, 1979, p. 373; Dolich, 1969; Landon, 1974; Delozier & Tillman, 1972). This is a summation of the process by which persons evaluate their own selves to form their self-concept. Newcomb (1956) elaborated, defining self-concept as the individual as perceived by that individual in a socially determined frame of reference. This definition allows for the context of environment to influence the formation of self-concept, and it





39


also hints at the importance of the external environment in the determination of self-concept.

Several researchers have noted the difference between one's individual

self-concept and the self as observed by other people. In other words, an individual's view of himself or herself may not be the same as how other people view that person. One reason for this is because people do not always behave consistently with their self-concepts. One's overt behavior is often linked with one's self-concept, but inconsistencies may exist between the behavioral component of self and one's actual self-concept. For example, a person might think that he is an excellent student, yet this same person may not study diligently and complete his academic assignments. Therefore, his self-concept, or perceptions of himself, would not be consistent with his behaviors. This could lead to observers viewing a person differently than that person views himself or herself (Hattie, 1992; Baughman & Walsh, 1962). While the self-concept is likely instrumental in determining future behaviors, the self-concept is still a cognitive measure, and it measures one's own perception of reality. This may be inconsistent both with an objective reality and with the ways in which others perceive someone (Loundon & Bitta, 1979; Frey & Carlock, 1989; Felker, 1974). Felker (1974) explained this in his definition of the self-concept, as follows,

self-concept is the sum total of the view which an individual has of
himself. Self-concept is a unique set of perceptions, ideas, and attitudes
which an individual has about himself. The view which an individual has about himself is unique and to varying degrees is different from the view
that anyone else has about him. (p. 2)

All the aforementioned definitions of self-concept refer to a single self-construct known as the actual self-concept, or the real self-concept (Sirgy, 1983; Dolich, 1969). However, the self-concept actually has multiple components. The number of measurable components of self-concept has varied greatly among researchers (Sirgy, 1983; Hattie, 1992; Loudon & Bitta, 1979; Runyon, 1977). For this research project, two constructs





40


of self-concept are addressed. They are as follows, with the corresponding definitions for the purposes of this research:



1) The Real Self: An individual's perception of how he actually is (Dolich,

1969; Birdwell, 1964; Ross, 1971; Runyon, 1977; Loundon & Bitta, 1979).

This refers to the most common self-concept, the self-concept which has been described in the previous paragraphs. This construct has also been called the

"actual self-concept" because it refers to the individual's self-perceptions as he

or she actually sees them (Sigry, 1983). For this research, this construct does not take into account a person's perceptions of how others view him or her, known as

the "social self-concept" (Sirgy, 1983). Instead, this refers only to the

individual's own perceptions of himself or herself.

2) The Ideal Self: An individual's perception of himself as he would like to be

(Delozier & Tillman, 1972; Loundon & Bitta, 1979; Runyon, 1977;

Baughman & Welsh, 1962; Ross, 1971).


These two components are used as separate measures of self-concept and together make up the construct of "self-concept" for this research. Domains of Self-Concept

The construct "self-concept," whether it be real or ideal self-concept, is often

regarded as a singular measurement, combining self-perceptions on a variety of different areas together into one construct. According to this classification, individual perceptions toward oneself as an athlete would not be differentiated from perceptions toward oneself as an academic. In other words, measurements of self-concept would not allow the individual to differentiate between distinct areas of himself or herself. Although widely used, this means of assessing self-concept is limited and somewhat incomplete (Harter, 1982; Runyon, 1977). Therefore, this research examines self-concept not as one





41


measurement, but rather as a measurement of several distinct domains, all of which are part of self-concept.

Harter (1982) pointed out distinct problems with using a single measure of self-concept in a critique of Coopersmith's (1967) self-esteem inventory, a test of 58 questions which make up one global evaluation of self-worth, and the Piers-Harris Self-Concept Scale (1969), a similar test of 80 questions used to evaluate overall self-concept. Harter wrote,

items on such popular scales as the Coopersmith Self-Esteem Inventory
and the Piers-Harris Self-Concept Scale tap a range of diverse content
including cognitive competencies, physical skills, popularity, acceptance
by parent, morality, personality traits, physical characteristics, and
affective reactions. Responses to these heterogeneous items are then summed, and the total score is interpreted as an index of global selfregard. In employing such a procedure, Coopersmith has assumed that
children do not make distinctions among the domains in their lives. This
assumption was seriously questioned in the scale-construction efforts
reported here. (1982, p. 87)

Early research on self-concept did not break down this construct into specific

domains. However, theorists did find that self-concept could differ in varying situations, such as a work environment as opposed to a social or an athletic environment (Loundon & Bitta, 1979; Newcomb, 1956). This was a precursor to the idea that there may actually be several different components of self-concept, and these different components accounted for a different self-concept in the context of different environments. Researchers subsequently determined that the general measurement of self-concept consisted of several components, including one's perception of one's athletic ability, morality, academic accomplishment, and physical appearance (Dolich, 1961; Krech et al., 1962; Loundon & Bitta, 1979; Reynolds et. al, 1980). Furthermore, these components of self-concept could change with cognitive development and aging (Frey & Carlock, 1989; Felker, 1974; Hattie, 1992; Katona-Sallay, 1993; Harter, 1982; Harter, 1988). Similarly, Felker (1974) identified separate categories of self-concept that applied





42


specifically to adolescents. These categories were behavior, anxiety, intellectual or school status, happiness, and satisfaction (Felker, 1974). Although self-concept was still measured as a single construct, this understanding of the separate components of self-concept lead to the eventual measurement of separate domains of self-concept.

Susan Harter (1982) conducted research on the separate domains of self-concept and subsequently created tests to measure these separate domains of self-concept. This research provided the foundation for the collection of data on self-concept for this study. Harter created the Perceived Competency Scale for Children, a self-concept test measuring four separate subscales of self-concept. These subscales are cognitive, social, physical, and general self-worth (Harter, 1982). Harter later developed the Self-Perception Profile for Adolescents, a similar self-concept measure for adolescents that assesses the following eight domains: scholastic competence, social acceptance, athletic competence, physical appearance, job competence, romantic appeal, behavioral conduct, and close friendship (Harter, 1988). Because this research examines self-concept and behaviors of adolescents, these are the domains used as the subscales of self-concept for this research.

The use of domains for research on self-concept and intentions to role model sports heroes is crucial because different domains of self-concept may relate quite differently to role modeling intentions of sports heroes. For example, the athletic domain of self-concept may have a vastly different relationship with the decision to role model a sports figure than behavioral conduct or romantic appeal has with such role modeling. Additionally, as discussed earlier, modern sports heroes are diverse figures, and different domains of self-concept may relate in different ways to these different qualities of sports heroes. Popular athletes such as NBA stars Charles Barkley and Latrell Sprewell, both whom are excellent athletes but who, on numerous occasions, have participated in illegal and unethical activity, can display seemingly contradictory behaviors and characteristics in different areas of their life (Starr & Samuels, 1997; Farrey, 1997; Malone, 1993).





43


The ability to separate these different aspects of the individual is paramount to this research, and this is why the use of Harter's research on the separate domains of self-concept for adolescents guides this examination of self-concept.

While Harter and others have identified specific domains for self-concept, this

does not mean that everyone places equal importance on each of the specific domains. In fact, each individual likely holds different opinions of the importance and salience of the various facets of self-concept (Hattie, 1992; Harter, 1987; Gergen, 1971). For example, Hattie (1992) pointed out, "A minister is more likely to have a more clearly delineated and richer moral self-concept than an athlete. The athlete may have a clearer physical self-concept" (p. 55). Individuals may deem particular domains important for a variety of reasons, including the amount of familiarity or learning with a particular domain, the presence of a need or motivation toward a particular domain of self-concept (ex. peers may give approval to those who are proficient in sports, yet may not give the same approval to high marks in academics), competence in a domain, and particular social situations which may cause particular domains to become increasingly important (ex. frequently attending parties where social interaction is important) (Hattie, 1992; Gergen, 1971; Videbeck, 1958). While any of these factors may help dictate which domains are perceived as more important than others for the individual, the importance of any particular domain may be different for each individual. Therefore, because domain importance can vary both by domain and for each individual, it is extremely important to measure self-concept as a function of domains instead of an overall measure. The importance of each domain should also be measured, as is done in this research project. Further explanation of the importance of domain specific self-concept evaluation will be given during the following section of self-concept formation and enhancement. Development and Maintenance of Self-Concept

Self-concept is formed through life experiences and interactions. No one is born with a self-concept, but rather it is learned (Felker, 1974; Frey & Carlock, 1989;





44


Baughman & Welsh, 1962; Wells & Marwell, 1976; Orenstein, 1994; Krech et al., 1962). Self-concept is developed through self-appraisal, reflecting upon one's behaviors; reflected appraisal, reflecting upon cues from others; social comparison, comparing oneself to others; and biased scanning, seeking out information to confirm one's opinions while filtering out contradictory information (Loundon & Bitta, 1979; Runyon, 1977; Baughman & Welsh, 1962; Krech et al., 1962; Burnstein et al., 1962). One may look to a variety of people and sources in the formation of self-concept. These include parents, peers, teachers, siblings, and other persons deemed significant to the individual (Loundon & Bitta, 1979; Hattie, 1992; Frey & Carlock, 1989; Covington & Beery, 1976; Orenstein, 1994; Baughman & Welsh, 1962; Krech et al., 1962; Felker, 1974; Katona-Sallay, 1993).

Although many influences on self-concept are personally known acquaintances, mass media figures, and specifically sports heroes, can also influence the formation and maintenance of self-concept (French, 1991; Katona-Sallay, 1993; Caughley, 1984). A fantasy relationship with a media figure can serve to confirm one's sense of his or her own identity. Caughley (1984) wrote in reference to such a perceived relationship between fan and media star, "Landing such a widely adored celebrity figure confirms your self worth; it makes you somebody" (p. 50). To create such a mythical relationship, a fan may actually fantasize that the media star chooses the individual over millions of other viewers, picking that person to be a special friend or even a lover or wife. Caughley (1984) explained the allure of these fantasy relationships, explaining that "through fantasy, a media love relationship is exquisitely turned not to the needs of the celebrity, but to the needs of the self' (p. 51). The relationships with mass-mediated, celebrity figures can be extremely powerful in shaping personal identity and self-concept.

Media fans also can make social comparisons with popular mass media figures, and such comparisons, both negative and positive, can lead to further development of self-concept (Oppenheimer & Oosterwegel, 1993). This process is similar to





45


self-evaluations that can occur with traditional, non-mass mediated models (Burnstein et al., 1962). Therefore, personal interaction is not necessary to impact self-concept, and media figures can help to shape the self-concept of a viewer. This may be particularly important to adolescents who, in an effort to establish their own individual identity, often turn from more traditional sources of persons, such as parents, and look toward new sources, such as mass media figures, to define their self-concepts (French, 1991; Katona-Sallay, 1993). Furthermore, most people form their self-concepts during childhood and adolescence, and self-concept becomes more resilient to change during adult years (Krech et al., 1962; Orenstein, 1994; Hattie, 1992; Katona-Sallay, 1993; Frey & Carlock, 1994). Therefore, research into adolescent self-concept formation is crucial, because this represents a period when self-concept is still volatile and changing and is a time when one seeks new sources and people, including media figures, who may help form one's self-concept.

Self-Concept Maintenance

A primary human objective is to maintain and enhance real self-concept, and this goal influences human behaviors and interactions (Hattie, 1992; Runyon, 1977; Grubb & Grathwohl, 1967; Felker, 1974; Frey & Carlock, 1989; Delozier & Tillman, 1972; Sirgy, 1983; Wells & Marwell, 1976; Loundon & Bitta, 1979; Orenstein, 1994; Krech et al., 1962). One prominent means of achieving this goal is to bring one's real self-concept, or one's view of oneself, closer to one's ideal self-concept. People's behavioral intentions and choices of friends and role models are often formed in an effort to maintain or improve their real self-concepts. Therefore, people act in such a way as to gain higher opinions of themselves or to preserve the opinion they have of themselves. Such self-maintenance and self-improvement occurs because it can result in a higher sense of self-worth (Orenstein, 1994; Harter, 1987; Harter, 1982; Hattie, 1992; Felker, 1974). In other words, improving one's self-concept can be a primary behavioral influence.





46


However, this view of self-concept enhancement is likely a simplification of the process. Instead of viewing self-concept as a singular construct, self-concept maintenance and enhancement should be studied in terms of the many distinct domains of self-concept. This is because each person places a different level of importance on each particular domain, or, as described by Hattie (1992), "The saliency of various dimensions of self-concept differs across individuals" (p. 169). This difference in saliency among individuals leads to each specific domain contributing differently to one's sense of self-worth (Harter, 1982). While people likely attempt to maintain and improve real self-concept in all areas, people are more likely to do so in their most important domains of self-concept (Harter, 1987; Harter, 1982; Hattie, 1992; Runyon, 1977). Hattie (1992) explained, "An individual's sense of worth may be in a variety of areas: the body (beautiful) is of the highest importance to some, but for others it may be more important to be academically able, to have a happy family life, to gain respect for each other, or to have a desirable personality" (p. 55). Therefore, it is these particular areas for which the individual will have the most interest in maintaining and improving the real self-concept, because these areas/domains will contribute the most to one's sense of value. Self-Concept and Behaviors

The focus area for this research project is the potential link between self-concept and behavior. The desire to maintain real-self concept and reach towards one's ideal self-concept by behavioral intention has been examined primarily through studies of consumer behavior and product selection. While this may differ from the topic researched in this study (role modeling of sports heroes), prior marketing research should help build a foundation of knowledge and guide the research for this study.

Two of the earliest examinations of self-concept as a behavioral influence were

done by Rogers (1951) and Maslow (1954). Rogers explained that each person will strive to actualize, maintain, and enhance himself or herself, and these needs will guide people's actions (Rogers, 1951; Felker, 1974). Maslow (1954) introduced similar ideas in





47


describing the process of self-actualization, which refers to the process of achieving one's potential. According to this process, people are motivated to behave and act in order to achieve their potential. However, according to Maslow's theory, this need for self-actualization is the last in a long list of human needs. Therefore, self-actualization will be the last factor to impact behavioral decisions. Since Maslow's theory of connecting self-concept and behaviors is deeply embedded within a larger theory of human needs and existence, it is somewhat difficult to isolate this concept within Maslow's work.

Much of the subsequent research relating self-concept to behaviors involved

marketing and product selection. According to such research, consumers choose products which are consistent with their own image, or self-concept, and this can lead to a variety of purchase decisions. Kretch et. al (1962) explained this relationship by describing how self-concept motivated middle and upper-class Europeans to choose a lower quality of bread due to the lofty image of this product.

the very foods that an individual chooses to eat must not only satisfy his hunger but also be congruent with his conception of himself as a certain
kind of person. Thus the less nutritious white bread came to be preferred to the more nutritious dark bread because the dark bread was reminiscent
of the European peasant's black bread. The color of the bread had
become a status symbol and therefore, a determinant of how well it would
assuage the status seeker's hunger. (pp. 83-84)

The choice of bread was motivated by congruence between the white bread and self-concept. Because white bread was perceived as a high status product, it probably was indicative of the ideal self-concept of these Europeans. Food was selected not only to satisfy hunger, but also due to its congruence with oneself as a person.

Several studies have found consumers to choose products consistent with their self-concepts (Landon, 1974; Sirgy, 1983; Loundon & Bitta, 1979; Runyon, 1977; Hattie, 1992; Delozier & Tillman, 1972; Birdwell, 1964; Ross, 1971; Felker, 1974; Dolich, 1961; Grubb & Grathwohl, 1967; Krech et al., 1962). Loundon and Bitta (1979)





48


explained, "Products and brands are considered as objects that consumers purchase either to maintain or to enhance their self-images. The choice of which brand to buy depends on how similar (or consistent) the consumer perceives the brand to be with his or her self-image" (p. 376). This theory has been examined and tested thoroughly, and it helps explain the motivations for certain behaviors and behavioral intentions.

Theorists have not definitively established whether real self-concept or ideal

self-concept is more instrumental in guiding consumer choices (Loundon & Bitta, 1979; Hattie, 1992; Delozier & Tillman, 1972; Sirgy, 1983; Runyon, 1977; Landon, 1974; Grubb & Grathwohl, 1967; Ross, 1971). It has been suggested that visible products, such as formal clothing or expensive wines, might reflect the ideal self-concept, while privately used products, such as toothpaste and soap, might reflect the real self-concept. Additionally, researchers have suggested certain products, such as a luxury/sports car, are purchased more for status appeal and will be chosen for congruence with the ideal self-concept. In contrast, other products, such as a station wagon, are purchased for their utility value and will be chosen for congruence with the real self-concept. However, researchers have yet to establish exactly under which conditions real self-concept is used or when ideal self-concept is used. Grubb and Grathwohl (1967) explained,

further research is needed in terms of specific consumer decision
situations to determine to what extent self-enhancement involves a
conformity concept or an ideal self-image concept. For example, are
consumers, through their consuming behavior and the interaction process, seeking support for their self-concept as they now perceive themselves, or are they seeking reactions that will promote the attainment of a more ideal
self? (p. 208)

While both self-concepts have been identified as important in determining consumer behaviors, the specific instances and motivations for particular behaviors is not yet fully understood.

The importance of the individual domains, or domain importance, of self-concept may also help determine consumer behavior. In other words, it is possible that choice of





49


consumer products also depends upon which domains of self-concept are most important to the individual and how the selected consumer product relates to this particular domain. For example, the purchase of a calculator may be more related to academic self-concept, and the selection of athletic shoes may be more related to the physical self-concept. While not specifically addressed in any known research, researchers hinted at this potential relationship when they explained that different types of people (ex. housewives, tennis players, aspiring executives) may seek different types of consumer products (ex. linens, white shorts, leather brief cases) in reference to their individual self-concepts (Hattie, 1992; Grubb & Grathwohl, 1967; Loundon & Bitta, 1979). Therefore, the areas of self-concept that are most important to the individual will guide the selection of consumer products. The selection of a brief case may not correlate at all to the self-concept of a housewife because the related domains (professional accomplishment, etc.) may be unimportant to this individual. Because this area of study is largely unstudied, domain importance and its correlation with behaviors is further examined in this research project.

Self-Concept and Mass Media Figures

Little research has addressed the selection and adoption of behaviors of mass

media figures with respect to self-concept or other related constructs. Caughley (1984) addressed the perceived relationship between a viewer and an admired media figure, writing, "The appeal is often complex, but the admired figure is typically felt to have qualities that the person senses in himself but desires to develop further. The admired figure represents an ideal self-image" (p. 54). In other words, the observer chooses a media figure who represents an ideal self-concept, particularly in areas that are important to the observer. Accordingly, domain importance is a crucial factor in the relationship between media heroes and fans. Through this complex relationship, the fan can attempt to develop the most appealing characteristics of his or her favorite celebrity figure, allowing the fan to strive towards the most valued areas of his or her ideal self-concept.





50


Even less is known about the relationship between the sports hero and the

observer in reference to the fan's individual self-concept. Tener's (1987) dissertation work found that the hero choices of an adolescent can be used to predict self-concept, suggesting a consistency between the hero and the observer. However, this work did not deal with the real and ideal self-concepts, but rather just the strength of an adolescent's real self-concept. Additionally, this research focused on the importance of choosing a personally known vs. an unknown hero, and it did not address the different characteristics of chosen heroes.

While not addressing the specific construct of self-concept, several authors have suggested that fans may choose and role model their favorite sports figures based on their perceived similarities between themselves and the athlete (Wilson & Sparks, 1996; Cole, 1996; Kellner, 1996; Harris, 1994a; Simons, 1997; "Role models", 1989). The work in this area has not addressed the individual's perception of himself, or self-concept, but rather studied observable characteristics of the individual, including race, playing style, and gender. Therefore, one primary goal of this research is to examine modeling of sports heroes in relation to individual self-concept, not general group characteristics or qualities.

Adolescent Self-Concept

This study deals specifically with self-concept of adolescents. Adolescence

represents the beginning of drastic change to one's self-concept, marked by an increased awareness of self-concept (Orenstein, 1994; Harter, 1988; Frey & Carlock, 1989; Hattie, 1992); Felker (1974) explained, "During adolescence, there is an increased drive for change in the individual. The increased change and corresponding lack of stability brings with it a need for reorganization of the self-concept" (p. 104). Such reorganization has been labeled a "crisis of identity," a process of drastic change and instability in self-concept during adolescence (Hattie, 1992, p. 132; Erikson, 1950).





51


Frey and Carlock (1989) described adolescence as a movement from a period of dependency to a period marked by increasing separateness and independence. They wrote, "The movement away from dependency during adolescence is characterized by such questions as 'Who am I? How do I want to be?'" (p. 285). Additionally, adolescents typically become aware of a variety of expectations and standards, including standards for academics, social standing, and athletic competence, against which the self can be measured (Hattie, 1992; Frey and Carlock, 1989; Orenstein, 1994). Therefore, this ongoing process of measurement creates a greater interest in and awareness of one's own perceived competence and ability.

However, most authors have acknowledged that while adolescence marks a period of change for the self-concept, it is generally a gradual, less rebellious period during which adolescents "make the bid for autonomy gradually and these requests meet reasonable consideration and deference from parents who ally themselves with the child's need to grow" (Douvan & Gold, 1969, p. 485; Hattie, 1992; Wylie, 1961). In other words, parents and children typically engage in a give-and-take as adolescents attempt to form their own identity. It is reasoned that adolescents often shift from using family as their primary reference and instead rely more on peers, teachers, and mass media to establish their own identity (Hattie, 1992; Felker, 1974; Katona-Sallay, 1993; Pekrun, 1993; Frey & Carlock, 1974). This can create an instability in self-concept that is only increased by the ongoing physical changes throughout adolescence (Orenstein, 1994; Frey & Carlock, 1974; Felker, 1974).

A study by McGuire and McGuire (1981) examined the types of people

adolescents named when asked to describe themselves in relative terms. Compared to younger children, adolescents were more likely to shift from parents to teachers and from siblings to friends when defining themselves. In other words, adolescents were more likely to seek external sources to define themselves during this period of self-concept awareness and development. Research done by Pekrun (1993) indicated that while the





52


self-concepts of adolescents are still influenced by family support, adolescents also look to other sources, particularly teachers, to formulate self-concept. Pekrun further added that different sources may impact different areas, or domains, of self-concept. Pekrun wrote, "School factors have an impact on domain-specific, school related self-concept development. Furthermore, it may be that peers exert primarily influences on the development of domain-specific self-concepts of social competencies" (1993, p. 118). This research not only indicates the variety of influences on adolescent self-concept, but also stresses the importance of using separate domains of self-concept for the study of adolescent self-concept.

Adolescents also may begin to look toward mass media figures and popular

figures in reference to their own self-concept (Orenstein, 1994; Frey and Carlock, 1989; Hattie, 1992). Unfortunately, despite the prevalence of media figures in the lives of adolescents, relatively little research addresses the influence of such figures on adolescent self-concept. Adams-Price and Green (1990) found that a crush on or an attachment to a celebrity figure is an important aspect in self-concept development during adolescence. This process of development often occurs during solitary use of media, allowing an adolescent to discover his or her new sense of self (Reed, 1995). Such solitary involvement with media reflects the increased awareness and individuality of self-concept expected during adolescence.

The majority of the recent literature on television and development of self-concept has focused on adolescent girls and their perceptions of their bodies. Due to the prevalence of thin, attractive females in mass media, adolescent female media users often develop lower self-concepts of their own physical appearance, seeing themselves as overweight and unattractive in comparison to these media figures (Schneider, 1996; Rodin, 1992; Harrison, 1998; Tiggermann and Pickering, 1996; Goode, 1999; Botta, 1999; Becker and Popenoe, 1998). These images also have caused many female adolescent viewers to develop much higher, and likely unachievable, ideal self-concepts





53


for their physical appearance (Smith, 1985; Becker and Popenoe, 1998; Goode, 1999; Harrison, 1998). According to dissertation work by Harrison (1998), viewing of television with conspicuously thin characters resulted in higher body dissatisfaction, while viewing of television with overweight characters did not have the same effect on physical self-concept. Similarly, Becker and Popenoe (1998) found that the recent introduction of television and its images of attractive, thin lead characters to a remote South Pacific island led to a decrease in individual self-image of the female adolescent viewers. Additionally, these adolescent viewers described shifts in their ideal body image from that of a robust, rounded body, typical of the adults of the island, to that of a thin body, typical of the televised figures. The changed image resulting from watching media figures has lead to changed behaviors of adolescent females, most notably eating disorders (Goode, 1999; Becker and Popenoe, 1998; Schneider, 1996; Harrison, 1998; Smith, 1985). In one study, 15 percent of surveyed teenage girls admitted to dieting or exercising to look more like an image seen on television (Schneider, 1996). While these studies have focused primarily on adolescent females and the physical component of self-concept, they do show examples of adolescents developing their self-concept based on television figures. Furthermore, some of these studies detailed behaviors resulting from such social comparison to media images. This behavioral component of mass media social comparison is a focal point of this study, and these studies therefore provide sound foundation for the theoretical base of this research.

The following section describes congruity theory, a theory predicting the

relationships between self-concept congruity and behavior. Congruity theory builds on the previously reviewed literature and is the theoretical foundation for the research in this study.





54


Congruity Theory


The reviewed literature has established a link between self-concept and behaviors. In order to logically establish hypotheses for such a link in this research, a theoretical framework must be established. For this research, congruity theory provides such a framework.

Congruity theory, proposed by Sirgy (1980), thus far has been used only in reference to self-concept and consumer behaviors. However, for this research, the framework of congruity theory has been extended to study self-concept and role modeling of sports heroes. According to congruity theory, individuals assign images to consumer products, much like they assign self-concepts to themselves. Congruity, or congruence, refers to the relationship between one's self image and the image one assigns to a consumer product. Specifically, congruity increases as the distance between self-concept and product image decreases. Conversely, an increase in this distance results in less congruity. This distance is also referred to as a "distance score." (Distance scores have an inverse relationship with congruity. Therefore, as congruity increases, the distance score decreases.) According to congruity theory, the subsequent intention to purchase a product is based on the congruity between one's self-concept and one's individually assigned product image.

Sirgy (1982) has identified two conditions that predict one's intention to use a

product. The first involves real (actual) self-concept and product image congruity. Sirgy wrote, (1982), "The relationship between actual self-image/product-image congruity (self-congruity) and brand preference has been supported by numerous studies" (p. 165). According to Sirgy (1982), as the two images become more similar, or more congruous (lower distance score), the resulting affect will be cognitive consistency, which will lead to subsequent purchase of that product. Conversely, a growing distance between these two images, or decreasing congruity (higher distance score), will lead to cognitive





55


dissonance (consistency affect), which will likely result in the individual avoiding such consumer products.

Numerous marketing studies have supported this relationship. These include

studies by Dolich (1969), Delozier and Tillman (1972), Grubb and Grathwohl (1967), and Ross (1971). In each of these studies, researchers found that the real self-concept was more closely matched to the most preferred brands and less closely matched to the least preferred brands. These studies support the first component of Sirgy's (1982) congruity theory.

The second condition involving self-concept and brand image as a predictor for purchase intention deals with the ideal self-concept. Sirgy (1982) wrote, "The relationship between ideal self-image/product-image congruity (ideal congruity) has been generally supported. .. Ideal congruity will lead to positive product evaluation, whereas ideal incongruity may be responsible for negative product evaluation" (p. 167). In other words, a higher congruity between ideal self-concept and an individual's image of a product will lead to a greater likelihood of purchase intentions. (The distance between ideal self-concept and the image of a product is also called an "ideal distance score." As congruity increases, or as the ideal self-concept is closer to the image of a product, the ideal distance score decreases.)

While theorists have supported this aspect of congruity theory, not as many studies have addressed ideal self-concept as have addressed real self-concept. Dolich (1969) concluded that both real self-image and ideal self-image are predictive of brand preference, and purchase intentions for some products tend to be more closely related to real self-image, while purchase intentions for other products tend to be more closely related to ideal self-image (Dolich, 1969; Ross, 1971; Loundon & Bitta, 1979). Loundon and Bitta (1979) provided a general assessment of the relationships for both real and ideal self-concept predicted by congruity theory, writing, "The consumer behavior of an individual will be directed towards the furthering and enhancing of his self-concept





56


through the consumption of goods as symbols" (p. 377). In other words, people will use products both to maintain self-concept, further establishing the real self, and to enhance self-concept, moving closer toward the ideal self. Therefore, both the real and ideal self-concepts are useful as predictors of consumer behavior, and both should be considered in the design of academic research. Application of Theoretical Model to this Study

Congruity theory has significantly added to the understanding of human behavior and the importance of self-concept in determining behavior. According to congruity theory, self-concept has an impact upon consumer behaviors. Consumers use objects that are congruent in image with either their real self-concept, their real ideal self-concept, or both self-concepts, as this will satisfy the inherent need to either maintain or improve self-concept. Congruity theory, therefore, explains the relationship between self-concept and some behaviors.

However, while theorists have indicated that self-concept has a major influence upon many human behaviors, including the choice of friends and effort placed into academic work, little (if any) research exists outside of the study of consumer behaviors (Hattie, 1992; Orenstein, 1994). Because marketing studies have been successful in identifying behavioral intentions, academics should attempt to apply congruity theory and the study of self-concept to other important areas of behavioral influence. This is the emphasis of this research project. It is hoped that the application of sound theory to the study of important societal behaviors, in this case role modeling of sports heroes, will be a valuable addition to the academic community and will further the understanding of the impact of sports heroes on American adolescents.

Prior research studies have been used to provide the framework for this research. However, because subsequent studies examined consumer behaviors, and this research project examines role modeling of sports heroes, obvious alterations have been made. Most obviously, the behavioral component of prior studies, purchasing consumer





57


products, has been replaced with role modeling behaviors of the sports hero. Popular, mass-mediated sports figures have become so entrenched with marketing products and mass-media promotions, often for products named after them, it is not such a extension to view sports figures as products themselves (Katz, 1994; "Michael Jordan's", 1991).

Perhaps the most important justification for this study design is the lack of scientific research in the area of self-concept and role modeling. Few studies of role modeling have examined the influence of self-concept. Because congruity theory provides a sound research framework, a somewhat liberal alteration of marketing studies using this theory provides the most logical approach for this dissertation. Self-Concept Domains as Part of Research Design

Prior studies (Dolich, 1969; Ross, 1971) of congruity theory have used matching questionnaires to measure the individual self-concept, both real and ideal, and the product image. From these matched surveys, researchers searched for statistical congruity between the answers from the two surveys. In other words, distance scores between self-concept and product image were calculated. Researchers in these studies then used different questionnaires to measure the behavioral component, purchase intention. To test congruity theory, distance scores (or congruity between self-concept and product image) were compared to intentions to buy the product. According to congruity theory, as distance scores decrease (higher congruity), purchase intentions should increase.

This dissertation follows a similar design. However, in this study, questionnaires for real self-concept, ideal self-concept, and image of the sports hero address eight distinct domains of self-concept (as outlined in Harter's [1988] Self-Perception Profile for Adolescents). Instead of looking at a single measure of congruity between self-concept (real and ideal) and the image of the behavioral object (sports hero), this study examines each of these relationships across the eight distinct domains of self-concept. As addressed earlier, the eight domains of self-concept from Harter's Self-Perception Profile for Adolescents are scholastic competence, social acceptance,





58


athletic competence, physical appearance, job competence, romantic appeal, behavioral conduct, and close friendship.

This use of domains in this research allows for more specific results. As

suggested previously, fans of a sports hero may attempt to copy certain aspects of the athlete's behavior or personality and not other aspects (Simons, 1997; "Role models", 1989). Therefore, addressing role modeling across the eight separate domains further clarifies the particular areas in which the role model will be copied. For example, an individual may role model his sports hero much more in the domain of athletic competence than physical appearance. The congruence between one's self-concept and the image of the sports hero should be analyzed across those particular domains instead of using a more general, overall measure. This provides a more accurate and useful analysis of the importance of congruity in role modeling sports heroes.

Additionally, as addressed in the literature review, certain domains of self-concept may be more important to an individual than other domains (ex. physical appearance may be more important than job competence to an individual). The literature review has suggested that individuals will copy role models in areas that are most important to them. Therefore, this research measures the importance of each domain to see if increasing domain importance leads to increased role modeling of one's sports hero.



Hypotheses


The review of prior research has lead to the formation of four hypotheses, all four of which are listed and explained in this section. These four hypotheses address the research questions presented at the end of Chapter 1 and are examined through the collection of survey data. The description of this analysis is outlined in Chapter 3, the Methods section.





59


Hypothesis I. For each domain of self-concept, role modeling of the sports hero
will be positively related to the congruity between real self-concept and the image
of the sports hero.


The reader is reminded of the positive relationship theorized through congruity theory; as the distance (difference) between self-concept and the image of the sports hero decreases, congruity increases. For this study, distance scores for each domain of self-concept are calculated, and these distance scores reflect the distance between self-concept and the image of the sports hero in each domain of self-concept. As a distance score increases, congruity between self-concept and image of a sports hero decreases. As a distance score decreases, congruity between self-concept and the image of the sports hero increases. Therefore, a positive relationship between role modeling and the congruity of self-concept/image of the sports hero would be realized through a negative relationship between role modeling and distance scores. HI deals specifically with real distance scores, or the distance between real self-concept and the image of the sports hero. This hypothesis, along with the other three hypotheses, will be written to reflect the negative relationship with distance scores and role modeling described above, as this negative relationship, and not the inverse positive relationship with congruity and role modeling, will be the focal point of the ensuing statistical analysis. This is reflected in the following restatement of HI.


Hypothesis I. For each domain of self-concept, role modeling of the sports hero will be negatively related to the distance between real self-concept and the image
of the sports hero, also know as the real distance scores.


The second hypothesis also addresses the relationship between role modeling and the distance between self-concept and the image of the sports hero. However, H2 deals specifically with ideal self-concept, and the distances between ideal self-concept and the image of the sports hero. These distances, calculated for each domain of self-concept, are





60


know as the ideal distance scores. Similar to H 1, a positive relationship between role modeling and the congruity of ideal self-concept and the image of the sports hero is predicted. This is reflected in a predicted negative relationship between role modeling and ideal distance scores, as stated in H2.


Hypothesis II. For each domain of self-concept, role modeling of the sports hero
will be negatively related to the distance between ideal self-concept and the
image of the sports hero, also know as the ideal distance scores.


Hypotheses III and IV build upon H 1 and H2 by incorporating domain importance into the independent variables (distance scores), creating new dependent variables known as distance worth scores. Distance worth incorporates both the distance between self-concept and the image of the sports hero (distance score) in a domain of self-concept and the importance of that domain of self-concept. Therefore, this new variable is created by combining two other variables. A detailed explanation of distance worth scores and a description of the calculation of distance worth scores can be found in the proceeding Methods section. A brief description of this construct follows below.

For this study, distance worth scores are created by incorporating the importance of a domain (domain importance score) with the distance between self-concept and the image of the sports hero (distance score). Distance worth scores reflect the distance score for a domain of self-concept multiplied by the importance of that domain. However, domain importance scores have an inverse relationship with distance worth scores. In other words, distance scores are actually divided by domain importance scores to create distance worth scores. Therefore, the lower the domain importance score is for a domain (domain deemed unimportant), the greater the distance worth score will be. Higher domain importance scores (domains deemed important) will result in smaller distance worth scores. In other words, the less important a domain of self-concept is, the more the distance worth score for that domain will increase. Domain importance serves to further





61


qualify the relationship between distance scores, or the congruity between self-concept and the image of the sports hero, and role modeling. By creating distance worth scores through the incorporation of domain importance and distance scores, it is predicted that the less important a domain of self-concept is to an individual, the less likely that the congruity between self-concept and the image of a sports hero in that domain will have a significant impact upon role modeling that sports hero. As a domain becomes more important to the individual, increased congruity (decreasing distance scores) for that domain will be more likely to have a positive relationship with role modeling of a sports hero. Hypotheses III deals specifically with real distance worth scores, a set of variables that incorporate real distance scores in each domain and domain importance scores for the corresponding domain.


Hypothesis III. For each domain of self-concept, role modeling of the sports hero
will be negatively related to real distance worth.


The fourth hypothesis also addresses the relationship between role modeling and distance worth scores. However, H4 deals specifically with ideal distance worth, which is created in the same manner as real distance worth, except that ideal distance scores, not real distance scores, are divided by domain importance scores for each corresponding domain. As with H3, a negative relationship between role modeling and ideal distance worth is predicted. This predicted relationship is reflected in Hypothesis IV, as listed below.



Hypothesis IV. For each domain of self-concept, role modeling of the sports hero
will be negatively related to ideal distance worth.


The following chapter, Methods, addresses the specific methods and statistical analysis used in this dissertation.













CHAPTER 3
METHODS



The methods chapter covers the following topics crucial to the collection of data and testing of stated hypotheses: the subjects for this study; the measures administered for self-concept and the image of the sports hero; creation of the role modeling questionnaire; pretesting the measures used in this study; the collection of data (the administration of questionnaires); statistical analysis of collected data, addressing each of the stated hypotheses; and potential limitations of this analysis. Validity and reliability are discussed throughout this chapter as necessary.



Subjects for Study


This dissertation was designed to analyze the adolescent self-concept, the relationship between this self-concept and perceptions of sports heroes, and the subsequent likelihood of adolescents role modeling the behaviors of sports heroes. Therefore, all subjects were high school students in grades nine and ten, approximately aged 14-16. Additionally, this dissertation only examined the behavioral intentions of male adolescents, so no females were included in the pool of subjects. There are three reasons for this restriction. First, such a research design prevents gender from being a confounding variable in data analysis. Second, researchers have suggested that male adolescents are more likely to look to mass media figures, including athletic heroes, as role models than are their female counterparts, who are more inclined than males to continue to view parents and personal acquaintances as role models (McEvoy & Erikson,



62





63


1981). Third, as the vast majority of popular mass media sports figures are male, adolescent males have many more available sports hero choices of the same gender. Because perceived similarity has been suggested as a potential factor for role modeling, adolescent males may be more likely to find role models in sports heroes than female adolescents. Such an approach is consistent with research in the area of sports heroes as role models, where researchers have suggested that popular sports figures, such as NBA basketball players, are more likely to serve as role models for male adolescents ("Role models", 1989; Wilson & Sparks, 1996; Simons, 1997).

The particular subjects for this dissertation consisted of ninth-grade and

tenth-grade males from Houston, Texas. Of the 172 valid subjects used for data analysis in this study, 120 subjects were students in a suburban private school. All of those 120 students were participants in school athletics. The decision to use only student-athletes was made due to the increased likelihood that student-athletes have strong ties and potential perceived similarities between themselves and their sports heroes.

The decision to use adolescents from this particular school as subjects was largely a pragmatic one. First and foremost, the principal of the high school willingly agreed to allow his students to participate in this study. Second, because the school is private, the researcher was able to avoid much of the difficulty involved with using subjects from the public school system. Finally, because all students are male, the research process could avoid the perception of being exclusionary to female students in a coed school, as the female students would not be able to participate as male students did.

The students from this school were predominantly white, with a small percentage of minorities (Asian, Hispanic, African-American). Most students live in stable upper-middle-class family environments. Because the student body is not as large as that of many public schools in the greater Houston area, a relatively large percentage of students participate in school athletics, and such participation is encouraged and





64


promoted by school faculty and leadership. Therefore, athletics, at least in the participatory sense, is a part of the daily lives of many of this school's students.

The remaining 52 subjects were participants in a sports tournament in Houston run through a local community center. These subjects, of the same grade and age parameters as the first 120 subjects, also were participants in school athletics. The majority of these subjects participated in baseball, but other selected sports included soccer, basketball, and track. These subjects share similar demographics traits with the first 120 students surveyed (white, upper-middle-class, supportive family environment), and in fact the data collected from the two groups were virtually identical. Again this group of subjects was selected largely for pragmatic reasons (availability, access, age appropriateness). Using two groups of subjects and receiving similar results helped to strengthen the validity of this research. Therefore, the data analysis for this study and the subsequent analyses of the stated hypotheses were based on the responses of 172 valid subjects. While more than 200 subjects participated in the study, surveys from approximately 30 subjects could not be used due to subject error in completing the surveys. The majority of these errors involved subjects not completing all parts of the survey questionnaires.

One criticism of this study may be that the students do not represent a diverse sample, decreasing external validity. Similarly, because most of these students come from stable family environments with many available role models, these students may not use sports heroes as role models as much as adolescents who do not have many other available role models. However, this study will not claim to be generalizable to all populations, nor is it able to address issues of race, socioeconomic status, and family/home environment. Additionally, because this study addresses each person's individual self-concept and allows each person to identify his individual choice of sports hero, the focus of this research is not on societal patterns, but rather individual choices based on individual relationships between self-concept and the image of their sports hero.





65


In other words, this research is designed to enable better understanding of how self-concept affects individual choices to role model a sports hero, not the impact of external sources on role modeling, and therefore demographic factors (race, family, etc.), while important in the study of role modeling sports heroes, are not paramount to this research.



Tests and Test Administration


The following sections detail the survey instruments used for data collection in this study. This includes the selection of existing instruments, the alterations of such instruments, and the construction of new instruments. Self-Concept and Image of the Sports Hero

For the measurement of real and ideal self-concept across all domains, the

Adolescent Self-Perception Profile, created by Susan Harter (1988), was used. This test, an upward extension of Harter's Self-Perception Profile for Children (1982), measures self-concept in eight specific domains that are most important and relevant to adolescents. Harter's rationale behind using such a domain-specific format, as opposed to a singular score of self-concept, is as follows: "An instrument providing separate measures of perceived competence or adequacy in different domains.., would provide a richer and more differentiated picture than those instruments providing only a single score" (Harter, 1988, p. 2). This test was designed for students in approximately grades nine through twelve.

The test measures self-concept in eight specific domains. Each domain is described below, as reported in Harter's manual for this test (1988):





66


1) Scholastic Competence: The adolescent's perception of his/her competence in

scholastic performance, achievement in classwork, perception of one's

intelligence.

2) Social Acceptance: The degree to which one is accepted by peers, popularity,

number of friends, feels that he/she is easy to like.

3) Athletic Competence: Adolescent's perceptions of his/her athletic ability,

feelings that one is good at sports.

4) Physical Appearance: Degree to which the adolescent is happy with how

he/she looks, likes one's body, feels he/she is good looking.

5) Job Competence: Extent to which the adolescent feels he/she has job skills, prepared to succeed at part-time jobs, feels he/she is doing well at current jobs.

6) Romantic Appeal: Adolescent's perceptions that he/she is romantically

attractive to those in whom he/she is interested, dating people he/she wants to

date, feels he/she is fun and interesting on a date.

7) Behavioral Conduct: The degree to which an adolescent likes the way he/she

behaves, does the right thing, acts in appropriate manner, and avoids trouble.

8) Close Friendship: Adolescent's perceptions of his/her ability to make close

friends to share personal thoughts and secrets with.


These eight domains make up the test of self-concept.

Five questions address each domain of self-concept (40 questions overall). The original test has 45 questions because it includes a ninth scale that measures overall self-concept. This ninth scale was not used in this study, so these questions were eliminated. Each question is rated on a 1 to 4 scale, and approximately half of all questions for each domain are counter-balanced such that some have the most adequate description on the left (scored 4 to 1) and some have the most adequate description on the right (scored 1 to 4). (Please see Appendix A and Appendix B for actual tests used.)





67


The reliability of Harter's test of self-concept has been determined through

repeated use and examination of this instrument. Reliability refers to the reproducibility of a measure (Graziano & Raulin, 1993). Of primary importance in this study, particularly for the tests of self-concept, is internal consistency reliability. Internal consistency reliability is relevant when several observations are made to obtain a score for each subject (Graziano & Raulin, 1993). Internal consistency reliability is high when several items measuring the same thing (in this case, each of the eight domains of self-concept) correlate highly with each other. Several questions (5) are asked for each domain of self-concept, and therefore an internal consistency reliability score, reported as Cronbach's Alpha, is reported for each. Harris (1988) reported the subscale internal consistency reliability for each subscale, as based on Cronbach's Alpha. They are as follows (for each domain):

1) Scholastic: .81

2) Athletic: .92

3) Acceptance: .78

4) Close Friend: .83

5) Romance: .80

6) Appearance: .86

7) Conduct: .78

8) Job Competence: .74

The high correlation of the items of each domain indicates a high internal consistency reliability (Graziano & Raulin, 1993). Additionally, Harter's (1988) pretests reported mean scores for each domain ranging from 2.5 to 3.3, depending on the sex and age of the subjects. The standard deviations for each domain ranged from approximately .41 to .8, again depending on the subjects.





68


Multiple Use/Administration of Harter's Self-Concept Test

As explained in the literature review, five separate measurements were made for this research (real self-concept, ideal self-concept, image of sports hero, domain importance, and behavioral intention to role model one's sports hero). All but the last two constructs, domain importance and behaviors, were measured using Harter's (1988) test of self-concept, but slight variations of the test were made for measurement purposes. In other words, Harter's test was altered so that the same questions addressing the real self-concept would instead address either the ideal self-concept or the perceived image of the participant's sports hero. Such a procedure for altering an existing test in this manner is derived from the previously reviewed marketing studies that examine self-concept, product image, and purchase intentions (Dolich, 1969; Delozier & Tillman, 1972; Grubb & Grathwohl, 1967; Ross, 1971; Landon, 1974). In these studies, the same (or extremely similar) tests of self-concept were used to measure the real self-concept, the ideal self-concept, and the perceived image of the brand/product. For example, Dolich (1969) used the same semantic differential scale to measure all three constructs, as explained: "The semantic differential was used to measure evaluations by respondents of real-self image, ideal-self image, and brand images. It is described as a combination of controlled association and scaling procedures and satisfies the requirement of measuring several concepts with the same instrument" (p. 81). While these examples use such a scale to rate a brand image as opposed to the image of a sports hero, the process is quite similar and represents a logical and sound research design.

In order to use such a design, Harter's (1988) test was administered three times. (See Appendix A and Appendix B for actual test alterations.) The first administration measured the real self-concept. This was done with the original format of the questionnaire (subtitled "What I Am Like") (Harter, 1988). The second administration measured the ideal self-concept. In measuring this, references to self (me) were changed to the ideal self (How I Would Like to Be). The third administration measured the





69


perceived image of the sports hero. In measuring this, references to self (me) were changed to each subject's individual choice of mass mediated sports hero.

For each of the three variations of this test used in this research, the scaling,

question order, and wording of the survey questions were identical. The only difference was that the first administration addressed the real self, the second administration addressed the ideal self, and the third administration addressed the sports hero. Measurement of Domain Importance

In order to measure domain importance, this study used Harter's (1988) test of domain importance (subtitled "How Important Are Each of These Things to You?"), as included in Harter's adolescent self-perception profile. (See Appendix A and Appendix B for actual test.) This questionnaire follows a similar format to the test of self-concept, with a rating scale of 1 to 4, with half of the questions keyed in reverse. This test contains 16 questions, 2 questions for each of the 8 domains. Data collected with this questionnaire were used for the creation of distance worth scores for analysis in Hypothesis III and Hypothesis IV.

Measurement of Role Modeling

The final questionnaire measured the adolescent's intention to role model his sports hero. In many of the questions, behavioral intentions were used instead of behavioral measures, as discussed earlier in the literature review. Such a process of measurement also has been copied from marketing studies, which often measure purchase intentions instead of actual purchase behaviors. For example, in a study of real self-concept, ideal self-concept, and consumer purchase intentions, Landon (1974) measured purchase intentions with a five-point scale, with the fifth (most extreme answer) representing an intention never to purchase the product. (See Appendix C for actual questionnaire for role modeling one's sports hero.) This same structure has been used in this dissertation for the measurement of role modeling one's sports hero. A





70


five-point scale, with one (1) representing a strong negative response and five (5) representing a strong positive response, was used to measure behavioral intentions.

Two processes were used in creating the actual questionnaire to measure role modeling of one's sports hero. First, specific potential types of role modeling were identified through literature review and examination of popular sports media (magazines, television, newspapers, etc.). Included in this examination was the purchase of many types of athletic goods and commercial products associated with popular sports figures, particularly those geared toward adolescent males. Second, the researcher asked several adolescent males to identify role modeling behaviors typical of themselves and their peer group. Both a focus group of seven teenage males and open-ended written surveys (sent to approximately ten adolescent males) were used to identify aspects of role modeling that would be relevant for this study. From the areas of role modeling one's sports hero identified through these procedures, a Likert scale survey instrument was formulated. This instrument included the examples of role modeling most frequently identified by adolescents and relevant literature. The instrument consisted of 23 Likert scale questions, 22 of which addressed specific behaviors of role modeling a sports hero and one overall role modeling question, used primarily to examine the validity of the questionnaire. Pretesting of Questionnaires

The instrument for role modeling and the other previously described measurement instruments (self-concept, image of sports hero) were further examined through pretesting. This pretest was conducted to allow for possible adjustments to the survey instruments before the actual test administration for this study. The pretest was conducted with 20 eighth-grade males from a private junior high school in Houston, Texas. These students completed all components of this study, just as was to be done in the actual research. This particular pretest population was chosen due to its similarity to the actual population that would be examined in this study.





71


Students completed all questionnaires in the same order and manner as would be done for the actual study. Upon completion, students were given the opportunity to provide both written and oral critiques of the survey instruments. Two issues were of particular interest in pretesting these instruments. The first issue dealt with the age appropriateness of the language used in the questionnaires. In other words, did the pretest subjects fully understand the wording of each question, and was this wording appropriate for this age and demographic group? The second issue addressed whether all relevant areas of adolescent role modeling had been included in the questionnaire of role modeling. In addition to addressing these questions, the pretest allowed the researcher to practice survey administration and to observe the amount of time required by subjects to finish all questionnaires.

While no major adjustments were necessary based on the pretest, minor

alterations were made to the wording of a few questions about role modeling. All of the role modeling questions can be found in Appendix C. Test Administration

Five tests (two tests of self-concept, a test of domain importance, a test of the image of the sports hero, and a test of role modeling of one's sports hero) were administered to each subject. Tests were administered in either school classrooms or gymnasiums. The total testing time, including instructions and time for questions, lasted approximately one hour. Each student completed all five tests in one session. Tests were administered to small groups of approximately 10-20 students for each session, and the researcher was present at all testing sessions, along with either a teacher or coach. After giving instructions about how the questionnaires should be completed and showing a brief example, the researcher provided an instruction packet for students instructions beyond those provided verbally by the researcher. The researcher was available to answer questions during the entire testing period. No incentives were given to subjects for participation in the study.





72


After the instructional session, subjects were instructed to pick the one athlete

they most considered to be their sports hero. As discussed in the literature review, while each subject was allowed to choose his own sports hero, individual choices were required to meet the following two criteria: the athlete must be, or must have been, covered heavily by mass media (particularly television), and the athlete could not be a personal acquaintance of the subject. This allowed for a wide range of hero choices but kept the focus of this study on mass mediated figures as potential role models, as opposed to personal acquaintances, such as family, friends, and coaches as role models. Subjects then were told that their selected sports hero would be the basis for all subsequent questions concerning sports heroes for the study (image, role modeling). In other words, subjects were instructed to use their individual choice of sports hero as replacement for the generic "sports hero" of the questionnaires.

After choosing a sports hero, subjects completed the tests for real self-concept, ideal self-concept, and image of the sports hero at the same time. Having subjects complete the three similar tests together was done so students would consider their real self in juxtaposition to their ideal self and the image of their sports hero. In other words, the real self-concept was meant to be used as an "anchor" or "standard" by which the ideal self-concept and the image of the sports hero would be considered.

These three tests were given using one questionnaire and one answer packet, in which students would place their answers for each question of real self-concept next to the counterpart answers for the same question on each of the other two constructs (ideal self, sport hero). For example, subjects would place their answer for Question 1 for their real self (who I am), followed by their answer for Question 1 for their ideal self (who I want to be), followed by their answer for Question 1 for the image of their sports hero (what my sports hero is like). After completing all three components of Question 1, students moved to Question 2 and followed the same procedure. This allowed for subjects to examine and contemplate their answers for real self in relation to ideal self





73


and the image of their sports hero. The sequence of test administration was identical for all subjects.

Because the same order was used to answer each question (real self answered first, then ideal self, and finally image of the hero), it is possible that this order affected the way in which each question was answered. In other words, contemplating the real self-concept first for each question might have impacted the scores given for the ideal self-concept and, subsequently, the image of the sports hero. The completed questionnaires were examined to ensure that such order effect did not occur. In particular, irregular answer patterns were looked for, such as a series of ascending answers for each question. The literature review has suggested that the image of the sports hero likely would fall between the real self-concept and the ideal self-concept, so inappropriate variations from this pattern would suggest potential order effect. However, upon review of the completed questionnaires, order does appear to have impacted the answers given for each question.

Conducting all three similar tests during the same testing period should have eliminated the slight possibility of "history" as a confounding variable (Graziano & Raulin, 1993). If a week were to transpire between the administration of these three tests, it is possible, although unlikely, that an experience could alter the real or ideal self-concept of an individual or an individual's image of his sports hero, and this would impact the ensuing analysis based on these three constructs.

Following the tests of self-concept and image of the sports hero, subjects

answered the 16 questions addressing domain importance. These were answered on the same question guide as the self-concept/image questions, and, upon completion, the entire answer page and question guide were given to the test administrator.

Subjects then were given the role modeling instrument. Before answering the 23 Likert scale questions on role modeling, students again wrote the name of their chosen sports hero at the top of the questionnaire and were reminded to focus answers on their





74


particular choice of sports hero. This test took much less time to complete than the previous tests of self-concept/hero image, and subjects were finished with their participation in the study when they completed and turned in this questionnaire.

The process of analyzing role modeling through survey questionnaires posed an obvious threat to the internal validity of this study, specifically that subjects may have been inclined to give socially desirable responses for such questions as the behavioral intentions to role model their sports hero. In other words, subjects may not have been willing to admit, or even be fully aware, that they would model such behaviors (ex. style of dress, dating behaviors, etc.). However, these threats to validity were reduced by allowing subjects to choose their own personal sports hero (as opposed to forcing any particular hero on all students) and by reminding students no question had either a correct or incorrect answer. Additionally, by using Likert scale questions for subject response (similar to the marketing studies that provide the framework for this research), this threat to internal validity was reduced, as subjects were allowed to consider such behaviors in degrees instead of merely being offered a "yes" or "no" option.

Because of the length of the testing session and the large number of questions

presented in the questionnaires, subject fatigue was a potential confounding variable for this study. Specifically, there was concern that subjects might not carefully consider answers given in the latter parts of the study questionnaires. To help reduce this risk, subjects were only given one of the two questionnaires at a time, which made the survey process less overwhelming for participants. Subjects first were given only the self-concept/image of the sports hero questionnaire. After completing and turning in this questionnaire, subjects were allowed to take a brief break before taking the role modeling questionnaire, which was the less time consuming of the two survey instruments. Additionally, completed surveys were examined by the researcher to look for signs of subject fatigue as represented by irregular or inconsistent answer patterns in the latter





75


parts of the questionnaires. Upon review of the completed questionnaires, subject fatigue does not appear to have impacted the answers given for this study.

Upon completion of all questionnaires, the collected data for all 172 valid subjects were input into SPSS for statistical analysis. The procedures for data analysis are discussed in a later section of this chapter.


Issues of Validity for this Study


The following section addresses several types of validity, threats to validity and confounding variables, the relevance of different types of validity to this study, and the means by which this study attempted to maximize validity.

While several different types of validity and organizations of these concepts have been outlined by various theorists (Graziano & Raulin, 1993), the most frequently described types of validity and those most relevant to this study are listed and outlined below.



1) Content Validity: "Content validity refers to the extent to which a

measure thoroughly and appropriately assesses the skills or characteristics

it is intended to measure" (Fink, 1991, p. 50). Content validity is

maximized by assuring that the measurements used in the study actually

address the constructs in question. For this particular study, the constructs include self-concept and role modeling of sports heroes. Content validity

is often maximized by adequately defining the variables in question and by

using questionnaires that address the construct in question (Fink, 1991).

This study has maximized content validity by using valid instruments

designed to assess the particular constructs in question. For the

questionnaire to measure role modeling, content validity was increased by





76


fully defining role modeling and by the use of adolescent focus groups, who provided information as to which aspects of the behavior of adolescents and role modeling were most appropriate to include. 2) Face Validity: Face validity refers to the surface characteristics of a measure, and, unlike content validity, does not rely on established theory (Fink, 1991). Face validity is increased by asking all appropriate questions and using language that is appropriate for the subject population. The face validity for this study was increased by pretesting questionnaires on adolescents similar to those in the study population and asking for appropriate feedback on the language of the questionnaires and whether any aspect of role modeling behaviors had been omitted. 3) Statistical Validity: Statistical validity refers to whether one's results are due to a systematic factor (the independent variable) or merely to chance variations (Graziano & Raulin, 1993). This is particularly important in experimental research with a manipulation of the independent variable. Statistical validity has been increased for this study by using reliable measures, such as Harter's test of self-concept. Violation of the assumptions of a statistical test can decrease statistical validity. Because regression analysis was a primary statistical test used in this research, violations of the following assumptions could have decreased statistical validity: normal distribution, a linear relationship between dependent and independent variables, an assumption of homoscedasticity, and assumptions that measurements of Y values are independent of each other (Licht, 1996; Kleinbaum & Kupper, 1978). Statistical validity was increased by examining a scatter plot of responses before statistical analysis to assure that statistical procedures such as the potential use of cross products to account for interaction effects was not required.





77


4) Concurrent Validity: Concurrent validity is maximized when a new measurement tool compares favorably with an existing measurement tool that is already considered valid (Fink, 1991). Concurrent validity increases with high correlation between the new survey instrument and the existing measurement tool. For this research, the survey instruments used to assess self-concept and the image of the sports hero are adaptations of Harter's self-concept test, and therefore new tests have not been created to replace existing, more valid measures. For the created measurement tool of role modeling intentions, no known existing measurement tool was found. Therefore, correlation between this new tool and an existing tool cannot be calculated, and concurrent validity could not be established for this measure.

5) Construct Validity: "Construct validity refers to how well the study's results support the theory or constructs behind the research and asks whether the theory supported by the findings provides the best available theoretical explanation of the results" (Graziano & Raulin, 1993, p. 171). Therefore, a focal component is to eliminate rival hypotheses and other theoretical explanations for the research results. Threats to construct validity are reduced by clearly stating definitions and building hypotheses on well-validated constructs (Graziano & Raulin, 1993). For this study, threats to construct validity have been reduced through building the hypotheses on a sound theoretical foundation (congruity theory) that has been used successfully in marketing research. Construct validity also refers to the ability of a survey to distinguish between persons who have certain characteristics and those who do not (Fink, 1991). One means of establishing construct validity of a survey instrument is to check for discriminant validity with another measure. In other words, construct





78


validity is increased by insuring that one measurement tool does not correlate highly with another measurement tool that is used to measure a different, dissimilar construct. For this study, content validity has been established by checking that real self-concept and ideal self-concept do not have a high correlation (discriminant validity). A statistical comparison of the collected data for real self-concept and ideal self-concept can be found in the results section of this dissertation. 6) External Validity: External validity, often know as generalizability, refers to the extent to which the results of a study can be generalized to different populations and conditions (Graziano & Raulin, 1993). Threats to external validity are reduced through careful selection of participants, including random sampling and attempting to choose a representative sample. However, for this study, subjects were not randomly selected. Instead, this study used convenience samples, and therefore results must be generalized very tentatively, likely only to those who have characteristics similar to those in the study. It is more likely that the results cannot truly be generalized at all to other populations (Graziano & Raulin, 1993). However, it should be noted that this study examined relationships between individual self-concepts, one's individual choice of a sports hero, and one's individual intentions to role model. Therefore, while this study cannot be truly generalized to the adolescent male population at-large, the results present relationships that can be studied and researched in other, broader adolescent populations. 7) Internal Validity: Internal validity deals primarily with causality, particularly whether the independent variable is actually responsible for the observed changes in the dependent variable (Graziano & Raulin, 1993). For this study, most of the threats to internal validity (maturation





79


of the subjects, attrition, instrumentation, etc.) did not affect this study. It is possible that the sequence of the administration of exams could potentially affect the answers given on the questionnaires. This is often controlled by using more than one order of administration of questionnaires (Graziano & Raulin, 1993). However, for this research, it was important that subjects used the questionnaire of real self-concept to anchor the other tests, and therefore one order of administration, with the real self-concept questionnaire completed first, was used. 8) Subject and Experimenter Effects: Subject effects can threaten the validity of a study when subjects give socially desirable answers or unintentionally (or intentionally) try to satisfy the demands of the researcher through their answers (Graziano & Raulin, 1993). For this research, it is possible that participants might report higher or lower levels of behavioral intentions to model a sports hero depending on what they deem more socially desirable. Additionally, they may name a sports hero who may not actually be their sports hero, but rather a person they feel would be a more socially desirable sports hero. These threats to validity have been reduced by assuring participants that there are no correct or incorrect answers and by limiting the questions of role modeling to those that were more likely to be answered honestly. Therefore, questions concerning anti-social behaviors such as domestic abuse and recreational drug use were not included. Experimenter effects could occur in this study if the experimenter tried to influence the answers of the subjects (Graziano & Raulin, 1993). This was reduced by having teachers administer the actual surveys and by closely monitoring the procedure. Because there are no observational data (as in experimental research), the threats to validity from experimenter effects are minimal in this study.





80


Data Analysis


This section reviews the data analysis for this study and the statistical analysis of the four hypotheses of this study. All valid collected data were input into SPSS for analysis. The topics to be discussed include the following: role modeling statistics, self-concept and calculation of distance scores, and analysis of stated hypotheses. All results of this data analysis and further explanations are found in the following "Results" chapter.

Role Modeling Data

In order to further utilize the results of the role modeling questionnaire as the

dependent variables for this study, reduction of the role modeling data was necessary. To accomplish this, a factor analysis of the 22 individual role modeling behavior variables (not including the general overall role modeling variable) was conducted. Factor analysis, which identifies patterns among the variations in the values of large groups of variables, is effective in identifying patterns of interrelationships among these variables and, subsequently, reducing this large number of variables into a smaller number of variables known as factors (Agresti & Finlay, 1986; Rummel, 1970; Vogt, 1993).

Several rotations of this factor analysis were conducted to find the best fit for the role modeling variables used in this factor analysis. Rotations are typically done to identify factors that are more clearly named and distinct in their identity (SPSS, 1999; Vogt, 1993). In other words, by rotating this factor analysis, more distinct and specific areas of role modeling were identified as factors. For the variables of role modeling for this study, a varimax rotation was used, and four significant factors of role modeling were identified. A significant factor was defined as having an eigenvalue greater than one (1) and at least three variables with significant factor loadings on that factor.

For each of the significant factors of the factor analysis, an index was created. An index is a group of individual measures that, when combined, represent a more general





81


characteristic (Vogt, 1993). For this study, each of the significant factors was used to create an index of role modeling behaviors. Indexes were created instead of merely using the factors as the more general variables of role modeling because factors place different weights on each variable in the factor and include components of other variables that are not strongly related to the targeted role modeling behavior of that particular factor. Each index, however, only includes the chosen variables that specifically relate to the core role modeling behavior of the more general index.

An index was created for each factor by summing and then averaging across each subject's responses on the variables shown to load significantly on that factor. Significant variables were identified as those with a factor loading greater than .4 on that factor only. This was done for two reasons. First, this factor loading (.4) represented a high level of correlation between the variables within an index. Second, upon examination of the significant factors, this value served as a logical break point for the selection of variables to each index. An overall index also was created from all variables that were included in each of the four indexes. Finally, two variables not included in any of the role modeling indexes (getting a tattoo or earring like a sports hero or taking Andro like a sports hero) were used in the test of the stated hypotheses. While these two variables did have a significant factor loading (greater than .4) on factors with eigenvalues greater than one (1), both of these factors had fewer than three significant variables and therefore were not used to create indexes.

These seven variables (four role modeling indexes, one overall role modeling index, and two independent role modeling variables) were the dependent variables for role modeling used in the four hypotheses of this study. Self-Concept and Image of the Sports Hero

Data for real self-concept were collected using Harter's Self-Perception Profile for Adolescents (1988). The data for the ideal self-concept and the image of the sports hero were collected using the altered forms of Harter's test, as described earlier in this





82


chapter. For each of these three constructs, a mean score was calculated in each of the eight domains of self-concept. Each of these 24 domains had five questions. Before a mean score could be calculated, the results of all questions asked in a negative direction had to be inverted such that the results from each question ranged from lowest rating of self-concept/image of sports hero (1) to highest rating (4). (See the Appendix for questionnaires and directions for individual questions.)

It should be noted that while these mean scores for both self-concept and the

image of the sports hero were reported in the results for informative purposes, these mean scores were not used in the calculation of the distance scores between self-concept and the image of the sports hero. Instead, distance scores were calculated from the individual questions, not the mean scores, to provide a more accurate determination of the specific distances between the constructs. An explanation of this calculation follows in the next section.

Distance Scores

Distance scores are a measure of the distance between two concepts, and they can be used to measure the similarity between two such concepts (Osgood et al., 1957) For this study, distance scores were calculated to measure the distance between self-concept, both real and ideal, and the image of the sports hero. In other words, these distance scores indicate the similarity between self-concept (real and ideal) and the image of the sports hero. These two separate sets of distance scores, real distance and ideal distance, have been calculated for each of the eight domains. This resulted in 16 distance scores (8 real, 8 ideal) for use in this study.

Several equations can be used to calculate distance scores. For this study, the difference squared model, which squares the difference between each paired set of questions and sums these differences, has been used (Sirgy, 1983; Osgood et al., 1957). The difference squared model assured a positive value for the distance between each set of values, regardless of which value, the self-concept or the image of the sports hero, is





83


greater. This formula for calculating distance scores between the image of the sports hero and either the real or the ideal self-concept in any of the eight domains can be represented as follows:


Distance (in each domain) = (Q1, Sp Hero Q1, SC)2 + (Q2, Sp hero Q2, SC)2 + (Q3, Sp hero Q3, SC)2 + (Q4, Sp hero Q4, SC)2 + (Q5, Sp hero Q5, SC)2.


Qn, Sp Hero = Question n from the test of the image of the sports hero.

Qn, SC = Question n from the test of self-concept.


This same formula was used in each domain for real distance and ideal distance. The sum of squared differences for each question was used instead of the summed difference of the overall construct scores as this provides a more accurate account of the descriptions between the sports hero image and the respective self-concept scores. In other words, this formula assures that differences on specific paired questions from each domain were not lost or ignored when combined for an overall score for that domain (Osgood et al., 1957). The lower the distance score for each domain, the closer the particular domain of self-concept is to the image of the sports hero in that domain. Lower distance scores indicate closer proximity between self-concept and sports hero image and a higher level of congruity between the two constructs. The higher the distance score, the less congruity that exists between self-concept and the image of the sports hero.

These distance scores were used as the independent variables for H1 and H2 of this study.

Hypothesis I and Hypothesis II

H1 and H2, which predict a negative relationship between role modeling and

distance scores, were statistically examined with multiple regression analysis. Multiple





84


regression is useful for the following types of analysis important to the testing of HI and H2:



1) Multiple regression analysis can "characterize the relationship between the

dependent and independent variables in the sense of determining the extent,

direction, and strength of the association among these variables" (Licht, 1995,

p. 34).
2) Multiple regression analysis can determine which of several variables are

important for predicting a dependent variable, and which of these variables are

unimportant in predicting this dependent variable (Licht, 1995; Weisberg et al.,

1989; Kleinbaum & Kupper, 1978).

3) Multiple regression analysis can determine the best model for describing the

relationships between various predictor variables and a dependent variable (Licht,

1995; Weisberg et al., 1989; Kleinbaum & Kupper, 1978; Vogt, 1993).



For this study, regression analysis was used to determine the relationships between the predictor variables (independent variables), distance scores, and the dependent variable, role modeling. The dependent and predictor variables were assumed to have a linear relationship, and an examination of the pretest data and actual data from this study confirmed this relationship. Therefore, linear regression analysis was used in this study.

Although regression does not necessarily imply causality, regression can give a better understanding of the nature of a phenomenon by identifying factors with which it co-occurs (Licht, 1995; Weisberg et al., 1989). In other words, while this regression analysis cannot necessarily determine whether congruence between self-concept and the sports hero causes role modeling of a sports hero, regression can determine whether





85


congruence co-occurs with such role modeling, providing a better understanding of role modeling of sports heroes and the possible conditions under which it is likely to occur.

For each of the seven role modeling variables (four indexes, one overall index,

and two role modeling variables), three regression analyses were completed. The first set of analysis used all eight real distance scores as the independent/predictor variables. The second set of analyses used all eight ideal distance scores as independent/predictor variables. The third set of analyses used all 16 distances scores (real and ideal) as independent/predictor variables. One particular function of regression is to determine the best model for describing a relationship between a dependent variable and independent variables (Licht, 1995). Therefore, this third set of regression analyses was done in attempt to create a more predictive regression equation for each of the role modeling indexes and variables than may be found in the regression analyses using only real distance scores or only ideal distance scores, and to account for a larger percentage of the variance of each of these dependent variables (in other words, to increase the amount of variance explained for each role modeling index or variable). The exclusion of important predictor variables can drastically affect the results of regression analysis, so it was hoped that including all the distance scores (real and ideal) would increase the prospect of achieving statistically significant results (Licht, 1995). Finally, placing all 16 distance scores in one regression allows for a simple comparison of the impact of all distance scores by examining the standardized beta weights of each of these 16 dependent variables (Kleinbaum & Kupper, 1978; Licht, 1995). One common function of regression analysis is to better understand which predictor variables are more important than others, and this set of regression analyses has enabled that understanding.

However, theorists do warn of using an excessively large number of predictor

variables in a regression analysis, and, in fact, one potential goal of regression analysis is to find the best and most predictive model while minimizing the number of predictor variables (Kleinbaum & Kupper, 1978; Licht, 1995; Weisberg et al., 1989; Fink, 1995).





86


This therefore presents the challenge of minimizing the number of predictor variables while still recognizing the importance of including important potential predictor variables, even those which do not make significant partial contributions to the dependent variable, as suggested by the theoretical framework of the study (Agresti & Finlay, 1986). For this study, while theory requires all distance scores to be included as potential predictor variables, even those that do not significantly contribute to role modeling, practice indicates that such a process will weaken the stability of these regression analyses.

This inclusive set of regression analyses, using all 16 distance scores as predictor variables, does create difficulties in accurate assessment of the relationships between role modeling and distance scores. First, using all 16 distance scores does not tell the variance

(R2) of role modeling accounted for by either the real distance scores or the ideal distance scores independently, but rather gives the variance from all distance scores. Therefore, these regressions do not indicate whether real distance scores or ideal distance scores independently serve as significant predictors of role modeling behaviors. Because that question is a focus of this study, particularly in attempting to better understand congruity theory and the differing impact of real vs. ideal distance scores, these inclusive regression analyses do not fully answer all key questions raised in this study.

Additionally, using all 16 distance scores increases the possibility for

multicollinearity among these independent variables. Multicollinearity refers to the intercorrelation between the independent/predictor variables in a regression analysis (Licht, 1995; Kleinbaum & Kupper, 1978; Weisberg et al., 1989). As muliticolliniarity increases, regression analysis becomes less stable, and it becomes more difficult to assess the statistical significance of individual predictor variables when such variables are highly intercorrelated with other predictor variables in the regression. For this study, high correlation could exist between several of the ideal and real distance scores, particularly between the two scores from the same domain. Such correlation would result





87


in high multicollinearity, making the regression analyses with all 16 distance scores as predictor variables less stable than analyses using fewer predictor variables.

For these reasons, the regression analyses using only the eight real distance scores (or only the eight ideal distance scores) as predictor variables are more crucial to the study of the first hypothesis. These less complex analyses likely give a more accurate and stable assessment of the impact of distance scores on role modeling behaviors. The seven regressions using all 16 distance scores serve to supplement this assessment and add information for further discussion.

Therefore, 21 separate regression analyses were completed to test H and H2. A stepwise regression procedure was used for these analyses. Stepwise regression, unlike forward selection and backward elimination regressions, finds the best fit for a regression equation by entering the predictor variables in a variety of orders (Licht, 1995; Vogt, 1993; SPSS, 1999). This type of regression drops predictor variables from a regression model if they lose their significance when other variables are added to the regression equation, and this makes stepwise regression the most effective method when correlation among the independent variables is strong (Agresti & Finlay, 1986; Weisberg et al., 1989; SPSS, 1999). Given the potential for strong correlation (or multicollinearity) between the distance scores used as the predictor variables for H1 and H2, the stepwise method was the logical choice for the regression analyses for this study.

The F value of each regression model was examined, and only those regressions with a p less than .05 were deemed to support H 1 or H2. As several regressions addressed each of the two hypotheses, an insignificant regression equation does not reject either HI or H2, but rather indicates that a particular role modeling behavior does not have a significant relationship with the distance scores (real or ideal) used in that particular regression equation. For either H1 or H2 to be fully rejected, none of the regressions for either of the two hypotheses could be significant. By using several regression analyses for each hypothesis, this study identifies which of the particular role





88


modeling behaviors have significant relationships with distance scores, and which of the role modeling behaviors do not have such significant relationships. The regressions with significant F statistics indicate which indexes/variables of role modeling support H 1 and H2.

For each of the significant regressions, the t statistic, the significance of this

statistic, and the Beta statistic for each of the individual predictor/independent variables (distance scores) were examined. These statistics clarified which of the distance scores had significant relationships with specific role modeling indexes/variables and the relative strength of these relationships. To determine significance of each of the predictor variables, the test statistic (t) was used. For any predictor variable to be deemed to have a significant relationship with a role modeling index/variable, the p value of the test statistic must be less than .05 (p less than .05). The beta weights (B), or standardized regression coefficients, of each significant relationship were examined to explain the strength of the relationship between individual distance scores and role modeling indexes/variables. This statistic not only explained the extent to which any of the individual distance scores related to role modeling, but it also explained which of the distance scores had the strongest relative relationships with role modeling.

An examination of the results of the 21 completed regression analyses provide an explanation of H 1 and H2 and selective support (or lack of support) for these hypotheses, which predict significant relationships between role modeling and distance scores. Domain Importance and Distance Worth Scores

Domain importance for each of the eight domains of self-concept was calculated from Harter's test of domain importance as part the Self-Perception Profile for Adolescents (1988). (See Appendix A for the actual test questions.) Two questions addressed each of the eight domains of self-concept, and a mean score was calculated for each domain. Domain importance was measured on a scale from one (1) (least important) to four (4) (most important). These domain importance scores were used, along with the





89


calculated distance scores used for H1 and H2, to create distance worth scores. These scores were used as the predictor/independent variables for H3 and H4. Before the calculation of this new variable can adequately be described, it is necessary to explain fully the proposed relationship between domain importance scores and role modeling of a sports hero.

As a domain of self-concept becomes more important (domain importance scores increase), one should be more likely to role model behaviors of one's hero, particularly in behaviors that are related to that domain of self-concept. Additionally, higher domain importance should increase the likelihood that one considers the congruity between one's self-concept and the image of one's sports hero in influencing role modeling. Therefore, it is predicted that as domain importance scores increase, role modeling of the sports hero also will increase. In other words, domain importance and role modeling should be positively correlated.

The following examples illustrate this predicted relationship. Two hypothetical persons, Frank and Mark, are used for illustration purposes in these examples. Frank and Mark each has his own individual sports hero. Frank places high importance on his athletic self-concept. Therefore, his domain importance score for the athletic self-concept is high (for this example, Frank's domain importance score is four). Mark places less importance on his athletic self-concept (domain importance score is two). According to the predicted relationship for this research, Frank should be more likely to role model his sports hero than Mark, especially for behaviors that are related to the athletic self-concept. Additionally, Frank's self-concept congruity with his sports hero in the athletic domain should be more likely to influence his decision to role model his hero. Therefore, as domain importance scores increase, role modeling also should increase.

For ease of further analysis and to make distance worth scores smaller and more easily read, the domain importance scores were reduced by dividing the scores from each domain by four (4). The new, reduced domain importance scores ranged from .25 (least





90


important) to one (1) (most important). In other words, a domain importance score of "l" became a score of".25", while a score of "4" became a score of "l". These reduced scores represent the same values and ratios as the original scores, and this data alteration did not change the nature of this construct nor the relative value of domain importance from one domain to any of the other seven domains. These reduced distance scores (eight real, eight ideal) were used in the creation of the 16 distance worth scores used as independent variables for H3 and H4.

Distance worth scores incorporate both the distance scores used as the

independent variables for H1 and H2 and domain importance scores. Therefore, this variable represents both (a) the congruity between individual self-concept and the image of a sports hero in a domain and (b) the importance of that domain to an individual. However, the statistical calculation of this new variable requires specific data alterations due to the nature of distance scores and domain importance scores. As predicted for this research, distance scores should have a negative correlation with role modeling. This is because increased congruity between self-concept and the image of the sports hero in a domain results in decreased distance scores for that domain. Conversely, it has been predicted that domain importance scores have a positive relationship with role modeling. Therefore, to create distance worth scores, it is necessary to make statistical alterations to one of the two variables such that they both have a relationship with role modeling in the same direction. In other words, because distance worth has been created as a variable which combines congruity and domain importance to predict role modeling, the relationships of congruity (represented by distance scores) and domain importance with the dependent variable role modeling must be in the same direction for the creation of this new variable.

The creation of distance worth scores requires distance scores to be multiplied by domain importance scores. However, to satisfy the above conditions, domain importance scores (the reduced domain importance scores have been used) have been inverted before





91


being multiplied. For this calculation (see formula below), each of the distance scores was multiplied by the inverse of the reduced domain importance score for that particular domain. In other words, distance scores were actually divided by domain importance scores to create distance worth scores. This was done 16 times, resulting in eight real distance worth scores and eight ideal distance worth scores. (Each reduced domain importance score was used two times -- once for the real distance worth score for that domain, and once for the ideal distance score). These 16 distance worth scores (eight real, eight ideal) were used as the predictor/independent variables for H3 and H4. To calculate distance worth scores, the following formula was used:



Distance Worth = Distance Score x (1 / Reduced Domain Importance Score)


According to this formula, higher distance scores result in higher distance worth scores, while lower distance scores result in lower distance worth scores. In this respect, distance worth scores are identical to distance scores in their proposed correlation with role modeling behaviors. Conversely, due to the inversion of domain importance scores, higher domain importance scores result in lower distance worth scores, while lower domain importance scores result in higher distance worth scores. As a domain is rated more important (domain importance score increases), the distance worth score for that domain will decrease for both real and ideal distance worth. Therefore, this formula has allowed for the negative correlation between role modeling and distance scores and the positive correlation between role modeling and domain importance scores to be represented in one variable (distance worth).

Due to the reduction and subsequent inversion of domain importance scores, virtually all distance worth scores will be greater than their corresponding distance scores. The greater the domain importance score, the less the distance worth score will increase from its corresponding distance score. The lower the domain importance score





92


(less importance for a domain), the more the distance worth scores will increase from the distance score. Therefore, higher domain importance scores result in lower relative distance worth scores. In other words, if two distance scores are equal but the domain importance scores for that domain are different, the lower relative distance worth score will be the one with the higher domain importance score.

The following hypothetical examples of the creation of two distance worth scores further explain this concept.

Example of the Creation of Distance Worth Scores

In Example A, Frank and John are two hypothetical subjects. Both subjects have identical real distance scores in the athletic domain (distance score = 5). In other words, both Frank and John have the same congruity between their real athletic self-concept and their athletic image of their individual sports hero. However, Frank has a higher domain importance score than John for the athletic domain. Frank's domain importance score is three (3), while John's domain importance score is two (2). In other words, Frank places greater value on the athletic domain of self-concept than John does. Using the equation presented in the previous section, distance worth scores have been calculated for the two hypothetical subjects, and these scores are represented in the following table.



Table 3-1. Example A of the Calculation of Distance Worth Scores.
Real Athletic Athletic Domain Reduced Domain Real Athletic Dist
Subject Distance Score Importance Score Importance Score Worth Score
Frank 5 3 0.75 6.67 John 5 2 0.5 10



In this example, while both subjects had identical distance scores, the subject who placed greater importance on the athletic domain (Frank) had a lower relative distance worth score for the athletic domain. John placed less value on this domain and had a lower domain importance score, so his resulting distance worth score was higher.





93


Because H3 and H4 predict that distance worth scores will have a negative correlation with role modeling, this research predicts that Frank would role model his sports hero more than John, who had equal congruity yet a lower domain importance score. Therefore, distance worth scores incorporate both the importance of a domain and the congruity of that domain in an effort to more accurately predict and assess role modeling of the sports hero. It should be noted that, as explained in the previous section, both Frank's and John's distance worth scores are greater than their corresponding distance scores. This is due to the fact that domain importance scores are reduced before being inverted in the process of calculation. However, this numerical increase from distance scores to distance worth scores does not affect the statistical analysis of the hypotheses for this research. Instead, it is the relative value between distance worth scores, not the relative value between distance worth scores and corresponding distance scores, that guides the analyses of H3 and H4.

In Example B, real distance worth scores are again calculated for Frank and John, this time for the social domain. In this example, the subjects have identical domain importance scores for the social domain (domain importance score = 3), but Frank has a lower real social distance score (distance score = 4) than John (distance score = 7). In other words, while both subjects place equal importance on the social domain, Frank has greater real self-concept congruity with the image of his sports hero in this domain than does John. The following table details the creation of distance worth scores for this scenario.



Table 3-2. Example B of the Calculation of Distance Worth Scores.
Real Social Distance Social Domain Reduced Domain Real Social Dist
Subject Score Importance Score Importance Score Worth Score
Frank -4 3 0.75 5.33 John 7 3 0.75 9.33




Full Text
27
celebrity. Rollin (1983) clarified the influence of mass media on the modem hero,
writing, Publicizing the heroic act and celebrating the performer, therefore, are crucial..
. All heroes then are celebrities; no hero is not a celebrity (p. 14). Unlike the traditional
hero, the modem hero must receive publicity by mass media in order to gain heroic
consideration. Publicity is crucial in acceptance as a modem hero, regardless of the
characteristics of the individual. However, unlike the celebrity figure, publicity in itself
is not enough to create the modem hero, and this figure must also exhibit certain heroic
qualities. This dissertation takes the perspective of these modem theorists, asserting that
American heroes still do exist and that mass media have sped the evolutionary process of
these heroic figures.
While the modem American hero can take several different forms, theorists have
attempted to define some parameters of this modem heroic figure. These include
attractiveness, winning, style, risk-taking behavior, individualism, and skill in a particular
field (Browne, 1983; Rollin, 1983; Fishwick, 1985; Crepeau, 1985; Harris, 1994;
How to, 1995). Americans have become less likely to require moral greatness from
their heroes than was required by previous generations. Additionally, due to an
increasingly fragmented American society and an increasing number of heroic choices
available through television, Americans of today are less likely than Americans of former
generations to unite behind one particular hero (Walden, 1986; Fishwick, 1983; Rollin,
1983; Crepeau, 1985).
This research deals with one specific type of hero, the sports hero, and this is the
focus of the following section.
The Sports Hero
Sports has become a popular and vital area in which Americans now find their
heroes, and this trend has grown through the 20th century (Heroes of, 1990; Ryan,
1995; Nixon, 1984; Oriard, 1982; Simons, 1997; Role models, 1989). One potential
reason for this is that sports remains one area where true greatness and superior beauty


23
individual perceptions of the self, or self-concept, and the influence self-concept has in
determining role modeling behaviors. Of primary importance for this research are the
relationships between self-concept and the image one has of his role model (for this
research, the sports hero). In other words, this research attempts to better understand
whether ones behavioral intentions to model the behaviors of a role model are influenced
by how similar or dissimilar ones self-concept is to his or her perceptions of his or her
sports hero. This approach of examining self-concept has not been widely used to
investigate behavioral intentions of role modeling. However, this research approach has
been used quite frequently in marketing research, and this dissertation will borrow
investigative strategies from such marketing studies. (See later section on self-concept
for a detailed description of this construct and a review of research and theory in this
area.)
Marketing researchers have provided a foundation for the use of self-concept in
studies of behaviors and behavioral intentions (specifically, purchase intentions). First,
they have established that self-concept maintenance and enhancement is a factor in
determining behavior (Sirgy, 1983; Landon, 1974; Birdwell, 1968; Dolich, 1969;
Grubb & Grathwohl, 1967; Ross, 1971; Delozier & Tillman, 1972; Loundon & Bitta,
1979; Krech et al., 1962; Wells & Marwell, 1976; Baughman & Welsh, 1962; Felker,
1974; Frey & Carlock, 1989). Several studies have established that consumers choose to
purchase certain products based on the similarity between their reported self-concept and
their perceived image of a product (Dolich, 1969; Ross, 1971; Delozier & Tillman,
1972; Grubb & Grathwohl, 1967; Hattie, 1992; Baughman & Welsh, 1962). These
studies have shown that individual perceptions of self (self-concept) can influence
behavioral intentions when viewed in reference to the individual perceptions of a
consumer product. Therefore, each individual has perceptions that influence his or her
behaviors, and these perceptions may be completely different than the perceptions of


17
model has been described as someone whose life and activities influenced the
respondent in specific life decisions (Bassoon & Owe, 1980, p. 559). This description
does not address actual copying of behaviors or intentions to copy behaviors, but rather
presents the role model as one influence in such intentions. Other definitions assert that
role models are adults who are worthy of imitation in some area of their life (Pleiss &
Feldhusen, 1995, p. 163) or someone who demonstrates the appropriate behavior for a
specific role or relationship with another person (Jung, 1986, p. 528). Role models
generally exert influence over one of three areas of life: educational, occupational, or
personal (Bell, 1970; Pleiss & Feldhusen, 1995).
Many of the descriptions of the role model imply the role model is a positive
influence whose actions should be copied and emulated. In other words, people who
display negative behaviors, such as wife abuse, drug use, or poor performance in
employment or school, are not role models (Malone, 1993; Role models, 1989;
Richards, 1998; Golden, 1986). However, this research adopts the perspective that role
models can exhibit both positive and negative behaviors and characteristics. A
description written by Hendricks (1992) gives insight into this position, noting, We all
have role models; some positive, some not so positive. They serve as archetypes,
proxies we scrutinize as guides for our own actions and perhaps too as templates for
possible selves (p. 7).
Role modeling is a component of the more general term modeling. In fact, the
term model is often interchanged with role model. While role modeling certainly
involves the processes of modeling, a clear distinction will be made to allow for sufficient
definition of this term for this research. Role modeling refers to a more specific process
than the more generic process of modeling of behaviors. Specifically, modeling is a
measure of the influence of observed behavior on an observer (Bandura & Walters, 1963;
Bandura, 1977; Bandura, 1994; Jung, 1986; Muuss, 1975). This process refers to a
specific set of behaviors that are not necessarily tied to a specific role, and this process


85
congruence co-occurs with such role modeling, providing a better understanding of role
modeling of sports heroes and the possible conditions under which it is likely to occur.
For each of the seven role modeling variables (four indexes, one overall index,
and two role modeling variables), three regression analyses were completed. The first set
of analysis used all eight real distance scores as the independent/predictor variables. The
second set of analyses used all eight ideal distance scores as independent/predictor
variables. The third set of analyses used all 16 distances scores (real and ideal) as
independent/predictor variables. One particular function of regression is to determine the
best model for describing a relationship between a dependent variable and independent
variables (Licht, 1995). Therefore, this third set of regression analyses was done in
attempt to create a more predictive regression equation for each of the role modeling
indexes and variables than may be found in the regression analyses using only real
distance scores or only ideal distance scores, and to account for a larger percentage of the
variance of each of these dependent variables (in other words, to increase the amount of
variance explained for each role modeling index or variable). The exclusion of important
predictor variables can drastically affect the results of regression analysis, so it was hoped
that including all the distance scores (real and ideal) would increase the prospect of
achieving statistically significant results (Licht, 1995). Finally, placing all 16 distance
scores in one regression allows for a simple comparison of the impact of all distance
scores by examining the standardized beta weights of each of these 16 dependent
variables (Kleinbaum & Kupper, 1978; Licht, 1995). One common function of
regression analysis is to better understand which predictor variables are more important
than others, and this set of regression analyses has enabled that understanding.
However, theorists do warn of using an excessively large number of predictor
variables in a regression analysis, and, in fact, one potential goal of regression analysis is
to find the best and most predictive model while minimizing the number of predictor
variables (Kleinbaum & Kupper, 1978; Licht, 1995; Weisberg et al., 1989; Fink, 1995).


58
athletic competence, physical appearance, job competence, romantic appeal, behavioral
conduct, and close friendship.
This use of domains in this research allows for more specific results. As
suggested previously, fans of a sports hero may attempt to copy certain aspects of the
athletes behavior or personality and not other aspects (Simons, 1997; Role models,
1989). Therefore, addressing role modeling across the eight separate domains further
clarifies the particular areas in which the role model will be copied. For example, an
individual may role model his sports hero much more in the domain of athletic
competence than physical appearance. The congruence between ones self-concept and
the image of the sports hero should be analyzed across those particular domains instead of
using a more general, overall measure. This provides a more accurate and useful analysis
of the importance of congruity in role modeling sports heroes.
Additionally, as addressed in the literature review, certain domains of self-concept
may be more important to an individual than other domains (ex. physical appearance
may be more important than job competence to an individual). The literature review has
suggested that individuals will copy role models in areas that are most important to them.
Therefore, this research measures the importance of each domain to see if increasing
domain importance leads to increased role modeling of ones sports hero.
Hypotheses
The review of prior research has lead to the formation of four hypotheses, all four
of which are listed and explained in this section. These four hypotheses address the
research questions presented at the end of Chapter 1 and are examined through the
collection of survey data. The description of this analysis is outlined in Chapter 3, the
Methods section.


158
Summary of Results of H3 and H4
Again, several clarifications can be made about the data analysis related to these
hypotheses. First, the support for both of these hypotheses is limited. This is particularly
true for H4 (ideal distance worth scores). Of all the regression analyses, only ideal
romantic distance worth was found to have a significant negative relationship with any of
the role modeling behaviors (social communication), and this relationship is limited.
Additionally, three ideal distance worth scores had significant relationships in the
opposite direction as predicted. Therefore, it can be concluded that ideal distance worth
scores were not predictive of role modeling behaviors for the subjects in this study as
predicted.
Greater support was found for H3 (real distance worth scores); selected real
distance worth scores were significantly related to four role modeling indexes (prosocial
behaviors, athletic behaviors, social communication, and overall behaviors) and one
variable (getting a tattoo or earring like ones sports hero). However, of all eight domains
of real distance worth, only two real distance worth scores, real athletic distance worth
and real physical distance worth, had significant negative relationships with any of these
indexes/variable. Distance worth scores from the other domains were not significantly
related to role modeling behaviors. These results are similar to those for H1, which did
not incorporate domain importance. The most notable differences between the results for
the two hypotheses came in the domain of job competence. Unlike real job distance
scores, real job distance worth scores were not significantly related to any of the role
modeling indexes/behaviors. This would indicate that adding measures of the domain
importance of job competence did not help predict role modeling.
Placed into greater perspective, few significant relationships were found for all the
real distance worth scores. Not surprisingly, the limited support for H3 is found almost


53
for their physical appearance (Smith, 1985; Becker and Popenoe, 1998; Goode, 1999;
Harrison, 1998). According to dissertation work by Harrison (1998), viewing of
television with conspicuously thin characters resulted in higher body dissatisfaction,
while viewing of television with overweight characters did not have the same effect on
physical self-concept. Similarly, Becker and Popenoe (1998) found that the recent
introduction of television and its images of attractive, thin lead characters to a remote
South Pacific island led to a decrease in individual self-image of the female adolescent
viewers. Additionally, these adolescent viewers described shifts in their ideal body image
from that of a robust, rounded body, typical of the adults of the island, to that of a thin
body, typical of the televised figures. The changed image resulting from watching media
figures has lead to changed behaviors of adolescent females, most notably eating
disorders (Goode, 1999; Becker and Popenoe, 1998; Schneider, 1996; Harrison, 1998;
Smith, 1985). In one study, 15 percent of surveyed teenage girls admitted to dieting or
exercising to look more like an image seen on television (Schneider, 1996). While these
studies have focused primarily on adolescent females and the physical component of
self-concept, they do show examples of adolescents developing their self-concept based
on television figures. Furthermore, some of these studies detailed behaviors resulting
from such social comparison to media images. This behavioral component of mass
media social comparison is a focal point of this study, and these studies therefore provide
sound foundation for the theoretical base of this research.
The following section describes congruity theory, a theory predicting the
relationships between self-concept congruity and behavior. Congruity theory builds on
the previously reviewed literature and is the theoretical foundation for the research in this
study.


104
Table 4-3. Results of Factor Analysis of Role Modeling Variables.
Significant Factors
1
2
3
4
5
6
SINfcAK (Wear sneakers like
hero)
.158
.168
-.033
.594
.309
-.132
HAIR (Get hairstyle like hero)
.388
.179
-.102
.258
.595
-.047
UNIF (Wear uniform like hero)
.102
.398
.086
.557
.298
-.155
TATT (Get tattoo or earring
like hero)
-.151
-.044
.092
.035
.815
.112
POSIT (Play same position as
hero)
-.098
.704
.145
.238
-.017
.109
MOVES (Copy athletic moves
of hero)
.044
.746
.042
.142
.127
.255
TRAIN (Train harder because
of hero)
.352
.636
.095
-.019
-.074
-.066
TECH (Learn athletic
techniques of hero)
.237
.611
.085
.150
.058
-.005
COL (Go to same college as
hero)
.259
.341
-.005
.223
.182
.347
ANDRO (Take Andro/drugs
like hero)
-.018
.038
.094
-.003
-.020
.830
FAM (Relationships with
family like hero)
.216
.036
.610
-.026
.368
.320
FRIEND (Support friends like
hero)
.186
.078
.814
.096
-.132
-.008
COMM (Communicate with
coaches like hero)
.238
.177
.775
.175
.043
.127
POLITE (Polite to family like
hero)
.722
-.006
.380
.121
.037
-.005
VOL (Volunteer and charity
work like hero)
.799
.018
.155
.150
.091
.165
GIRLS (Popular with girls like
hero)
.194
.230
.189
TOO
.387
.476
TRASH (On-court behaviors
like hero)
.270
.387
.451
-.055
.156
.023
STUDY (Study harder
because of hero)
.699
.303
.114
.082
-.087
.035
OBS (Overcome obstacles like
hero)
.699
.282
.161
.107
.018
-.070
ACT (Make more friends by
acting like hero)
.578
.052
.245
.282
.185
.294
ENDORSE (Buy products
endorsed by hero)
.302
.202
.119
.599
.001
.378
SHIRT (Wear sports shirt of
hero)
.116
.074
.139
.800
-.083
.155
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.


Table 4-51. Regression Analysis for Ideal Self-Worth
Distance Scores on Tattoo or Earring Modeling Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant]
rw
TT"
12.999
MT
Ideal Athletic Dist Worth
.018
.009
.203
1.919
.057
Ideal Behavior Dist Worth
-.002
.007
-.024
-.255
.799
Ideal Friend Dist Worth
.020
.010
.216
2.101
.037
Ideal Job Dist Worth
.010
.009
.134 _
1.078
.283
Ideal Physical Dist Worth
-.002
.007
-.026 <
-.286
.776
Ideal Romantic Dist Worth
-.009
.010
-.096
-.816
.416
Ideal Scholastic Dist Worth
-.005
.006
-.076
-.758
.450
Ideal Social Dist Worth
-.016
.010
-.169
-1.596
.113
R = .447
R Square = .199
F = 2.149
Significance of F = .009.
The only distance worth score with a significant relationship with tattoo or earring
modeling behaviors was ideal friend distance worth (Beta = .21 *, NOTE: This
relationship is the opposite direction as predicted).


14
Critics and fans of sports heroes alike have suggested that popular American
sports heroes serve as role models for American adolescents (Hams, 1994a; Hams,
1994b; Role models, 1989; Malone, 1993; Simons, 1997; Wilson & Sparks, 1996;
Parker & Lord, 1993; Taylor, 1991; Michael Jordans, 1991; Goodman, 1993). Little
scientific research has addressed the motivations for such role modeling or the conditions
under which such role modeling is most likely to occur for the individual adolescent.
Socioeconomic status, race, and parental and peer influences have been suggested as
potential factors affecting such role modeling (Harris, 1994a; Role models, 1989;
Simons, 1997; Wilson & Sparks, 1996; Michael Jordans, 1991; Goodman, 1993).
However, these factors do not address the individual role modeling decisions of each
adolescent and the motivations for the individual to model the behaviors of his or her own
particular sports hero. One area of research which has examined motivations for
individual behavioral intentions is the study of individual self-concept and how this
self-concept relates to the images of other objects (Dolich, 1969; Grubb & Grathwohl,
1967; Ross, 1971; Hattie, 1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon&
Bitta, 1979; Krech et al., 1962). This research on self-concept has primarily been used to
better understand consumer behaviors and individual purchase decisions. However,
self-concept research also presents a potential means to better understand a variety of
behaviors, including the emulation of sports heroes addressed in this research.
This theoretical approach to studying adolescent role modeling of sports heroes
has lead to the following research questions:
1) Among adolescent males, how do the relationships between self-concept and
perceived images of a sports hero relate to behavioral intentions to role model a
sports hero? In other words, how do the perceived similarities and differences
between ones individual self-concept and the image of a sports hero relate to
ones intentions to emulate that sports hero?


95
either H3 or H4, but rather indicates a particular role modeling behavior does not have a
significant relationship with the real or ideal distance worth scores used in that particular
regression equation. The regressions with significant F statistics indicate which
indexes/variables of role modeling had significant relationships with distance worth
scores, and therefore which such indexes/variables would support H3 and H4.
For each of the significant regressions, the t statistic and the Beta statistic for each
of the individual predictor/independent variables (distance worth scores) were examined.
To determine significance of each of the predictor variables, the test statistic (t) was used.
For any predictor variable to have a significant relationship with a role modeling
index/variable, the p value of the test statistic must be less than .05 (p less than .05). The
beta weights (B), or standardized regression coefficients, of each significant relationship
were examined to explain the strength of the relationship between individual distance
worth scores and role modeling indexes/variables. This statistic not only explained the
extent to which any of the individual distance worth scores related to role modeling, but it
also explained which of the distance worth scores had the strongest relative relationships
with role modeling.
An examination of the results of the 21 completed regression analyses provided
explanation of H3 and H4 and displayed selective support (or lack of support) of these
hypotheses, which predict significant relationships between role modeling and distance
worth scores.
The following chapter, Results, details the results of the described analysis. These
results address the four stated hypotheses and the potential support for these hypothesis
based on the collected data for this study.


176
of a domain of self-concept, particularly in conjunction with a closer proximity between
self-concept and the image of the sports hero for that domain, would correlate with role
modeling behaviors. This was not supported from the data collected in this study.
Interestingly, it is possible that incorporating domain importance actually
detracted from the predictive nature of distance scores, and the creation of distance worth
scores actually provided less information on role modeling of sports heroes than the
original distance scores used in HI and H2. Several distance worth scores had
relationships with role modeling in the opposite direction as predicted. For example,
several domains of self-concept distance worth scores, both real and ideal, were
negatively correlated with tattoo or earring modeling behaviors. These relationships were
the opposite as predicted in H3 and H4. Furthermore, these relationships were not found
in the analysis of HI and H2. It is possible that increasing importance in an area of
self-concept actually made these subjects less likely to look towards their sports heroes as
role models. As an area of self-concept became more important, these subjects likely
considered other motivations for deciding whether to model certain behaviors, such as
getting a tattoo or earring like a sports hero. Given the strong ties between domain
importance and self-worth (in contrast to self-concept, which does not incorporate
domain importance), perhaps a low sense of self-worth contributed to the decision to get
an earring or tattoo of a sports hero, while increasing self-worth makes this decision less
likely. As respondents felt better about themselves, they became less likely to get a tattoo
or earring like a sports hero.
Another explanation for this relationship between domain importance of
self-concept and role modeling is provided by the elaboration likelihood model of
persuasion (ELM), a theory that addresses the effectiveness of persuasive messages (Petty
& Cacioppo, 1986; Petty & Priester, 1994). According to ELM, messages can be
processed two ways. The more involved route, the central route, involves effortful
cognitive activity whereby the person draws upon prior experience and knowledge in


87
in high multicollinearity, making the regression analyses with all 16 distance scores as
predictor variables less stable than analyses using fewer predictor variables.
For these reasons, the regression analyses using only the eight real distance scores
(or only the eight ideal distance scores) as predictor variables are more crucial to the
study of the first hypothesis. These less complex analyses likely give a more accurate
and stable assessment of the impact of distance scores on role modeling behaviors. The
seven regressions using all 16 distance scores serve to supplement this assessment and
add information for further discussion.
Therefore, 21 separate regression analyses were completed to test HI and H2. A
stepwise regression procedure was used for these analyses. Stepwise regression, unlike
forward selection and backward elimination regressions, finds the best fit for a regression
equation by entering the predictor variables in a variety of orders (Licht, 1995; Vogt,
1993; SPSS, 1999). This type of regression drops predictor variables from a regression
model if they lose their significance when other variables are added to the regression
equation, and this makes stepwise regression the most effective method when correlation
among the independent variables is strong (Agresti & Finlay, 1986; Weisberg et al.,
1989; SPSS, 1999). Given the potential for strong correlation (or multicollinearity)
between the distance scores used as the predictor variables for HI and H2, the stepwise
method was the logical choice for the regression analyses for this study.
The F value of each regression model was examined, and only those regressions
with a p less than .05 were deemed to support HI or H2. As several regressions
addressed each of the two hypotheses, an insignificant regression equation does not reject
either HI or H2, but rather indicates that a particular role modeling behavior does not
have a significant relationship with the distance scores (real or ideal) used in that
particular regression equation. For either HI or H2 to be fully rejected, none of the
regressions for either of the two hypotheses could be significant. By using several
regression analyses for each hypothesis, this study identifies which of the particular role


213
Porter, D. (1983). Americas greatest sports figures. In R. Browne & M. Fishwick
(Eds.), The hero in transition (pp. 248-259). Bowling Green, OH: Bowling
Green University Popular Press.
Prebish, C. (1984). Heavenly father, divine goalie: Sport and religion. The Antioch
Review, 42, 312.
Rada, J. (1997). Colorblind-sided: Racial bias in network televisions coverage of
professional football games. In S. Biagi & M. Kem-Foxworth (Eds.), Facing
differences: Race, gender, and mass media (pp. 23-28). Thousand Oaks, CA:
Pine Forge Press.
Raglan, L. (1956). The hero: A study in tradition, myth, and drama. New York:
Vintage.
Raine, H. (1997, March 24). The glory and cost of childhood dreams. U.S. News and
World Report, p. 6.
Rao, B. (1979). The American fictional hero. New Delhi: Bahri Publications.
Reed, L. (1995). Secrets in the bedroom: Adolescentsprivate use of media. Journal of
Youth and Adolescence, 24, 535-550.
Reynolds, W., Ramirez, M., Magrina, A., & Allen, J. (1980). Initial development and
validation of the academic self-concept scale. Educational and Psychological
Measurement, 40, 1013-1016.
Richards, B. (1998, October 2). Nike says it plans to slash SI00 million on spending
for star endorsements. Wall Street Journal, Section B, p. 6.
Rogers, E., & Shoemaker, F. (1971). Communication on innovations: A cross-cultural
approach (2nd ed.). New York: Free Press.
Role models, sports and youth. (1985). Report from the National School Safety Center,
Malibu, CA.
Rollin, R. (1983). The Lone Ranger and Larry Skutnik: The hero as popular culture.
In R. Browne & M. Fishwick (Eds.), The hero in transition (pp. 14-45).
Bowling Green, OH: Bowling Green University Popular Press.
Rodin, J. (1992, January/February). Body mania. Psychology Today, pp. 56-60.
Rodman signs endorsement deal with Converse. (1997, January 26). Associated Press.


125
modeling indexes, the overall role modeling index, or ANDRO or TATT served as the
dependent variable. By including all distance scores as independent variables, these
regression analyses helped to determine whether real or ideal distance scores were
stronger predictors of each individual role modeling index/behavior. Additionally, these
regression analyses should create a more predictive regression equation for each of the
role modeling indexes/variables by accounting for a larger percentage of the variance of
each of these independent variables (in other words, to increase the of each regression
analysis). Finally, because the exclusion of important predictor variables can drastically
affect the results of regression analysis, it was hoped that including all the distance scores
(real and ideal) would increase the prospect of achieving statistically significant results
(Licht, 1995).
The results of all of the regression analyses for these two hypotheses are detailed
in the following tables. The tables are organized by dependent variable (role modeling
index/variable), not by hypotheses. Therefore, all three regression analyses for each of
the seven dependent variables are grouped together. A more general discussion of the
individual hypotheses follows these results. For each of the significant regression
analyses that produced a significant F-value (p less than .05), one table details both the
overall significance of the regression (including the R and R^ values and the F-value) and
the significance of the separate independent variables (distance scores). Following each
table is a description of the significant relationship(s) between the independent variable(s)
and the role modeling index/variable. This includes the Beta coefficient for each
relationship. For each relationship, indicates significance at p less than .05, and **
indicates significance at p less than .01.
For those regressions whose F Statistic is not significant at the .05 level, only a
brief text description of the analysis is given. In the proceeding section, for each role
modeling index/variable examined in this research, the regression analysis with only real
distance scores is explained first, followed by the regression analysis with only ideal


142
Table 4-42. Reduced Domain Importance.
N
Mean
Std. Deviation
Scholastic
nr

.17624
Friendship
172
.82049
.20149
Job competence
172
.80523
.20689
Romantic
172
.78997
.19152
Behavior
172
.76599
.18317
Athletic
172
.75073
.19937
Social
172
.74709
.18235
Physical
172
.73910
.19111
Vlinimum = .25;
Maximum
= 1.
Distance Worth Scores
While the domain importance scores in themselves are useful tools to better
understand adolescent behavior and self-concept, for this study, domain importance
scores were used along with distance scores to create a new construct. By incorporating
both distance scores for each domain and the importance of each of those domains, it was
hoped that this new construct, distance worth, would serve as a strong predictive factor
for the role modeling behaviors for this study. Distance worth scores were calculated for
both real self-concept and ideal self-concept across all eight domains, resulting in 16 new
variables for analysis.
To calculate distance worth scores, the individual distance scores were divided by
the normalized domain important scores. Accordingly, eight scores have been calculated
using real distance scores (real distance worth scores), and eight scores have been
calculated using ideal distance scores (ideal distance worth scores). These scores are the
independent variables in the regression analyses testing H3 and H4. The distance scores
were divided instead of multiplied by domain importance scores due to the negative
relationship proposed between distance scores and role modeling. This is in contrast to
the positive relationship proposed between domain importance and role modeling.
Therefore, to create this new construct (distance worth scores), the variables were divided


160
importance scores with distance scores would create more accurate predictors of role
modeling, this did not appear true for this research, as evidenced by the lack of significant
results found for H3 and H4 as compared to HI and H2.
The final chapter, Conclusions, addresses the stated results, discusses the
importance and relevance of these results, and explains potential reasons these results
were found. Additionally, weaknesses of the study and recommendations for future
research are addressed.


54
Congruitv Theory
The reviewed literature has established a link between self-concept and behaviors.
In order to logically establish hypotheses for such a link in this research, a theoretical
framework must be established. For this research, congruity theory provides such a
framework.
Congruity theory, proposed by Sirgy (1980), thus far has been used only in
reference to self-concept and consumer behaviors. However, for this research, the
framework of congruity theory has been extended to study self-concept and role modeling
of sports heroes. According to congruity theory, individuals assign images to consumer
products, much like they assign self-concepts to themselves. Congruity, or congruence,
refers to the relationship between ones self image and the image one assigns to a
consumer product. Specifically, congruity increases as the distance between self-concept
and product image decreases. Conversely, an increase in this distance results in less
congruity. This distance is also referred to as a distance score. (Distance scores have
an inverse relationship with congruity. Therefore, as congruity increases, the distance
score decreases.) According to congruity theory, the subsequent intention to purchase a
product is based on the congruity between ones self-concept and ones individually
assigned product image.
Sirgy (1982) has identified two conditions that predict ones intention to use a
product. The first involves real (actual) self-concept and product image congruity. Sirgy
wrote, (1982), The relationship between actual self-image/product-image congruity
(self-congruity) and brand preference has been supported by numerous studies (p. 165).
According to Sirgy (1982), as the two images become more similar, or more congruous
(lower distance score), the resulting affect will be cognitive consistency, which will lead
to subsequent purchase of that product. Conversely, a growing distance between these
two images, or decreasing congruity (higher distance score), will lead to cognitive


187
combined with interviews and focus groups, a qualitative approach, to allow subjects to
explain the observed behaviors. This research approach would reduce the potential for
socially desirable responses given for surveys and interview questions. Through
observation, researchers would be able to study role modeling behaviors that may not be
addressed through survey research, behaviors that may be overlooked during the creation
of survey instruments. Although quantitative research is the most appropriate approach
for theory testing, a more qualitative research method would give greater access to the
range of behaviors that are modeled and the conditions under which such role modeling
occurs.
The role modeling of televised sports heroes by adolescents, particularly in
conjunction with self-concept theory, is largely unexplored by academia. With the
impressive growth of sports media and the emphasis on sports and sports heroes in
modem society, future research should address a variety of topics and employ a range of
methodologies. Additionally, given the ongoing changes in the heroic construct and the
evolution in media technologies and coverage of sports by mass media, researchers
should continue to examine and reexamine the qualities of the mass mediated sports hero,
the relationships between these athletes and adolescent fans, and the various impacts of
these sports heroes on adolescents. Such a research agenda would help to provide insight
into an important area of modem society and help guide media critics, parents, educators,
and adolescents in better understanding the social impact of the modem American sports
hero in their individual lives.


166
In reviewing the individual role modeling variables, several worthwhile
observations should be noted. First, the range on a 5-point Likert scale from the lowest
rated role modeling variable (getting a tattoo or earring like ones hero = 1.44) to the
highest rated role modeling variable (train harder because of ones hero = 3.84) is
considerable. While these adolescents appear very likely to model the training techniques
of their sports hero, they are very unlikely to get a tattoo like their sports hero. It is
similarly unlikely that these adolescents will get the same hair style as their sports hero
(HAIR = 1.46).
In more general terms, five of the six variables with the highest mean scores deal
with on-the-court behaviors (training, trash talking on the court, playing a particular
position, training techniques, and athletic maneuvers on the court). Besides the variable
dealing with trash talking (TRASH = 3.27), all of these variables deal specifically with
athletic behaviors and training. This indicates that these subjects value the athletic
actions of their heroes above other behaviors, and often they indicated intentions to model
these actions. Such would be consistent with the reported ratings of the image of their
sports heroes, where these subjects ranked the athletic and job performance domains as
two of the highest rated domains of this image. In other words, these adolescents seemed
more interested with the athletic components and qualities of their sports heroes than with
the many other facets of their sports heroes.
Conversely, many of the variables with the lowest mean scores deal either with
negative behaviors (getting a tattoo, hairstyle, taking Andr) or major life decisions
(choosing a college). It is predictable that these adolescents would not want to model
their sports heros college choice (COL = 1.62) given their relatively low opinion of the
scholastic image of their sports heroes. Somewhat contusing and contradictory, however,
is the relatively high mean score for the role modeling variable dealing with studying
harder (STUDY = 2.82). However, the choice to study harder is a much less intrusive
decision than choosing a college, and this distinction may account for the vast difference


163
closest in distance to the image of the sports hero. In other words, this explains in which
domains the adolescent felt closest to his particular sports hero. From this information,
one can gain insight into the perceived similarity of todays sports heroes to adolescent
fans and perhaps even understand why certain types of popular athletes are chosen as
sports heroes.
Not surprisingly, the ideal athletic distance score had the lowest mean score
(3.70). In other words, these subjects viewed their sports heroes as closer to their ideal
self in athletics than for any other domain of self-concept, real or ideal. According to
these results, the sports hero, first and foremost, embodies ideal athletic abilities,
confirming the notion that sports heroes must be supreme athletes above all else (good
looks, good behavior, etc.) to be considered sports heroes. Although the media may
stress a variety of characteristics of popular athletes, these adolescents placed a premium
on athletic performance. This emphasis should help to further define the current state of
the evolving modem American sports hero.
The relative order of the 16 calculated distance scores (eight real distance, eight
ideal distance) gives insight into the qualities of the modem American sports hero. Many
theorists have suggested that the sports hero has lost his luster, largely due to invasive
media coverage, and this heroic figure no longer represents societys ideal. However,
such a sentiment is not wholly supported by this research. In fact, according to the
calculated distance scores, the sports heroes chosen for this research were more closely
aligned to the ideal self-concept than the real self-concept. Of the 16 distance scores, the
six scores with the smallest mean score (lower score equaling closer proximity between
hero image and self-concept) were ideal distance scores. The ideal distance scores for
athletic ability, physical appearance, and job competence, all domains related to athletic
and physical characteristics, were three of the four smallest distance scores. Such
proximity between the ideal self and the sports hero in these domains was expected due to
the athletic excellence and the physical appearance of the sports hero achieved from years


184
Other future research could address the impact of mass mediated sports heroes, as
studied in this dissertation, vs. personally known sports heroes. This would give insight
into the impact of mass media in providing role models for adolescents. Similar research
projects should address the impact of mass mediated sports heroes vs. traditional,
personally known role models, including teachers, peers, and parents. Again,
comparisons of this nature would further examine the relative influence of mass media in
providing role models and could also speak to the importance of traditional role modeling
influences in the lives of adolescents.
Self-Concept
This dissertation examined two distinct types of self-concept, real self-concept
and ideal self-concept, across eight separate domains. In doing so, knowledge was gained
into the relative impact of real self-concept congruity vs. ideal self-concept congruity in
influencing role modeling of sports heroes. In other words, adolescent subjects for this
research expressed whether they role model sports heroes who more closely represented
their real self or their ideal self and in which areas such role modeling was likely to occur.
However, given the prevalence of negative actions of star athletes and the coverage this
negative activity receives from mass media, future research should address the impact of
negative athletes, athletes not typically seen as sports heroes, and the impact these
athletes have on the behaviors of adolescent viewers. Specifically, research should
address whether viewers model the negative behaviors observed through the media or if
they instead actively avoid the negative behaviors of these athletes. Negative, non-heroic
athletes would not provide a model for how to act, but rather for how not to act.
To conduct such research, another component of self-concept should be
addressed, the negative self-concept. While researchers have typically limited studies to
real and ideal self-concept, particularly when addressing congruity theory, future research
could examine a negative component of self-concept (myself as I do not want to be) and
how this construct impacts avoidance behaviors. Research should address whether


131
Consumer Behaviors
Real distance scores. No significant relationships were found in the regression analysis
for real distance scores on consumer behaviors (p > .05).
Ideal distance scores. No significant relationships were found in the regression analysis
for ideal distance scores on consumer behaviors (p > .05).
Real and ideal distance scores. No significant relationships were found in the regression
analysis for real and ideal distance scores on consumer behaviors (p > .05).
Overall Behaviors
Table 4-36. Regression Analysis for Real Self-Concept
Distance Scores on Overall Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
2.990
.115
25.920'
OTT
Real Athletic Distance
-.032
.012
-.223
-2.754
.007
Real Behavior Distance
-.005
.010
-.049
-.538
.592
Real Friend Distance
.003
.010
.030
.306
.760
Real Job Distance
-.035
.013
-.246
-2.681
.008
Real Physical Distance
.023
.009
.215
2.476
.014
Real Romantic Distance
-.007
.011
-.060
-.588
.557
Real Scholastic Distance
-.001
.010
-.011
-.126
.900
Real Social Distance
-.003
.014
-.018
-.185
.854
R = .391
R Square = .153
F = 3.323
Significance of F = .002.
The distance scores with significant relationships with overall behaviors were real athletic
distance (Beta = -.23**), real job distance (Beta = -.22*), and real physical distance (Beta
= .21 *, NOTE: This relationship is the opposite direction as predicted).


29
turns spectator sports into observance of intimates, albeit a controlled
intimacy within the confines of a 19-inch screen, (p. 171)
The individual has become so paramount in televised sports that television ratings
often hinge upon whether a star athlete, such as basketball star Michael Jordan, plays in a
particular contest (Fatsis, 1998; Shapiro, 1999; Starr, 1999). Professional basketball
(NBA) in particular has employed a marketing strategy of promoting star players ahead
of the actual game (Starr, 1999). For the 1997-98 basketball season, NBCs ratings for
nationally televised Chicago Bulls games, featuring Michael Jordan, were 75% higher
than for the networks other nationally televised NBA contests (Fatsis, 1998). With
Jordans recent retirement from basketball, both league and network executives have
openly worried over the potential loss of league popularity and ratings of televised games.
Peter Roby, vice president of marketing for Reebok Athletics, asserted that the league
would never again reach the status attained through Jordans star appeal, saying, I dont
think the NBA will ever again attain the glory it did with Michael (Starr, 1999, p. 56).
To combat this potential loss, networks have embraced the opportunity to create
and publicize new athletic stars. NBC Sports chairman Dick Eberson said in reference to
the potential for decreased ratings upon Jordans retirement, We do have an opportunity
to reintroduce a whole new generation of stars (Shapiro, 1999, p. D5). In other words,
the individual star remains the prime emphasis for television sports, and the retirement of
one star leads to the creation of other stars. Nike corporation CEO Phil Knight, whose
company airs numerous commercials featuring star athletes, has stated that even his
television advertising campaigns alone are powerful enough to create popular sports
figures. For instance, in reference to then-NBA newcomer Alonzo Mourning, Knight
said, All wed have to do is take the money were using to advertise Charles Barkley
and Michael Jordan, put it into Alonzo, and a lot more people would be talking about
him (Katz, 1994, p. 253). With the excessive television coverage of sports stars and
the media influence on actually creating star appeal of accomplished athletes, the


185
negative behaviors of athletes, such as shoving coaches, fighting, and spouse abuse, are
actively modeled or actively avoided and whether the congruity between negative
self-concept and the image of a negative sports hero is predictive of this process.
Because cultural critics, parents, and educators alike have criticized mass mediated sports
heroes as poor role models due to their negative behaviors, this direction of research
would present a logical step in further addressing these concerns.
Another area for future research involves the importance of self-worth or
self-esteem in impacting role modeling of sports heroes. For this research, domain
importance scores, a primary component of determining self-worth, did not contribute to
the importance of self-concept congruity. Conversely, increased domain importance
scores, which lead to higher self-worth, often reduced the impact of congruity in
determining role modeling. Therefore, it is important to consider whether adolescents
with a high self-worth are less likely to role model sports heroes. In other words,
researchers should study whether adolescents with high opinions of themselves are less
likely to copy the behaviors of mass mediated sports heroes. Similarly, adolescents with
a low sense of self-worth might role model their sports heroes more frequently in an
effort to feel better about themselves. Research on adolescent females has suggested that
low self-worth can serve as a behavioral influence to adopt behaviors such as weight loss
techniques in attempt to achieve a higher sense of self-worth (Orenstein, 1994). Similar
research on adolescent role modeling of sports heroes and self-worth would give new
insight into the actual motivations for role modeling decisions.
Methodological Alterations
The final area for additional research deals with potential methods to study
adolescent role modeling of sports heroes. This dissertation used surveys and
quantitative statistics to measure and analyze role modeling of sports heroes. However,
while this approach is both practical and allows for access to a large subject pool, these
methods are not necessarily the most effective means to study this topic. Subjects may be


64
promoted by school faculty and leadership. Therefore, athletics, at least in the
participatory sense, is a part of the daily lives of many of this schools students.
The remaining 52 subjects were participants in a sports tournament in Houston
run through a local community center. These subjects, of the same grade and age
parameters as the first 120 subjects, also were participants in school athletics. The
majority of these subjects participated in baseball, but other selected sports included
soccer, basketball, and track. These subjects share similar demographics traits with the
first 120 students surveyed (white, upper-middle-class, supportive family environment),
and in fact the data collected from the two groups were virtually identical. Again this
group of subjects was selected largely for pragmatic reasons (availability, access, age
appropriateness). Using two groups of subjects and receiving similar results helped to
strengthen the validity of this research. Therefore, the data analysis for this study and the
subsequent analyses of the stated hypotheses were based on the responses of 172 valid
subjects. While more than 200 subjects participated in the study, surveys from
approximately 30 subjects could not be used due to subject error in completing the
surveys. The majority of these errors involved subjects not completing all parts of the
survey questionnaires.
One criticism of this study may be that the students do not represent a diverse
sample, decreasing external validity. Similarly, because most of these students come
from stable family environments with many available role models, these students may not
use sports heroes as role models as much as adolescents who do not have many other
available role models. However, this study will not claim to be generalizable to all
populations, nor is it able to address issues of race, socioeconomic status, and
family/home environment. Additionally, because this study addresses each persons
individual self-concept and allows each person to identify his individual choice of sports
hero, the focus of this research is not on societal patterns, but rather individual choices
based on individual relationships between self-concept and the image of their sports hero.


88
modeling behaviors have significant relationships with distance scores, and which of the
role modeling behaviors do not have such significant relationships. The regressions with
significant F statistics indicate which indexes/variables of role modeling support HI and
H2.
For each of the significant regressions, the t statistic, the significance of this
statistic, and the Beta statistic for each of the individual predictor/independent variables
(distance scores) were examined. These statistics clarified which of the distance scores
had significant relationships with specific role modeling indexes/variables and the
relative strength of these relationships. To determine significance of each of the predictor
variables, the test statistic (t) was used. For any predictor variable to be deemed to have a
significant relationship with a role modeling index/variable, the p value of the test
statistic must be less than .05 (p less than .05). The beta weights (B), or standardized
regression coefficients, of each significant relationship were examined to explain the
strength of the relationship between individual distance scores and role modeling
indexes/variables. This statistic not only explained the extent to which any of the
individual distance scores related to role modeling, but it also explained which of the
distance scores had the strongest relative relationships with role modeling.
An examination of the results of the 21 completed regression analyses provide an
explanation of HI and H2 and selective support (or lack of support) for these hypotheses,
which predict significant relationships between role modeling and distance scores.
Domain Importance and Distance Worth Scores
Domain importance for each of the eight domains of self-concept was calculated
from Harters test of domain importance as part the Self-Perception Profile for
Adolescents (1988). (See Appendix A for the actual test questions.) Two questions
addressed each of the eight domains of self-concept, and a mean score was calculated for
each domain. Domain importance was measured on a scale from one (1) (least important)
to four (4) (most important). These domain importance scores were used, along with the


216
Suci, G., & Tannenbaum, P. (1957). The measurement of meaning. Urbana, IL:
University of Illinois Press.
T.V. sports. (1996, May 13). Broadcasting and Cable, p. 34.
Tapin, I. (1988). Why we need heroes to be heroic. Journal of Popular Culture, 22,
133-142.
Taylor, P. (1991, November 25). A model role model. Sports Illustrated, pp. 78-81.
Television and growing up; The impact of televised violence. (1972). Report to the
Surgeon General, United States Public Health Service. Rockville, MD: National
Institute of Mental Health.
Temple, K. (1992). Interview. In J. Coakley (1994), Sport in society, St. Louis: Mosby.
Tener, R. (1987). The significance of hero choice as related to self-concept and
ego strength in adolescence. Doctoral dissertation, University of Detroit.
Thomson, R. (1993). Sport and ideology in contemporary society. In A. Yiannakis,
M. Melnick, & T. McIntyre (Eds.), Sport sociology: Contemporary themes
(4th ed.) (pp. 90-96). Dubuque, IA: Kendall/Hunt.
Tiggerman, M., & Pickering, A. (1996). Role of television in adolescent womens body
dissatisfaction and drive for thinness. The International Journal of Eating
Disorders, 20, 199.
TKO: Tyson gets year in jail for assault. (1999, February 6). Associated Press.
Tollefson, T. (1993, May/June). Is a hero really nothing but a sandwich? Utne Reader,
pp. 102-103.
Vesser, L. (1998, January 24). [Interview with Charles Groden]. The Charles Grodin
Show. CNBC Network.
Videbeck, R. (1958). Properties of the concept role. Midwest Sociologist, 20, 104-108.
Voigt, D. (1978). Myths after baseball: Notes on myths in sports. Quest: Literature
Myth, 30, 46-57.
Vogt, W. (1993). Dictionary of statistics and methodology. Newbury Park, CA: Sage.
Walden, D. (1986, January). Where have all our heroes gone? USA Today Magazine,
pp. 20-25.


50
Even less is known about the relationship between the sports hero and the
observer in reference to the fans individual self-concept. Teners (1987) dissertation
work found that the hero choices of an adolescent can be used to predict self-concept,
suggesting a consistency between the hero and the observer. However, this work did not
deal with the real and ideal self-concepts, but rather just the strength of an adolescents
real self-concept. Additionally, this research focused on the importance of choosing a
personally known vs. an unknown hero, and it did not address the different characteristics
of chosen heroes.
While not addressing the specific construct of self-concept, several authors have
suggested that fans may choose and role model their favorite sports figures based on their
perceived similarities between themselves and the athlete (Wilson & Sparks, 1996; Cole,
1996; Kellner, 1996; Harris, 1994a; Simons, 1997; Role models, 1989). The work in
this area has not addressed the individuals perception of himself, or self-concept, but
rather studied observable characteristics of the individual, including race, playing style,
and gender. Therefore, one primary goal of this research is to examine modeling of
sports heroes in relation to individual self-concept, not general group characteristics or
qualities.
Adolescent Self-Concept
This study deals specifically with self-concept of adolescents. Adolescence
represents the beginning of drastic change to ones self-concept, marked by an increased
awareness of self-concept (Orenstein, 1994; Harter, 1988; Frey & Carlock, 1989;
Hattie, 1992); Felker (1974) explained, During adolescence, there is an increased drive
for change in the individual. The increased change and corresponding lack of stability
brings with it a need for reorganization of the self-concept (p. 104). Such reorganization
has been labeled a crisis of identity, a process of drastic change and instability in
self-concept during adolescence (Hattie, 1992, p. 132; Erikson, 1950).


78
validity is increased by insuring that one measurement tool does not
correlate highly with another measurement tool that is used to measure a
different, dissimilar construct. For this study, content validity has been
established by checking that real self-concept and ideal self-concept do not
have a high correlation (discriminant validity). A statistical comparison of
the collected data for real self-concept and ideal self-concept can be found
in the results section of this dissertation.
6) External Validity: External validity, often know as generalizability,
refers to the extent to which the results of a study can be generalized to
different populations and conditions (Graziano & Raulin, 1993). Threats
to external validity are reduced through careful selection of participants,
including random sampling and attempting to choose a representative
sample. However, for this study, subjects were not randomly selected.
Instead, this study used convenience samples, and therefore results must
be generalized very tentatively, likely only to those who have
characteristics similar to those in the study. It is more likely that the
results cannot truly be generalized at all to other populations (Graziano &
Raulin, 1993). However, it should be noted that this study examined
relationships between individual self-concepts, ones individual choice of
a sports hero, and ones individual intentions to role model. Therefore,
while this study cannot be truly generalized to the adolescent male
population at-large, the results present relationships that can be studied
and researched in other, broader adolescent populations.
7) Internal Validity: Internal validity deals primarily with causality,
particularly whether the independent variable is actually responsible for
the observed changes in the dependent variable (Graziano & Raulin,
1993). For this study, most of the threats to internal validity (maturation


82
chapter. For each of these three constructs, a mean score was calculated in each of the
eight domains of self-concept. Each of these 24 domains had five questions. Before a
mean score could be calculated, the results of all questions asked in a negative direction
had to be inverted such that the results from each question ranged from lowest rating of
self-concept/image of sports hero (1) to highest rating (4). (See the Appendix for
questionnaires and directions for individual questions.)
It should be noted that while these mean scores for both self-concept and the
image of the sports hero were reported in the results for informative purposes, these mean
scores were not used in the calculation of the distance scores between self-concept and
the image of the sports hero. Instead, distance scores were calculated from the individual
questions, not the mean scores, to provide a more accurate determination of the specific
distances between the constructs. An explanation of this calculation follows in the next
section.
Distance Scores
Distance scores are a measure of the distance between two concepts, and they can
be used to measure the similarity between two such concepts (Osgood et al., 1957) For
this study, distance scores were calculated to measure the distance between self-concept,
both real and ideal, and the image of the sports hero. In other words, these distance
scores indicate the similarity between self-concept (real and ideal) and the image of the
sports hero. These two separate sets of distance scores, real distance and ideal distance,
have been calculated for each of the eight domains. This resulted in 16 distance scores (8
real, 8 ideal) for use in this study.
Several equations can be used to calculate distance scores. For this study, the
difference squared model, which squares the difference between each paired set of
questions and sums these differences, has been used (Sirgy, 1983; Osgood et al., 1957).
The difference squared model assured a positive value for the distance between each set
of values, regardless of which value, the self-concept or the image of the sports hero, is


TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS i
ABSTRACT vi
CHAPTERS
1 INTRODUCTION 1
Mass Media Research 3
Self-Concept Theory 11
Problem Statement 13
2 REVIEW OF LITERATURE 16
Role Models and Behavioral Modeling 16
The Sports Hero as a Role Model Figure -24
Self-Concept -38
Congruity Theory 54
Hypotheses 58
3 METHODS 62
Subjects for Study 62
Tests and Test Administration 65
Issues of Validity for this Study 75
Data Analysis 80
4 RESULTS AND DISCUSSION 96
Descriptive Statistics 97
Self Concept, Image of Sports Hero, and Distance Scores 110
Results for Hypothesis I and Hypothesis II 124
Summary Chart of Significant Results of Regression Analyses for HI and H2 135
Summary of Results of HI and H2 138
Domain Importance and Distance Worth 140
Results for Hypothesis III and Hypothesis IV 143
Summary Chart of Significant Results of Regression Analyses for H3 and H4 155
iv


101
Really Sort of
Thi* Thie
for Mo for Mo
Sort of
Thie
for Me
Realty
True
for Me
u


Some teenagers are popular
with others their age
BUT
Other teenagers are not
very popular.
c

n.


Some teenagers don't do well
at new outdoor games
BUT
Other teenagers are good at
new games right away.
c

31


Some teenagers think that
they are good looking
BUT
Other teenagers think that they
are not very good looking.


3X


Some teenagers feel like they
could do better at work they
do lor pay
BUT
Other teenagers feel that they
are doing really wen at work
they do lor pay.


30.


Some teenagers feel that they
are fun and interesting on
a date
BUT
Other teenagers wonder about
how fun and interesting they
are on a date


S'-


Some teenagers do things
they know they shouldn't do
BUT
Other teenagers hardly ever
do things they know they
shouldn't da


32.


t
Some teenagers find it hard
to make friends they can
really trust
BUT
Other teenagers are able
to make dose triends they
can really trust.


33.


Some teenagers feel that
they are pretty intelligent
BUT
Other teenagers question
whether they are intelligent.


3*


Some teenagers feel that they
are socially accepted
BUT
Other teenagers wished
that more people their age
accepted them.


35.


Some teenagers do not feel
that they are very athletic
BUT
Other teenagers feel that they
are very athletic

!

3i.


Some teenagers really like
their looks
BUT
Other teenagers wish they
looked different.

1

31.


Some teenagers feel that they
are really able to handle
the work on a paying job
BUT
Other teenagers wonder if they
are really doing as good a job
at work as they should be doing

1

31


Some teenagers usually don t
go out with the people they
would really like to date
BUT
Other teenagers do go out
with the people they really
want to date.

1

39.


Some teenagers usually act
the way they know they are
supposed to
BUT
Other teenagers often don't
act the way they are
supposed ta

1

Ho.


Some teenagers don't have
a friend that is close enough
to share really personal
thoughts with
BUT
Other teenagers do have a
close friend that they can share
personal thoughts and
feelings with.
1



120
Table 4-22. Reliability Statistics for Ideal Social Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance l
IW
.414
.163
.155
.466
Distance 2
.414
1.000
.263
.305
.389
Distance 3
.163
.263
1.000
.316
.316
Distance 4
.155
.305
.316
1.000
.499
Distance 5
.466
.389
.316
.499
1.000
N of Cases = 168
Alpha = .7066
Average Item to Total Correlation = .4673.
Table 4-23. Reliability Statistics for Ideal Scholastic Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance I
T"
.276
,irr
W
.205
Distance 2
.276
1.000
.360
.317
.361
Distance 3
.312
.360
1.000
.499
.393
Distance 4
.259
.317
.499
1.000
.341
Distance 5
.205
.361
.393
.341
1.000
N of Cases = 170
Alpha = .7125
Average Item to Total Correlation = .4719.
Table 4-24. Reliability Statistics for Ideal Athletic Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance 1
1.000
.456
.116
.135
.171
Distance 2
.456
1.000
.103
.305
.189
Distance 3
.116
.103
1.000
.197
.246
Distance 4
.135
.305
.197
1.000
.234
Distance 5
.171
.189
.246
.234
1.000
N of Cases = 169
Alpha = .5753
Average Item to Total Correlation = .3352.


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
THE MASS MEDIATED SPORTS HERO AS A ROLE MODEL FOR
ADOLESCENT MALES: MEASURING SELF-CONCEPT
CONGRUITY WITH THE PERCEIVED IMAGE OF THE HERO TO
PREDICT ROLE MODELING BEHAVIORS
By
Keith Andrew Strudler
May 2000
Chair: John Sutherland
Major Department: Journalism and Communications
The sports hero has become a popular and influential figure in modem American
society. The growth of this social phenomenon can be attributed largely to heavy mass
media coverage of sports, with a particular emphasis placed on the star athlete. This
emphasis by a variety of media outlets, most notably television, allows fans to learn more
about their favorite athletes, both positive and negative, than possible for previous
generations of fans. Such coverage has created concern over the potential role modeling
influence of star athletes, particularly such behavioral influence on adolescents. Two
primary areas of interest are the actual behaviors that are most likely to be role modeled
and the conditions under which such behavioral role modeling will occur. A theoretical
foundation to understanding human behaviors is provided by congruity theory.
According to congruity theory, the closer ones self-concept is to his or her image of a
consumer product, the more likely this person is to purchase this product.
vi


135
Ideal distance scores. No significant relationships were found in the regression analysis
for ideal distance scores on using Andr modeling behaviors (p > .05).
Real and ideal distance scores. No significant relationships were found in the regression
analysis for real and ideal distance scores on using Andr modeling behaviors (p > .05).
Summary Chart of Significant Results of Regression Analyses for HI and H2
The results of the above regression analyses are summarized in the following
table. A number in a cell indicates a significant relationship was found between that role
modeling index/variable and the corresponding distance score. The particular regressions
in which the significant relationships were found are explained by the key beneath the
table. In other words, the numbers indicate whether the relationship was found in the
regression including only real distance scores, the regression including only ideal distance
scores, or the regression including real and ideal distance scores.


THE MASS MEDIATED SPORTS HERO AS A ROLE MODEL
FOR ADOLESCENT MALES: MEASURING SELF-CONCEPT
CONGRUITY WITH THE PERCEIVED IMAGE OF THE HERO
TO PREDICT ROLE MODELING BEHAVIORS
By
KEITH ANDREW STRUDLER
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2000

For Seth, CJ, and X Butlers everywhere.

ACKNOWLEDGMENTS
Thanks to my parents and family for always believing that I could make it and to
my friends for not holding it against me if I did not. Thanks to all the professors who
made me think and all the staff who made me remember. Finally, a special thanks to
Dr. John Sutherland and Dr. Kim Walsh-Childers who, both in their own way, provided
the guidance, knowledge, and patience necessary to write this dissertation.
in

TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS i
ABSTRACT vi
CHAPTERS
1 INTRODUCTION 1
Mass Media Research 3
Self-Concept Theory 11
Problem Statement 13
2 REVIEW OF LITERATURE 16
Role Models and Behavioral Modeling 16
The Sports Hero as a Role Model Figure -24
Self-Concept -38
Congruity Theory 54
Hypotheses 58
3 METHODS 62
Subjects for Study 62
Tests and Test Administration 65
Issues of Validity for this Study 75
Data Analysis 80
4 RESULTS AND DISCUSSION 96
Descriptive Statistics 97
Self Concept, Image of Sports Hero, and Distance Scores 110
Results for Hypothesis I and Hypothesis II 124
Summary Chart of Significant Results of Regression Analyses for HI and H2 135
Summary of Results of HI and H2 138
Domain Importance and Distance Worth 140
Results for Hypothesis III and Hypothesis IV 143
Summary Chart of Significant Results of Regression Analyses for H3 and H4 155
iv

Summary of Results of H3 and H4 158
Summary of Results 159
5 CONCLUSIONS 161
Descriptive Statistics 161
Hypothesis I and Hypothesis II 168
Hypothesis III and Hypothesis IV 175
Limitations of the Study 178
Recommendations for Future Research 181
APPENDICES
A SELF-CONCEPT/IMAGE OF THE SPORTS HERO QUESTIONNAIRE 188
B ANSWER GUIDE FOR SELF-CONCEPT/IMAGE OF THE SPORTS HERO 193
C ROLE MODELING QUESTIONNAIRE 197
REFERENCES 200
BIOGRAPHICAL SKETCH 218
v

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
THE MASS MEDIATED SPORTS HERO AS A ROLE MODEL FOR
ADOLESCENT MALES: MEASURING SELF-CONCEPT
CONGRUITY WITH THE PERCEIVED IMAGE OF THE HERO TO
PREDICT ROLE MODELING BEHAVIORS
By
Keith Andrew Strudler
May 2000
Chair: John Sutherland
Major Department: Journalism and Communications
The sports hero has become a popular and influential figure in modem American
society. The growth of this social phenomenon can be attributed largely to heavy mass
media coverage of sports, with a particular emphasis placed on the star athlete. This
emphasis by a variety of media outlets, most notably television, allows fans to learn more
about their favorite athletes, both positive and negative, than possible for previous
generations of fans. Such coverage has created concern over the potential role modeling
influence of star athletes, particularly such behavioral influence on adolescents. Two
primary areas of interest are the actual behaviors that are most likely to be role modeled
and the conditions under which such behavioral role modeling will occur. A theoretical
foundation to understanding human behaviors is provided by congruity theory.
According to congruity theory, the closer ones self-concept is to his or her image of a
consumer product, the more likely this person is to purchase this product.
vi

This study adapts congruity theory to examine behavioral role modeling of mass
mediated sports heroes by American adolescent males. Surveys were given to 172 9th
and 10th grade male student athletes. These subjects first identified their individual
sports hero. Subjects then used Harters Self-Perception Profile for Adolescents to rate
their real and ideal self-concepts and the image of the their sports heroes. Finally,
subjects rated role modeling of their sports heroes for a variety of behaviors. Regression
analyses were completed to examine whether the proximity between self-concept and the
image of a sports hero, or congruity, correlated with role modeling behaviors of the sports
hero. According to the statistical analysis, behavioral role modeling correlated with
congruity between real self-concept and the image of the sports hero for a limited number
of behaviors. This correlation came primarily with sports-related behaviors. Almost no
correlation was found for the congruity between ideal self-concept and the image of the
sports hero. Additionally, the increasing importance of an area of self-concept did not
correlate with increased role modeling as predicted.
vii

CHAPTER 1
INTRODUCTION
Star athletes for years have assumed a heroic position in American society.
Historically, these star athletes were assumed to represent the best of American society.
As written by Lewis Lord in U.S. News & World Report as an obituary to the recently
deceased baseball hero Joe DiMaggio,
the recent death of baseball legend Joe DiMaggio recalls an era when a
sports hero embodied Americas hopes and dreams... Stories about and
pictures of DiMaggio are extremely evocative, representing as they do the
way America was -- or at least wished it was. (1999, p. 18)
DiMaggio played his career in an era when television and media did not heavily
influence and potentially overexpose star athletes and when such athletes did not
command the enormous salaries relative to the society at large. However, mass media,
particularly television, has had a profound influence in recent years on popular sports and
the fans perspectives of star athletes. Certainly, fans now have greater opportunity to
watch their favorite athletes perform. Fans also have greater insight into the personal
lives of star athletes, an insight that often has exposed a less than heroic side of these
stars. The money generated through media coverage has allowed star athletes to
command yearly salaries of millions of dollars, not including the income they derive from
advertising and endorsements. Obviously, the perspective of todays mass mediated
sports stars differs greatly from that of the sports heroes of past generations.
For a variety of reasons, many journalists and social critics have argued that the
sports hero no longer exists in modem American society. Popular journalist Mike Lupica
wrote in 1994,
1

2
Americas sports heroes were once larger than life, but today even the idea
of the sports hero is dead in America. A great number of sports heroes
have fallen off their pedestals in the public eye... Sports heroes have
destroyed themselves, but the fans and the media have helped them along.
The fans want to know too much about the players, and the media are only
too willing to give them what they want. (p. 62)
Whether the sports hero still exists in America or whether this figure has merely changed
with changing social conditions is debatable. However, it is obvious that the star athlete
assumes a visible and lofty position in American society.
Of greater importance than the labeling of star athletes is the potential influence of
these highly conspicuous persons. It is often assumed that popular star athletes, such as
basketball players Michael Jordan and Charles Barkley, are role models to many views,
particularly children and adolescents (Peart, 1996). If young sports fans model positive
behaviors of athletes, then this relationship has a positive influence on American society.
Positive behaviors of star athletes might include athletic skill, good sportsmanship, and
community involvement. However, if adolescents model the highly publicized negative
activities and characteristics of many star athletes, including drug use, greed, and violent
behavior, then this status is highly problematic for society at large (Benedict & Yaeger,
1998; Keteyian, 1998).
Therefore, the more important debate about sports heroes focuses not only on the
characteristics of these athletes, but rather whether these star athletes serve as role models
to Americas youth. This debate is complicated by the increasing diversity of American
society, the diversity of star athletes, and the huge number of sports stars accessible
through various media outlets. Some journalists and social critics have written that
athletes are no longer role models for American adolescents. Harry Edwards expressed
that opinion in Sport magazine (1994), writing,
athletes have long been expected to be role models, but this
notion is becoming outdated. The role model expectation is based on the
assumption that star athletes and their fans share a uniform group of
values, but this is no longer true. Indeed, the racial, ethnic, gender, and

3
lifestyle diversity of the United States has shattered all but the pretense of
consensus upon which this notion is founded. Athletes are sports
entertainment celebrities with no more special calling, capability, or
obligation to be role models than anyone else in the mass-oriented
entertainment business, (p. 32)
According to this viewpoint, not only have star athletes lost their heroic credential, but
due to the complexity of modem American society, these athletes are no longer role
models.
The question of whether sports heroes still serve as role models for young
American sports fans is of great interest to parents, educators, and society in general.
Further, it has become increasingly important to understand the specific behaviors that
are most likely to be modeled and the motivations for individual fans to model their own
favorite sports stars. This research project will address these topics to help gain a better
understanding of the function of the modem star athlete in the lives of American
adolescents. One area of study crucial to understanding the modem American sports hero
is mass media theory and research. This topic will be reviewed extensively throughout
this dissertation.
Mass Media Research
Years of academic research has focused on the vast impact mass media, and
specifically television, historically have had upon society. This school of research has
produced theories predicting a range of effects, varying from theories examining general
societal impact of media to those dealing with a narrow range of media types and
populations.
Obvious reasons exist for the ongoing interest in this research agenda. By 1980,
televisions were present in 98% of American households, a percentage that has held
steady since that time (Harris, 1994b; Andreasen, 1990). Television viewing comprised
an inordinate amount of American leisure time, more than four times more than any other

4
single activity (Cohen, 1993; Spring, 1993). Americans watch, on average, four hours of
television a day, and the television remains on for seven hours a day, on average, in the
American household (Harris, 1994b; Kubey & Csikszentmihalyi, 1990). Statistics on
television in the lives of children are even more staggering. American youths, on
average, watch approximately 1,500 hours of television a year, far greater than the 900
hours the American youth will spend in school annually (Harris, 1994b). With so much
of Americas time invested in television, it is natural and vital that researchers attempt to
understand the various impacts of television content on society.
Television and Society
Americas interest in television has produced a society somewhat dependent upon
television for its daily existence. Ball-Rokeach and DeFleur (1976) explained such a
relationship with their dependency theory, which suggests that, particularly in a
modem urban-industrial society, audiences have a high level of dependence on mass
media information (Severin & Tankard, 1992, p. 262). America, a highly industrial and
increasingly urban society, exemplifies such a reliance on mass media information. This
dependent relationship creates a need for continuous research on the ramifications of such
dependence on American society.
One area of particular interest for this research project is the potential for
television viewers to copy the behaviors and actions of characters they see on television
and to use television figures as role models. Much of the prior research has focused on
violent behaviors and models, including such notable research works as the Surgeon
Generals Report on Television Violence, Television and Growing Up (1972), which
established tentative relationships between viewing violent television content and
modeling of those behaviors (Lowry & Defleur, 1983). Other examples of research
concerning modeling of television content have shown that viewers may model
anti-social behaviors such as smoking and drinking (Charlton & Blair, 1989; Botvin et
ah, 1991; Conrad et al., 1987), prosocial behaviors (Cohen, 1993; Bandura, 1994), styles

5
of dress and fashion (Muuss, 1975; Wilson & Sparks, 1996; Harris, 1994b), verbal
expression (Muuss, 1975), and career selection (Muuss, 1975; Harris, 1994b).
Research has indicated that many television viewers look to television for role
models, and these televised role models have replaced traditional models such as family,
peers, and teachers (Jung, 1986; Hendricks, 1992; Harris, 1994a; Pleiss & Feldhusen,
1995; Role models, 1989; Duck, 1990). This may be particularly true during
adolescence, an age when children often reject traditional role models, such as parents,
and seek new role models to help establish their own identity (Jung, 1986; Orenstein,
1994; Krechetal., 1962; Hendricks, 1992; Felker, 1974). While one would hope
television role models would display positive behaviors, this may not always be the case,
and critics have suggested that much of television content portrays potential role models
engaged in negative activity (Harris, 1994a). These potentially negative role models on
television have drawn growing attention, and they are paramount in the study of the
questionable behaviors of popular athletes as role models.
Sports Television
Sports television, central to this research project, has received growing attention
by networks and viewers alike in recent years. While the first televised sporting event
was a college baseball game between Columbia and Princeton in 1939, sports did not
gain much attention from television networks until the 1960s (McPherson et al., 1989;
Davies, 1994; Coakley, 1994). Sports has been an increasingly vital part of the
television industry since that time.
Growing interest in sports television developed from a variety of factors,
including a gradual exodus from downtown living areas, where stadiums are located, to
the suburbs; realization by networks that sports offered an opportunity to boost ratings
during traditionally slow times, such as Sunday afternoons; and the ability for advertisers
to consistently attract young male viewers (Davies, 1994; Nixon, 1984; Harris, 1994a;
Harris, 1994b; Katz, 1996; McPherson et al., 1989; Coakley, 1994; Clarkin, 1990).

6
Sports programming has increased through to the 1980s, when, on average, sports
programming accounted for 15% of total programming on commercial television, a
percentage that has held steady through the mid-1990s (Harris, 1994a; Coakley, 1994).
Several television networks, including ESPN and Fox Sports, are now devoted solely to
sports content, 24 hours a day, making sports extremely accessible to television viewers.
Along with the increase of sports media coverage has come a change in the
content of televised sports. Much of this can be attributed to the enormous amounts of
money networks pay for the rights to televise amateur and professional sports. For
example, in 1998 four networks (ABC, CBS, ESPN, and FOX) agreed to pay the
National Football League (NFL) $17.6 billion over eight years for the right to broadcast
professional football (Goldberg, 1998). This amount far surpasses the amount teams
receive each year in ticket revenues. The National Basketball Association (NBA) has
signed a similarly lucrative contract with two networks (NBC and Turner), receiving $2.4
billion for four years (Farhi, 1997). This does not include revenues received from lesser
contracts signed with local networks for regional sports coverage. Given the dependence
of major sports leagues on television for financial success, popular sports are influenced
greatly by television networks and orchestrated for television viewing even more so than
for live spectating. This coverage may cause the television audience to believe that they
are watching the real game, not a media representation of the game (Temple, 1992;
Coakley, 1994; Kinkema & Harris, 1992; Nixon, 1994; Harris, 1994b). In other words,
viewers may be oblivious to the potential influence of networks to shape and influence
the sporting events regularly seen on television. Additionally, the exorbitant prices of
tickets to professional sporting contests (an average NBA ticket cost $36.32 in 1998; an
average NFL ticket cost $40.93 in 1998) has made live attendance of such professional
sporting contests virtually impossible for most Americans (Fehr, 1997; Heath &
Solomon, 1997). Therefore, professional sports, an American institution rivaling religion
in popularity and societal influence (Prebish, 1985; Novak. 1976; Sage, 1993), has

7
become a phenomenon Americans see almost exclusively on television (Temple, 1992;
Chandler, 1985). Coakley (1994) explained,
the emphasis comes in a form that makes it seem as if the real game is
the one on television, not the one seen in the stadium. In other words,
television has constructed sports and spectator experiences in important
ways. Its happened so smoothly that most people think that when they
watch a game on television they are experiencing sports in a natural
form! (p. 332)
What most Americans see and know about professional sports, they see and know from
television. Considering the potential extent to which television can impact society, sports
media should receive considerable attention from academics.
Television Stars and Sports Heroes
A focal point of televised sports and the focal component for this research project
is the star athletic performer. Through the process of mediation using narration,
visuals, intimacy with particular athletes, and constant fixture on star performers
television has placed emphasis on the best individual performers and their greatness
(Duncan & Brummett, 1987; Weisman, 1993; Katz, 1993; Harris, 1994b; Kinkema&
Harris, 1992; Hargreaves, 1986; Figler & Whitaker, 1995; Shugar, 1988). Harris
(1994b) wrote,
often the star mentality of television and the entertainment business in
general affects the presentation of sports in the media. The superstars are
exalted and glorified far more than, for example, a show of fine teamwork
on the field. Stories about sports tend to focus on individual performers,
much like stories about religion focus on the Pope or Jerry Falwell or
stories about government tend to focus on the president, (p. 129)
This emphasis on the star has been accomplished not only through the actual
coverage of regular sporting events, but also through other sports media, including sports
highlights programs, pre- and post-game programs, and entertainment-based programs
that highlight individual athletes over team performance (Kinkema & Harris, 1992;
Figler & Whitaker, 1995; Harris, 1994b). Networks have promoted individuals in their

8
ongoing effort to increase the popularity of their programming, and the respective sports
leagues hope to prosper from this coverage through the potential sale of merchandise
from these star performers and overall increased interest in their sports product. This
emphasis on the individual has created a society in which television sports viewers can
learn more about popular athletes than in prior times, and the star sports figure has
become one of the most prominent and well-known persons in American society
(McPherson, 1989; Nixon, 1984; Harris, 1994a; Harris, 1994b; Coakley, 1994;
Duncan & Brummett, 1987; Kinkema & Harris, 1992).
This focus on individual star athletes in sports television has promoted the
growing societal interest in the sports hero. As previously mentioned, the sports hero has
been a part of American society for years before the existence of television, but television
has contributed to changes in this construct (Porter, 1983; Smith, 1973; Voigt, 1978).
Before the growth of sports television, sports heroes such as baseball legend Babe Ruth
were accessible local figures, and many fans had the opportunity to meet their sports
heroes in person (Walden, 1986; Katz, 1993). However, the lack of heavy coverage and
scrutiny from mass media kept fans from knowing all the details, both good and bad, of
the lives of their sports heroes. Due to televisions current ubiquitous presence, while the
sports fan today will not personally know his or her sports heroes or even have the
opportunity to watch them play in person, these fans likely will know much more about
their heroes (Walden, 1986; Nixon, 1984; Harris, 1994a; Harris, 1994b; McPherson et
al., 1989; Malone, 1993; Paterno, 1994). Television has allowed greater access to sports
heroes than ever before, as Caughley (1984) explained: Even an average fan knows
more than a baseball players name. He also knows his team, uniform number, position,
batting average, salary, and something about his appearance, personality, medical history,
and off-the-field conduct (p. 32).
Despite the increasingly visible information about the negative actions of popular
sports stars, including rape, drug abuse, and a variety of criminal activities, which has

9
lead some theorists to question whether the sports hero still truly exists in American
society (Rollin, 1983; Crepeau, 1985; Walden, 1986; Smith, 1973; Fairlie, 1978;
Nixon, 1984; Harris, 1994a; Farrey, 1997; Ryan, 1995; Coakley, 1994; Nack&
Munson, 1995; Goldberg, 1995), sports figures have frequently been chosen as heroes in
surveys of America teens (Heroes of, 1990; Simons, 1997; Harris, 1994a; Harris,
1994b). Such survey results refute the idea of a recent extinction of the American sports
hero and instead infer that the hero exists as a constantly evolving construct (Fishwick,
1983; Fishwick, 1985; Rollins, 1983; Harris, 1994a). In other words, the qualities once
considered essential for the sports hero may be very different than those of the modem
sports hero.
Sports fans now seem willing to overlook negative, or perhaps unheroic, activity
in certain areas of the lives of their sports heroes. This is likely due to the impressive
physical skills of the popular sports figure. A star athlete, while performing spectacular,
seemingly impossible feats on the playing field or in the arena, transcends typical mortal
athletic potential and appears, if only for that brief period of time, almost god-like
(Nixon, 1984; Oriard, 1982; Goodman, 1993; Walden, 1986; Birrell & Cole, 1981;
Loy & Hesketh, 1984). Sports may be one of the few remaining areas of American
society in which one consistently can find greatness, and fans may be willing to overlook
negative activity to embrace this greatness in their sports heroes. This, coupled with the
astonishing growth of sports television, has made sports a very popular and frequent area
from which American society finds its heroes. As Joan Ryan wrote (1995), Every
culture has its gods, and ours hit baseballs, make baskets, and score touchdowns (p. 85).
Regardless of the views one takes of the heroic condition in American society, the fact
remains that the popular sports figure has continued to grow as a popular choice of hero
for American society, and television has contributed greatly to this growth.

10
Societal Functions of Sports Heroes
A more clouded but perhaps more important topic is the popular belief that
athletes inherently serve as role models, so much so that the terms hero and role
model are often used interchangeably (Role models, 1989; Malone, 1993; Harris,
1994b; Simons, 1997; Wilson & Sparks, 1996; Parker & Lord, 1993; Taylor, 1991;
Michael Jordans, 1991; Goodman, 1993). In a column authored for Sports Illustrated
magazine, NBA basketball star Karl Malone (1993) wrote, We dont choose to be role
models, we are chosen. Our only choice is whether to be a good role model or a bad one
(p. 84). A lesser held, contrary opinion is that while popular mass media figures such as
sports heroes may be idolized, they often do not serve as role models because they are not
perceived as similar enough to fans of the hero (Groller, 1995; Goodman, 1993). In
other words, while fans may enjoy watching their sports heroes, these fans cannot
actually view these heroes as role models due to the vast perceived differences between
themselves and these athletes. Yet another suggestion is that sports heroes can serve as
role models to certain people for certain areas of their life (Simons, 1997; Wilson &
Sparks, 1996; Harris, 1994a; Harris, 1994b; Dononhew, Helm, & Haas, 1989; Nixon,
1984). Research about sports heroes should address whether the sports hero is a role
model and the specific conditions under which a sports hero is most likely to serve as a
role model.
Authors have addressed the potential for mass media figures to serve as role
models. Caughley (1984) explained the potential relationship between media figures and
their fans, writing,
... relationships are those in which the media figure becomes the object
of intense admiration. Such relationships approximate the general
stereotypic conceptualization of the fan, but much more is involved than
esthetic appreciation. Characteristically, the admired figure comes to
represent some combination of idol, hero, alter ego, mentor, and role
model, (p. 53)
i

11
As popular sports heroes have become mass mediated figures, they too are subject
to the same types of relationships with their fans. Many athletes have become aware of
their potential influence on the viewing public. Weisman (1993) wrote on this topic,
athletes behavior, whether it involves drug use, the shoes they wear or
their views on society, will be emulated. They are taught, very early in
their careers, to be the best example of American manhood, or at the very
least to escape with all their defects undetected, (p. 166)
In order for such a definitive statement to be made about the influence of televised star
athletes, detailed research should address the specific impact of these mass mediated
figures. Critical insight and research in this area would help society to further understand
the impact of the positive and negative actions of sports heroes.
Self-Concept Theory
Importance of Self-Concept
One theoretical framework used to study motivations for human behavior is
self-concept theory (Grubb & Grathwohl, 1967; Ross, 1971; Dolich, 1969; Hattie,
1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon & Bitta, 1979; Krechetal.,
1962). Humans typically develop their wants and goals around maintaining and
improving their self-concept, which can be defined as myself as I see myself (Kretch et
al., 1962; Loundon & Bitta, 1979; Sirgy, 1983; Hattie, 1992). This desire to maintain
and improve self-concept is a strong predictor for purchase intentions (Grubb &
Grathwohl, 1967; Ross, 1971; Dolich, 1969; Hattie, 1992; Baughman & Welsh, 1962;
Sirgy, 1983; Loundon & Bitta, 1979; Krech et al., 1962). Self-concept theory also has
been linked to behaviors such as choosing friends and academic efforts in school (Felker,
1974; Hattie, 1992).
While most of the research on self-concept and behavior has focused on consumer
behaviors, the link between self-concept and behavior has been strongly established and

12
is a valuable tool for future investigation into human behavior. Grubb and Grathwohl
(1967) explained the importance of this topic, writing,
the self represents a totality which becomes a principal value around
which life revolves, something to be safeguarded and, if possible, to be
made still more valuable. An individuals evaluation of himself will
greatly influence his behavior, (p. 204)
Felker (1974) provided a more detailed explanation for the relationship between
self-concept and behaviors, writing,
the role of self-concept is three-fold. The self-concept operates as a
mechanism for maintaining inner consistency; the self-concept
determines how experiences are interpreted; and the self-concept
provides a set of expectancies. Each of these three factors is a powerful
determiner of behavior, (p. 7)
Although self-concept theory has not previously been used to examine role
modeling behaviors of sports heroes, use of this theory, which predicts that people
continuously try to maintain and improve self-concept, provides a promising new
framework to examine the motivations for such role modeling.
Role Modeling Sports Heroes and Self-Concept Theory
This research project will look at the behaviors involved in role modeling sports
heroes as related to self-concept. Similar types of studies have been done for consumer
behaviors, determining that similarity between self-concept and product image correlates
with purchase behaviors (Dolich, 1969; Grubb & Grathwohl, 1967; Ross, 1971; Hattie,
1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon & Bitta, 1979; Krechetal.,
1962). Similarly, researchers have suggested that mass media figures can serve as role
models, particularly for adolescents (French & Pina, 1991; Bandura, 1992; Harris,
1994b; Caughley, 1984). This research project incorporates these two concepts while
examining a increasingly popular, criticized, and critical aspect of American society, the
sports hero.

13
Problem Statement
Americans, and particularly American adolescents, historically have held up
popular sports figures as heroic figures. This is not surprising, as theorists have asserted
that the need for heroes is inherent in human societies (Fishwick, 1983; Walden, 1986;
Harris, 1994a; Schlesinger, 1968; Browne, 1983), and sports has become one of
Americas most celebrated pastimes and personifications of American dreams and values
(Nixon, 1984; Porter, 1983; Goodman, 1993; Oriad, 1982; Ryan, 1995; Walden, 1986;
Smith, 1973; Coakley, 1994; Crepeau, 1985; Harris, 1994a; Role models, 1989;
Michael Jordans, 1991).
Given the growing mass media interest in professional and major collegiate
sports, Americans now have more access to watch star athletes than American sports fans
of generations past (T.V. sports, 1996; McPherson, Curtis, & Loy, 1989; Coakley,
1994; Davies, 1994; Clarkin, 1990; Harris, 1994a; Harris, 1994b; Kellner, 1996;
Caughley, 1984; Hargreaves, 1986; Kinkema & Harris, 1992; Denzin, 1996).
Accordingly, the opportunity to find heroes in sports appears endless. In a society where
popular athletes have become incredibly wealthy figures about which spectators know
more than ever before, the issue of choosing a particular sports hero is a complex
phenomenon. While theorists have suggested that heroes no longer truly exist in America
(Boorstin, 1961; Klapp, 1962; Monaco, 1978; Lubin, 1968; Kelly, 1993; McLuhan,
1964), and many of todays popular athletes have displayed behaviors that certainly
appear unheroic, athletes are still popular choices as heroes among American youths
and adolescents (Heroes of, 1990; Role models, 1989; Wilson & Sparks, 1996;
Simons, 1997). Television has only aided this process, bringing heroes and hero
worshippers together more easily (Higgs, 1978). However, one is left to wonder exactly
what the impact of these sports heroes is on American adolescents.

14
Critics and fans of sports heroes alike have suggested that popular American
sports heroes serve as role models for American adolescents (Hams, 1994a; Hams,
1994b; Role models, 1989; Malone, 1993; Simons, 1997; Wilson & Sparks, 1996;
Parker & Lord, 1993; Taylor, 1991; Michael Jordans, 1991; Goodman, 1993). Little
scientific research has addressed the motivations for such role modeling or the conditions
under which such role modeling is most likely to occur for the individual adolescent.
Socioeconomic status, race, and parental and peer influences have been suggested as
potential factors affecting such role modeling (Harris, 1994a; Role models, 1989;
Simons, 1997; Wilson & Sparks, 1996; Michael Jordans, 1991; Goodman, 1993).
However, these factors do not address the individual role modeling decisions of each
adolescent and the motivations for the individual to model the behaviors of his or her own
particular sports hero. One area of research which has examined motivations for
individual behavioral intentions is the study of individual self-concept and how this
self-concept relates to the images of other objects (Dolich, 1969; Grubb & Grathwohl,
1967; Ross, 1971; Hattie, 1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon&
Bitta, 1979; Krech et al., 1962). This research on self-concept has primarily been used to
better understand consumer behaviors and individual purchase decisions. However,
self-concept research also presents a potential means to better understand a variety of
behaviors, including the emulation of sports heroes addressed in this research.
This theoretical approach to studying adolescent role modeling of sports heroes
has lead to the following research questions:
1) Among adolescent males, how do the relationships between self-concept and
perceived images of a sports hero relate to behavioral intentions to role model a
sports hero? In other words, how do the perceived similarities and differences
between ones individual self-concept and the image of a sports hero relate to
ones intentions to emulate that sports hero?

15
2) What specific role modeling behaviors are most strongly related to these
relationships between self-concept and the image of a sports hero?
3) Which areas of self-concept are most influential in predicting adolescent role
modeling of a sports hero?
Chapter 2, the Literature Review, addresses prior research related to these
questions. Specifically, each construct used in this study is defined and discussed. Prior
research on related subject matter is reviewed. Finally, Chapter 2 addresses areas of
research that have not been studied adequately as well as the ways that this research
project adds to the base of knowledge and contributes meaningful information to the
academic community.
Three additional chapters are included in this dissertation. Chapter 3, Methods,
addresses the specific methods and statistical analysis used in this dissertation. Chapter
4, Results and Discussion, details the results of the stated analysis. Finally, Chapter 5,
Conclusions, addresses the results, discusses the importance of these results, and explains
potential reasons these results were found.

CHAPTER 2
LITERATURE REVIEW
This chapter reviews relevant prior research. Additionally, important terms are
defined for use in this research. Finally, the contribution of this research project is
discussed. The topics reviewed in this chapter, in order, are as follows: role modeling,
characteristics of the sports hero, the sports hero as a role model, self-concept, and
self-concept congruity theory.
Role Models and Behavioral Modeling
Sports heroes are widely assumed to be role models, especially for children and
adolescents, although such assumptions are often made based on personal opinion and
lack concrete support (Malone, 1993; Harris, 1994a; Harris, 1994b; Wilson & Sparks,
1996; Nack & Munson, 1995; Taylor, 1991; Walden, 1986; Michael Jordans, 1991;
Simons, 1997; Role Models, 1989). One primary purpose of this research is to
determine how the relationships between self-concept and the image of sports hero
determine role modeling behaviors of this sports hero. Therefore, an adequate
understanding of the role model and its fimction and influence is necessary. This section
of the literature review defines role models, explains the process of imitating role
models, and outlines the conditions under which such modeling is most likely to occur.
Finally, issues specific to role models and adolescents are discussed.
The term role model, like self-concept, has been used vaguely in reference to a
variety of different constructs, so it is necessary to define this term specifically. A role
16

17
model has been described as someone whose life and activities influenced the
respondent in specific life decisions (Bassoon & Owe, 1980, p. 559). This description
does not address actual copying of behaviors or intentions to copy behaviors, but rather
presents the role model as one influence in such intentions. Other definitions assert that
role models are adults who are worthy of imitation in some area of their life (Pleiss &
Feldhusen, 1995, p. 163) or someone who demonstrates the appropriate behavior for a
specific role or relationship with another person (Jung, 1986, p. 528). Role models
generally exert influence over one of three areas of life: educational, occupational, or
personal (Bell, 1970; Pleiss & Feldhusen, 1995).
Many of the descriptions of the role model imply the role model is a positive
influence whose actions should be copied and emulated. In other words, people who
display negative behaviors, such as wife abuse, drug use, or poor performance in
employment or school, are not role models (Malone, 1993; Role models, 1989;
Richards, 1998; Golden, 1986). However, this research adopts the perspective that role
models can exhibit both positive and negative behaviors and characteristics. A
description written by Hendricks (1992) gives insight into this position, noting, We all
have role models; some positive, some not so positive. They serve as archetypes,
proxies we scrutinize as guides for our own actions and perhaps too as templates for
possible selves (p. 7).
Role modeling is a component of the more general term modeling. In fact, the
term model is often interchanged with role model. While role modeling certainly
involves the processes of modeling, a clear distinction will be made to allow for sufficient
definition of this term for this research. Role modeling refers to a more specific process
than the more generic process of modeling of behaviors. Specifically, modeling is a
measure of the influence of observed behavior on an observer (Bandura & Walters, 1963;
Bandura, 1977; Bandura, 1994; Jung, 1986; Muuss, 1975). This process refers to a
specific set of behaviors that are not necessarily tied to a specific role, and this process

18
can be replicated in laboratory settings. For example, learning the technical processes
involved in juggling from a model in a laboratory setting would be an example of
modeling. The person demonstrating the juggling behaviors would be the model.
However, role modeling, while it does involve the technical processes of modeling, also
involves an emotional attachment between the observer and role model. Jung (1986)
explained distinct differences in the two constructs: Modeling does not imply that there
is any emotional tie between the model and observer. In contrast, many instances of role
modeling ... seem to require some affectional bond for the model to be effective in
influencing the observer (p. 527). Role modeling involves a relationship between role
model and observer, and such relationships can influence behavior through learned roles
in specific situations (Hendricks, 1992; Brim, 1968; Bell, 1970; Jung, 1986; Miller,
1983). Therefore, the process of role modeling involves adopting and modeling
behaviors based on an emotional tie between observer and model, and this process of
modeling involves context and relationships with the role model. In order for role
modeling to occur, the role model and the observer must interact and the observer must
identity the role model as such (Bell, 1970; Basow & Howe, 1980).
One popular view of role model influence is that everyone has role models, and
these figures serve as guides for our own actions and possible templates for our own
selves (Hendricks, 1992; Gecas & Mortrimer, 1987). Such influence can be behavioral
and attitudinal. However, behavioral influence has been studied and discussed more
frequently than attitudinal influence by researchers and theorists (Basow & Howe, 1980;
Almquist & Angrist, 1971; Hendricks, 1992; Jung, 1986), and these influences are the
focal point of this research project.
Observers are not limited to one role model, and it is quite normal for observers to
have several role models for different aspects of their lives (Jung, 1986). Additionally,
some role models influence more aspects of observers lives than other role models
(Gecas & Mortimer, 1987; Basow & Howe, 1980). Hendricks (1992) explained this,

19
writing, Some models are relevant in specific realms only; that of the older, wiser
colleague at work, or the sports figure we seek to emulate. Others cut across specific
roles and have a more profound influence on our lives (p. 7). Most descriptions of role
modeling refer to the process of copying the behaviors of the role model, although they
also can, in some circumstances, refer to avoiding behaviors of the role model and
actively seeking alternative behavioral techniques to those displayed by the role model
(Pleiss & Feldhusen, 1995; Jung, 1986). While including such avoidance techniques
broadens the definition of the role model, this research specifically examines the intention
to model the behavior of a role model, not the intention to avoid certain behaviors.
Many available sources of role models exist. Persons who have frequent
interaction with and are well known by the observer, including parents, teachers, peers,
co-workers, and siblings, typically are popular role models for adolescents. (Schaefer,
1978; Muuss, 1975; Basow & Howe, 1980; Malgady et al., 1990; Hackett et al., 1989;
Almquist & Angrist, 1971; Betz & Fitzgerald, 1987). In addition to personal
acquaintances, mass media figures also can serve as role models (Caughley, 1985;
Schaefer, 1978; Role models, 1989; Jung, 1986; Pleiss & Feldhusen, 1995; Duck,
1990). In contrast to personally known role models such as parents and peers, mass
media figures, including famous athletes, actors, and politicians, have the potential to
serve as role models for large numbers of people due to their mass exposure.
Researchers and critics have taken varying positions on the potential influence of
mass-mediated figures such as sports hero. One popular belief is that popular,
mass-mediated figures, particularly those portrayed as positive or heroic in their
behaviors, inherently serve as role models due to their stature and position in society. In
fact, the terms public hero and role model are often used interchangeably without
regard to a distinction between the two (Pleiss & Feldhusen, 1995, p. 163; Malgady et
al., 1990; Hunter, 1983). Conversely, some have concluded that a public hero cannot
truly be a role model because there is almost no potential for interaction between the role

20
model and the observer (Singer, 1991; Hall & Post-Kammer, 1987). Both of these
contradictory opinions are too inflexible to fully summarize the potential influence of the
mass media figure, and the true influence of this figure lies somewhere inbetween these
viewpoints. Research has indicated that mass-mediated figures, particularly those who
are considered heroes, can serve as role models (Pleiss & Feldhusen, 1995; Role
models, 1989; Schaefer, 1978; Bandura, 1981; Michael Jordans, 1991; Malone,
1993; Kellner, 1996; Weisman, 1993; Wilson & Sparks, 1996). Parker and Lord (1993)
have echoed this sentiment in research with African-American males, noting that highly
visible and successful individuals (such as sports personalities or entertainers) make
excellent role models for young Black males (p. 102). Therefore, while popular media
figures are not inherently role models, they certainly have the potential to be role models.
Media users often develop intense imaginary relationships with their favorite
mass-media figures, and these relationships contribute to the role modeling potential of
such figures (Caughley, 1985; McEvoy & Erikson, 1981). McEvoy and Erikson (1981)
called these public heroes reference idols. Reference idols, although not personally
known by the observer, become part of the reference group of the observer, a
personification of values and behaviors perceived as important to the individual (p. 114).
The authors explained the potential influence of these mass media heroes, writing,
Reference idols, real or fictitious, can affect the formation of attitudes, beliefs, identities,
and behaviors (p. 114). In other words, mass media heroes provide example attitudes,
beliefs, identities, and behaviors that individuals use in creating, modifying, and/or
maintaining/reinforcing their self-concept. It is these potential influences of mass media
figures that will be the focus of research for this dissertation.
In summary, it is necessary to define the term role model for use in this
dissertation. For this study, the role model is a mass mediated figure (in this instance, a
sports figure). The processes of role modeling involve modeling of specific behaviors of
an admired figure. Finally, an affectionate bond must exist between the observer and the

21
role model, unlike laboratory models. Therefore, role modeling will occur as part of
daily life, making the study of role modeling one of naturalistic inquiry, not laboratory
observation.
Conditions of Role Modeling
Many researchers have studied the conditions or factors which make observers
more likely to adopt the behaviors of a role model. These conditions are reviewed in this
section. One primary factor is perceived personal similarity between observer and role
model (Hackett et al., 1989; Jung, 1986; Decker & Nathan, 1985; Owens & Ascione,
1991; Wilson & Sparks, 1996; Muuss, 1975; Bandura, 1994; Role models, 1989;
Malgady et al., 1990; Bumstein et al., 1961; Stotland & Patchen, 1961; Stotland et al.,
1962; Rosekrans, 1967). People are more likely to adopt the behaviors of a role model
whom they perceive as having similar attributes to themselves. Specific attributes
include sex, race, and cultural identification. Wilson and Sparks (1996) provided
examples of this through interviews with African-American teens in reference to their
sports heroes. They specifically studied African-American teenagers and their role
modeling of African-American basketball stars. They wrote, in reference to potential
modeling of clothing styles, Many of the Black adolescents gave responses
acknowledging the celebrity Black athletes definitive influence on popular style through
athletic apparel and the athletic-apparel commercial. This influence was manifested most
often in the respondents desire to look like the celebrity athlete (p. 414). These
adolescents were more likely to use African-American athletes as their role models due to
perceptions of racial similarity, cultural identity, and masculine identity. Therefore, the
perceived similarities between popular male, African-American athletes and male,
African-American adolescent fans contributed greatly to this selection of role models.
Perceived self-efficacy of the observer also can affect the choice of a role model
(Simons, 1997; Weisman, 1991; Role models, 1989; Bandura, 1994). Self-efficacy
refers to the a belief that one can successfully model an action (Bandura, 1994). As the

22
individual becomes more confident in his or her ability to perform a behavior, he or she
will be more likely to model such behaviors.
Another factor influencing modeling is the perceived benefits the role model gains
from his or her behaviors (Ostlund, 1974; Midgley, 1976). Role models who are
rewarded for their actions are more likely to be imitated than those who are not rewarded.
This has lead to the belief that popular, mass-mediated role models are often very
appealing to observers because they are well compensated with fame and wealth (Simons,
1997; Wilson & Sparks, 1996). Viewers, therefore, will attempt to role model those
behaviors perceived as instrumental in achieving such fame and wealth. Simons (1997)
explained this appeal of wealthy, mass-mediated figures as potential role models to
certain groups of adolescents, using basketball star Michael Jordan as an example.
Simons wrote,
a lot of them (low-income, African-American adolescents) will tell you
they want to be like Mike.. In this context, being like Mike doesnt mean
becoming an entrepreneur, a corporate spokesman, or a college graduate.
It means being a high-flying, windmill slammer of a ballplayer, (p. 50)
In other words, the most glamorous aspects of Michael Jordan (athletic fame, financial
riches from athletic accomplishment) made Jordan appealing as a role model to these
adolescents.
Self-Concept as a Condition for Role Modeling
Most research on the impact of role models, and specifically athletes as role
models, has focused on the characteristics of the potential role model; the societal role of
the role model; and easily observable characteristics of the observer, such as race,
socio-economic status, and perceived positive or negative value to society (Simons, 1997;
Malone, 1993; Role models, 1989; Wilson & Sparks, 1996; Chen, 1997; Golden,
1996; Walden, 1986; Pleiss & Feldhusen, 1995). This dissertation addresses the
relationship between role model and observer from a slightly different perspective.
Instead of measuring the aforementioned characteristics, this research examines

23
individual perceptions of the self, or self-concept, and the influence self-concept has in
determining role modeling behaviors. Of primary importance for this research are the
relationships between self-concept and the image one has of his role model (for this
research, the sports hero). In other words, this research attempts to better understand
whether ones behavioral intentions to model the behaviors of a role model are influenced
by how similar or dissimilar ones self-concept is to his or her perceptions of his or her
sports hero. This approach of examining self-concept has not been widely used to
investigate behavioral intentions of role modeling. However, this research approach has
been used quite frequently in marketing research, and this dissertation will borrow
investigative strategies from such marketing studies. (See later section on self-concept
for a detailed description of this construct and a review of research and theory in this
area.)
Marketing researchers have provided a foundation for the use of self-concept in
studies of behaviors and behavioral intentions (specifically, purchase intentions). First,
they have established that self-concept maintenance and enhancement is a factor in
determining behavior (Sirgy, 1983; Landon, 1974; Birdwell, 1968; Dolich, 1969;
Grubb & Grathwohl, 1967; Ross, 1971; Delozier & Tillman, 1972; Loundon & Bitta,
1979; Krech et al., 1962; Wells & Marwell, 1976; Baughman & Welsh, 1962; Felker,
1974; Frey & Carlock, 1989). Several studies have established that consumers choose to
purchase certain products based on the similarity between their reported self-concept and
their perceived image of a product (Dolich, 1969; Ross, 1971; Delozier & Tillman,
1972; Grubb & Grathwohl, 1967; Hattie, 1992; Baughman & Welsh, 1962). These
studies have shown that individual perceptions of self (self-concept) can influence
behavioral intentions when viewed in reference to the individual perceptions of a
consumer product. Therefore, each individual has perceptions that influence his or her
behaviors, and these perceptions may be completely different than the perceptions of

24
another observer. This is the most important justification for the use of self-concept to
examine behavioral intentions for this dissertation.
In the aforementioned consumer studies, subjects rated their perceptions of
consumer products, and these ratings were compared to the individual self-concepts of
these subjects. From these comparisons, researchers examined the role of self-concept in
influencing consumer behaviors toward selected products. However, this dissertation
looks at perceptions and influence of sports heroes, not consumer products. Therefore, it
is necessary to review the characteristics of popular, mass-mediated sports heroes and
their potential to serve as role models for adolescents. This is the focus of the next
section.
The Sports Hero as a Role Model Figure
The Hero
The term hero is a heavily cited yet frequently changing construct in American
society. However, the hero has been a vital part of societies throughout history, and a
brief review of this evolution will assist in better understanding the qualities of the
modem hero and, more specifically, the modem sports hero.
The term hero dates back to ancient Greece and originally was defined as a
superior man, embodiment of composite ideals (Fishwick, 1979, p. 340). Qualities that
the Greeks associated with the hero included an intimacy with gods; superiority to other
humans; and an abundance of the best of human characteristics, such as strength and
courage (Bowra, 1969; Brombert, 1969; Whitman, 1969; Fishwick, 1983). The ancient
hero was a pre-destined character, and he was inherently greater than the normal man
(Macmechan, 1840; Lehman, 1928). Another historical name for the heroic figure is the
Great Man (Hook, 1957; Wecter, 1941).

25
According to Carlyles Theory of the Hero, the hero is one who is destined for
greatness from a divine source (Carlyle, 1841; Lehman, 1928; Smith, 1973; Wecter,
1941; Fishwick, 1985). Following this theoretical perspective, heroes were predestined
by God as leaders to move the world forward (Carlyle, 1941; Lehman, 1928; Campbell,
1949; Wecter, 1941). Examples of Great Men in history include Moses, Frederick the
Great, and Napoleon.
The evolution of the hero has been quite pronounced in the 20th century,
particularly in a relatively youthful American society. The youth of this society allowed
Americans to define the characteristics of their own heroic types (Hook, 1957; Lubin,
1968; Wecter, 1941; Lemer, 1957). Theorists of this period asserted the hero was not
pre-determined, but instead was a product of his society and those who followed him
(Hook, 1957; Klapp, 1962). Additionally, an increasingly complex society has allowed
for a variety of different heroic types and characteristics. In other words, 20th century
America experienced a fragmentation of heroic types, allowing for greater heroic
diversity and more heroic choices than in ancient Greek society.
Instead of possessing God-like qualities, the American hero of the mid-20th
century had a closer relationship with the common man (Wecter, 1941; Hook, 1957).
This presented a strong contrast with European Great Men of that time period, including
Hitler and Mussolini. Americans, increasingly fearful of such powerful and destructive
heroic figures, did not embrace Great Men as their heroic choices. They instead chose
more familiar, less intimidating heroic types. (Hook, 1957; Fishwick, 1985).
The increasing influence of mass media in American society in the late 20th
century further changed the dynamic of the American hero. Theorists of the 1960s and
1970s described a new type of figure, the celebrity, which had become prominent in
American society (Boorstin, 1961; Klapp, 1962; Lubin, 1968; Bowra, 1969; Monaco,
1978; Fishwick, 1979; Browne, 1983; Rollin, 1983; Hoagland, 1974; Harris, 1994).
The celebrity, first described by Boorstin (1961), does not gain fame through greatness or

26
accomplishment, but instead due to mass exposure and visibility. Boorstin provided the
following description of the American celebrity as:
a person who is known for his well-knownness. His qualities or lack of
qualities illustrate our particular problems. He is neither good nor bad,
great nor petty. He is the human pseudo-event. He has been fabricated on
purpose to satisfy our exaggerated expectations of human greatness.
(1961, pp.57-58)
Unlike heroes, celebrities do not display a model of greatness. They are a product of the
media as much as they are a product of their accomplishments. Celebrities do not
consistently display excellence, and they exhibit ordinary behaviors and accomplishments
(Klapp, 1962; Kelly, 1993; Monaco, 1978).
Therefore, many theorists of the 1960s and 1970s believed, largely due to
Americas increased reliance on mass media and television, that the celebrity had
replaced the hero in American society, and the heroic figure no longer existed in America
(Schlesinger, 1968; Monaco, 1978; Hoagland, 1974; Kelly, 1993; Harris, 1994; How
to, 1995). This cynical view blamed television for initiating the extinction of the heroic
figure. This viewpoint also expressed a static view of the hero. Accordingly, this heroic
figure did evolve over time and was eventually replaced by the celebrity. However, many
theorists of the 1980s and 1990s have refuted these arguments, writing instead about the
ongoing evolution of the hero into what can be called the modem American hero
(Fishwick, 1979; Fishwick, 1983; Fishwick, 1985; Browne, 1983; Walden, 1986;
Harris, 1994). These authors argued that heroes have not disappeared, but instead heroic
qualities have changed with the growing influence of mass media. In particular, because
mass media have made heroes more available to the public, Americans are given a larger
number of heroic choices, and these heroes cannot maintain the longevity of heroes of
past generations.
Mass media exposure has become a critical component of modem heroic stature,
and therefore the modem hero does share characteristics with the less substantial

27
celebrity. Rollin (1983) clarified the influence of mass media on the modem hero,
writing, Publicizing the heroic act and celebrating the performer, therefore, are crucial..
. All heroes then are celebrities; no hero is not a celebrity (p. 14). Unlike the traditional
hero, the modem hero must receive publicity by mass media in order to gain heroic
consideration. Publicity is crucial in acceptance as a modem hero, regardless of the
characteristics of the individual. However, unlike the celebrity figure, publicity in itself
is not enough to create the modem hero, and this figure must also exhibit certain heroic
qualities. This dissertation takes the perspective of these modem theorists, asserting that
American heroes still do exist and that mass media have sped the evolutionary process of
these heroic figures.
While the modem American hero can take several different forms, theorists have
attempted to define some parameters of this modem heroic figure. These include
attractiveness, winning, style, risk-taking behavior, individualism, and skill in a particular
field (Browne, 1983; Rollin, 1983; Fishwick, 1985; Crepeau, 1985; Harris, 1994;
How to, 1995). Americans have become less likely to require moral greatness from
their heroes than was required by previous generations. Additionally, due to an
increasingly fragmented American society and an increasing number of heroic choices
available through television, Americans of today are less likely than Americans of former
generations to unite behind one particular hero (Walden, 1986; Fishwick, 1983; Rollin,
1983; Crepeau, 1985).
This research deals with one specific type of hero, the sports hero, and this is the
focus of the following section.
The Sports Hero
Sports has become a popular and vital area in which Americans now find their
heroes, and this trend has grown through the 20th century (Heroes of, 1990; Ryan,
1995; Nixon, 1984; Oriard, 1982; Simons, 1997; Role models, 1989). One potential
reason for this is that sports remains one area where true greatness and superior beauty

28
still can be found in a complex society (Goodman, 1993). Regardless of the negative
perceptions Americans may have of politicians, actors, and other frequently tarnished
public figures, a star athlete, unlike other mass mediated figures, will have rare moments
during which they appear to surpass typical mortal limitations through spectacular,
seemingly impossible athletic feats. He will appear, if for only that brief period of time,
almost god-like (Nixon, 1984; Oriard, 1982).
While the sports hero has been part of American culture since well before the turn
of the 20th century, a fantastic growth and emphasis on the American sports hero
occurred beginning in the 1960s with the rapid growth of sports television (McPherson,
1989; Davies, 1994; Harris, 1994; Nixon, 1984). By the mid-1980s, sports
programming accounted for 15 percent of total programming on commercial television,
including networks, such as ESPN, devoted solely to sports programming (Harris, 1994b;
Davies, 1994; Nixon, 1984; Katz, 1996). This growth in sports media allowed for
increased exposure and popularity of the sports hero to the general public. In addition to
publicizing the modem sports hero, the current focus of the content of sports television
has contributed also to the evolution of the American sports hero. To increase fan interest
and provide more exciting television material, American sports programs are presented to
emphasize heroic actions of the individual and individual emotions and personalities of
the star athletes (Kinkema & Harris, 1992; Coakley, 1994; Hargreaves, 1986; Hilliard,
1984; Sabo & Jensen, 1992). This focus on the individual has helped to create a strong
and unique relationship between the viewer and the individual star athletes. Duncan and
Brummett (1987) explained medias ability to create such relationships, writing,
during sporting events, television tends to focus more on the individual
than the group, thus establishing a kind of intimacy between the audience
and the player. This intimacy is created by close-ups of particular athletes
before, during, and after the competition and shots capturing the athletes
expressions and reactions at critical moments, by the announcers creation
of imaginary conversations, but their use of the players first names, and
by personal interviews with the athletes or coaches. Television, therefore

29
turns spectator sports into observance of intimates, albeit a controlled
intimacy within the confines of a 19-inch screen, (p. 171)
The individual has become so paramount in televised sports that television ratings
often hinge upon whether a star athlete, such as basketball star Michael Jordan, plays in a
particular contest (Fatsis, 1998; Shapiro, 1999; Starr, 1999). Professional basketball
(NBA) in particular has employed a marketing strategy of promoting star players ahead
of the actual game (Starr, 1999). For the 1997-98 basketball season, NBCs ratings for
nationally televised Chicago Bulls games, featuring Michael Jordan, were 75% higher
than for the networks other nationally televised NBA contests (Fatsis, 1998). With
Jordans recent retirement from basketball, both league and network executives have
openly worried over the potential loss of league popularity and ratings of televised games.
Peter Roby, vice president of marketing for Reebok Athletics, asserted that the league
would never again reach the status attained through Jordans star appeal, saying, I dont
think the NBA will ever again attain the glory it did with Michael (Starr, 1999, p. 56).
To combat this potential loss, networks have embraced the opportunity to create
and publicize new athletic stars. NBC Sports chairman Dick Eberson said in reference to
the potential for decreased ratings upon Jordans retirement, We do have an opportunity
to reintroduce a whole new generation of stars (Shapiro, 1999, p. D5). In other words,
the individual star remains the prime emphasis for television sports, and the retirement of
one star leads to the creation of other stars. Nike corporation CEO Phil Knight, whose
company airs numerous commercials featuring star athletes, has stated that even his
television advertising campaigns alone are powerful enough to create popular sports
figures. For instance, in reference to then-NBA newcomer Alonzo Mourning, Knight
said, All wed have to do is take the money were using to advertise Charles Barkley
and Michael Jordan, put it into Alonzo, and a lot more people would be talking about
him (Katz, 1994, p. 253). With the excessive television coverage of sports stars and
the media influence on actually creating star appeal of accomplished athletes, the

30
distinction between sports hero and sports celebrity appears to be a confusing and blurred
line. However, this research follows the perspective that mass media, specifically
television, are now necessary to publicize sports heroes. Additionally, the fact that sports
figures have primarily become mass-mediated figures does not mean all popular athletes
must be celebrities instead of heroes, but instead reflects the evolution of the modem
sports hero.
Due to the growth of media coverage of sports figures, Americans now know
more about popular sports figures than ever before, including both their on-the-field and
off-the-field activities. As Caughley (1984) wrote, Even an average fan knows more
than a baseball players name. He also knows his team, uniform number, position,
batting average, salary, and something about his appearance, personality, medical history,
and off-the-field conduct (p. 32). The popular sports hero has been demystified, and
fans now see the imperfections as well as the greatness (Hargreaves, 1986; Harris,
1994a; Hoagland, 1974; Coakley, 1994; McPherson et al., 1989). Such imperfections
include wife/girlffiend abuse, drug and alcohol use, sports gambling, fighting, and
fathering of illegitimate children (Harris, 1994; Paterno, 1992; Johnson, 1995; Messner
& Solomon, 1994; Long, 1991; Michael Jordans, 1991; Nack & Munson, 1995;
Starr & Samuels, 1997; Wilson & Sparks, 1996; Farhi, 1997; Franks, 1996; Howard,
1995). These imperfections must be considered in addressing the potential influence of
popular sports figures, particularly their potential influence as role models for
adolescents.
Despite these imperfections, many theorists still believe the modem,
mass-mediated sports figure can be a hero, and they have identified several characteristics
that are commonly associated with this modem sports hero. These qualities include the
following: supreme athleticism on the field or court, high winning percentages and the
potential to win important games and championship matches, statistical records and
greatness throughout a career, flair and charisma in style of play and personality, and

31
confidence in ones abilities (Nixon, 1984; Harris, 1994a; Harris, 1994b; Crepeau,
1985; Goodman, 1993; Smith, 1973; Porter, 1983; Starr & Samuels, 1997). According
to surveys of American teenagers, particular sports-related behaviors, such as
sportsmanship and fair play, are also important to some sports fans (Harris, 1994a).
While not necessarily related to the athletic performance, financial success and lucrative
commercial endorsement deals are commonly identified qualities of the sports hero,
particularly to adolescent boys who aspire to reach similar financial heights through
professional athletics (Katz, 1994; Michael Jordans, 1991; Weisman, 1993; Wilson
& Sparks, 1996; Simons, 1997). Finally, theorists also have identified several
non-performance-related characteristics, including civic and community involvement,
academic accomplishment, strong family ties, and avoiding illegal and immoral
behaviors, as components of the modem heroic qualification (Walden, 1986; Smith,
1973; Harris, 1994a; Harris, 1994b; Hoagland, 1974; Nixon, 1984; Coakley, 1994).
While the off-the-court actions of the sports star may have some impact on heroic
classification, on-the-court excellence has been identified as more instrumental in
determining heroic status. Nixon explained this, writing, Wayward athletes may be
excused by fans... in their lifestyle off the field as long as they work hard and produce
on the field and... their behavior on or off the field does not depart too much from
conventional standards (1984, p. 174).
This dissertation accepts the perspective that the sports hero is a non-static,
evolving construct. Additionally, each individual may have his or her own views of the
characteristics that are most important in being a sports hero. For this research, however,
the sports hero will be defined as adhering to the following guidelines:
1) The sports hero must be well known. Therefore, he must be shown
prominently on television. This perspective acknowledges the impact of celebrity
on the heroic condition in America.

32
2) The sports hero must display athletic excellence on the field or court. This
can be displayed through record or notable performances (scoring, defense, etc.),
flamboyant or outstanding style of play, or other feats that set an athlete apart
from his or her peers on the field or court.
3) The sports hero must display the ability to be a winner on the field or court.
4) The sports hero may display the ability to overcome difficult odds against
success in sports, such as injury, physical limitations, or underprivileged family
life as a youth. The sports hero may also display extreme perseverance in
reaching athletic goals.
5) The sports hero may display exemplary behavior on and off the court,
although this is less important than exemplary performance on the court.
Therefore, it is possible to have a sports hero who is quite flawed from a
behavioral perspective, as long as he or she performs well in athletics.
6) The sports hero will likely be extremely wealthy through high salaries
and sports marketing and endorsement deals.
7) While all the previously listed qualities are important in the definition of the
sports hero, this research also will accept the idea that each individual may have
his or her own priorities as to the most important characteristics of the American
sports hero.
The two concepts, the role model and the sports hero, described in the previous two
sections have been synthesized in the following section, which describes the most critical
concept for this research project, the sports hero as a role model.
The Sports Hero as a Role Model
The popular sports figure is both praised and criticized as a potential role model,
although limited research addresses these viewpoints. The function of star athletes as role
models is often taken for granted and assumed to be true despite a lack of specific

33
evidence (Jung, 1986). The limited research that has been dQne on this topic is addressed
below.
The popular sports figure can serve as a role model in several different areas of
adolescent life. Some of these roles are positive, while others are negative. One
identified area of role modeling is style of play and thletic performance. Observers may
pattern aspects of their athletic participation after the athletic accomplishments of their
sports role models (Wilson & Sparks, 1996; Kellner, 1996; Simons, 1997; Harris,
1994b). For example, golfers reported a noticeable slowing of play at their local courses
when golf tournaments were first televised (Harris, 1994b). This was because amateurs
began lining up their putts as the pros did on television, despite the fact that many of
them had no idea why this was done or the effectiveness of this procedure.
Similarly, observers model the athletic moves of NBA basketball stars (Harris,
1994; Wilson & Sparks, 1996; Simons, 1997; Nixon, 1984; Michael Jordans, 1991).
Observers may attempt to adopt aspects of their athletic role models play style, whether
it be the flamboyant dribbling style of Julius Erving or the physical play of Karl Malone.
The influence of professional athletes often extends from one generation of athlete to the
next, as many current stars list former athletic stars as their role models in developing
playing styles (Nixon, 1984; Wilson & Sparks, 1996).
Non-sports-related on-court behaviors also can be modeled by observers. These
behaviors can be positive, such as exhibiting grace during defeat or helping players from
other teams. They can also be negative, including throwing temper tantrums during
competition, fighting with other players and officials, and abusing equipment (Weisman,
1993; Wilson & Sparks, 1996; Starr & Samuels, 1997; Harris, 1994). Several popular
athletes, including Charles Barkley, Joe Namath, and Deon Sanders, have become well
known for their brash and arrogant behavior, and adolescent observers may copy such
behaviors (Hams, 1994b). Athletic role models also can display positive behaviors

34
during play, and observers can imitate these behaviors as well (Coakley, 1994; Role
models, 1989; Goodman, 1993; Harris, 1994b).
Theorists have suggested that athletes can serve as positive role models when
observers imitate positive work behaviors of successful sports heroes. These behaviors
include hard work toward reaching a goal, practice, desire to win, and prosocial behaviors
such as charitable work and sportsmanship (Kellner, 1996; Wilson & Sparks, 1996;
Coakley, 1994; Malone, 1993; Weisman, 1993; Michael Jordans, 1991; Nixon,
1984; Role models, 1989; Harris, 1994a).
One prominent potential area of role modeling of popular mass-mediated athletes
involves clothing style and athletic footwear, including athlete-endorsed shoes costing in
excess of $ 100 per pair. While observers may not be able to perform the athletic feats of
their role models, they can wear or hope to wear the clothing endorsed by role models
(Weisman, 1993; Katz, 1994; Michael Jordans, 1991; Kellner, 1996; Simons,
1997). Mass media advertising campaigns have contributed to the marketing power of
the athletic role model, with companies telling the consumer that shoes are a key
component to being more like their athletic hero, whether it be Michael Jordan or Bo
Jackson (Weisman, 1993; Wilson & Sparks, 1996; Simons, 1997). As Wilson and
Sparks (1996) wrote about young African American males and their favorite NBA
players, Many of the Black adolescents gave responses acknowledging the celebrity
Black athletes definitive influence on popular style through athletic apparel and the
athletic-apparel commercial. This influence was manifest most often in the respondents
desire to look like the celebrity athlete (p. 414).
The coinciding occurrence of imitation of athletic play style and choice of athletic
shoes points to the influence of the athletic role model. Katz (1994) described this in his
chronicle of the Nike corporation, writing,

35
by 1991, a visitor to any playground in the world could find hefty young
Charles Barkleys and towering David Robinsons playing Charles-style or
Mr. (David) Robinson-style in their Air Force shoes. Lean and athletic
Scottie Pippens flew in Air Flights; little Bo Jacksons pumped iron in
thousands of gyms, and, by then, miniature Andre Agassis lit up tennis
courts in their peacock-colored Nike clothes and shoes, (p. 148)
Fans could couple their role modeling of athletic behaviors of their sports hero with
purchases of equipment of their sports hero, certainly an inviting concept to athletic
companies marketing such endorsed sports products.
Popular athletes also may serve as role models of economic success. Professional
athletes receive exorbitant salaries; NBA salaries average more than $2 million a season,
not including endorsement income (Simons, 1997). Salaries of top athletes far exceed
this figure, and such wealth is often accomplished at a very young age. Theorists have
suggested that American youths use famous athletes as role models of economic success
and mobility (Simons, 1997; Weisman, 1993; Wilson & Sparks 1996; Kellner, 1996;
Harris, 1994; Michael Jordans, 1991; Nixon, 1984; Starr & Samuels, 1997).
Therefore, such youths may desire to copy this financial success instead of imitating more
realistic role models, such as teachers and local business persons. This occurs despite the
fact that only one of every 50,000 high school basketball players actually reaches the
NBA (Simons, 1997). In other words, an adolescents chances of ever achieving fame
and fortune through professional athletics are quite slim.
Additionally, many popular athletes have exhibited negative behaviors, both on
and off-the-court, that possibly could be role modeled by fans. It is this type of role
modeling that is most worrisome to critics of the modem sports hero. Although certainly
not all popular sports stars display such behavioral tendencies, the list of athletes who
have done so is long and covers a wide range of sports and teams. A few specific
examples of such well publicized negative behaviors are described below.
Many of these negative behaviors occur during competition, referred to as
on-court behaviors. These behaviors often involve arguments and confrontations

36
between players; perhaps the most notable of these occurred when boxer Mike Tyson bit
Evander Holifields ear during a title fight. These behaviors also include athlete
confrontations with fans, coaches, and officials (TKO: Tyson, 1999). For example, in
a much publicized case, NBA star Latrell Sprewell, then of the Golden State Warriors,
attacked his coach, P. J. Carlisimo, twice during a practice session (Starr and Samuels,
1997). In recent years, several players have been suspended in recent from league play
for physically attacking a referee. These include Nick Van Exel and Dennis Rodman of
the NBA, Orlando Brown of the NFL, and Roberto Alomar of Major League Baseball,
who spit on an umpire after a disputed strike call (Collins, 1999). Rodman also has
incurred fines and a suspension for kicking a TV cameraman during a game (Starr and
Samuels, 1997).
Star athletes also have had conflicts with fans during games. Houston Rockets
guard Vernon Maxwell punched a heckling fan in Portland during a 1995 game, costing
Maxwell $20,000 and a ten-game suspension (Nance, 1995). Even professional coaches,
who themselves have become highly paid, popular sports figures, have displayed
extremely inappropriate behaviors during games. NFL coach Mike Ditka made an
obscene gesture to home fans during a home New Orleans football game (Collins, 1999).
Such negative on-court behaviors, including fighting, yelling at other players, shoving
and arguing with officials, kicking cameramen, and, in the most extreme of cases, even
physically harming a coach or fan, have become a somewhat regular occurrence for many
popular sports figures (Wilson & Sparks, 1996; Weisman, 1993; Katz, 1994; Harris,
1994; Harris, 1994b; Starr & Samuels, 1997; Smith, 1973; Walden, 1986; Coakley,
1994).
Negative off-the-court behaviors of popular athletes have become even more
troubling than such on-court behaviors, and many of these behaviors have resulted in
much more serious consequences. One such example is the case of Ray Carruth, wide
receiver for the NFL Carolina Panthers, who was indicted in December 1999 for

37
conspiring to murder his wife (Brady, 1999). Unlike many other athletes who have
committed serious criminal offenses, Carruth did not have a prior record of negative and
illegal behaviors. Other athletes, such as Mike Tyson and Dennis Rodman, became well
known as regular law breakers. Among other offenses, Tyson has been sentenced to jail
twice, once for rape and once for assault, and Rodman has been arrested for sexual
assault, domestic abuse, and drug use (TKO: Tyson, 1999; Rodman sued, 1999).
Although this paper cannot effectively address the many examples of the different
varieties of negative off-the-court behaviors, some behaviors of popular athletes that are
frequently reported by mass media include wife and girlfriend abuse, drug abuse, alcohol
abuse, and assault (Howard, 1995; Taylor, 1991; Star & Samuels, 1997; Donohew,
Helm, & Haas, 1989; Paterno, 1994; Messner & Solomon, 1994; Goldberg, 1995).
Given the invasive media coverage of the off-the-field behaviors of athletes, these
negative behaviors hold serious ramifications for any study of role modeling of popular
athletes.
The previous section addressed the sports hero as a potential role model and some
specific types of role modeling that could result from the actions of mass-mediated sports
figures. This study, however, addresses not only different types of role modeling, but
also the motivations for such role modeling and conditions under which role modeling
might occur. Specifically, this study examines adolescent self-concept and the proximity
of such self-concept to perceived images of sports heroes as a factor in determining a
variety of role modeling behaviors of sports heroes. Therefore, the next section addresses
self-concept and its relationship with behavior and role modeling.

38
Self-Concept
Introduction and Definitions
Much of the literature and debate dealing with the choice of sports heroes has
focused on the qualities of the sports figure, not the individual qualities of the sports fans.
For example, theorists have debated whether the sports hero is determined solely through
athletic feats, or whether he must also personify outstanding qualities in his life outside of
sports (Harris, 1994; Walden, 1986; Smith, 1973). The limited discussion of the
qualities of sports fans typically has focused on group affiliation and characterization of
sports fans, including race, nationality, socio-economic status, and gender, and how these
general characteristics might influence the choice of a sports hero. (Harris, 1994a;
Wilson & Sparks, 1996; Lapchick & Stuckey, 1993; Klein, 1991). However, these
perspectives ignore the fact that selection of a sports hero is an individual choice based on
the individual characteristics and needs of each person. One potential means of
addressing such individual characteristics in a standardized and measurable method is to
examine self-concept in relation to the choice of a sports hero.
Self-concept is an extension of the term self that has been defined as the
individuals attitudes toward self (Frey & Carlock, 1989, p. 2; Secord & Backman,
1964). The self serves as the reference point for each individuals view of the world.
From self, theorists have established self-concept. Self-concept has been defined with
several slight variations, as explained below.
The simplest definition of self-concept is myself as I see myself (Loundon &
Bitta, 1979, p. 373; Dolich, 1969; Landon, 1974; Delozier & Tillman, 1972). This is a
summation of the process by which persons evaluate their own selves to form their
self-concept. Newcomb (1956) elaborated, defining self-concept as the individual as
perceived by that individual in a socially determined frame of reference. This definition
allows for the context of environment to influence the formation of self-concept, and it

39
also hints at the importance of the external environment in the determination of
self-concept.
Several researchers have noted the difference between ones individual
self-concept and the self as observed by other people. In other words, an individuals
view of himself or herself may not be the same as how other people view that person.
One reason for this is because people do not always behave consistently with their
self-concepts. Ones overt behavior is often linked with ones self-concept, but
inconsistencies may exist between the behavioral component of self and ones actual
self-concept. For example, a person might think that he is an excellent student, yet this
same person may not study diligently and complete his academic assignments.
Therefore, his self-concept, or perceptions of himself, would not be consistent with his
behaviors. This could lead to observers viewing a person differently than that person
views himself or herself (Hattie, 1992; Baughman & Walsh, 1962). While the
self-concept is likely instrumental in determining future behaviors, the self-concept is still
a cognitive measure, and it measures ones own perception of reality. This may be
inconsistent both with an objective reality and with the ways in which others perceive
someone (Loundon & Bitta, 1979; Frey & Carlock, 1989; Felker, 1974). Felker(1974)
explained this in his definition of the self-concept, as follows,
self-concept is the sum total of the view which an individual has of
himself. Self-concept is a unique set of perceptions, ideas, and attitudes
which an individual has about himself. The view which an individual has
about himself is unique and to varying degrees is different from the view
that anyone else has about him. (p. 2)
All the aforementioned definitions of self-concept refer to a single self-construct
known as the actual self-concept, or the real self-concept (Sirgy, 1983; Dolich, 1969).
However, the self-concept actually has multiple components. The number of measurable
components of self-concept has varied greatly among researchers (Sirgy, 1983; Hattie,
1992; Loudon & Bitta, 1979; Runyon, 1977). For this research project, two constructs

40
of self-concept are addressed. They are as follows, with the corresponding definitions for
the purposes of this research:
1) The Real Self: An individuals perception of how he actually is (Dolich,
1969; Birdwell, 1964; Ross, 1971; Runyon, 1977; Loundon & Bitta, 1979).
This refers to the most common self-concept, the self-concept which has been
described in the previous paragraphs. This construct has also been called the
actual self-concept because it refers to the individuals self-perceptions as he
or she actually sees them (Sigry, 1983). For this research, this construct does not
take into account a persons perceptions of how others view him or her, known as
the social self-concept (Sirgy, 1983). Instead, this refers only to the
individuals own perceptions of himself or herself.
2) The Ideal Self: An individuals perception of himself as he would like to be
(Delozier & Tillman, 1972; Loundon & Bitta, 1979; Runyon, 1977;
Baughman & Welsh, 1962; Ross, 1971).
These two components are used as separate measures of self-concept and together make
up the construct of self-concept for this research.
Domains of Self-Concept
The construct self-concept, whether it be real or ideal self-concept, is often
regarded as a singular measurement, combining self-perceptions on a variety of different
areas together into one construct. According to this classification, individual perceptions
toward oneself as an athlete would not be differentiated from perceptions toward oneself
as an academic. In other words, measurements of self-concept would not allow the
individual to differentiate between distinct areas of himself or herself. Although widely
used, this means of assessing self-concept is limited and somewhat incomplete (Harter,
1982; Runyon, 1977). Therefore, this research examines self-concept not as one

41
measurement, but rather as a measurement of several distinct domains, all of which are
part of self-concept.
Harter (1982) pointed out distinct problems with using a single measure of
self-concept in a critique of Coopersmiths (1967) self-esteem inventory, a test of 58
questions which make up one global evaluation of self-worth, and the Piers-Harris
Self-Concept Scale (1969), a similar test of 80 questions used to evaluate overall
self-concept. Harter wrote,
items on such popular scales as the Coopersmith Self-Esteem Inventory
and the Piers-Harris Self-Concept Scale tap a range of diverse content
including cognitive competencies, physical skills, popularity, acceptance
by parent, morality, personality traits, physical characteristics, and
affective reactions. Responses to these heterogeneous items are then
summed, and the total score is interpreted as an index of global self-
regard. In employing such a procedure, Coopersmith has assumed that
children do not make distinctions among the domains in their lives. This
assumption was seriously questioned in the scale-construction efforts
reported here. (1982, p. 87)
Early research on self-concept did not break down this construct into specific
domains. However, theorists did find that self-concept could differ in varying situations,
such as a work environment as opposed to a social or an athletic environment (Loundon
& Bitta, 1979; Newcomb, 1956). This was a precursor to the idea that there may actually
be several different components of self-concept, and these different components
accounted for a different self-concept in the context of different environments.
Researchers subsequently determined that the general measurement of self-concept
consisted of several components, including ones perception of ones athletic ability,
morality, academic accomplishment, and physical appearance (Dolich, 1961; Krech et
al., 1962; Loundon & Bitta, 1979; Reynolds et. al, 1980). Furthermore, these
components of self-concept could change with cognitive development and aging (Frey &
Carlock, 1989; Felker, 1974; Hattie, 1992; Katona-Sallay, 1993; Harter, 1982; Harter,
1988). Similarly, Felker (1974) identified separate categories of self-concept that applied

42
specifically to adolescents. These categories were behavior, anxiety, intellectual or
school status, happiness, and satisfaction (Felker, 1974). Although self-concept was still
measured as a single construct, this understanding of the separate components of
self-concept lead to the eventual measurement of separate domains of self-concept.
Susan Harter (1982) conducted research on the separate domains of self-concept
and subsequently created tests to measure these separate domains of self-concept. This
research provided the foundation for the collection of data on self-concept for this study.
Harter created the Perceived Competency Scale for Children, a self-concept test
measuring four separate subscales of self-concept. These subscales are cognitive, social,
physical, and general self-worth (Harter, 1982). Harter later developed the
Self-Perception Profile for Adolescents, a similar self-concept measure for adolescents
that assesses the following eight domains: scholastic competence, social acceptance,
athletic competence, physical appearance, job competence, romantic appeal, behavioral
conduct, and close friendship (Harter, 1988). Because this research examines
self-concept and behaviors of adolescents, these are the domains used as the subscales of
self-concept for this research.
The use of domains for research on self-concept and intentions to role model
sports heroes is crucial because different domains of self-concept may relate quite
differently to role modeling intentions of sports heroes. For example, the athletic domain
of self-concept may have a vastly different relationship with the decision to role model a
sports figure than behavioral conduct or romantic appeal has with such role modeling.
Additionally, as discussed earlier, modem sports heroes are diverse figures, and different
domains of self-concept may relate in different ways to these different qualities of sports
heroes. Popular athletes such as NBA stars Charles Barkley and Latrell Sprewell, both
whom are excellent athletes but who, on numerous occasions, have participated in illegal
and unethical activity, can display seemingly contradictory behaviors and characteristics
in different areas of their life (Starr & Samuels, 1997; Farrey, 1997; Malone, 1993).

43
The ability to separate these different aspects of the individual is paramount to this
research, and this is why the use of Harters research on the separate domains of
self-concept for adolescents guides this examination of self-concept.
While Harter and others have identified specific domains for self-concept, this
does not mean that everyone places equal importance on each of the specific domains. In
fact, each individual likely holds different opinions of the importance and salience of the
various facets of self-concept (Hattie, 1992; Harter, 1987; Gergen, 1971). For example,
Hattie (1992) pointed out, A minister is more likely to have a more clearly delineated
and richer moral self-concept than an athlete. The athlete may have a clearer physical
self-concept (p. 55). Individuals may deem particular domains important for a variety of
reasons, including the amount of familiarity or learning with a particular domain, the
presence of a need or motivation toward a particular domain of self-concept (ex. peers
may give approval to those who are proficient in sports, yet may not give the same
approval to high marks in academics), competence in a domain, and particular social
situations which may cause particular domains to become increasingly important (ex.
frequently attending parties where social interaction is important) (Hattie, 1992; Gergen,
1971; Videbeck, 1958). While any of these factors may help dictate which domains are
perceived as more important than others for the individual, the importance of any
particular domain may be different for each individual. Therefore, because domain
importance can vary both by domain and for each individual, it is extremely important to
measure self-concept as a function of domains instead of an overall measure. The
importance of each domain should also be measured, as is done in this research project.
Further explanation of the importance of domain specific self-concept evaluation will be
given during the following section of self-concept formation and enhancement.
Development and Maintenance of Self-Concept
Self-concept is formed through life experiences and interactions. No one is bom
with a self-concept, but rather it is learned (Felker, 1974; Frey & Carlock, 1989;

44
Baughman & Welsh, 1962; Wells & Marwell, 1976; Orenstein, 1994; Krechetal.,
1962). Self-concept is developed through self-appraisal, reflecting upon ones behaviors;
reflected appraisal, reflecting upon cues from others; social comparison, comparing
oneself to others; and biased scanning, seeking out information to confirm ones
opinions while filtering out contradictory information (Loundon & Bitta, 1979; Runyon,
1977; Baughman & Welsh, 1962; Krech et al., 1962; Bumstein et al., 1962). One may
look to a variety of people and sources in the formation of self-concept. These include
parents, peers, teachers, siblings, and other persons deemed significant to the individual
(Loundon & Bitta, 1979; Hattie, 1992; Frey & Carlock, 1989; Covington & Beery,
1976; Orenstein, 1994; Baughman & Welsh, 1962; Krechetal., 1962; Felker, 1974;
Katona-Sallay, 1993).
Although many influences on self-concept are personally known acquaintances,
mass media figures, and specifically sports heroes, can also influence the formation and
maintenance of self-concept (French, 1991; Katona-Sallay, 1993; Caughley, 1984). A
fantasy relationship with a media figure can serve to confirm ones sense of his or her
own identity. Caughley (1984) wrote in reference to such a perceived relationship
between fan and media star, Landing such a widely adored celebrity figure confirms
your self worth; it makes you somebody (p. 50). To create such a mythical relationship,
a fan may actually fantasize that the media star chooses the individual over millions of
other viewers, picking that person to be a special friend or even a lover or wife. Caughley
(1984) explained the allure of these fantasy relationships, explaining that through
fantasy, a media love relationship is exquisitely turned not to the needs of the celebrity,
but to the needs of the self (p. 51). The relationships with mass-mediated, celebrity
figures can be extremely powerful in shaping personal identity and self-concept.
Media fans also can make social comparisons with popular mass media figures,
and such comparisons, both negative and positive, can lead to further development of
self-concept (Oppenheimer & Oosterwegel, 1993). This process is similar to

45
self-evaluations that can occur with traditional, non-mass mediated models (Bumstein et
al., 1962). Therefore, personal interaction is not necessary to impact self-concept, and
media figures can help to shape the self-concept of a viewer. This may be particularly
important to adolescents who, in an effort to establish their own individual identity, often
turn from more traditional sources of persons, such as parents, and look toward new
sources, such as mass media figures, to define their self-concepts (French, 1991;
Katona-Sallay, 1993). Furthermore, most people form their self-concepts during
childhood and adolescence, and self-concept becomes more resilient to change during
adult years (Krech et al., 1962; Orenstein, 1994; Hattie, 1992; Katona-Sallay, 1993;
Frey & Carlock, 1994). Therefore, research into adolescent self-concept formation is
crucial, because this represents a period when self-concept is still volatile and changing
and is a time when one seeks new sources and people, including media figures, who may
help form ones self-concept.
Self-Concept Maintenance
A primary human objective is to maintain and enhance real self-concept, and this
goal influences human behaviors and interactions (Hattie, 1992; Runyon, 1977; Grubb
& Grathwohl, 1967; Felker, 1974; Frey & Carlock, 1989; Delozier & Tillman, 1972;
Sirgy, 1983; Wells & Marwell, 1976; Loundon & Bitta, 1979; Orenstein, 1994; Krech
et al., 1962). One prominent means of achieving this goal is to bring ones real
self-concept, or ones view of oneself, closer to ones ideal self-concept. Peoples
behavioral intentions and choices of friends and role models are often formed in an effort
to maintain or improve their real self-concepts. Therefore, people act in such a way as to
gain higher opinions of themselves or to preserve the opinion they have of themselves.
Such self-maintenance and self-improvement occurs because it can result in a higher
sense of self-worth (Orenstein, 1994; Harter, 1987; Harter, 1982; Hattie, 1992; Felker,
1974). In other words, improving ones self-concept can be a primary behavioral
influence.

46
However, this view of self-concept enhancement is likely a simplification of the
process. Instead of viewing self-concept as a singular construct, self-concept
maintenance and enhancement should be studied in terms of the many distinct domains of
self-concept. This is because each person places a different level of importance on each
particular domain, or, as described by Hattie (1992), The saliency of various dimensions
of self-concept differs across individuals (p. 169). This difference in saliency among
individuals leads to each specific domain contributing differently to ones sense of
self-worth (Harter, 1982). While people likely attempt to maintain and improve real
self-concept in all areas, people are more likely to do so in their most important domains
of self-concept (Harter, 1987; Harter, 1982; Hattie, 1992; Runyon, 1977). Hattie (1992)
explained, An individuals sense of worth may be in a variety of areas: the body
(beautiful) is of the highest importance to some, but for others it may be more important
to be academically able, to have a happy family life, to gain respect for each other, or to
have a desirable personality (p. 55). Therefore, it is these particular areas for which the
individual will have the most interest in maintaining and improving the real self-concept,
because these areas/domains will contribute the most to ones sense of value.
Self-Concept and Behaviors
The focus area for this research project is the potential link between self-concept
and behavior. The desire to maintain real-self concept and reach towards ones ideal
self-concept by behavioral intention has been examined primarily through studies of
consumer behavior and product selection. While this may differ from the topic
researched in this study (role modeling of sports heroes), prior marketing research should
help build a foundation of knowledge and guide the research for this study.
Two of the earliest examinations of self-concept as a behavioral influence were
done by Rogers (1951) and Maslow (1954). Rogers explained that each person will strive
to actualize, maintain, and enhance himself or herself, and these needs will guide peoples
actions (Rogers, 1951; Felker, 1974). Maslow (1954) introduced similar ideas in

47
describing the process of self-actualization, which refers to the process of achieving ones
potential. According to this process, people are motivated to behave and act in order to
achieve their potential. However, according to Maslows theory, this need for
self-actualization is the last in a long list of human needs. Therefore, self-actualization
will be the last factor to impact behavioral decisions. Since Maslows theory of
connecting self-concept and behaviors is deeply embedded within a larger theory of
human needs and existence, it is somewhat difficult to isolate this concept within
Maslows work.
Much of the subsequent research relating self-concept to behaviors involved
marketing and product selection. According to such research, consumers choose products
which are consistent with their own image, or self-concept, and this can lead to a variety
of purchase decisions. Kretch et. al (1962) explained this relationship by describing how
self-concept motivated middle and upper-class Europeans to choose a lower quality of
bread due to the lofty image of this product.
the very foods that an individual chooses to eat must not only satisfy his
hunger but also be congruent with his conception of himself as a certain
kind of person. Thus the less nutritious white bread came to be preferred
to the more nutritious dark bread because the dark bread was reminiscent
of the European peasants black bread. The color of the bread had
become a status symbol and therefore, a determinant of how well it would
assuage the status seekers hunger, (pp. 83-84)
The choice of bread was motivated by congruence between the white bread and
self-concept. Because white bread was perceived as a high status product, it probably
was indicative of the ideal self-concept of these Europeans. Food was selected not only
to satisfy hunger, but also due to its congruence with oneself as a person.
Several studies have found consumers to choose products consistent with their
self-concepts (Landon, 1974; Sirgy, 1983; Loundon & Bitta, 1979; Runyon, 1977;
Hattie, 1992; Delozier & Tillman, 1972; Birdwell, 1964; Ross, 1971; Felker, 1974;
Dolich, 1961; Grubb & Grathwohl, 1967; Krechetal., 1962). Loundon and Bitta (1979)

48
explained, Products and brands are considered as objects that consumers purchase either
to maintain or to enhance their self-images. The choice of which brand to buy depends
on how similar (or consistent) the consumer perceives the brand to be with his or her
self-image (p. 376). This theory has been examined and tested thoroughly, and it helps
explain the motivations for certain behaviors and behavioral intentions.
Theorists have not definitively established whether real self-concept or ideal
self-concept is more instrumental in guiding consumer choices (Loundon & Bitta, 1979;
Hattie, 1992; Delozier & Tillman, 1972; Sirgy, 1983; Runyon, 1977; Landon, 1974;
Grubb & Grathwohl, 1967; Ross, 1971). It has been suggested that visible products,
such as formal clothing or expensive wines, might reflect the ideal self-concept, while
privately used products, such as toothpaste and soap, might reflect the real self-concept.
Additionally, researchers have suggested certain products, such as a luxury/sports car, are
purchased more for status appeal and will be chosen for congruence with the ideal
self-concept. In contrast, other products, such as a station wagon, are purchased for their
utility value and will be chosen for congruence with the real self-concept. However,
researchers have yet to establish exactly under which conditions real self-concept is used
or when ideal self-concept is used. Grubb and Grathwohl (1967) explained,
further research is needed in terms of specific consumer decision
situations to determine to what extent self-enhancement involves a
conformity concept or an ideal self-image concept. For example, are
consumers, through their consuming behavior and the interaction process,
seeking support for their self-concept as they now perceive themselves, or
are they seeking reactions that will promote the attainment of a more ideal
self? (p. 208)
While both self-concepts have been identified as important in determining consumer
behaviors, the specific instances and motivations for particular behaviors is not yet fully
understood.
The importance of the individual domains, or domain importance, of self-concept
may also help determine consumer behavior. In other words, it is possible that choice of

49
consumer products also depends upon which domains of self-concept are most important
to the individual and how the selected consumer product relates to this particular domain.
For example, the purchase of a calculator may be more related to academic self-concept,
and the selection of athletic shoes may be more related to the physical self-concept.
While not specifically addressed in any known research, researchers hinted at this
potential relationship when they explained that different types of people (ex. housewives,
tennis players, aspiring executives) may seek different types of consumer products (ex.
linens, white shorts, leather brief cases) in reference to their individual self-concepts
(Hattie, 1992; Grubb & Grathwohl, 1967; Loundon & Bitta, 1979). Therefore, the areas
of self-concept that are most important to the individual will guide the selection of
consumer products. The selection of a brief case may not correlate at all to the
self-concept of a housewife because the related domains (professional accomplishment,
etc.) may be unimportant to this individual. Because this area of study is largely
unstudied, domain importance and its correlation with behaviors is further examined in
this research project.
Self-Concept and Mass Media Figures
Little research has addressed the selection and adoption of behaviors of mass
media figures with respect to self-concept or other related constructs. Caughley (1984)
addressed the perceived relationship between a viewer and an admired media figure,
writing, The appeal is often complex, but the admired figure is typically felt to have
qualities that the person senses in himself but desires to develop further. The admired
figure represents an ideal self-image (p. 54). In other words, the observer chooses a
media figure who represents an ideal self-concept, particularly in areas that are important
to the observer. Accordingly, domain importance is a crucial factor in the relationship
between media heroes and fans. Through this complex relationship, the fan can attempt
to develop the most appealing characteristics of his or her favorite celebrity figure,
allowing the fan to strive towards the most valued areas of his or her ideal self-concept.

50
Even less is known about the relationship between the sports hero and the
observer in reference to the fans individual self-concept. Teners (1987) dissertation
work found that the hero choices of an adolescent can be used to predict self-concept,
suggesting a consistency between the hero and the observer. However, this work did not
deal with the real and ideal self-concepts, but rather just the strength of an adolescents
real self-concept. Additionally, this research focused on the importance of choosing a
personally known vs. an unknown hero, and it did not address the different characteristics
of chosen heroes.
While not addressing the specific construct of self-concept, several authors have
suggested that fans may choose and role model their favorite sports figures based on their
perceived similarities between themselves and the athlete (Wilson & Sparks, 1996; Cole,
1996; Kellner, 1996; Harris, 1994a; Simons, 1997; Role models, 1989). The work in
this area has not addressed the individuals perception of himself, or self-concept, but
rather studied observable characteristics of the individual, including race, playing style,
and gender. Therefore, one primary goal of this research is to examine modeling of
sports heroes in relation to individual self-concept, not general group characteristics or
qualities.
Adolescent Self-Concept
This study deals specifically with self-concept of adolescents. Adolescence
represents the beginning of drastic change to ones self-concept, marked by an increased
awareness of self-concept (Orenstein, 1994; Harter, 1988; Frey & Carlock, 1989;
Hattie, 1992); Felker (1974) explained, During adolescence, there is an increased drive
for change in the individual. The increased change and corresponding lack of stability
brings with it a need for reorganization of the self-concept (p. 104). Such reorganization
has been labeled a crisis of identity, a process of drastic change and instability in
self-concept during adolescence (Hattie, 1992, p. 132; Erikson, 1950).

51
Frey and Carlock (1989) described adolescence as a movement from a period of
dependency to a period marked by increasing separateness and independence. They
wrote, The movement away from dependency during adolescence is characterized by
such questions as Who am I? How do I want to be? (p. 285). Additionally,
adolescents typically become aware of a variety of expectations and standards, including
standards for academics, social standing, and athletic competence, against which the self
can be measured (Hattie, 1992; Frey and Carlock, 1989; Orenstein, 1994). Therefore,
this ongoing process of measurement creates a greater interest in and awareness of ones
own perceived competence and ability.
However, most authors have acknowledged that while adolescence marks a period
of change for the self-concept, it is generally a gradual, less rebellious period during
which adolescents make the bid for autonomy gradually and these requests meet
reasonable consideration and deference from parents who ally themselves with the childs
need to grow (Douvan & Gold, 1969, p. 485; Hattie, 1992; Wylie, 1961). In other
words, parents and children typically engage in a give-and-take as adolescents attempt to
form their own identity. It is reasoned that adolescents often shift from using family as
their primary reference and instead rely more on peers, teachers, and mass media to
establish their own identity (Hattie, 1992; Felker, 1974; Katona-Sallay, 1993; Pekrun,
1993; Frey & Carlock, 1974). This can create an instability in self-concept that is only
increased by the ongoing physical changes throughout adolescence (Orenstein, 1994;
Frey & Carlock, 1974; Felker, 1974).
A study by McGuire and McGuire (1981) examined the types of people
adolescents named when asked to describe themselves in relative terms. Compared to
younger children, adolescents were more likely to shift from parents to teachers and from
siblings to friends when defining themselves. In other words, adolescents were more
likely to seek external sources to define themselves during this period of self-concept
awareness and development. Research done by Pekrun (1993) indicated that while the

52
self-concepts of adolescents are still influenced by family support, adolescents also look
to other sources, particularly teachers, to formulate self-concept. Pekrun further added
that different sources may impact different areas, or domains, of self-concept. Pekrun
wrote, School factors have an impact on domain-specific, school related self-concept
development... Furthermore, it may be that peers exert primarily influences on the
development of domain-specific self-concepts of social competencies (1993, p. 118).
This research not only indicates the variety of influences on adolescent self-concept, but
also stresses the importance of using separate domains of self-concept for the study of
adolescent self-concept.
Adolescents also may begin to look toward mass media figures and popular
figures in reference to their own self-concept (Orenstein, 1994; Frey and Carlock, 1989;
Hattie, 1992). Unfortunately, despite the prevalence of media figures in the lives of
adolescents, relatively little research addresses the influence of such figures on adolescent
self-concept. Adams-Price and Green (1990) found that a crush on or an attachment to a
celebrity figure is an important aspect in self-concept development during adolescence.
This process of development often occurs during solitary use of media, allowing an
adolescent to discover his or her new sense of self (Reed, 1995). Such solitary
involvement with media reflects the increased awareness and individuality of self-concept
expected during adolescence.
The majority of the recent literature on television and development of self-concept
has focused on adolescent girls and their perceptions of their bodies. Due to the
prevalence of thin, attractive females in mass media, adolescent female media users often
develop lower self-concepts of their own physical appearance, seeing themselves as
overweight and unattractive in comparison to these media figures (Schneider, 1996;
Rodin, 1992; Harrison, 1998; Tiggermann and Pickering, 1996; Goode, 1999; Botta,
1999; Becker and Popenoe, 1998). These images also have caused many female
adolescent viewers to develop much higher, and likely unachievable, ideal self-concepts

53
for their physical appearance (Smith, 1985; Becker and Popenoe, 1998; Goode, 1999;
Harrison, 1998). According to dissertation work by Harrison (1998), viewing of
television with conspicuously thin characters resulted in higher body dissatisfaction,
while viewing of television with overweight characters did not have the same effect on
physical self-concept. Similarly, Becker and Popenoe (1998) found that the recent
introduction of television and its images of attractive, thin lead characters to a remote
South Pacific island led to a decrease in individual self-image of the female adolescent
viewers. Additionally, these adolescent viewers described shifts in their ideal body image
from that of a robust, rounded body, typical of the adults of the island, to that of a thin
body, typical of the televised figures. The changed image resulting from watching media
figures has lead to changed behaviors of adolescent females, most notably eating
disorders (Goode, 1999; Becker and Popenoe, 1998; Schneider, 1996; Harrison, 1998;
Smith, 1985). In one study, 15 percent of surveyed teenage girls admitted to dieting or
exercising to look more like an image seen on television (Schneider, 1996). While these
studies have focused primarily on adolescent females and the physical component of
self-concept, they do show examples of adolescents developing their self-concept based
on television figures. Furthermore, some of these studies detailed behaviors resulting
from such social comparison to media images. This behavioral component of mass
media social comparison is a focal point of this study, and these studies therefore provide
sound foundation for the theoretical base of this research.
The following section describes congruity theory, a theory predicting the
relationships between self-concept congruity and behavior. Congruity theory builds on
the previously reviewed literature and is the theoretical foundation for the research in this
study.

54
Congruitv Theory
The reviewed literature has established a link between self-concept and behaviors.
In order to logically establish hypotheses for such a link in this research, a theoretical
framework must be established. For this research, congruity theory provides such a
framework.
Congruity theory, proposed by Sirgy (1980), thus far has been used only in
reference to self-concept and consumer behaviors. However, for this research, the
framework of congruity theory has been extended to study self-concept and role modeling
of sports heroes. According to congruity theory, individuals assign images to consumer
products, much like they assign self-concepts to themselves. Congruity, or congruence,
refers to the relationship between ones self image and the image one assigns to a
consumer product. Specifically, congruity increases as the distance between self-concept
and product image decreases. Conversely, an increase in this distance results in less
congruity. This distance is also referred to as a distance score. (Distance scores have
an inverse relationship with congruity. Therefore, as congruity increases, the distance
score decreases.) According to congruity theory, the subsequent intention to purchase a
product is based on the congruity between ones self-concept and ones individually
assigned product image.
Sirgy (1982) has identified two conditions that predict ones intention to use a
product. The first involves real (actual) self-concept and product image congruity. Sirgy
wrote, (1982), The relationship between actual self-image/product-image congruity
(self-congruity) and brand preference has been supported by numerous studies (p. 165).
According to Sirgy (1982), as the two images become more similar, or more congruous
(lower distance score), the resulting affect will be cognitive consistency, which will lead
to subsequent purchase of that product. Conversely, a growing distance between these
two images, or decreasing congruity (higher distance score), will lead to cognitive

55
dissonance (consistency affect), which will likely result in the individual avoiding such
consumer products.
Numerous marketing studies have supported this relationship. These include
studies by Dolich (1969), Delozier and Tillman (1972), Grubb and Grathwohl (1967), and
Ross (1971). In each of these studies, researchers found that the real self-concept was
more closely matched to the most preferred brands and less closely matched to the least
preferred brands. These studies support the first component of Sirgys (1982) congruity
theory.
The second condition involving self-concept and brand image as a predictor for
purchase intention deals with the ideal self-concept. Sirgy (1982) wrote, The
relationship between ideal self-image/product-image congruity (ideal congruity) has been
generally supported... Ideal congruity will lead to positive product evaluation, whereas
ideal incongruity may be responsible for negative product evaluation (p. 167). In other
words, a higher congruity between ideal self-concept and an individuals image of a
product will lead to a greater likelihood of purchase intentions. (The distance between
ideal self-concept and the image of a product is also called an ideal distance score. As
congruity increases, or as the ideal self-concept is closer to the image of a product, the
ideal distance score decreases.)
While theorists have supported this aspect of congruity theory, not as many
studies have addressed ideal self-concept as have addressed real self-concept. Dolich
(1969) concluded that both real self-image and ideal self-image are predictive of brand
preference, and purchase intentions for some products tend to be more closely related to
real self-image, while purchase intentions for other products tend to be more closely
related to ideal self-image (Dolich, 1969; Ross, 1971; Loundon & Bitta, 1979).
Loundon and Bitta (1979) provided a general assessment of the relationships for both real
and ideal self-concept predicted by congruity theory, writing, The consumer behavior of
an individual will be directed towards the furthering and enhancing of his self-concept

56
through the consumption of goods as symbols (p. 377). In other words, people will use
products both to maintain self-concept, further establishing the real self, and to enhance
self-concept, moving closer toward the ideal self. Therefore, both the real and ideal
self-concepts are useful as predictors of consumer behavior, and both should be
considered in the design of academic research.
Application of Theoretical Model to this Study
Congruity theory has significantly added to the understanding of human behavior
and the importance of self-concept in determining behavior. According to congruity
theory, self-concept has an impact upon consumer behaviors. Consumers use objects that
are congruent in image with either their real self-concept, their real ideal self-concept, or
both self-concepts, as this will satisfy the inherent need to either maintain or improve
self-concept. Congruity theory, therefore, explains the relationship between self-concept
and some behaviors.
However, while theorists have indicated that self-concept has a major influence
upon many human behaviors, including the choice of friends and effort placed into
academic work, little (if any) research exists outside of the study of consumer behaviors
(Hattie, 1992; Orenstein, 1994). Because marketing studies have been successful in
identifying behavioral intentions, academics should attempt to apply congruity theory and
the study of self-concept to other important areas of behavioral influence. This is the
emphasis of this research project. It is hoped that the application of sound theory to the
study of important societal behaviors, in this case role modeling of sports heroes, will be
a valuable addition to the academic community and will further the understanding of the
impact of sports heroes on American adolescents.
Prior research studies have been used to provide the framework for this research.
However, because subsequent studies examined consumer behaviors, and this research
project examines role modeling of sports heroes, obvious alterations have been made.
Most obviously, the behavioral component of prior studies, purchasing consumer

57
products, has been replaced with role modeling behaviors of the sports hero. Popular,
mass-mediated sports figures have become so entrenched with marketing products and
mass-media promotions, often for products named after them, it is not such a extension to
view sports figures as products themselves (Katz, 1994; Michael Jordans, 1991).
Perhaps the most important justification for this study design is the lack of
scientific research in the area of self-concept and role modeling. Few studies of role
modeling have examined the influence of self-concept. Because congruity theory
provides a sound research framework, a somewhat liberal alteration of marketing studies
using this theory provides the most logical approach for this dissertation.
Self-Concept Domains as Part of Research Design
Prior studies (Dolich, 1969; Ross, 1971) of congruity theory have used matching
questionnaires to measure the individual self-concept, both real and ideal, and the product
image. From these matched surveys, researchers searched for statistical congruity
between the answers from the two surveys. In other words, distance scores between
self-concept and product image were calculated. Researchers in these studies then used
different questionnaires to measure the behavioral component, purchase intention. To
test congruity theory, distance scores (or congruity between self-concept and product
image) were compared to intentions to buy the product. According to congruity theory,
as distance scores decrease (higher congruity), purchase intentions should increase.
This dissertation follows a similar design. However, in this study, questionnaires
for real self-concept, ideal self-concept, and image of the sports hero address eight
distinct domains of self-concept (as outlined in Harters [1988] Self-Perception Profile
for Adolescents). Instead of looking at a single measure of congruity between
self-concept (real and ideal) and the image of the behavioral object (sports hero), this
study examines each of these relationships across the eight distinct domains of
self-concept. As addressed earlier, the eight domains of self-concept from Harters
Self-Perception Profile for Adolescents are scholastic competence, social acceptance,

58
athletic competence, physical appearance, job competence, romantic appeal, behavioral
conduct, and close friendship.
This use of domains in this research allows for more specific results. As
suggested previously, fans of a sports hero may attempt to copy certain aspects of the
athletes behavior or personality and not other aspects (Simons, 1997; Role models,
1989). Therefore, addressing role modeling across the eight separate domains further
clarifies the particular areas in which the role model will be copied. For example, an
individual may role model his sports hero much more in the domain of athletic
competence than physical appearance. The congruence between ones self-concept and
the image of the sports hero should be analyzed across those particular domains instead of
using a more general, overall measure. This provides a more accurate and useful analysis
of the importance of congruity in role modeling sports heroes.
Additionally, as addressed in the literature review, certain domains of self-concept
may be more important to an individual than other domains (ex. physical appearance
may be more important than job competence to an individual). The literature review has
suggested that individuals will copy role models in areas that are most important to them.
Therefore, this research measures the importance of each domain to see if increasing
domain importance leads to increased role modeling of ones sports hero.
Hypotheses
The review of prior research has lead to the formation of four hypotheses, all four
of which are listed and explained in this section. These four hypotheses address the
research questions presented at the end of Chapter 1 and are examined through the
collection of survey data. The description of this analysis is outlined in Chapter 3, the
Methods section.

59
Hypothesis I. For each domain of self-concept, role modeling of the sports hero
will be positively related to the congruity between real self-concept and the image
of the sports hero.
The reader is reminded of the positive relationship theorized through congruity theory; as
the distance (difference) between self-concept and the image of the sports hero decreases,
congruity increases. For this study, distance scores for each domain of self-concept are
calculated, and these distance scores reflect the distance between self-concept and the
image of the sports hero in each domain of self-concept. As a distance score increases,
congruity between self-concept and image of a sports hero decreases. As a distance score
decreases, congruity between self-concept and the image of the sports hero increases.
Therefore, a positive relationship between role modeling and the congruity of
self-concept/image of the sports hero would be realized through a negative relationship
between role modeling and distance scores. HI deals specifically with real distance
scores, or the distance between real self-concept and the image of the sports hero. This
hypothesis, along with the other three hypotheses, will be written to reflect the negative
relationship with distance scores and role modeling described above, as this negative
relationship, and not the inverse positive relationship with congruity and role modeling,
will be the focal point of the ensuing statistical analysis. This is reflected in the following
restatement of HI.
Hypothesis I. For each domain of self-concept, role modeling of the sports hero
will be negatively related to the distance between real self-concept and the image
of the sports hero, also know as the real distance scores.
The second hypothesis also addresses the relationship between role modeling and the
distance between self-concept and the image of the sports hero. However, H2 deals
specifically with ideal self-concept, and the distances between ideal self-concept and the
image of the sports hero. These distances, calculated for each domain of self-concept, are

60
know as the ideal distance scores. Similar to HI, a positive relationship between role
modeling and the congruity of ideal self-concept and the image of the sports hero is
predicted. This is reflected in a predicted negative relationship between role modeling
and ideal distance scores, as stated in H2.
Hypothesis II. For each domain of self-concept, role modeling of the sports hero
will be negatively related to the distance between ideal self-concept and the
image of the sports hero, also know as the ideal distance scores.
Hypotheses III and IV build upon HI and H2 by incorporating domain importance
into the independent variables (distance scores), creating new dependent variables known
as distance worth scores. Distance worth incorporates both the distance between
self-concept and the image of the sports hero (distance score) in a domain of self-concept
and the importance of that domain of self-concept. Therefore, this new variable is created
by combining two other variables. A detailed explanation of distance worth scores and a
description of the calculation of distance worth scores can be found in the proceeding
Methods section. A brief description of this construct follows below.
For this study, distance worth scores are created by incorporating the importance
of a domain (domain importance score) with the distance between self-concept and the
image of the sports hero (distance score). Distance worth scores reflect the distance score
for a domain of self-concept multiplied by the importance of that domain. However,
domain importance scores have an inverse relationship with distance worth scores. In
other words, distance scores are actually divided by domain importance scores to create
distance worth scores. Therefore, the lower the domain importance score is for a domain
(domain deemed unimportant), the greater the distance worth score will be. Higher
domain importance scores (domains deemed important) will result in smaller distance
worth scores. In other words, the less important a domain of self-concept is, the more the
distance worth score for that domain will increase. Domain importance serves to further

61
qualify the relationship between distance scores, or the congruity between self-concept
and the image of the sports hero, and role modeling. By creating distance worth scores
through the incorporation of domain importance and distance scores, it is predicted that
the less important a domain of self-concept is to an individual, the less likely that the
congruity between self-concept and the image of a sports hero in that domain will have a
significant impact upon role modeling that sports hero. As a domain becomes more
important to the individual, increased congruity (decreasing distance scores) for that
domain will be more likely to have a positive relationship with role modeling of a sports
hero. Hypotheses III deals specifically with real distance worth scores, a set of variables
that incorporate real distance scores in each domain and domain importance scores for the
corresponding domain.
Hypothesis III. For each domain of self-concept, role modeling of the sports hero
will be negatively related to real distance worth.
The fourth hypothesis also addresses the relationship between role modeling and distance
worth scores. However, H4 deals specifically with ideal distance worth, which is created
in the same manner as real distance worth, except that ideal distance scores, not real
distance scores, are divided by domain importance scores for each corresponding domain.
As with H3, a negative relationship between role modeling and ideal distance worth is
predicted. This predicted relationship is reflected in Hypothesis IV, as listed below.
Hypothesis IV. For each domain of self-concept, role modeling of the sports hero
will be negatively related to ideal distance worth.
The following chapter, Methods, addresses the specific methods and statistical analysis
used in this dissertation.

CHAPTER 3
METHODS
The methods chapter covers the following topics crucial to the collection of data
and testing of stated hypotheses: the subjects for this study; the measures administered
for self-concept and the image of the sports hero; creation of the role modeling
questionnaire; pretesting the measures used in this study; the collection of data (the
administration of questionnaires); statistical analysis of collected data, addressing each of
the stated hypotheses; and potential limitations of this analysis. Validity and reliability
are discussed throughout this chapter as necessary.
Subjects for Study
This dissertation was designed to analyze the adolescent self-concept, the
relationship between this self-concept and perceptions of sports heroes, and the
subsequent likelihood of adolescents role modeling the behaviors of sports heroes.
Therefore, all subjects were high school students in grades nine and ten, approximately
aged 14-16. Additionally, this dissertation only examined the behavioral intentions of
male adolescents, so no females were included in the pool of subjects. There are three
reasons for this restriction. First, such a research design prevents gender from being a
confounding variable in data analysis. Second, researchers have suggested that male
adolescents are more likely to look to mass media figures, including athletic heroes, as
role models than are their female counterparts, who are more inclined than males to
continue to view parents and personal acquaintances as role models (McEvoy & Erikson,
62

63
1981). Third, as the vast majority of popular mass media sports figures are male,
adolescent males have many more available sports hero choices of the same gender.
Because perceived similarity has been suggested as a potential factor for role modeling,
adolescent males may be more likely to find role models in sports heroes than female
adolescents. Such an approach is consistent with research in the area of sports heroes as
role models, where researchers have suggested that popular sports figures, such as NBA
basketball players, are more likely to serve as role models for male adolescents (Role
models, 1989; Wilson & Sparks, 1996; Simons, 1997).
The particular subjects for this dissertation consisted of ninth-grade and
tenth-grade males from Houston, Texas. Of the 172 valid subjects used for data analysis
in this study, 120 subjects were students in a suburban private school. All of those 120
students were participants in school athletics. The decision to use only student-athletes
was made due to the increased likelihood that student-athletes have strong ties and
potential perceived similarities between themselves and their sports heroes.
The decision to use adolescents from this particular school as subjects was largely
a pragmatic one. First and foremost, the principal of the high school willingly agreed to
allow his students to participate in this study. Second, because the school is private, the
researcher was able to avoid much of the difficulty involved with using subjects from the
public school system. Finally, because all students are male, the research process could
avoid the perception of being exclusionary to female students in a coed school, as the
female students would not be able to participate as male students did.
The students from this school were predominantly white, with a small percentage
of minorities (Asian, Hispanic, African-American). Most students live in stable
upper-middle-class family environments. Because the student body is not as large as that
of many public schools in the greater Houston area, a relatively large percentage of
students participate in school athletics, and such participation is encouraged and

64
promoted by school faculty and leadership. Therefore, athletics, at least in the
participatory sense, is a part of the daily lives of many of this schools students.
The remaining 52 subjects were participants in a sports tournament in Houston
run through a local community center. These subjects, of the same grade and age
parameters as the first 120 subjects, also were participants in school athletics. The
majority of these subjects participated in baseball, but other selected sports included
soccer, basketball, and track. These subjects share similar demographics traits with the
first 120 students surveyed (white, upper-middle-class, supportive family environment),
and in fact the data collected from the two groups were virtually identical. Again this
group of subjects was selected largely for pragmatic reasons (availability, access, age
appropriateness). Using two groups of subjects and receiving similar results helped to
strengthen the validity of this research. Therefore, the data analysis for this study and the
subsequent analyses of the stated hypotheses were based on the responses of 172 valid
subjects. While more than 200 subjects participated in the study, surveys from
approximately 30 subjects could not be used due to subject error in completing the
surveys. The majority of these errors involved subjects not completing all parts of the
survey questionnaires.
One criticism of this study may be that the students do not represent a diverse
sample, decreasing external validity. Similarly, because most of these students come
from stable family environments with many available role models, these students may not
use sports heroes as role models as much as adolescents who do not have many other
available role models. However, this study will not claim to be generalizable to all
populations, nor is it able to address issues of race, socioeconomic status, and
family/home environment. Additionally, because this study addresses each persons
individual self-concept and allows each person to identify his individual choice of sports
hero, the focus of this research is not on societal patterns, but rather individual choices
based on individual relationships between self-concept and the image of their sports hero.

65
In other words, this research is designed to enable better understanding of how
self-concept affects individual choices to role model a sports hero, not the impact of
external sources on role modeling, and therefore demographic factors (race, family, etc.),
while important in the study of role modeling sports heroes, are not paramount to this
research.
Tests and Test Administration
The following sections detail the survey instruments used for data collection in
this study. This includes the selection of existing instruments, the alterations of such
instruments, and the construction of new instruments.
Self-Concept and Image of the Sports Hero
For the measurement of real and ideal self-concept across all domains, the
Adolescent Self-Perception Profile, created by Susan Harter (1988), was used. This test,
an upward extension of Harters Self-Perception Profile for Children (1982), measures
self-concept in eight specific domains that are most important and relevant to adolescents.
Harters rationale behind using such a domain-specific format, as opposed to a singular
score of self-concept, is as follows: An instrument providing separate measures of
perceived competence or adequacy in different domains... would provide a richer and
more differentiated picture than those instruments providing only a single score (Harter,
1988, p. 2). This test was designed for students in approximately grades nine through
twelve.
The test measures self-concept in eight specific domains. Each domain is
described below, as reported in Harters manual for this test (1988):

66
1) Scholastic Competence: The adolescents perception of his/her competence in
scholastic performance, achievement in classwork, perception of ones
intelligence.
2) Social Acceptance: The degree to which one is accepted by peers, popularity,
number of friends, feels that he/she is easy to like.
3) Athletic Competence: Adolescents perceptions of his/her athletic ability,
feelings that one is good at sports.
4) Physical Appearance: Degree to which the adolescent is happy with how
he/she looks, likes ones body, feels he/she is good looking.
5) Job Competence: Extent to which the adolescent feels he/she has job skills,
prepared to succeed at part-time jobs, feels he/she is doing well at current jobs.
6) Romantic Appeal: Adolescents perceptions that he/she is romantically
attractive to those in whom he/she is interested, dating people he/she wants to
date, feels he/she is fun and interesting on a date.
7) Behavioral Conduct: The degree to which an adolescent likes the way he/she
behaves, does the right thing, acts in appropriate manner, and avoids trouble.
8) Close Friendship: Adolescents perceptions of his/her ability to make close
friends to share personal thoughts and secrets with.
These eight domains make up the test of self-concept.
Five questions address each domain of self-concept (40 questions overall). The
original test has 45 questions because it includes a ninth scale that measures overall
self-concept. This ninth scale was not used in this study, so these questions were
eliminated. Each question is rated on a 1 to 4 scale, and approximately half of all
questions for each domain are counter-balanced such that some have the most adequate
description on the left (scored 4 to 1) and some have the most adequate description on the
right (scored 1 to 4). (Please see Appendix A and Appendix B for actual tests used.)

67
The reliability of Harters test of self-concept has been determined through
repeated use and examination of this instrument. Reliability refers to the reproducibility
of a measure (Graziano & Raulin, 1993). Of primary importance in this study,
particularly for the tests of self-concept, is internal consistency reliability. Internal
consistency reliability is relevant when several observations are made to obtain a score
for each subject (Graziano & Raulin, 1993). Internal consistency reliability is high when
several items measuring the same thing (in this case, each of the eight domains of
self-concept) correlate highly with each other. Several questions (5) are asked for each
domain of self-concept, and therefore an internal consistency reliability score, reported as
Cronbachs Alpha, is reported for each. Harris (1988) reported the subscale internal
consistency reliability for each subscale, as based on Cronbachs Alpha. They are as
follows (for each domain):
1) Scholastic: .81
2) Athletic: .92
3) Acceptance: .78
4) Close Friend: .83
5) Romance: .80
6) Appearance: .86
7) Conduct: .78
8) Job Competence: .74
The high correlation of the items of each domain indicates a high internal consistency
reliability (Graziano & Raulin, 1993). Additionally, Harters (1988) pretests reported
mean scores for each domain ranging from 2.5 to 3.3, depending on the sex and age of the
subjects. The standard deviations for each domain ranged from approximately .41 to .8,
again depending on the subjects.

68
Multiple Use/Administration of Harters Self-Concept Test
As explained in the literature review, five separate measurements were made for
this research (real self-concept, ideal self-concept, image of sports hero, domain
importance, and behavioral intention to role model ones sports hero). All but the last
two constructs, domain importance and behaviors, were measured using Harters (1988)
test of self-concept, but slight variations of the test were made for measurement purposes.
In other words, Harters test was altered so that the same questions addressing the real
self-concept would instead address either the ideal self-concept or the perceived image of
the participants sports hero. Such a procedure for altering an existing test in this manner
is derived from the previously reviewed marketing studies that examine self-concept,
product image, and purchase intentions (Dolich, 1969; Delozier & Tillman, 1972; Grubb
& Grathwohl, 1967; Ross, 1971; Landon, 1974). In these studies, the same (or
extremely similar) tests of self-concept were used to measure the real self-concept, the
ideal self-concept, and the perceived image of the brand/product. For example, Dolich
(1969) used the same semantic differential scale to measure all three constructs, as
explained: The semantic differential was used to measure evaluations by respondents of
real-self image, ideal-self image, and brand images. It is described as a combination of
controlled association and scaling procedures and satisfies the requirement of measuring
several concepts with the same instrument (p. 81). While these examples use such a
scale to rate a brand image as opposed to the image of a sports hero, the process is quite
similar and represents a logical and sound research design.
In order to use such a design, Harters (1988) test was administered three times.
(See Appendix A and Appendix B for actual test alterations.) The first administration
measured the real self-concept. This was done with the original format of the
questionnaire (subtitled What I Am Like) (Harter, 1988). The second administration
measured the ideal self-concept. In measuring this, references to self (me) were changed
to the ideal self (How I Would Like to Be). The third administration measured the

69
perceived image of the sports hero. In measuring this, references to self (me) were
changed to each subjects individual choice of mass mediated sports hero.
For each of the three variations of this test used in this research, the scaling,
question order, and wording of the survey questions were identical. The only difference
was that the first administration addressed the real self, the second administration
addressed the ideal self, and the third administration addressed the sports hero.
Measurement of Domain Importance
In order to measure domain importance, this study used Harters (1988) test of
domain importance (subtitled How Important Are Each of These Things to You?), as
included in Harters adolescent self-perception profile. (See Appendix A and Appendix
B for actual test.) This questionnaire follows a similar format to the test of self-concept,
with a rating scale of 1 to 4, with half of the questions keyed in reverse. This test
contains 16 questions, 2 questions for each of the 8 domains. Data collected with this
questionnaire were used for the creation of distance worth scores for analysis in
Hypothesis III and Hypothesis IV.
Measurement of Role Modeling
The final questionnaire measured the adolescents intention to role model his
sports hero. In many of the questions, behavioral intentions were used instead of
behavioral measures, as discussed earlier in the literature review. Such a process of
measurement also has been copied from marketing studies, which often measure purchase
intentions instead of actual purchase behaviors. For example, in a study of real
self-concept, ideal self-concept, and consumer purchase intentions, Landon (1974)
measured purchase intentions with a five-point scale, with the fifth (most extreme
answer) representing an intention never to purchase the product. (See Appendix C for
actual questionnaire for role modeling ones sports hero.) This same structure has been
used in this dissertation for the measurement of role modeling ones sports hero. A

70
five-point scale, with one (1) representing a strong negative response and five (5)
representing a strong positive response, was used to measure behavioral intentions.
Two processes were used in creating the actual questionnaire to measure role
modeling of ones sports hero. First, specific potential types of role modeling were
identified through literature review and examination of popular sports media (magazines,
television, newspapers, etc.). Included in this examination was the purchase of many
types of athletic goods and commercial products associated with popular sports figures,
particularly those geared toward adolescent males. Second, the researcher asked several
adolescent males to identify role modeling behaviors typical of themselves and their peer
group. Both a focus group of seven teenage males and open-ended written surveys (sent
to approximately ten adolescent males) were used to identify aspects of role modeling
that would be relevant for this study. From the areas of role modeling ones sports hero
identified through these procedures, a Likert scale survey instrument was formulated.
This instrument included the examples of role modeling most frequently identified by
adolescents and relevant literature. The instrument consisted of 23 Likert scale questions,
22 of which addressed specific behaviors of role modeling a sports hero and one overall
role modeling question, used primarily to examine the validity of the questionnaire.
Pretesting of Questionnaires
The instrument for role modeling and the other previously described measurement
instruments (self-concept, image of sports hero) were further examined through
pretesting. This pretest was conducted to allow for possible adjustments to the survey
instruments before the actual test administration for this study. The pretest was
conducted with 20 eighth-grade males from a private junior high school in Houston,
Texas. These students completed all components of this study, just as was to be done in
the actual research. This particular pretest population was chosen due to its similarity to
the actual population that would be examined in this study.

71
Students completed all questionnaires in the same order and manner as would be
done for the actual study. Upon completion, students were given the opportunity to
provide both written and oral critiques of the survey instruments. Two issues were of
particular interest in pretesting these instruments. The first issue dealt with the age
appropriateness of the language used in the questionnaires. In other words, did the pretest
subjects fully understand the wording of each question, and was this wording appropriate
for this age and demographic group? The second issue addressed whether all relevant
areas of adolescent role modeling had been included in the questionnaire of role
modeling. In addition to addressing these questions, the pretest allowed the researcher to
practice survey administration and to observe the amount of time required by subjects to
finish all questionnaires.
While no major adjustments were necessary based on the pretest, minor
alterations were made to the wording of a few questions about role modeling. All of the
role modeling questions can be found in Appendix C.
Test Administration
Five tests (two tests of self-concept, a test of domain importance, a test of the
image of the sports hero, and a test of role modeling of ones sports hero) were
administered to each subject. Tests were administered in either school classrooms or
gymnasiums. The total testing time, including instructions and time for questions, lasted
approximately one hour. Each student completed all five tests in one session. Tests were
administered to small groups of approximately 10-20 students for each session, and the
researcher was present at all testing sessions, along with either a teacher or coach. After
giving instructions about how the questionnaires should be completed and showing a
brief example, the researcher provided an instruction packet for students instructions
beyond those provided verbally by the researcher. The researcher was available to
answer questions during the entire testing period. No incentives were given to subjects
for participation in the study.

72
After the instructional session, subjects were instructed to pick the one athlete
they most considered to be their sports hero. As discussed in the literature review, while
each subject was allowed to choose his own sports hero, individual choices were required
to meet the following two criteria: the athlete must be, or must have been, covered
heavily by mass media (particularly television), and the athlete could not be a personal
acquaintance of the subject. This allowed for a wide range of hero choices but kept the
focus of this study on mass mediated figures as potential role models, as opposed to
personal acquaintances, such as family, friends, and coaches as role models. Subjects
then were told that their selected sports hero would be the basis for all subsequent
questions concerning sports heroes for the study (image, role modeling). In other words,
subjects were instructed to use their individual choice of sports hero as replacement for
the generic sports hero of the questionnaires.
After choosing a sports hero, subjects completed the tests for real self-concept,
ideal self-concept, and image of the sports hero at the same time. Having subjects
complete the three similar tests together was done so students would consider their real
self in juxtaposition to their ideal self and the image of their sports hero. In other words,
the real self-concept was meant to be used as an anchor or standard by which the
ideal self-concept and the image of the sports hero would be considered.
These three tests were given using one questionnaire and one answer packet, in
which students would place their answers for each question of real self-concept next to
the counterpart answers for the same question on each of the other two constructs (ideal
self, sport hero). For example, subjects would place their answer for Question 1 for their
real self (who I am), followed by their answer for Question 1 for their ideal self (who I
want to be), followed by their answer for Question 1 for the image of their sports hero
(what my sports hero is like). After completing all three components of Question 1,
students moved to Question 2 and followed the same procedure. This allowed for
subjects to examine and contemplate their answers for real self in relation to ideal self

73
and the image of their sports hero. The sequence of test administration was identical for
all subjects.
Because the same order was used to answer each question (real self answered
first, then ideal self, and finally image of the hero), it is possible that this order affected
the way in which each question was answered. In other words, contemplating the real
self-concept first for each question might have impacted the scores given for the ideal
self-concept and, subsequently, the image of the sports hero. The completed
questionnaires were examined to ensure that such order effect did not occur. In
particular, irregular answer patterns were looked for, such as a series of ascending
answers for each question. The literature review has suggested that the image of the
sports hero likely would fall between the real self-concept and the ideal self-concept, so
inappropriate variations from this pattern would suggest potential order effect. However,
upon review of the completed questionnaires, order does appear to have impacted the
answers given for each question.
Conducting all three similar tests during the same testing period should have
eliminated the slight possibility of history as a confounding variable (Graziano &
Raulin, 1993). If a week were to transpire between the administration of these three tests,
it is possible, although unlikely, that an experience could alter the real or ideal
self-concept of an individual or an individuals image of his sports hero, and this would
impact the ensuing analysis based on these three constructs.
Following the tests of self-concept and image of the sports hero, subjects
answered the 16 questions addressing domain importance. These were answered on the
same question guide as the self-concept/image questions, and, upon completion, the
entire answer page and question guide were given to the test administrator.
Subjects then were given the role modeling instrument. Before answering the 23
Likert scale questions on role modeling, students again wrote the name of their chosen
sports hero at the top of the questionnaire and were reminded to focus answers on their

74
particular choice of sports hero. This test took much less time to complete than the
previous tests of self-concept/hero image, and subjects were finished with their
participation in the study when they completed and turned in this questionnaire.
The process of analyzing role modeling through survey questionnaires posed an
obvious threat to the internal validity of this study, specifically that subjects may have
been inclined to give socially desirable responses for such questions as the behavioral
intentions to role model their sports hero. In other words, subjects may not have been
willing to admit, or even be fully aware, that they would model such behaviors (ex. style
of dress, dating behaviors, etc.). However, these threats to validity were reduced by
allowing subjects to choose their own personal sports hero (as opposed to forcing any
particular hero on all students) and by reminding students no question had either a correct
or incorrect answer. Additionally, by using Likert scale questions for subject response
(similar to the marketing studies that provide the framework for this research), this threat
to internal validity was reduced, as subjects were allowed to consider such behaviors in
degrees instead of merely being offered a yes or no option.
Because of the length of the testing session and the large number of questions
presented in the questionnaires, subject fatigue was a potential confounding variable for
this study. Specifically, there was concern that subjects might not carefully consider
answers given in the latter parts of the study questionnaires. To help reduce this risk,
subjects were only given one of the two questionnaires at a time, which made the survey
process less overwhelming for participants. Subjects first were given only the
self-concept/image of the sports hero questionnaire. After completing and turning in this
questionnaire, subjects were allowed to take a brief break before taking the role modeling
questionnaire, which was the less time consuming of the two survey instruments.
Additionally, completed surveys were examined by the researcher to look for signs of
subject fatigue as represented by irregular or inconsistent answer patterns in the latter

75
parts of the questionnaires. Upon review of the completed questionnaires, subject fatigue
does not appear to have impacted the answers given for this study.
Upon completion of all questionnaires, the collected data for all 172 valid subjects
were input into SPSS for statistical analysis. The procedures for data analysis are
discussed in a later section of this chapter.
Issues of Validity for this Study
The following section addresses several types of validity, threats to validity and
confounding variables, the relevance of different types of validity to this study, and the
means by which this study attempted to maximize validity.
While several different types of validity and organizations of these concepts have
been outlined by various theorists (Graziano & Raulin, 1993), the most frequently
described types of validity and those most relevant to this study are listed and outlined
below.
1) Content Validity: Content validity refers to the extent to which a
measure thoroughly and appropriately assesses the skills or characteristics
it is intended to measure (Fink, 1991, p. 50). Content validity is
maximized by assuring that the measurements used in the study actually
address the constructs in question. For this particular study, the constructs
include self-concept and role modeling of sports heroes. Content validity
is often maximized by adequately defining the variables in question and by
using questionnaires that address the construct in question (Fink, 1991).
This study has maximized content validity by using valid instruments
designed to assess the particular constructs in question. For the
questionnaire to measure role modeling, content validity was increased by

76
fully defining role modeling and by the use of adolescent focus groups,
who provided information as to which aspects of the behavior of
adolescents and role modeling were most appropriate to include.
2) Face Validity: Face validity refers to the surface characteristics of a
measure, and, unlike content validity, does not rely on established theory
(Fink, 1991). Face validity is increased by asking all appropriate
questions and using language that is appropriate for the subject population.
The face validity for this study was increased by pretesting questionnaires
on adolescents similar to those in the study population and asking for
appropriate feedback on the language of the questionnaires and whether
any aspect of role modeling behaviors had been omitted.
3) Statistical Validity: Statistical validity refers to whether ones results
are due to a systematic factor (the independent variable) or merely to
chance variations (Graziano & Raulin, 1993). This is particularly
important in experimental research with a manipulation of the independent
variable. Statistical validity has been increased for this study by using
reliable measures, such as Harters test of self-concept. Violation of the
assumptions of a statistical test can decrease statistical validity. Because
regression analysis was a primary statistical test used in this research,
violations of the following assumptions could have decreased statistical
validity: normal distribution, a linear relationship between dependent and
independent variables, an assumption of homoscedasticity, and
assumptions that measurements of Y values are independent of each other
(Licht, 1996; Kleinbaum & Kupper, 1978). Statistical validity was
increased by examining a scatter plot of responses before statistical
analysis to assure that statistical procedures such as the potential use of
cross products to account for interaction effects was not required.

77
4) Concurrent Validity: Concurrent validity is maximized when a new
measurement tool compares favorably with an existing measurement tool
that is already considered valid (Fink, 1991). Concurrent validity
increases with high correlation between the new survey instrument and the
existing measurement tool. For this research, the survey instruments used
to assess self-concept and the image of the sports hero are adaptations of
Harters self-concept test, and therefore new tests have not been created to
replace existing, more valid measures. For the created measurement tool
of role modeling intentions, no known existing measurement tool was
found. Therefore, correlation between this new tool and an existing tool
cannot be calculated, and concurrent validity could not be established for
this measure.
5) Construct Validity: Construct validity refers to how well the studys
results support the theory or constructs behind the research and asks
whether the theory supported by the findings provides the best available
theoretical explanation of the results (Graziano & Raulin, 1993, p. 171).
Therefore, a focal component is to eliminate rival hypotheses and other
theoretical explanations for the research results. Threats to construct
validity are reduced by clearly stating definitions and building hypotheses
on well-validated constructs (Graziano & Raulin, 1993). For this study,
threats to construct validity have been reduced through building the
hypotheses on a sound theoretical foundation (congruity theory) that has
been used successfully in marketing research. Construct validity also
refers to the ability of a survey to distinguish between persons who have
certain characteristics and those who do not (Fink, 1991). One means of
establishing construct validity of a survey instrument is to check for
discriminant validity with another measure. In other words, construct

78
validity is increased by insuring that one measurement tool does not
correlate highly with another measurement tool that is used to measure a
different, dissimilar construct. For this study, content validity has been
established by checking that real self-concept and ideal self-concept do not
have a high correlation (discriminant validity). A statistical comparison of
the collected data for real self-concept and ideal self-concept can be found
in the results section of this dissertation.
6) External Validity: External validity, often know as generalizability,
refers to the extent to which the results of a study can be generalized to
different populations and conditions (Graziano & Raulin, 1993). Threats
to external validity are reduced through careful selection of participants,
including random sampling and attempting to choose a representative
sample. However, for this study, subjects were not randomly selected.
Instead, this study used convenience samples, and therefore results must
be generalized very tentatively, likely only to those who have
characteristics similar to those in the study. It is more likely that the
results cannot truly be generalized at all to other populations (Graziano &
Raulin, 1993). However, it should be noted that this study examined
relationships between individual self-concepts, ones individual choice of
a sports hero, and ones individual intentions to role model. Therefore,
while this study cannot be truly generalized to the adolescent male
population at-large, the results present relationships that can be studied
and researched in other, broader adolescent populations.
7) Internal Validity: Internal validity deals primarily with causality,
particularly whether the independent variable is actually responsible for
the observed changes in the dependent variable (Graziano & Raulin,
1993). For this study, most of the threats to internal validity (maturation

79
of the subjects, attrition, instrumentation, etc.) did not affect this study. It
is possible that the sequence of the administration of exams could
potentially affect the answers given on the questionnaires. This is often
controlled by using more than one order of administration of
questionnaires (Graziano & Raulin, 1993). However, for this research, it
was important that subjects used the questionnaire of real self-concept to
anchor the other tests, and therefore one order of administration, with the
real self-concept questionnaire completed first, was used.
8) Subject and Experimenter Effects: Subject effects can threaten the
validity of a study when subjects give socially desirable answers or
unintentionally (or intentionally) try to satisfy the demands of the
researcher through their answers (Graziano & Raulin, 1993). For this
research, it is possible that participants might report higher or lower levels
of behavioral intentions to model a sports hero depending on what they
deem more socially desirable. Additionally, they may name a sports hero
who may not actually be their sports hero, but rather a person they feel
would be a more socially desirable sports hero. These threats to validity
have been reduced by assuring participants that there are no correct or
incorrect answers and by limiting the questions of role modeling to those
that were more likely to be answered honestly. Therefore, questions
concerning anti-social behaviors such as domestic abuse and recreational
drug use were not included. Experimenter effects could occur in this study
if the experimenter tried to influence the answers of the subjects (Graziano
& Raulin, 1993). This was reduced by having teachers administer the
actual surveys and by closely monitoring the procedure. Because there are
no observational data (as in experimental research), the threats to validity
from experimenter effects are minimal in this study.

80
Data Analysis
This section reviews the data analysis for this study and the statistical analysis of
the four hypotheses of this study. All valid collected data were input into SPSS for
analysis. The topics to be discussed include the following: role modeling statistics,
self-concept and calculation of distance scores, and analysis of stated hypotheses. All
results of this data analysis and further explanations are found in the following Results
chapter.
Role Modeling Data
In order to further utilize the results of the role modeling questionnaire as the
dependent variables for this study, reduction of the role modeling data was necessary. To
accomplish this, a factor analysis of the 22 individual role modeling behavior variables
(not including the general overall role modeling variable) was conducted. Factor
analysis, which identifies patterns among the variations in the values of large groups of
variables, is effective in identifying patterns of interrelationships among these variables
and, subsequently, reducing this large number of variables into a smaller number of
variables known as factors (Agresti & Finlay, 1986; Rummel, 1970; Vogt, 1993).
Several rotations of this factor analysis were conducted to find the best fit for the
role modeling variables used in this factor analysis. Rotations are typically done to
identify factors that are more clearly named and distinct in their identity (SPSS, 1999;
Vogt, 1993). In other words, by rotating this factor analysis, more distinct and specific
areas of role modeling were identified as factors. For the variables of role modeling for
this study, a varimax rotation was used, and four significant factors of role modeling were
identified. A significant factor was defined as having an eigenvalue greater than one (1)
and at least three variables with significant factor loadings on that factor.
For each of the significant factors of the factor analysis, an index was created. An
index is a group of individual measures that, when combined, represent a more general

81
characteristic (Vogt, 1993). For this study, each of the significant factors was used to
create an index of role modeling behaviors. Indexes were created instead of merely using
the factors as the more general variables of role modeling because factors place different
weights on each variable in the factor and include components of other variables that are
not strongly related to the targeted role modeling behavior of that particular factor. Each
index, however, only includes the chosen variables that specifically relate to the core role
modeling behavior of the more general index.
An index was created for each factor by summing and then averaging across each
subjects responses on the variables shown to load significantly on that factor.
Significant variables were identified as those with a factor loading greater than .4 on that
factor only. This was done for two reasons. First, this factor loading (.4) represented a
high level of correlation between the variables within an index. Second, upon
examination of the significant factors, this value served as a logical break point for the
selection of variables to each index. An overall index also was created from all variables
that were included in each of the four indexes. Finally, two variables not included in any
of the role modeling indexes (getting a tattoo or earring like a sports hero or taking Andr
like a sports hero) were used in the test of the stated hypotheses. While these two
variables did have a significant factor loading (greater than .4) on factors with
eigenvalues greater than one (1), both of these factors had fewer than three significant
variables and therefore were not used to create indexes.
These seven variables (four role modeling indexes, one overall role modeling
index, and two independent role modeling variables) were the dependent variables for
role modeling used in the four hypotheses of this study.
Self-Concept and Image of the Sports Hero
Data for real self-concept were collected using Harters Self-Perception Profile
for Adolescents (1988). The data for the ideal self-concept and the image of the sports
hero were collected using the altered forms of Harters test, as described earlier in this

82
chapter. For each of these three constructs, a mean score was calculated in each of the
eight domains of self-concept. Each of these 24 domains had five questions. Before a
mean score could be calculated, the results of all questions asked in a negative direction
had to be inverted such that the results from each question ranged from lowest rating of
self-concept/image of sports hero (1) to highest rating (4). (See the Appendix for
questionnaires and directions for individual questions.)
It should be noted that while these mean scores for both self-concept and the
image of the sports hero were reported in the results for informative purposes, these mean
scores were not used in the calculation of the distance scores between self-concept and
the image of the sports hero. Instead, distance scores were calculated from the individual
questions, not the mean scores, to provide a more accurate determination of the specific
distances between the constructs. An explanation of this calculation follows in the next
section.
Distance Scores
Distance scores are a measure of the distance between two concepts, and they can
be used to measure the similarity between two such concepts (Osgood et al., 1957) For
this study, distance scores were calculated to measure the distance between self-concept,
both real and ideal, and the image of the sports hero. In other words, these distance
scores indicate the similarity between self-concept (real and ideal) and the image of the
sports hero. These two separate sets of distance scores, real distance and ideal distance,
have been calculated for each of the eight domains. This resulted in 16 distance scores (8
real, 8 ideal) for use in this study.
Several equations can be used to calculate distance scores. For this study, the
difference squared model, which squares the difference between each paired set of
questions and sums these differences, has been used (Sirgy, 1983; Osgood et al., 1957).
The difference squared model assured a positive value for the distance between each set
of values, regardless of which value, the self-concept or the image of the sports hero, is

83
greater. This formula for calculating distance scores between the image of the sports hero
and either the real or the ideal self-concept in any of the eight domains can be represented
as follows:
Distance (in each domain) = (Ql, Sp Hero Ql, SC)2 + (Q2, Sp hero Q2, SC)2
+ (Q3, Sp hero Q3, SC)2 + (Q4, Sp hero Q4, SC)2 + (Q5, Sp hero Q5, SC)2.
Qn, Sp Hero = Question n from the test of the image of the sports hero.
Qn, SC = Question n from the test of self-concept.
This same formula was used in each domain for real distance and ideal distance.
The sum of squared differences for each question was used instead of the summed
difference of the overall construct scores as this provides a more accurate account of the
descriptions between the sports hero image and the respective self-concept scores. In
other words, this formula assures that differences on specific paired questions from each
domain were not lost or ignored when combined for an overall score for that domain
(Osgood et al., 1957). The lower the distance score for each domain, the closer the
particular domain of self-concept is to the image of the sports hero in that domain. Lower
distance scores indicate closer proximity between self-concept and sports hero image and
a higher level of congruity between the two constructs. The higher the distance score, the
less congruity that exists between self-concept and the image of the sports hero.
These distance scores were used as the independent variables for H1 and H2 of
this study.
Hypothesis I and Hypothesis II
H1 and H2, which predict a negative relationship between role modeling and
distance scores, were statistically examined with multiple regression analysis. Multiple

84
regression is useful for the following types of analysis important to the testing of HI and
H2:
1) Multiple regression analysis can characterize the relationship between the
dependent and independent variables in the sense of determining the extent,
direction, and strength of the association among these variables (Licht, 1995,
p. 34).
2) Multiple regression analysis can determine which of several variables are
important for predicting a dependent variable, and which of these variables are
unimportant in predicting this dependent variable (Licht, 1995; Weisberg et al.,
1989; Kleinbaum & Kupper, 1978).
3) Multiple regression analysis can determine the best model for describing the
relationships between various predictor variables and a dependent variable (Licht,
1995; Weisberg et al., 1989; Kleinbaum & Kupper, 1978; Vogt, 1993).
For this study, regression analysis was used to determine the relationships
between the predictor variables (independent variables), distance scores, and the
dependent variable, role modeling. The dependent and predictor variables were assumed
to have a linear relationship, and an examination of the pretest data and actual data from
this study confirmed this relationship. Therefore, linear regression analysis was used in
this study.
Although regression does not necessarily imply causality, regression can give a
better understanding of the nature of a phenomenon by identifying factors with which it
co-occurs (Licht, 1995; Weisberg et al., 1989). In other words, while this regression
analysis cannot necessarily determine whether congruence between self-concept and the
sports hero causes role modeling of a sports hero, regression can determine whether

85
congruence co-occurs with such role modeling, providing a better understanding of role
modeling of sports heroes and the possible conditions under which it is likely to occur.
For each of the seven role modeling variables (four indexes, one overall index,
and two role modeling variables), three regression analyses were completed. The first set
of analysis used all eight real distance scores as the independent/predictor variables. The
second set of analyses used all eight ideal distance scores as independent/predictor
variables. The third set of analyses used all 16 distances scores (real and ideal) as
independent/predictor variables. One particular function of regression is to determine the
best model for describing a relationship between a dependent variable and independent
variables (Licht, 1995). Therefore, this third set of regression analyses was done in
attempt to create a more predictive regression equation for each of the role modeling
indexes and variables than may be found in the regression analyses using only real
distance scores or only ideal distance scores, and to account for a larger percentage of the
variance of each of these dependent variables (in other words, to increase the amount of
variance explained for each role modeling index or variable). The exclusion of important
predictor variables can drastically affect the results of regression analysis, so it was hoped
that including all the distance scores (real and ideal) would increase the prospect of
achieving statistically significant results (Licht, 1995). Finally, placing all 16 distance
scores in one regression allows for a simple comparison of the impact of all distance
scores by examining the standardized beta weights of each of these 16 dependent
variables (Kleinbaum & Kupper, 1978; Licht, 1995). One common function of
regression analysis is to better understand which predictor variables are more important
than others, and this set of regression analyses has enabled that understanding.
However, theorists do warn of using an excessively large number of predictor
variables in a regression analysis, and, in fact, one potential goal of regression analysis is
to find the best and most predictive model while minimizing the number of predictor
variables (Kleinbaum & Kupper, 1978; Licht, 1995; Weisberg et al., 1989; Fink, 1995).

\
86
This therefore presents the challenge of minimizing the number of predictor variables
while still recognizing the importance of including important potential predictor
variables, even those which do not make significant partial contributions to the dependent
variable, as suggested by the theoretical framework of the study (Agresti & Finlay, 1986).
For this study, while theory requires all distance scores to be included as potential
predictor variables, even those that do not significantly contribute to role modeling,
practice indicates that such a process will weaken the stability of these regression
analyses.
This inclusive set of regression analyses, using all 16 distance scores as predictor
variables, does create difficulties in accurate assessment of the relationships between role
modeling and distance scores. First, using all 16 distance scores does not tell the variance
(R.2) of role modeling accounted for by either the real distance scores or the ideal distance
scores independently, but rather gives the variance from all distance scores. Therefore,
these regressions do not indicate whether real distance scores or ideal distance scores
independently serve as significant predictors of role modeling behaviors. Because that
question is a focus of this study, particularly in attempting to better understand congruity
theory and the differing impact of real vs. ideal distance scores, these inclusive regression
analyses do not fully answer all key questions raised in this study.
Additionally, using all 16 distance scores increases the possibility for
multicollinearity among these independent variables. Multicollinearity refers to the
intercorrelation between the independent/predictor variables in a regression analysis
(Licht, 1995; Kleinbaum & Kupper, 1978; Weisberg et al., 1989). As muliticolliniarity
increases, regression analysis becomes less stable, and it becomes more difficult to assess
the statistical significance of individual predictor variables when such variables are
highly intercorrelated with other predictor variables in the regression. For this study,
high correlation could exist between several of the ideal and real distance scores,
particularly between the two scores from the same domain. Such correlation would result

87
in high multicollinearity, making the regression analyses with all 16 distance scores as
predictor variables less stable than analyses using fewer predictor variables.
For these reasons, the regression analyses using only the eight real distance scores
(or only the eight ideal distance scores) as predictor variables are more crucial to the
study of the first hypothesis. These less complex analyses likely give a more accurate
and stable assessment of the impact of distance scores on role modeling behaviors. The
seven regressions using all 16 distance scores serve to supplement this assessment and
add information for further discussion.
Therefore, 21 separate regression analyses were completed to test HI and H2. A
stepwise regression procedure was used for these analyses. Stepwise regression, unlike
forward selection and backward elimination regressions, finds the best fit for a regression
equation by entering the predictor variables in a variety of orders (Licht, 1995; Vogt,
1993; SPSS, 1999). This type of regression drops predictor variables from a regression
model if they lose their significance when other variables are added to the regression
equation, and this makes stepwise regression the most effective method when correlation
among the independent variables is strong (Agresti & Finlay, 1986; Weisberg et al.,
1989; SPSS, 1999). Given the potential for strong correlation (or multicollinearity)
between the distance scores used as the predictor variables for HI and H2, the stepwise
method was the logical choice for the regression analyses for this study.
The F value of each regression model was examined, and only those regressions
with a p less than .05 were deemed to support HI or H2. As several regressions
addressed each of the two hypotheses, an insignificant regression equation does not reject
either HI or H2, but rather indicates that a particular role modeling behavior does not
have a significant relationship with the distance scores (real or ideal) used in that
particular regression equation. For either HI or H2 to be fully rejected, none of the
regressions for either of the two hypotheses could be significant. By using several
regression analyses for each hypothesis, this study identifies which of the particular role

88
modeling behaviors have significant relationships with distance scores, and which of the
role modeling behaviors do not have such significant relationships. The regressions with
significant F statistics indicate which indexes/variables of role modeling support HI and
H2.
For each of the significant regressions, the t statistic, the significance of this
statistic, and the Beta statistic for each of the individual predictor/independent variables
(distance scores) were examined. These statistics clarified which of the distance scores
had significant relationships with specific role modeling indexes/variables and the
relative strength of these relationships. To determine significance of each of the predictor
variables, the test statistic (t) was used. For any predictor variable to be deemed to have a
significant relationship with a role modeling index/variable, the p value of the test
statistic must be less than .05 (p less than .05). The beta weights (B), or standardized
regression coefficients, of each significant relationship were examined to explain the
strength of the relationship between individual distance scores and role modeling
indexes/variables. This statistic not only explained the extent to which any of the
individual distance scores related to role modeling, but it also explained which of the
distance scores had the strongest relative relationships with role modeling.
An examination of the results of the 21 completed regression analyses provide an
explanation of HI and H2 and selective support (or lack of support) for these hypotheses,
which predict significant relationships between role modeling and distance scores.
Domain Importance and Distance Worth Scores
Domain importance for each of the eight domains of self-concept was calculated
from Harters test of domain importance as part the Self-Perception Profile for
Adolescents (1988). (See Appendix A for the actual test questions.) Two questions
addressed each of the eight domains of self-concept, and a mean score was calculated for
each domain. Domain importance was measured on a scale from one (1) (least important)
to four (4) (most important). These domain importance scores were used, along with the

89
calculated distance scores used for H1 and H2, to create distance worth scores. These
scores were used as the predictor/independent variables for H3 and H4. Before the
calculation of this new variable can adequately be described, it is necessary to explain
fully the proposed relationship between domain importance scores and role modeling of a
sports hero.
As a domain of self-concept becomes more important (domain importance scores
increase), one should be more likely to role model behaviors of ones hero, particularly in
behaviors that are related to that domain of self-concept. Additionally, higher domain
importance should increase the likelihood that one considers the congruity between ones
self-concept and the image of ones sports hero in influencing role modeling. Therefore,
it is predicted that as domain importance scores increase, role modeling of the sports hero
also will increase. In other words, domain importance and role modeling should be
positively correlated.
The following examples illustrate this predicted relationship. Two hypothetical
persons, Frank and Mark, are used for illustration purposes in these examples. Frank and
Mark each has his own individual sports hero. Frank places high importance on his
athletic self-concept. Therefore, his domain importance score for the athletic self-concept
is high (for this example, Franks domain importance score is four). Mark places less
importance on his athletic self-concept (domain importance score is two). According to
the predicted relationship for this research, Frank should be more likely to role model his
sports hero than Mark, especially for behaviors that are related to the athletic
self-concept. Additionally, Franks self-concept congruity with his sports hero in the
athletic domain should be more likely to influence his decision to role model his hero.
Therefore, as domain importance scores increase, role modeling also should increase.
For ease of further analysis and to make distance worth scores smaller and more
easily read, the domain importance scores were reduced by dividing the scores from each
domain by four (4). The new, reduced domain importance scores ranged from .25 (least

90
important) to one (1) (most important). In other words, a domain importance score of 1
became a score of .25, while a score of 4 became a score of 1. These reduced
scores represent the same values and ratios as the original scores, and this data alteration
did not change the nature of this construct nor the relative value of domain importance
from one domain to any of the other seven domains. These reduced distance scores (eight
real, eight ideal) were used in the creation of the 16 distance worth scores used as
independent variables for H3 and H4.
Distance worth scores incorporate both the distance scores used as the
independent variables for HI and H2 and domain importance scores. Therefore, this
variable represents both (a) the congruity between individual self-concept and the image
of a sports hero in a domain and (b) the importance of that domain to an individual.
However, the statistical calculation of this new variable requires specific data alterations
due to the nature of distance scores and domain importance scores. As predicted for this
research, distance scores should have a negative correlation with role modeling. This is
because increased congruity between self-concept and the image of the sports hero in a
domain results in decreased distance scores for that domain. Conversely, it has been
predicted that domain importance scores have a positive relationship with role modeling.
Therefore, to create distance worth scores, it is necessary to make statistical alterations to
one of the two variables such that they both have a relationship with role modeling in the
same direction. In other words, because distance worth has been created as a variable
which combines congruity and domain importance to predict role modeling, the
relationships of congruity (represented by distance scores) and domain importance with
the dependent variable role modeling must be in the same direction for the creation of this
new variable.
The creation of distance worth scores requires distance scores to be multiplied by
domain importance scores. However, to satisfy the above conditions, domain importance
scores (the reduced domain importance scores have been used) have been inverted before

91
being multiplied. For this calculation (see formula below), each of the distance scores
was multiplied by the inverse of the reduced domain importance score for that particular
domain. In other words, distance scores were actually divided by domain importance
scores to create distance worth scores. This was done 16 times, resulting in eight real
distance worth scores and eight ideal distance worth scores. (Each reduced domain
importance score was used two times once for the real distance worth score for that
domain, and once for the ideal distance score). These 16 distance worth scores (eight
real, eight ideal) were used as the predictor/independent variables for H3 and H4. To
calculate distance worth scores, the following formula was used:
Distance Worth = Distance Score x (1 / Reduced Domain Importance Score)
According to this formula, higher distance scores result in higher distance worth scores,
while lower distance scores result in lower distance worth scores. In this respect, distance
worth scores are identical to distance scores in their proposed correlation with role
modeling behaviors. Conversely, due to the inversion of domain importance scores,
higher domain importance scores result in lower distance worth scores, while lower
domain importance scores result in higher distance worth scores. As a domain is rated
more important (domain importance score increases), the distance worth score for that
domain will decrease for both real and ideal distance worth. Therefore, this formula has
allowed for the negative correlation between role modeling and distance scores and the
positive correlation between role modeling and domain importance scores to be
represented in one variable (distance worth).
Due to the reduction and subsequent inversion of domain importance scores,
virtually all distance worth scores will be greater than their corresponding distance
scores. The greater the domain importance score, the less the distance worth score will
increase from its corresponding distance score. The lower the domain importance score

92
(less importance for a domain), the more the distance worth scores will increase from the
distance score. Therefore, higher domain importance scores result in lower relative
distance worth scores. In other words, if two distance scores are equal but the domain
importance scores for that domain are different, the lower relative distance worth score
will be the one with the higher domain importance score.
The following hypothetical examples of the creation of two distance worth scores
further explain this concept.
Example of the Creation of Distance Worth Scores
In Example A, Frank and John are two hypothetical subjects. Both subjects have
identical real distance scores in the athletic domain (distance score = 5). In other words,
both Frank and John have the same congruity between their real athletic self-concept and
their athletic image of their individual sports hero. However, Frank has a higher domain
importance score than John for the athletic domain. Franks domain importance score is
three (3), while Johns domain importance score is two (2). In other words, Frank places
greater value on the athletic domain of self-concept than John does. Using the equation
presented in the previous section, distance worth scores have been calculated for the two
hypothetical subjects, and these scores are represented in the following table.
Table 3-1. Example A of the Calculation of Distance Worth Scores.
Real Athletic
Athletic Domain
Reduced Domain
Real Athletic Dist
Subject
Distance Score
Importance Score
Importance Score
Worth Score
Frank
5
3
0.75
6.67
John
5
2
0.5
10
In this example, while both subjects had identical distance scores, the subject who
placed greater importance on the athletic domain (Frank) had a lower relative distance
worth score for the athletic domain. John placed less value on this domain and had a
lower domain importance score, so his resulting distance worth score was higher.

93
Because H3 and H4 predict that distance worth scores will have a negative correlation
with role modeling, this research predicts that Frank would role model his sports hero
more than John, who had equal congruity yet a lower domain importance score.
Therefore, distance worth scores incorporate both the importance of a domain and the
congruity of that domain in an effort to more accurately predict and assess role modeling
of the sports hero. It should be noted that, as explained in the previous section, both
Franks and Johns distance worth scores are greater than their corresponding distance
scores. This is due to the fact that domain importance scores are reduced before being
inverted in the process of calculation. However, this numerical increase from distance
scores to distance worth scores does not affect the statistical analysis of the hypotheses
for this research. Instead, it is the relative value between distance worth scores, not the
relative value between distance worth scores and corresponding distance scores, that
guides the analyses of H3 and H4.
In Example B, real distance worth scores are again calculated for Frank and John,
this time for the social domain. In this example, the subjects have identical domain
importance scores for the social domain (domain importance score = 3), but Frank has a
lower real social distance score (distance score = 4) than John (distance score = 7). In
other words, while both subjects place equal importance on the social domain, Frank has
greater real self-concept congruity with the image of his sports hero in this domain than
does John. The following table details the creation of distance worth scores for this
scenario.
Table 3-2. Example B of the Calculation of Distance Worth Scores.
Real Social Distance
Social Domain
Reduced Domam
Real Social Dist
Subject
Score
Importance Score
Importance Score
Worth Score
Frank
4
3
0.75
5.33
John
7
3
0.75
9.33

94
Although the domain importance scores are identical, the differences in the
distance scores result in differences in the calculated distance worth scores. Again, both
distance worth scores are greater than their corresponding distance scores. While
Example A showed how differences in domain importance lead to differences in distance
worth scores, this example shows how distance worth scores also reflect differences in
subjects distance scores, or congruity with ones sports hero.
This calculation of distance worth scores has allowed for two concepts, (a) the
distance between self-concept and the image of the sports hero in a domain and (b) the
importance of that domain to be statistically represented in one variable. This variable,
distance worth (for both real and ideal self-concept), has a predicted negative relationship
with role modeling, and this was examined through statistical analysis of hypothesis III
and hypothesis IV.
Hypothesis III and Hypothesis IV
H3 and H4, which predict a negative relationship between role modeling and
distance worth scores, were statistically examined with multiple regression analysis. This
process of analysis was identical to that used to test H1 and H2, except that the distance
scores used for H1 and H2 were replaced with distance worth scores. Twenty-one
separate regression analyses were completed. Three regression analyses were done for
each of the seven role modeling indexes/variables. For each of the dependent variables,
the first regression analysis used the eight real distance worth scores as
independent/predictor variables, the second analysis used the eight ideal distance worth
scores as independent/predictor variables, and the third analysis used all 16 real and ideal
distance worth scores as independent/predictor variables.
The results of these analyses were examined just as was done with the analyses
for HI and H2. The F value of each regression was examined, and only those regressions
with a p value less than .05 were used in support of H3 and H4. As several regressions
addressed each of the two hypotheses, an insignificant regression equation did not reject

95
either H3 or H4, but rather indicates a particular role modeling behavior does not have a
significant relationship with the real or ideal distance worth scores used in that particular
regression equation. The regressions with significant F statistics indicate which
indexes/variables of role modeling had significant relationships with distance worth
scores, and therefore which such indexes/variables would support H3 and H4.
For each of the significant regressions, the t statistic and the Beta statistic for each
of the individual predictor/independent variables (distance worth scores) were examined.
To determine significance of each of the predictor variables, the test statistic (t) was used.
For any predictor variable to have a significant relationship with a role modeling
index/variable, the p value of the test statistic must be less than .05 (p less than .05). The
beta weights (B), or standardized regression coefficients, of each significant relationship
were examined to explain the strength of the relationship between individual distance
worth scores and role modeling indexes/variables. This statistic not only explained the
extent to which any of the individual distance worth scores related to role modeling, but it
also explained which of the distance worth scores had the strongest relative relationships
with role modeling.
An examination of the results of the 21 completed regression analyses provided
explanation of H3 and H4 and displayed selective support (or lack of support) of these
hypotheses, which predict significant relationships between role modeling and distance
worth scores.
The following chapter, Results, details the results of the described analysis. These
results address the four stated hypotheses and the potential support for these hypothesis
based on the collected data for this study.

CHAPTER 4
RESULTS AND DISCUSSION
This chapter details the results of administered surveys and ensuing data analysis.
In addition to a statistical review of the four stated hypotheses, the chapter includes a
statistical review of the collected descriptive statistics of the study. These statistics were
used as a foundation for analysis of the hypotheses. The following areas are addressed:
1) A detailed list of the athletes chosen as sports heroes, the sports played by
these athletes, and the frequency of selection of these athletes.
2) Results of the participants responses to the questionnaires on role modeling
and intentions to role model sports heroes behaviors; factor analysis of the
results of the role modeling questionnaires; and the creation of role modeling
indexes, which were used as the dependent variables of role modeling for the tests
of the hypotheses.
3) Results of the tests of self-concept, both real and ideal, and the test of
the image of the sports hero. These results are reported by the eight domains
of self-concept.
4) Results of the calculated distance scores. These results include both real
distance scores (distance between real self-concept and the image of the sports
hero) and ideal distance scores (distance between ideal self-concept and the
image of the sports hero), reported for each of the eight domains of self-concept.
These distance scores were used as the independent variables for Hypothesis I and
Hypothesis II.
96

97
5) Results of the 21 regression analyses completed to test HI and H2, predicting
a negative relationship between the dependent variable (role modeling) and the
independent variables (real and ideal distance scores). These results include
detailed descriptions of which particular distance scores had negative
relationships with role modeling and for which of the particular role modeling
indexes/variables these significant relationships exist. In other words, the
discussion describes the particular areas of role modeling and the particular
distance scores that bore significant negative relationships to each other.
6) Results of the domain importance questionnaire, reported for each of the eight
domains of self-concept, and the data reduction conducted on the results of these
questionnaires.
7) Analysis of the creation of distance worth scores, both real and ideal,
which incorporate both the distance scores and the corresponding domain
importance score for that domain. These distance scores are reported for each
of the eight domains of self-concept, and they were used as the independent
variables for Hypothesis III and Hypothesis IV.
8) Results of the 21 regression analyses completed to test H3 and H4, predicting
a negative relationship between the dependent variable (role modeling) and the
independent variables (real and ideal distance worth scores). These results were
reported in the same fashion as was done for HI and H2.
7) Summary of the results of the hypotheses.
Descriptive Statistics
Much of the data collected for this study could not be used to test the hypotheses
for this study until undergoing statistical transformations. However, this data provides
important information about the sports heroes chosen in this study and the actual role

98
modeling intentions of the subjects in this study. Specifically, this section will detail the
athletes chosen as sports heroes for this study, the results of the role modeling
questionnaires, and the creation of role modeling indexes through factor analysis.
Chosen Sports Heroes
The actual athletes chosen as sports heroes in this study and the frequency with
which they were chosen are detailed in the following table (Table 4-1). All athletes who
were selected by more than one subject are named in the table. The data for athletes
selected only once are summarized at the bottom of the table. Also included are the
sports played by the selected sports heroes. It should be remembered that each subject
could only choose one athlete. Further, this athlete had to be well known through
television and other mass media outlets, eliminating local athletes or athletes only known
through personal acquaintance.

99
Table 4-1. Athletes Chosen as Sports Hero, Listed in Descending
Order by Frequency.
Frequency
Percent
Cumulative Percent
Michael Jordan (basketball)
16
9.3
9.3
Craig Biggio (baseball)
6
3.5
12.8
Mark McGuire (baseball)
6
3.5
16.3
Ken Griffey Jr (baseball)
6
3.5
19.8
Jeff Bagwell (baseball)
5
2.9
22.7
Steve Austin (pro wrestling)
4
2.3
25.0
Bill Goldberg (pro wrestling)
4
2.3
27.3
Ken Caminetti (baseball)
4
2.3
29.7
Ricky Williams (football)
4
2.3
32.0
Tiger Woods (golf)
3
1.7
33.7
Barry Sanders (football)
3
1.7
35.5
Jason Williams (basketball)
3
1.7
37.2
Charles Barkley (basketball)
3
1.7
39.0
Mike Tyson (boxing)
3
1.7
40.7
Sammy Sosa (baseball)
3
1.7
42.4
Pete Sampras (tennis)
3
1.7
44.2
Allen Iverson (basketball)
2
1.2
45.3
Tony Hawk (extreme sports)
2
1.2
46.5
Cal Ripkin Jr (baseball)
2
1.2
47.7
Tim Duncan (basketball)
2
1.2
48.8
Patrick Roy (hockey)
2
1.2
50.0
Greg Maddux (baseball)
2
1.2
51.2
Jerry Rice (football)
2
1.2
52.3
Jurgen Klinsmann (soccer)
2
1.2
53.5
Michael Owen (soccer)
2
1.2
54.7
Cobi Jones (soccer)
2
1.2
55.8
Kevin Garnett (basketball)
2
1.2
57.0
John Elway (football)
2
1.2
58.1
Emmitt Smith (football)
2
1.2
59.3
70 Athletes with 1 Vote
Each;
70
40.7
100.0
Total
172
100.0
Athletes with 1 vote: 23 baseball, 15 basketball, 8 football, 4 hockej
soccer, 4 track and field, 3 golf, 8 other sports.

100
Ninety-nine different athletes were chosen as sports heroes for the 172 subjects,
demonstrating a diversity of selections. Of the athletes, Michael Jordan was selected by
the most subjects by an overwhelming margin, with 16 subjects choosing the basketball
star. This is not surprising given his athletic excellence, extremely heavy media coverage
and marketing associations, and the generally positive conception of Jordan both on and
off the court. Broken down by sport, however, baseball players were chosen by the
largest number of subjects (57), followed by basketball (43) and football players (21).
Because many of the subjects for this study were participants in a baseball tournament,
the large number of subjects selecting baseball players is not surprising. These three
major sports had far more sports heroes selected than all other sports in this study. A
distant fourth in number of heroes chosen was professional wrestling, with eight subjects
identifying wrestlers as their sports hero. Given the high television ratings for
professional wrestling and the emphasis placed on the character development of wrestlers
throughout the staged productions, it is not surprising that this fake sport had more hero
selections than established sports such as hockey, tennis, track and field, and golf. This is
true despite the fact that many subjects participated in these sports at school. However,
these sports do not receive the same media coverage as the three major sports in America
(football, baseball, and basketball), which accounted for the majority of athletes chosen in
this study, providing support for the contention that television and other mass media are
instruments in providing hero choices for adolescents. Only one of the 172 male subjects
chose a female sports hero, soccer player Mia Hamm, stressing the importance of
perceived similarity in the selection of a sports hero. Additionally, because male athletics
typically receives far greater media coverage than female athletics, these subjects likely
have had much more exposure to star male athletes than their female counterparts. This
also may have influenced the dominant selection of male sports heroes by subjects in this
study.

101
Although it may be difficult to find commonalties in the hero selections, virtually
all athletes chosen currently are or once were stars in their sports. A select few may not
have been stars, but did overcome overwhelming odds to achieve success. One such
example is Rudy Rudiger (selected by one subject), an undersized former Notre Dame
football player who achieved his dream of playing football for Notre Dame despite his
obvious lack of physical talent and size. Additionally, most of the identified athletes
have displayed positive behaviors on and off the court throughout their careers. There
were a few notable exceptions, such as basketball player Charles Barkley (chosen by
three subjects) and boxer Mike Tyson (chosen by three subjects), both of whom have
displayed negative behaviors and encountered legal trouble for their behaviors off the
court.
Finally, the sheer number of different athletes chosen (99) is notable. This
suggests a large number of available sports heroes for adolescents and refutes the idea
that only a small group of popular athletes are chosen as heroes. Subjects chose a wide
range of athletes as their heroes, suggesting the individual nature of the selection of a
sports hero, and even the most popular of athletes was selected by only 16 subjects. In
other words, these subjects did not merely chose a sports hero because he was the most
popular athlete in a particular sport. Therefore, it should be considered that adolescents,
who may learn about their sports heroes from television and mass media, still play an
active part in the selection of the particular televised athletes they choose as heroes.
Role Modeling
The results of participants responses to the role modeling questionnaire are
detailed in the following table. The role modeling variables and corresponding variable
descriptions are listed in descending order from highest mean score (most likely to
model) to lowest mean score (least likely to model). (5 = highest role modeling
intentions, 1 = lowest role modeling intentions).

102
Table 4-2. Role Modeling Intentions, Listed by Individual Role Modeling
Variable.
N
Mean
Std.
Deviation
1KA1IN (1 rain harder because ot hero)
172
1.17
FRIEND (Support friends like hero)
172
3.28
1.42
TRASH (On-court behaviors like hero)
171
3.27
1.38
POSIT (Play same position as hero)
171
3.01
1.53
TECH (Learn athletic techniques from hero)
172
2.97
1.32
MOVES (Copy athletic moves of hero)
172
2.95
1.44
GIRLS (Popular with girls like hero)
172
2.83
1.52
STUDY (Study harder because of hero)
171
2.82
1.47
SHIRT (Wear sports shirt of hero)
171
2.78
1.62
OVERALL (General overall modeling)
169
2.72
.78
OBS (Overcome obstacles like hero)
171
2.65
1.43
COMM (Communicate with coaches like hero)
172
2.64
1.29
FAM (Relationships with family like hero)
172
2.49
1.40
POLITE (Polite to family like hero)
171
2.47
1.48
ENDORSE (Buy products endorsed by hero)
171
2.34
1.31
VOL (Volunteer and charity work like hero)
172
2.29
1.31
UNIF (Wear uniform like hero)
172
2.27
1.30
ACT (Make more friends by acting like hero)
171
2.08
1.31
ANDRO (Take Andro/drugs like hero)
172
1.81
1.18
SNEAK (Wear sneakers like hero)
172
1.70
1.07
COL (Go to same college as hero)
172
1.62
1.09
HAIR (Get hairstyle like hero)
172
1,46
.98
TATT (Get tattoo or earring like hero)
172
1.44
.91
Minimum = 1; Maximum = 5.
Mean scores for each variable ranged from 3.84 for TRAIN, or the desire to train
like ones sports hero, to 1.44 for TATT, or the desire to get a tattoo or earring like ones
sports hero. The six role modeling variables with the highest mean scores, or the six
behaviors that are reportedly most likely to be role modeled, deal with athletic behaviors,
such as training techniques or playing a particular position, or other on-the-court
behaviors, such as trash talking while playing or supporting ones teammates. Note
that a mean score (the variable ROLE, mean = 2.50) was calculated from all role
modeling variables. Additionally, a general question of role modeling intentions (the
variable OVERALL, mean = 2.72) was included in the questionnaire. Neither of these

103
variables were used in further analysis of hypothesis, but rather were used to further
check the reliability of the results of this questionnaire. A correlation analysis among the
two variables was conducted to further examine the relationships between these variables.
According to the Pearson correlation analysis, the two variables had a significant
correlation coefficient of .982 (p < .01).
The high correlation between these variables suggests that subjects answered
questionnaires consistently and further supports this questionnaire as a reliable measure
of role modeling intentions. Given these results, the questionnaire was deemed reliable
enough to use in creating indexes to be used to test the stated hypotheses.
Factor Analysis of Role Modeling Variables
To create indexes of role modeling, a factor analysis of all role modeling variables
was completed, using varimax rotation. This resulted in six factors with eigenvalues
greater than one (1). These six factors and the loading of each role modeling variable on
each factor are detailed in the following table. Significant factor loadings are highlighted
in bold italic print.

104
Table 4-3. Results of Factor Analysis of Role Modeling Variables.
Significant Factors
1
2
3
4
5
6
SINfcAK (Wear sneakers like
hero)
.158
.168
-.033
.594
.309
-.132
HAIR (Get hairstyle like hero)
.388
.179
-.102
.258
.595
-.047
UNIF (Wear uniform like hero)
.102
.398
.086
.557
.298
-.155
TATT (Get tattoo or earring
like hero)
-.151
-.044
.092
.035
.815
.112
POSIT (Play same position as
hero)
-.098
.704
.145
.238
-.017
.109
MOVES (Copy athletic moves
of hero)
.044
.746
.042
.142
.127
.255
TRAIN (Train harder because
of hero)
.352
.636
.095
-.019
-.074
-.066
TECH (Learn athletic
techniques of hero)
.237
.611
.085
.150
.058
-.005
COL (Go to same college as
hero)
.259
.341
-.005
.223
.182
.347
ANDRO (Take Andro/drugs
like hero)
-.018
.038
.094
-.003
-.020
.830
FAM (Relationships with
family like hero)
.216
.036
.610
-.026
.368
.320
FRIEND (Support friends like
hero)
.186
.078
.814
.096
-.132
-.008
COMM (Communicate with
coaches like hero)
.238
.177
.775
.175
.043
.127
POLITE (Polite to family like
hero)
.722
-.006
.380
.121
.037
-.005
VOL (Volunteer and charity
work like hero)
.799
.018
.155
.150
.091
.165
GIRLS (Popular with girls like
hero)
.194
.230
.189
TOO
.387
.476
TRASH (On-court behaviors
like hero)
.270
.387
.451
-.055
.156
.023
STUDY (Study harder
because of hero)
.699
.303
.114
.082
-.087
.035
OBS (Overcome obstacles like
hero)
.699
.282
.161
.107
.018
-.070
ACT (Make more friends by
acting like hero)
.578
.052
.245
.282
.185
.294
ENDORSE (Buy products
endorsed by hero)
.302
.202
.119
.599
.001
.378
SHIRT (Wear sports shirt of
hero)
.116
.074
.139
.800
-.083
.155
Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.

105
For a factor to be used as the foundation of an index for this analysis, that factor
had to have both an eigenvalue greater than one (1) and at least three significant factor
loadings, as defined by a factor loading of greater than .4 on that factor and that factor
only. By these criteria, four factors were used to create indexes (Factors 1 through 4).
Factor 5 and Factor 6, which both had an eigenvalue greater than 1, only had two
significant factor loadings per factor and therefore were not used to created indexes.
Two role modeling variables, TATT (getting a tattoo or earring like a sports hero)
and ANDRO (modeling Andr or drug use of sports hero), loaded highly (.81 and .83
respectively) on Factors 5 and 6. Given the general importance in understanding the
potential role modeling of these behaviors and the potential negative results of such role
modeling, it was deemed worthwhile to include these variables in further analysis.
Instead of being included in one of the role modeling indexes, these variables were used
individually in the analysis of the stated hypotheses. They were not included in any of
the four created indexes.
Role Modeling Indexes
Each of the four factors that met the above stated criteria was used to create a
separate role modeling index. To create each index, the role modeling variables with
significant loading (greater than .4 on that factor only) from each of the four factors were
identified. (See chart below for significant variables used from each factor.) For each
factor, a mean score was calculated from the identified significant variables. The mean
score calculated from each of the four factors became an index of role modeling. These
four created indexes of role modeling were used as dependent variables to test the stated
hypotheses of the study. Additionally, an overall role modeling index was created by
calculating a mean score from all 17 role modeling variables used in creating the four
individual indexes. The individual indexes and a titled description of each index are
detailed in the following table. The indexes are listed in descending order of mean score,

106
from highest mean score (most likely to role model) to lowest mean score (least likely to
role model). (5 = highest role modeling intentions, 1 = lowest role modeling intentions.)
Table 4-4. Role Modeling Indexes Created from Factor
Analysis.
N
Mean
Std. Deviation
Athletic Behaviors
171
IMT
1.0060
Social Communications
171
2.9284
1.0281
Prosocial Behaviors
170
2.4635
1.0893
Consumer Behaviors
171
2.2719
.9670
Overall Indexed Behaviors
172
2.7069
.7810
Minimum = 1; Maximum = 5.
A brief description of each index and the variables included in each variable is as
follows:
1) ATHLETIC BEHAVIORS (Behaviors related to athletic training and
competition created from Factor 2): POSIT (play same position as hero),
MOVES (copy athletic moves of hero), TRAIN (train harder because of hero),
TECH (learn athletic techniques from hero).
2) SOCIAL COMMUNICATION (Behaviors relating to interactions with others,
both on and off the court, created from Factor 3): FAM (relationships with family
like hero), FRIEND (support friend like hero), COMM (communicate with
coaches like hero), TRASH (on-court behaviors like hero).
3) PROSOCIAL BEHAVIORS (Positive behaviors, particularly in behaviors not
related to athletics, created from Factor 1): POLITE (polite to family like hero),
VOL (volunteer and charity work like hero), STUDY (study harder because of
hero), OBS (overcome obstacles because of hero), ACT (make more friends by
acting like hero).

107
4) CONSUMER BEHAVIORS (Behaviors relating to using and purchasing
products used and promoted by the sports hero, created from Factor 4): SNEAK
(wear sneakers like hero), UNIF (wear uniform like hero), ENDORSE (buy
products endorsed by hero), SHIRT (wear sports shirt of hero).
Reliability of the Role Modeling Indexes
With the combination of multiple variables to create role modeling indexes, a
statistical analysis of the reliability of each index was necessary. Several statistics have
been calculated from the data collected from the 172 subjects to substantiate the
reliability of each index. First, the internal consistency reliability as based on Cronbachs
Alpha coefficient was calculated for each index. Often, this is the sole statistic used to
examine the reliability of a measure. However, other statistical checks of reliability can
also be used. These additional statistics are most beneficial when the Cronbachs Alpha
coefficient is too low to ensure the reliability of the measure. Two additional tests of
reliability have been examined for each role modeling index. Both of these reliability
checks are described below.
One of the statistics helpful in examining reliability is the correlation between
each of the individual questions used to create each overall measure (in this case, each
individual index of role modeling). This statistic indicates the strength of the relationship
between the individual questions of the measure. Consistently high positive correlation
between the questions increases the likelihood that these questions form a reliable
measure. An extremely weak or negative correlation between questions for a measure,
particularly when multiple examples of a weak or negative correlation exist for a
measure, indicates that certain questions within a measure may not accurately assess the
intended area of research as do the other questions of the measure. Therefore, some
individual questions may weaken the reliability of a particular measure.
Another revealing statistic is the set of average item-to-total correlation for each
of the variables used to create each measure. This statistic indicates the average of the

108
relationships of each individual question to the overall measure. Higher item-to-total
correlation indicates greater reliability of the measure (in this case, each individual index
of role modeling). Correlation values greater than .3 are desirable to ensure the reliability
of the measure.
These three statistics, internal consistency reliability, correlation between
individual variables, and average item-to-total correlation, are detailed in the following
four tables. Each individual table outlines the reliability statistics for one specific index
of role modeling. Each table includes a correlation matrix for the variables used to create
the index, Cronbachs Alpha coefficient for the index, and the average item-to-total
correlation for the index.
Table 4-5. Reliability Statistics for Role Modeling Questions
of the Athletic Behavior Index.
POSIT
MOVES
TRAIN
TECH
EOS1 l (play position ot Hero)
nror
.564
W
.256
MOVES (copy athletic
.564
1.000
.313
.395
TRAIN (train harder)
.303
.313
1.000
.488
I'ECH (learn techniques)
.256
.395
.488
1.000
N of Cases = 171
Alpha = .7138
Average Item to Total Correlation = .5042.
Table 4-6. Reliability Statistics for Role Modeling Questions
of the Social Communication Index.
FAM
FRIEND
COMM
TRASH
TAM (relations witn tamily)
T"
.405
.505
"''TIT
TRIEND (support friends)
.405
1.000
.588
.325
COMM (communicate)
.505
oo
00
1.000
.392
1 RASH (on-court behaviors)
.312
.325
.392
1.000
N of Cases = 171
Alpha = .7420
Average Item to Total Correlation = .5387.

109
Table 4-7. Reliability Statistics for Role Modeling Questions
of the Prosocial Behavior Index.
POLITE
VOL
STUDY
OBS
ACT
PULI 11 (polite to tamily)
rw
.642
389
.456
.493"
VOL (volunteer and charity)
.642
1.000
.508
.491
.541
STUDY (study harder)
.489
.508
1.000
.546
.416
OBS (overcome obstacles)
.456
.491
.546
1.000
.457
ACT (make more friends)
.493
.541
.416
.457
1.000
N of Cases = 170
Alpha = .8342
Average Item to Total Correlation = .6359.
Table 4-8. Reliability Statistics for Role Modeling Questions
of the Consumer Behavior Index.
SNEAK
UNIF
ENDORSE
SHIRT
SNEAK, (wear shoes)
nr
.383"
.326
.299
UNIT (uniform wearing style)
.383
1.000
.370
.366
ENDORSE (buy products)
.326
.370
1.000
.471
SHIRT (wear sport shirt)
.299
.366
.471
1.000
'4 of Cases = 171
Alpha = .6953
Average Item to Total Correlation = .4866.
The Cronbachs Alpha coefficients ranged from .6953 for Consumer Behaviors to
.8342 for Prosocial Behavior. Further examination of the reliability statistics reveal
positive correlations between all of the variables used to create each of the individual
indexes. Additionally, the average item to total correlation for each index ranged from
.4866 for Consumer Behaviors to .6359 for Prosocial Behaviors. Although higher
Cronbachs Alpha coefficients of reliability would be preferable to ensure the reliability
of each index, the positive correlation of all variables within each index and the relatively
high average item-to-total correlation for each index indicate that these four created role
modeling indexes were reliable measures of their respective behavioral categories.

110
In addition to the four role modeling indexes, an Overall Behavior Index was
created from all 17 of the role modeling variables used to create the four described
indexes. Additionally, the two role modeling variables TATT (get a tattoo or earring like
ones hero) and ANDRO (take Andr or drugs like ones hero) were used individually in
further analysis of role modeling, although they were not included in any of the role
modeling indexes. These seven variables (five indexes and two role modeling variables)
were the dependent variables for the tests of all the stated hypotheses.
Self-Concept, Image of Sports Hero, and Distance Scores
Self-Concept
The real and ideal self-concept scores and the image of the sports hero were
calculated as instructed by Harters Test of Adolescent Self-Concept, resulting in eight
domain scores for each of the three constructs. These results for the tests of self-concept,
both real and ideal, are detailed in the two following tables. Each of the eight domains of
self-concept are listed, in descending order, from the highest mean score (highest rating
of personal competence) to the lowest mean score (lowest rating of personal competence).
Table 4-9. Real Self-Concept, Listed by Domain.
N
Mean
Std. Deviation
Social
172
3.0677
"3925
Scholastic
172
3.0497
.6146
Friendship
172
3.0387
.7063
Job Competence
172
3.0134
.5592
Athletics
172
3.0087
.6310
Romantic
172
2.7907
.6613
Physical
172
2.7020
.6300
Behavior
172
2.6919
.6635
Minimum = 1; Maximum = 4.

Ill
Table 4-10. Ideal Self-Concept, Listed by Domain.
N
Mean
Std. Deviation
Social
172
T5T7T"
Scholastic
172
3.5119
.5385
Romantic
172
3.4171
.5804
Athletic
172
3.4169
.6010
Friendship
172
3.3846
.6342
Physical
172
3.3520
.5775
Job Competence
172
3.3201
.5973
Behavior
172
3.2852
.6366
Minimum = 1; Maximum = 4.
For both real and ideal self-concept, the social domain has the highest mean score
of all eight domains, followed by the scholastic domain. In other words, subjects viewed
themselves as more competent in the social domain than in the other seven domains of
self concept. They also rated the social domain as the area in which they ideally would
hold the highest level of competence. Additionally, the behavior domain has the lowest
mean score for all eight domains in both self-concepts, indicating a low relative sense of
competency in this area self-concept. In all of the eight domains, the mean score for ideal
self-concept is higher than the mean score for real self-concept, as would be expected.
This suggested the subjects perceived level of competence was below their ideal level of
competence. T-Tests were completed to confirm a significant difference between real
and ideal self-concept in each domain. Real self-concept scores from each of the eight
domains were compared to ideal self-concept scores from the same domain, resulting in
eight separate T-Tests. The results of these T-Tests are detailed in the following table.

112
Table 4-11. Differences Between Real Self-Concept and Ideal Self-Concept,
Listed by Domain of Self-Concept.
Paired Differences
t
Sig.
(2-tailed)
Mean
Std.
Deviation
Keaiyideal Athletic Sell-Concept
mr
.5675
-9.432'
IT
Real/Ideal Behavior Self-Concept
-.5933
.7018
-11.088
.000
Real/Ideal Friend Self-Concept
-.3459
.5732
-7.915
.000
Real/Ideal Job Self-Concept
-.3067
.5207
-7.724
.000
Real/Ideal Physical Self-Concept
-.6500
.7620
-11.187
.000
Real/Ideal Romantic Self-Concept
-.6264
.7292
-11.264
.000
Real/Ideal Scholastic Self-Concept
-.4622
.5845
-10.371
.000
Real/Ideal Social Self-Concept
-.4695
.5824
-10.573
.000
For each of the eight domains, real self-concept was significantly lower than ideal
self-concept. Therefore, these subjects have not achieved their ideal self, and there is a
significant difference between how they perceive themselves and how they would like to
be.
Image of the Sports Hero
Similar to the results of the tests for self-concept, these results for the tests of the
image of the sports hero are detailed in the following table. Each of the eight domains of
the image are listed, in descending order, from the highest mean score (highest image) to
the lowest mean score (lowest image).

113
Table 4-12. Image of the Sports Hero, Listed by
Domain.
N
Mean
Std. Deviation
Athletic
172
3.3096
.6005
Job Competence
172
3.2081
.6195
Social
172
3.0677
.5925
Scholastic
172
3.0497
.6146
Friendship
172
3.0387
.7063
Romantic
172
2.7907
.6613
Physical
172
2.7020
.6300
Behavior
172
2.6919
.6635
Minimum = 1; Maximum = 4.
For the image of the sports hero, the athletic domain had the highest mean score,
followed by job competence, which, given the nature of the sports hero and his
employment, also relates to athletics. Therefore, athletic components of the sports hero
were given the highest mean scores. As with both types of self-concept, the behavior
domain received the lowest mean score. Such results suggest a view of sports heroes
which place a premium on supreme athletic competence, yet allow for lower levels of
competence in non-athletic areas, particularly in the ability to behave in the right way and
avoid getting into trouble. In other words, supreme athleticism is the leading
characteristic of the modem sports hero, above other social characteristics of the
traditional hero, an idea suggested in the literature review.
When viewing the image of the sports hero in relation to the real and ideal
self-concepts, the image of the sports hero fell in-between the real and ideal self-concepts
for seven of the eight domains. The only domain for which this was not true was the
behavior domain, where the mean score of the sports hero fell below both the ideal and
the real self-concepts. Therefore, these subjects felt the sports hero typically falls
somewhere between who they are and who they ideally would like to be for all areas
except for the behavioral domain, which measures the ability to behave in the proper

114
manner and avoid getting in trouble. The sports hero may not necessarily be the ideal for
these subjects, but he is a step from the real self toward the ideal.
Distance Scores
Distance scores were calculated in each of the eight domains for the following
two relationships: distance between the real self-concept and the image of the sports hero
(real distance) and distance between the ideal self-concept and the image of the sports
hero (ideal distance). This resulted in 16 distance scores (eight real distance scores and
eight ideal distance scores). As described earlier, these distance scores reflect the
distances between the individual self-concept (both real and ideal) and the image of ones
sports hero in each of the eight domains. The formula for calculating distance scores and
an explanation of this formula can be found in the preceding Methods chapter. As
calculated for this study, the lower the distance score for each domain, the closer the
particular domain of self-concept is to the image of the sports hero in that domain. In
other words, lower distance scores indicate closer proximity between self-concept and the
image of the sports hero.
The results of these calculations are detailed in the following table. The distance
scores for both real distance and ideal distance are listed, in ascending order, from lowest
distance score (closest distance from self-concept to sports hero image) to highest
distance score (furthest distance from self-concept to sports hero image).

115
Table 4-13. Distance Scores of Self-Concept to Sports Hero:
Real and Ideal, Listed by Domain.
N
Mean
Std. Deviation
ideal Athletic Distance
169
3.7041
5.0102
Ideal Social Distance
168
3.9643
5.6842
Ideal Physical Distance
167
4.9162
5.4854
Ideal Job Distance
168
5.0417-
6.8999
Ideal Friendship Distance
171
5.2047
6.8117
Ideal Romantic Distance
171
5.2865
6.2746
Real Athletic Distance
169
5.8935
5.8676
Real Social Distance
168
6.0119
5.9529
Real Job Distance
168
6.0179
5.6202
Real Behavior Distance
168
6.2440
7.0639
Real Friendship Distance
171
6.6491
7.7037
Real Behavior Distance
168
7.3274
7.2364
Real Scholastic Distance
170
7.6824
6.8045
Real Physical Distance
168
7.8750
7.6762
Real Romantic Distance
171
7.9883
7.6535
Ideal Scholastic Distance
170
8.0941
8.0533
Of the 16 distance scores (eight real, eight ideal), the six domain scores with the
lowest mean scores (least distance) were ideal distance scores. Additionally, nine of the
ten distance scores with the highest mean scores (furthest distance) were real distance
scores, the exception being Ideal Scholastic Distance, which had the highest mean score
of all 16 distance scores. The athletic and social distance scores were lowest for both the
ideal and real distance scores, suggesting that these subjects perceived themselves as
closest to their sports heroes in these two areas and that they aspired to be close to their
sports heroes the most in these two areas. These 16 distance scores were used as the
independent variables for Hypothesis I and Hypothesis II.
Reliability of Real Distance Scores and Ideal Distance Scores
Because the creation of each of the 16 distance scores involved the combining of
five separate measures of the distance between self-concept and the image of the sports
hero to create one overall distance score for each domain of both real and ideal
self-concept, an analysis of the reliability of these distance scores was necessary. This

116
statistical analysis is detailed in the following section. (Please refer to the Methods
section for a description of the creation of distance scores.)
For each distance score, three statistical checks of reliability (internal consistency
reliability, correlation between individual variables, and average item-to-total correlation)
were examined. These three statistics are detailed in the following tables, and each
individual table outlines the reliability statistics for one distance score. Each table
includes a correlation matrix for the five separate distances between self-concept and the
image of the sports hero used to create the overall measure, Cronbachs Alpha coefficient
for the measure, and the average item-to-total correlation for the measure. The first eight
tables below detail the reliability for the real distance score measures.
Table 4-14. Reliability Statistics for Real Social Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance 1
nr
399"
.237
.228
.3112
Distance 2
.399
1.000
.257
.164
.335
Distance 3
.237
.257
1.000
.102
.159
Distance 4
.228
.164
.102
1.000
.261
Distance 5
.312
.335
.159
.261
1.000
N of Cases =
m
Alpha = .6210
Average Item to Total Correlation = .3759.
Table 4-15. Reliability Statistics for Real Scholastic Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance I
nr
.102
.158
.060
.051
Distance 2
.102
1.000
.320
.404
.259
Distance 3
.158
.320
1.000
.289
.245
Distance 4
.060
.404
.289
1.000
.342
Distance 5
.051
.259
.245
.342
1.000
N of Cases = 170
Alpha = .5893
Average Item to Total Correlation = .3504.

117
Table 4-16. Reliability Statistics for Real Athletic Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Uistance l
rmr
33TT
TIT
.159
.ITT
Distance 2
.430
1.000
.138
O
Os
OO
.238
Distance 3
.373
.138
1.000
.189
.137
Distance 4
.159
.068
.189
1.000
.349
Distance 5
.241
.238
.137
.349
1.000
N of Cases = 169
Alpha = .6032
Average Item to Total Correlation = .3589.
Table 4-17. Reliability Statistics for Real Physical Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Uistance 1
mr
.344
.262
TW*
.38 r
Distance 2
.344
1.000
.493
.249
.258
Distance 3
.262
.493
1.000
.311
.260
Distance 4
.350
.249
.311
1.000
.524
Distance 5
.381
.258
.260
.524
1.000
N of Cases = 168
Alpha = .7221
Average Item to Total Correlation = .4819.
Table 4-18. Reliability Statistics for Real Job Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Uistance 1
rw
"".'l 85"
.119
wr
.054
Distance 2
.185
1.000
.202
.228
.262
Distance 3
.119
.202
1.000
.195
.221
Distance 4
.071
.228
.195
1.000
.313
Distance 5
.054
.262
.221
.313
1.000
N of Cases = 168
Alpha = .5263
Average Item to Total Correlation = .2966.

118
Table 4-19. Reliability Statistics for Real Romantic Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance 1
nr
327"
.452"
372"
.28(1
Distance 2
.227
1.000
.314
.300
.343
Distance 3
.452
.314
1.000
00
rn
.279
Distance 4
.372
.300
CO
oo
1.000
.296
Distance 5
O
OC
CM
.343
.279
.296
1.000
N of Cases = 171
Alpha= .7041
Average Item to Total Correlation = .4617.
Table 4-20. Reliability Statistics for Real Behavior Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance 1
mr
.1o
.291
.168
.442
Distance 2
.270
1.000
.273
.310
.275
Distance 3
.291
.273
1.000
.265
.371
Distance 4
.168
.310
.265
1.000
.296
Distance 5
.442
.275
.371
.296
1.000
N of Cases = 168
Alpha= .6779
Average Item to Total Correlation = .4323.
Table 4-21. Reliability Statistics for Real Friend Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance 1
TIT
.356
323"
355"
.490
Distance 2
.356
1.000
.503
.408
.393
Distance 3
.323
.503
1.000
.394
.409
Distance 4
.298
.408
.394
1.000
.377
Distance 5
.490
.393
.409
.377
1.000
N of Cases = 171
Alpha= .7655
Average Item to Total Correlation = .5350.
The Cronbachs Alpha coefficients of reliability ranged from .5263 for Real Job
Distance to .7655 for Real Friend Distance. All of the distance scores had average
item-to-total correlations greater than .3 except for Real Job Distance, which had an

119
average item-to-total correlation of .2966. The reliability statistics also reveal positive
correlations between all of the variables/distances used to create each of the individual
real distance score measures. A few slightly weaker correlations, although still positive,
came within the measures of Real Job Distance and Real Scholastic Distance. Because
many of the subjects for this study have had limited experience working in actual jobs,
they may not yet have formed a strong sense of their own self-perceptions in this area.
This likely accounted for the relatively weak reliability statistics for the measure of Real
Job Distance. Similarly, it is possible that these subjects did not have well developed
perceptions of the academic abilities of their heroes. This would be particularly true for
those subjects who selected their heroes based primarily on the heros athletic abilities
and accomplishments, not the heros abilities in other areas of life. This likely accounted
for the relatively weak reliability statistics for the measure of Real Scholastic Distance.
Although higher Cronbachs Alpha coefficients of reliability would be preferable
to ensure the reliability of each real distance score measure, the positive correlation of all
variables within each measure and the relatively high average item-to-total correlation for
each measure indicate that these eight distance scores were reliable measures of the
distance between real self-concept and the image of the sports hero.
The following eight tables outline the same reliability statistics (internal
consistency reliability, correlation between individual variables, and average item-to-total
correlations) for the eight ideal distance score measures.

120
Table 4-22. Reliability Statistics for Ideal Social Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance l
IW
.414
.163
.155
.466
Distance 2
.414
1.000
.263
.305
.389
Distance 3
.163
.263
1.000
.316
.316
Distance 4
.155
.305
.316
1.000
.499
Distance 5
.466
.389
.316
.499
1.000
N of Cases = 168
Alpha = .7066
Average Item to Total Correlation = .4673.
Table 4-23. Reliability Statistics for Ideal Scholastic Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance I
T"
.276
,irr
W
.205
Distance 2
.276
1.000
.360
.317
.361
Distance 3
.312
.360
1.000
.499
.393
Distance 4
.259
.317
.499
1.000
.341
Distance 5
.205
.361
.393
.341
1.000
N of Cases = 170
Alpha = .7125
Average Item to Total Correlation = .4719.
Table 4-24. Reliability Statistics for Ideal Athletic Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance 1
1.000
.456
.116
.135
.171
Distance 2
.456
1.000
.103
.305
.189
Distance 3
.116
.103
1.000
.197
.246
Distance 4
.135
.305
.197
1.000
.234
Distance 5
.171
.189
.246
.234
1.000
N of Cases = 169
Alpha = .5753
Average Item to Total Correlation = .3352.

121
Table 4-25. Reliability Statistics for Ideal Physical Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance I
nr
.133
.145
.16/
.256
Distance 2
.133
1.000
.306
.262
.258
Distance 3
.145
.306
1.000
OO
.271
Distance 4
.167
.262
.285
1.000
.512
Distance 5
.256
.258
.271
.512
1.000
N of Cases = 167
Alpha = .6392
Average Item to Total Correlation = .3936.
Table 4-26. Reliability Statistics for Ideal Job Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance l
T-
ITT
.277
.172
.337"
Distance 2
.428
1.000
.358
.188
.366
Distance 3
.277
.358
1.000
.269
.279
Distance 4
.172
.188
.269
1.000
.247
Distance 5
.337
.366
.279
.247
1.000
N of Cases = 168
Alpha = .6712
Average Item to Total Correlation = .4277.
Table 4-27. Reliability Statistics for Ideal Romantic Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance 1
T"
.213
.290
.254
.207
Distance 2
.215
1.000
.319
.170
.395
Distance 3
.290
.319
1.000
.384
.215
Distance 4
.254
.170
.384
1.000
.235
Distance 5
.207
.395
.215
.235
1.000
N of Cases = 171
Alpha= .6443
Average Item to Total Correlation = .3987.

122
Table 4-28. Reliability Statistics for Ideal Behavior Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance 1
nr
ITT
ITT
.278
.451
Distance 2
.223
1.000
.358
.394
.347
Distance 3
.312
.358
1.000
.119
.407
Distance 4
.278
.394
.119
1.000
.345
Distance 5
.451
.347
.407
.345
1.000
N of Cases = 168
Alpha = .7033
Average Item to Total Correlation = .4625.
Table 4-29. Reliability Statistics for Ideal Friend Distance Scores.
Distance 1
Distance 2
Distance 3
Distance 4
Distance 5
Distance l
TTOT
W
.222
.220'
.426
Distance 2
.340
1.000
.542
.224
.329
Distance 3
.222
.542
1.000
.389
.405
Distance 4
.220
.224
.389
1.000
.450
Distance 5
.426
.329
.405
.450
1.000
N of Cases = 171
Alpha = .7320
Average Item to Total Correlation = .4944.
The Cronbachs Alpha coefficients of reliability ranged from .5753 for Ideal
Athletic Distance to .7320 for Ideal Friend Distance. All of the ideal distance score
measures had average item-to-total correlations greater than .3. The reliability statistics
also reveal positive correlations between all of the variables/distances used to create each
of the individual ideal distance score measures.
As with the real distance score measures, higher Cronbachs Alpha coefficients of
reliability would be preferable to ensure the reliability of each ideal distance score
measure, but the positive correlation of all variables within each measure and the
relatively high average item-to-total correlation for each measure indicate that these eight
ideal distance scores were reliable measures of the distance between ideal self-concept
and the image of the sports hero.

123
Comparison of Real Distance Scores to Ideal Distance Scores
It was expected that the image of the sports hero would be closer to the ideal
self-concept than the real self-concept, or that these subjects would perceive their sports
hero as more similar to their ideal than to their real self. In other words, it would be
expected that ideal distance scores would be smaller than real distance scores. T-Tests
were calculated for each domain to confirm this assumption. Real distance scores from
each of the eight domains were compared to ideal distance scores from the same domain,
resulting in eight separate T-Tests. It was expected that there would be a significant
difference between ideal distance and real distance for each of the eight domains. The
results of these T-Tests are detailed in the following table.
Table 4-30. Differences Between Real Distance Scores and Ideal Distance
Scores, Listed by Domain of Distance Score.
Paired Differences
t
Sig.
(2-tailed)
Mean
Std.
Deviation
Keal/ldeal Athletic Distance
2.1893"
5.8483
4.867
nr
Real/Ideal Behavior Distance
1.0833
7.3358
1.914
.057
Real/Ideal Friend Distance
1.4444
6.1779
3.057
.003
Real/Ideal Job Distance
.9762
5.2082
2.429
.016
Real/Ideal Physical Distance
2.9940
8.6807
4.457
.000
Real/Ideal Romantic Distance
2.7018
7.3644
4.797
.000
Real/Ideal Scholastic Distance
-.4118
6.8953
-.779
.437
Real/Ideal Social Distance
2.0476
6.0673
4.374
.000
For six of the eight domains, the ideal distance scores are significantly smaller
than the real distance scores. Therefore, these subjects perceived their sports heroes as
closer to their ideal self than their real self in six of the eight domains. The only two
domains for which this is not true are the behavior domain and the scholastic domain. In
fact, the scholastic domain is the only one of the eight domains where the image of the

124
sports hero is closer to the real self-concept than the ideal self-concept, reflecting both a
high ideal academic self-concept and a correspondingly low image of the sports heros
competence in academic areas. For six of the areas of self-concept, however, it can be
stated that the sports hero more closely approximates the ideal self, or who these subjects
want to be, than the real self, or who these subjects currently perceive themselves to be.
Results for Hypothesis I and Hypothesis.!!
The first hypothesis (HI) predicted a negative relationship between real distance
and role modeling intentions. Therefore, it was expected that there would be a negative
and significant relationship between the individual indexes of role modeling and the real
distance scores. To test this, a separate regression analysis was done for each role
modeling index, the overall role modeling index, and the individual variables TATT
(getting a tattoo or earring like ones hero) and ANDRO (using Andr or drugs like
ones hero). For each regression analysis, the role modeling variable/index served as the
dependent variable, while each of the eight real distance scores for the eight domains
served as the independent variables. Therefore, seven separate regression analyses were
done to test H1.
The second hypothesis (H2) predicted a negative relationship between ideal
distance and role modeling intentions. The regression analyses for this hypothesis were
identical to those done for HI, except that ideal distance scores were used instead of real
distance scores as the independent variables.
A third group of regression analyses were done that included both real distance
scores and ideal distance scores. These analyses, which addressed both HI and H2, were
the same as those done for real distance and for ideal distance, except that all 16 distance
scores (eight real distance, eight ideal distance) were used as the independent variables
for each of the seven regression equations. For each equation, one of the four role

125
modeling indexes, the overall role modeling index, or ANDRO or TATT served as the
dependent variable. By including all distance scores as independent variables, these
regression analyses helped to determine whether real or ideal distance scores were
stronger predictors of each individual role modeling index/behavior. Additionally, these
regression analyses should create a more predictive regression equation for each of the
role modeling indexes/variables by accounting for a larger percentage of the variance of
each of these independent variables (in other words, to increase the of each regression
analysis). Finally, because the exclusion of important predictor variables can drastically
affect the results of regression analysis, it was hoped that including all the distance scores
(real and ideal) would increase the prospect of achieving statistically significant results
(Licht, 1995).
The results of all of the regression analyses for these two hypotheses are detailed
in the following tables. The tables are organized by dependent variable (role modeling
index/variable), not by hypotheses. Therefore, all three regression analyses for each of
the seven dependent variables are grouped together. A more general discussion of the
individual hypotheses follows these results. For each of the significant regression
analyses that produced a significant F-value (p less than .05), one table details both the
overall significance of the regression (including the R and R^ values and the F-value) and
the significance of the separate independent variables (distance scores). Following each
table is a description of the significant relationship(s) between the independent variable(s)
and the role modeling index/variable. This includes the Beta coefficient for each
relationship. For each relationship, indicates significance at p less than .05, and **
indicates significance at p less than .01.
For those regressions whose F Statistic is not significant at the .05 level, only a
brief text description of the analysis is given. In the proceeding section, for each role
modeling index/variable examined in this research, the regression analysis with only real
distance scores is explained first, followed by the regression analysis with only ideal

126
distance scores. The regression analysis with both real and ideal distance scores is
described last for each index/behavior.
Following the tables of the separate regression analyses, one overall table
summarizes the results of these analyses. This table indicates which role modeling
indexes/variables were significantly related to the distance scores and which particular
distance scores had significant relationships with each of these indexes/variables.
Prosocial Behaviors
Table 4-31. Regression Analysis for Real Self-Concept
Distance Scores on Prosocial Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
im~
.160
17.559
.000
Real Athletic Distance
-.026
.016
-.133
-1.601
.112
Real Behavior Distance
-.016
.013
-.108
-1.164
.246
Real Friend Distance
.022
.014
.152
1.516
.132
Real Job Distance
-.038
.019
-.190
-2.010
.046
Real Physical Distance
.034
.013
.234
2.644
.009
Real Romantic Distance
-.017
.016
-.112
-1.079
.282
Real Scholastic Distance
-.013
.014
-.087
-.957
.340
Real Social Distance
-.010
.020
-.049
-.492
.624
R = .364
R Square = .132
F = 2.765
Significance of F = .007.
The distance scores with significant relationships with prosocial behaviors were real job
distance (Beta = -.19*) and real physical distance (Beta = .23**, NOTE: This
relationship is the opposite direction as predicted).
Ideal distance scores. No significant relationships were found in the regression analysis
for ideal distance scores on prosocial behaviors (p > .05).

127
Table 4-32. Regression Analysis for Real and Ideal Self-Concept
Distance Scores on Prosocial Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
w
.175
16.758
MT
Real athletic distance
-.039
.018
-.203
-2.176
.031
Ideal athletic distance
.039
.023
.184
1.704
.091
Real behavior distance
-.016
.015
-.108
-1.046
.297
Ideal behavior distance
-.015
.016
-.104
-.950
.344
Real friend distance
.029
.018
.202
1.591
.114
Ideal friend distance
.001
.019
.004
.032
.975
Real job distance
-.039
.022
-.195
-1.737
.085
Ideal job distance
-.005
.022
-.031
-.214
.831
Real physical distance
.028
.013
.196
2.118
.036
Ideal physical distance
.006
.018
.031
.323
.747
Real romantic distance
-.021
.016
-.140
-1.284
.201
Ideal romantic distance
.021
.024
.124
.883
.379
Real scholastic distance
-.003
.017
-.018
-.160
.873
Ideal scholastic distance
-.025
.016
-.200
-1.580
.116
Real social distance
-.010
.022
-.053
-.472
.638
Ideal social distance
-.011
.023
-.058
-.451
.653
R .452
R Square = .204
F = 2.184
Significance of F = .008.
The distance scores with significant relationships with prosocial behaviors were real
athletic distance (Beta = -.20*) and real physical distance (Beta = .19*, NOTE: This
relationship is the opposite direction as predicted). Note that although no ideal distance
scores had significant relationships, their inclusion affected the relationships between real
distance scores and prosocial behaviors.

128
Athletic Behaviors
Table 4-33. Regression Analysis for Real Self-Concept
Distance Scores on Athletic Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
3.645
.150
24.222
MT
Real Athletic Distance
-.044
.015
-.238
-2.922
.004
Real Behavior Distance
.000
.013
.000
.003
.997
Real Friend Distance
-.003
.013
-.026
-.259
.796
Real Job Distance
-.044
.017
-.239
-2.601
.010
Real Physical Distance
.016
.012
.118
1.352
.178
Real Romantic Distance
-.006
.015
-.042
-.412
.681
Real Scholastic Distance
.011
.013
.078
.871
.385
Real Social Distance
-.015
.018
-.078
-.803
.423
R = .392
R Square = .153
F = 3.309
Significance of F = .002.
The distance scores with significant relationships with athletic behaviors were real
athletic distance (Beta = -.24**) and real job distance (Beta = -.24*)
Ideal distance scores. No significant relationships were found in the regression analysis
for ideal distance scores on athletic behaviors (p > .05).

129
Table 4-34. Regression Analysis for Real and Ideal Self-Concept
Distance Scores on Athletic Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
3.600
.166
21.624
.000
Real athletic distance
-.063
.017
-.341
-3.675
.000
Ideal athletic distance
.052
.021
.262
2.433
.016
Real behavior distance
-.006
.014
-.045
-.438
.662
Ideal behavior distance
.011
.015
.076
.701
.484
Real friend distance
-.002
.017
-.011
-.088
.930
Ideal friend distance
-.011
.018
-.073
-.607
.545
Real job distance
-.041
.021
-.221
-1.978
.050
Ideal job distance
-.004
.021
-.027
-.187
.852
Real physical distance
.014
.013
.099
1.083
.281
Ideal physical distance
.016
.017
.091
.967
.335
Real romantic distance
-.009
.015
-.066
-.609
.544
Ideal romantic distance
-.004
.022
-.025
-.181
.857
Real scholastic distance
.008
.016
.056
.509
.611
Ideal scholastic distance
-.002
.015
-.015
-.120
.905
Real social distance
.003
.021
.013
.121
.904
Ideal social distance
-.021
.022
-.120
-.938
.350
R = .456
R Square .208
F 2.249
Significance of F = .006.
The distance scores with significant relationships with athletic behaviors were real
athletic distance (Beta = -.34**), real job distance (Beta = -.22*), and ideal athletic
distance (Beta = .26*, NOTE: This relationship is the opposite direction as predicted).
Note that the inclusion of real distance scores affected the relationships between ideal
distance scores and athletic behaviors.
Social Communication
Real distance scores. No significant relationships were found in the regression analysis
for real distance scores on social communication (p > .05).

130
Ideal distance scores. No significant relationships were found in the regression analysis
for ideal distance scores on social communication (p > .05).
Table 4-35. Regression Analysis for Real and Ideal Self-Concept
Distance Scores on Social Communication.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
TW~
,I7l
18.775
.0(i
Real athletic distance
-.050
.018
-.267
-2.817
.006
Ideal athletic distance
.038
.022
.191
1.736
.085
Real behavior distance
-.001
.015
-.007
-.069
.945
Ideal behavior distance
-.003
.016
-.024
-.217
.828
Real friend distance
.011
.018
.082
.638
.524
Ideal friend distance
.012
.018
.078
.641
.523
Real job distance
-.024
.021
-.125
-1.097
.274
Ideal job distance
.008
.022
.054
.370
.712
Real physical distance
-.002
.013
-.016
-.169
.866
Ideal physical distance
.029
.017
.162
1.692
.093
Real romantic distance
.004
.016
.026
.236
.813
Ideal romantic distance
-.055
.023
-.340
-2.386
.018
Real scholastic distance
-.007
.017
-.045
-.397
.692
Ideal scholastic distance
-.017
.016
-.138
-1.072
.286
Real social distance
.028
.021
.150
1.316
.190
Ideal social distance
-.004
.023
-.022
-.170
.865
R = .417
R Square = .174
F = 1.803
Significance of F = .036.
The distance scores with significant relationships with social communication were real
athletic distance (Beta = -.26**) and ideal romantic distance (Beta = -.33*). Note that the
inclusion of both real and ideal distance scores affected the relationships between both
real distance scores and ideal distance scores with social communication.

131
Consumer Behaviors
Real distance scores. No significant relationships were found in the regression analysis
for real distance scores on consumer behaviors (p > .05).
Ideal distance scores. No significant relationships were found in the regression analysis
for ideal distance scores on consumer behaviors (p > .05).
Real and ideal distance scores. No significant relationships were found in the regression
analysis for real and ideal distance scores on consumer behaviors (p > .05).
Overall Behaviors
Table 4-36. Regression Analysis for Real Self-Concept
Distance Scores on Overall Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
2.990
.115
25.920'
OTT
Real Athletic Distance
-.032
.012
-.223
-2.754
.007
Real Behavior Distance
-.005
.010
-.049
-.538
.592
Real Friend Distance
.003
.010
.030
.306
.760
Real Job Distance
-.035
.013
-.246
-2.681
.008
Real Physical Distance
.023
.009
.215
2.476
.014
Real Romantic Distance
-.007
.011
-.060
-.588
.557
Real Scholastic Distance
-.001
.010
-.011
-.126
.900
Real Social Distance
-.003
.014
-.018
-.185
.854
R = .391
R Square = .153
F = 3.323
Significance of F = .002.
The distance scores with significant relationships with overall behaviors were real athletic
distance (Beta = -.23**), real job distance (Beta = -.22*), and real physical distance (Beta
= .21 *, NOTE: This relationship is the opposite direction as predicted).

132
Ideal distance scores. No significant relationships were found in the regression analysis
for ideal distance scores on overall behaviors (p > .05).
Table 4-37. Regression Analysis for Real and Ideal Self-Concept
Distance Scores on Overall Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
w
.125
24.063
MT
Real athletic distance
-.044
.013
-.313
-3.419
.001
Ideal athletic distance
.034
.016
.222
2.100
.038
Real behavior distance
-.005
.011
-.047
-.465
.643
Ideal behavior distance
-.003
.012
-.024
-.225
.822
Real friend distance
.006
.013
.058
.465
.643
Ideal friend distance
.006
.013
.051
.429
.668
Real job distance
-.035
.016
-.246
-2.225
.028
Ideal job distance
.002
.016
.018
.126
.900
Real physical distance
.016
.010
.150
1.656
.100
Ideal physical distance
.019
.013
.136
1.469
.144
Real romantic distance
-.005
.012
-.047
-.445
.657
Ideal romantic distance
-.013
.017
-.108
-.792
.429
Real scholastic distance
.001
.012
.012
.114
.909
Ideal scholastic distance
-.013
.012
-.145
-1.167
.245
Real social distance
.005
.016
.035
.320
.750
Ideal social distance
-.013
.017
-.097
-.771
.442
R = .475
R Square = .225
F = 2.511
Significance of F = .002.
The distance scores with significant relationships with overall behaviors were real athletic
distance (Beta = -.32**), real job distance (Beta = -.22*), and ideal athletic distance
(Beta = .22*, NOTE: This relationship is the opposite direction as predicted). Note that
the inclusion of both real and ideal distance scores affected the relationships between
both real distance scores and ideal distance scores with overall behaviors.

133
Tattoo or Earring Modeling Behaviors
Real distance scores. No significant relationships were found in the regression analysis
for real distance scores on tattoo or earring modeling behaviors (p > .05).
Table 4-38. Regression for Ideal Self-Concept Distance Scores
on Tattoo or Earring Modeling Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
OTT
.112
11.905'
IT
Ideal athletic distance
.043
.016
.247
2.662
.009
Ideal behavior distance
-.006
.012
-.045
-.467
.642
Ideal friend distance
.019
.014
.145
1.348
.180
Ideal job distance
.022
.016
.173
1.413
.160
Ideal physical distance
.000
.014
-.003
-.029
.977
Ideal romantic distance
-.036
.019
-.255
-1.937
.055
Ideal scholastic distance
-.004
.012
-.040
-.355
.723
Ideal social distance
-.007
.018
-.048
-.402
.688
R = .337
R Square = .114
F = 2.338
Significance of F = .022.
The only distance score with a significant relationship with tattoo or earring modeling
behaviors was ideal athletic distance (Beta = .24**, NOTE: This relationship is the
opposite direction as predicted).

134
Table 4-39. Regression Analysis for Real and Ideal Self-Concept
Distance Scores on Tattoo or Earring Modeling Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
ns3
.149
9.293
,0(J(T
Real athletic distance
-.019
.015
-.119
-1.255
.212
Ideal athletic distance
.056
.019
.319
2.908
.004
Real behavior distance
-.001
.013
-.009
-.083
.934
Ideal behavior distance
.001
.014
.006
.058
.954
Real friend distance
.002
.016
.017
.132
.895
Ideal friend distance
.018
.016
.135
1.108
.270
Real job distance
.007
.018
.046
.404
.687
Ideal job distance
.021
.019
.160
1.095
.276
Real physical distance
-.025
.011
-.209
-2.226
.028
Ideal physical distance
.003
.015
.020
.205
.838
Real romantic distance
.004
.014
.035
.317
.752
Ideal romantic distance
-.044
.020
-.315
-2.231
.027
Real scholastic distance
.019
.014
.147
1.315
.191
Ideal scholastic distance
-.014
.014
-.130
-1.014
.312
Real social distance
.018
.019
.110
.978
.330
Ideal social distance
-.016
.020
i
o
-.805
.422
R = .410
R Square = .168
F = 1.741
Significance of F = .046.
The distance scores with significant relationships with tattoo or earring modeling
behaviors were real physical distance (Beta = -.20*), ideal athletic distance (Beta = .31**,
NOTE: This relationship is the opposite direction as predicted), and ideal romantic
distance (Beta = -.31 *). Note that the inclusion of both real and ideal distance scores
affected the relationships between both real distance scores and ideal distance scores with
tattoo or earring modeling behaviors.
Using Andr Modeling Behaviors
Real distance scores. No significant relationships were found in the regression analysis
for real distance scores on using Andr modeling behaviors (p > .05).

135
Ideal distance scores. No significant relationships were found in the regression analysis
for ideal distance scores on using Andr modeling behaviors (p > .05).
Real and ideal distance scores. No significant relationships were found in the regression
analysis for real and ideal distance scores on using Andr modeling behaviors (p > .05).
Summary Chart of Significant Results of Regression Analyses for HI and H2
The results of the above regression analyses are summarized in the following
table. A number in a cell indicates a significant relationship was found between that role
modeling index/variable and the corresponding distance score. The particular regressions
in which the significant relationships were found are explained by the key beneath the
table. In other words, the numbers indicate whether the relationship was found in the
regression including only real distance scores, the regression including only ideal distance
scores, or the regression including real and ideal distance scores.

136
Table 4-40. Significant Results of Regression Analyses for HI and H2.
DISTANCE
SCORE
Athletic
Beh
Social
Comm
Prosocial
Beh
Consumer
Beh
Overall
Beh
Tattoo
&
Earring
Andr
&
Drugs
Real Athletic
1,3
3
3
1,3
Real Behavior
Real Friend
Real Job
1,3
1
1,3
Real Physical
4,6
4
3
Real Romantic
Real Scholastic
Real Social
Ideal Athletic
6
6
5,6
Ideal Behavior
Ideal Friend
Ideal Job
Ideal Physical
Ideal Romantic
3
3
Ideal Scholastic
Ideal Social
1 = Significant negative relationship found in regression for real distance scores
2 = Significant negative relationship found in regression for ideal distance scores
3 = Significant negative relationship found in regression for real and ideal
distance scores
4 = Significant positive relationship found in regression for real distance scores
(opposite direction as predicted)
5 = Significant positive relationship found in regression for ideal distance scores
(opposite direction as predicted)
6 = Significant positive relationship found in regression for real and ideal distance
scores (opposite direction as predicted).
As indicated in the above table, the proximity between real athletic self-concept
and the heros athletic image and the proximity between real job competence self-concept
and the heros job competence have the most significant relationships with the role
modeling indexes. This is particularly true for athletic behaviors, prosocial behaviors,
and overall behaviors, indexes with which both athletic and job distance scores had
significant relationships. Given the relevance of athletic performance in rating the job
competence of sports heroes, these results indicate the importance of athletics, the athletic
self-concept, and the athletic accomplishment of a sports hero in determining role

137
modeling behaviors of that hero. Additionally, these results indicate the importance of
the distance between the real self-concept and the sports hero over the distance between
the ideal self and the sports hero in determining role modeling behaviors. In other words,
these adolescents were more likely to role model their sports heroes based on a close
proximity between their heroes and who they really perceive themselves to be, not on
who they would like to be. These adolescents indicated role modeling of athletes who
were like themselves, not athletes who represented their ideal. Of all the ideal distance
scores, only the distance between ideal romantic self-concept and the image of the hero in
romantic behaviors had a significant negative relationship with any of the role modeling
indexes/variables. These relationships came with indexes/variables dealing with
non-athletic behaviors (social communications, getting a tattoo or earring).
Two of the role modeling indexes/variables (consumer behaviors, taking Andr or
drugs) did not have significant relationships with any of the distance scores, real or ideal.
Therefore, subjects in this study indicated a close proximity between self-concept and the
image of the sports hero did not correlate with their role modeling intentions for these
behaviors. However, there is another potential explanation for these results. Both of
these dependent variables (consumer behaviors and taking Andr) had relatively low
mean scores (consumer behaviors = 2.27; Andr = 1.81) and low variances (consumer
behaviors = .93; Andr = 1.38). Low mean scores for a role modeling variable and low
variances for that variable would make it difficult to find significant relationships with
any of the distance scores. In other words, because most of the subjects reported that they
did not role model their sports heroes in these two areas and there was little variation in
these results across the subjects, congruity between self-concept and the image of the
sports hero would not be a significant predictor these two behaviors.

138
Summary of Results of HI and H2
Several general observations can be made about the results of these regression
analyses. First, all of the scores, even those associated with significant F scores, are
fairly low. While scores of less than .5 are often looked upon with skepticism, scores
as low as those found in the described regression analyses (ranging from 11 to .22) can
be considered marginal in terms of the relationships between the independent variables
and their dependent variable. In other words, while several significant effects were
found, it is questionable as to the amount of importance these results truly represent. If
nothing else, it can be stated with confidence that the independent variables (distance
scores) have limited relationships with the dependent variables (role modeling), and
therefore these distance scores, while contributing factors to role modeling, are not likely
the primary predictors for role modeling behaviors.
Second, two of the dependent variables, marketing behaviors and Andr use, were
not significantly related to any of the self-concept distance scores. Therefore, in this
study, these role modeling behaviors cannot be predicted from real or ideal distance
scores.
Third, more real distance scores than ideal distance scores were found to be
significantly predictive. In fact, the ideal distance score that most often had a significant
relationship with role modeling behaviors, ideal athletic distance, had a relationship in the
opposite direction predicted by the hypothesis, indicating that the smaller the distance
between the participants ideal self-concept and his assessment of the athletes athletic
abilities, the less likely he was to role model that sports heros behaviors. The only other
ideal distance score that had a significant relationship with any of the role modeling
indexes/variables is ideal romantic distance. Therefore, little if any support was found for
H2. This is further confirmed by the marginal impact the ideal distance scores have on
the regression analyses which include both real and ideal distance scores. In other words,

139
when real and ideal distance scores were placed in one regression to examine their
relative predictive values, real distance scores were identified as the stronger predictors of
role modeling. The distance between ideal self-concept and the image of the sports hero
apparently has little relationship with role modeling of sports heroes for these subjects.
Instead, attention should be turned to real distance scores and role modeling.
Fourth, the real distance scores that most frequently had significant relationships
with the role modeling indexes/variables were real athletic distance and real job distance.
Both of these distance scores relate to the athletic performance of sports heroes, and for
this study, these areas of self-concept have shown the greatest relationship with role
modeling behaviors. These two distance scores, real athletic distance and real job
distance, both were significantly related to prosocial behaviors, athletic behaviors, and
overall behaviors. Real physical distance also was significantly related to some role
modeling indexes, but this relationship was in the opposite direction predicted in H1 and
H2. Perhaps the close proximity between real physical self-concept and the physical
image of the sports hero indicates a high level of satisfaction with ones physical
appearance. Therefore, subjects with a high level of satisfaction might be less inclined to
try to improve their physical appearance and physical self-concept by modeling their
sports hero. However, this positive relationship between physical distance scores and
role modeling is in the opposite direction as predicted for this research and therefore is
not supportive of the hypotheses of this study.
Finally, in summary, this study shows limited support for HI and negligible, if
any, support for H2. For HI, the closer the subjects perceived themselves to be to their
sports heroes in two areas, athletics and job competence, the more likely these subjects
were to role model their sports heroes. The relationships between these two distance
scores, real athletic distance and real job distance, were found for prosocial behaviors,
athletic behaviors, and an overall measure of behaviors. The proximity between real
athletic self-concept and the athletic image of the sports hero (real athletic distance) also

140
was significantly related with role modeling social communications behaviors.
Additionally, real physical distance, or the subjects perceptions of how similar they were
physically to their sports hero, had a minimally significant relationship with role
modeling the behaviors of getting a tattoo or an earring like ones sports hero. The more
similar the subjects thought they looked physically to their sports hero, the more likely
they were to get a tattoo or earring like their sports hero. Given the ties such behaviors
have to external physical appearance, this relationship to real physical distance scores is
not surprising. However, for all the relationships found to support H1, the low Rz
statistics are notable, indicating that distance scores are a relatively weak predictor of role
modeling behaviors in this study.
Domain Importance and Distance Worth
Domain Importance Scores
For the tests of H3 and H4, the measured construct of domain importance was
used in addition to distance scores and role modeling indexes/variables. These domain
importance scores were collected from Harters Test of Domain Importance as part of the
Self-Concept Inventory. The results of the test of domain importance are detailed in the
following table. Domain importance scores are listed by domain in descending order,
from highest mean score (most important domain) to lowest mean score (least important
domain). (4 = very important to me; 1 = not important to me).

141
Table 4-41. Domain Importance of Self-Concept.
N
Mean
Std. Deviation
Scholastic
172
3.43TI"
7030~
Friendship
172
3.2820
.8060
Job competence
172
3.2209
.8275
Romantic
172
3.1599
.7661
Behavior
172
3.0640
.7327
Athletic
172
3.0029
.7975
Social
172
2.9884
.7294
Physical
172
2.9564
.7644
Minimum = 1; Maximum = 4.
Although there is not much variance among the means of all the domain
importance scores, the order of importance and respective mean scores should be noted.
The scholastic domain has the highest domain score (mean = 3.43), followed by the
domain for close friendship (mean = 3.28). The domain with the lowest domain
importance score is the domain of physical appearance (mean = 2.96). However, all
domain importance mean scores are fairly close to one another, and, given the relatively
large standard deviations for each domain importance score, there does not appear to be
an overwhelming difference between the importance of the individual domains.
Although not crucial to this study, it should be noted that mean scores for domain
importance ranged from 2.96 to 3.43, indicating that all domains of self-concept were
rated as relatively important to subjects in this study. These subjects were obviously
concerned with all areas of self-concept. For the purpose of further analysis, these
domain importance scores were all divided by four, reducing them to a scale ranging
from .25 to 1. An explanation and description of this process can be found in the
preceding Methods chapter. The reduced score means, which represent the same values
as the raw score means, are detailed in the following table. (1 = very important to me;
.25 = not important to me).

142
Table 4-42. Reduced Domain Importance.
N
Mean
Std. Deviation
Scholastic
nr

.17624
Friendship
172
.82049
.20149
Job competence
172
.80523
.20689
Romantic
172
.78997
.19152
Behavior
172
.76599
.18317
Athletic
172
.75073
.19937
Social
172
.74709
.18235
Physical
172
.73910
.19111
Vlinimum = .25;
Maximum
= 1.
Distance Worth Scores
While the domain importance scores in themselves are useful tools to better
understand adolescent behavior and self-concept, for this study, domain importance
scores were used along with distance scores to create a new construct. By incorporating
both distance scores for each domain and the importance of each of those domains, it was
hoped that this new construct, distance worth, would serve as a strong predictive factor
for the role modeling behaviors for this study. Distance worth scores were calculated for
both real self-concept and ideal self-concept across all eight domains, resulting in 16 new
variables for analysis.
To calculate distance worth scores, the individual distance scores were divided by
the normalized domain important scores. Accordingly, eight scores have been calculated
using real distance scores (real distance worth scores), and eight scores have been
calculated using ideal distance scores (ideal distance worth scores). These scores are the
independent variables in the regression analyses testing H3 and H4. The distance scores
were divided instead of multiplied by domain importance scores due to the negative
relationship proposed between distance scores and role modeling. This is in contrast to
the positive relationship proposed between domain importance and role modeling.
Therefore, to create this new construct (distance worth scores), the variables were divided

143
to invert the relationship between domain importance and role modeling. In doing so,
those distance scores with higher domain importance scores increase less than those with
lower domain importance scores. A further explanation and description of this process is
found in the preceding Methods chapter.
Real and ideal distance worth scores, listed by domain in descending order of the
domain mean scores, are detailed in the following table.
Table 4-43. Distance Worth Scores (Real and Ideal) of
Self-Concept to Sports Hero, Listed by Domain.
N
Mean
Std.
Deviation
Ideal Social Distance Worth
168
5.9853
9.2 m
Ideal Athletic Distance Worth
169
6.0334
9.7188
Ideal Friendship Distance Worth
171
7.2283
9.6769
Ideal Romantic Distance Worth
171
7.6037
9.9812
Ideal Physical Distance Worth
167
7.7704
11.1627
Ideal Job Distance Worth
168
8.1672
14.2562
Real Social Distance Worth
168
8.6726
8.7455
Real Job Distance Worth
168
9.2197
12.1188
Ideal Behavior Distance Worth
168
9.2567
11.6727
Real Athletic Distance Worth
169
9.5646
11.5895
Real Scholastic Distance Worth
170
9.5921
9.1738
Real Friendship Distance Worth
171
9.5953
12.6077
Ideal Scholastic Distance Worth
170
10.4391
13.3187
Real Behavior Distance Worth
168
10.9579
12.7936
Real Romantic Distance Worth
171
11.7015
13.4617
Real Physical Distance Worth
168
12.3672
15.6517
Results for Hypothesis HI and Hypothesis IV
Procedures for testing H3 and H4 were similar to the testing of HI and H2, and
the description of these procedures is similar to that for HI and H2.
The third hypothesis (H3) predicted a negative relationship between real distance
worth and role modeling intentions. Therefore, it was expected to find a negative and

144
significant relationship between the individual indexes/variables of role modeling and the
real distance worth scores. To test this, a separate regression analysis was done for each
of the four role modeling indexes, the overall role modeling index, and the individual
variables TATT (getting a tattoo or earring like ones hero) and ANDRO (using Andr or
drugs like ones hero). For each of these regression analyses, the role modeling
index/variable served as the dependent variable, while each of the eight real distance
worth scores for the eight domains comprised the independent variables. Therefore,
seven separate regression analyses were done to test H3.
The fourth hypothesis (H4) predicted a negative relationship between ideal
distance worth and role modeling intentions. The regression analyses completed to test
this hypothesis were identical to those done for H3, except ideal distance worth scores
were used as the independent variables instead of real distance worth scores.
A third group of regression analyses were done that included both real distance
worth scores and ideal distance worth scores. These analyses, which address both H3 and
H4, were the same as those done for HI and H2 using all 16 real and ideal distance
scores, except that the 16 distance worth scores (eight real distance worth, eight ideal
distance worth) were used as the independent variables for each of the seven regression
equations. For each equation, one of the four role modeling indexes, the overall role
modeling index, or the role modeling variable ANDRO or TATT served as the
independent variable. As with the regression analyses done for HI and H2, by including
all distance worth scores as independent variables, these regression analyses helped to
determine whether real or ideal distance worth scores were stronger predictors of each
individual role modeling index/behavior. Additionally, these regression analyses should
create a more predictive regression equation for each of the role modeling
indexes/variables by accounting for a larger percentage of the variance of each of these
independent variables. In other words, this should increase the R^ of each regression
analysis. Finally, because the exclusion of important predictor variables can drastically

145
affect the results of regression analysis, it was hoped that including all the distance worth
scores (real and ideal) would increase the prospect of achieving statistically significant
results (Licht, 1995).
The results of all of the regression analyses for these two hypotheses are detailed
in the following tables. The tables are organized by dependent variable (role modeling
index/variable), not by hypotheses. Therefore, all three regression analyses for each of
the seven dependent variables are grouped together. A more general discussion of the
individual hypotheses follows these results. For each of the significant regression
analyses that produced a significant F-value (p less than .05), one table details both the
overall significance of the regression (including the R and values and the F-value) and
the significance of the separate independent variables (distance worth scores). Following
each table is a description of the significant relationship(s) between the independent
variable(s) and the role modeling index/variable. This includes the Beta coefficient for
each relationship. (For each relationship, indicates significance at p less than .05, and
** indicates significance at p less than .01.)
For those regressions whose F Statistic is not significant at the .05 level, only a
brief text description of the analysis is given. In the proceeding section, for each role
modeling index/variable examined in this research, the regression analysis with only real
distance worth scores is explained first, followed by the regression analysis with only
ideal distance worth scores. The regression analysis with both real and ideal distance
worth scores is described last for each index/behavior.
Following the tables of the separate regression analyses, one overall table
summarizes the results of these analyses. This table indicates which role modeling
indexes/variables were significantly related to the distance worth scores and which
particular distance worth scores had significant relationships with each of these
indexes/variables.

146
Prosocial Behaviors
Table 4-44. Regression Analysis for Real Self-Worth
Distance Scores on Prosocial Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
TTW
.143
19.505
MT
Real Athletic Dist Worth
-.019
.009
-.177
-2.043
.043
Real Behavior Dist Worth
-.007
.008
-.086
-.945
.346
Real Friend Dist Worth
.016
.008
.189
1.921
.057
Real Job Dist Worth
-.016
.011
-.140
-1.413
.160
Real Physical Dist Worth
.012
.006
.174
1.892
.061
Real Romantic Dist Worth
-.010
.008
-.120
-1.184
.238
Real Scholastic Dist Worth
-.009
.012
-.078
-.779
.437
Real Social Dist Worth
-.009
.013
-.074
-.735
.463
R = .347
R Square = .121
F = 2.488
Significance of F = .015.
The only distance worth score with a significant relationship with prosocial behaviors
was real athletic distance worth (Beta = -. 17*).
Ideal distance worth. No significant relationships were found in the regression analysis
for ideal distance worth scores on prosocial behaviors (p > .05).
Real and ideal distance worth. No significant relationships were found in the regression
analysis for real and ideal distance worth scores on prosocial behaviors (p > .05). Note
that the inclusion of ideal distance worth scores eliminated the significant relationship
between real athletic distance worth scores and prosocial behaviors.

147
Athletic Behaviors
Table 4-45. Regression Analysis of Real Self-Worth
Distance Scores on Athletic Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
M-
.136
76.769"
ror
Real Athletic Dist Worth
-.025
.009
-.252
-2.951
.004
Real Behavior Dist Worth
.000
.007
-.001
-.013
.989
Real Friend Dist Worth
-.001
.008
-.018
-.185
.853
Real Job Dist Worth
-.015
.010
-.140
-1.421
.158
Real Physical Dist Worth
.005
.006
.081
.885
.378
Real Romantic Dist Worth
-.005
.008
-.062
-.618
.538
Real Scholastic Dist Worth
.004
.011
.041
.421
.674
Real Social Dist Worth
-.008
.012
-.069
-.688
.492
R = .351
R Square = .123
F = 2.561
Significance of F = .012.
The only distance worth score with a significant relationship with athletic behaviors was
real athletic distance worth (Beta = -.25**).
Ideal distance worth. No significant relationships were found in the regression analysis
for ideal distance worth scores on athletic behaviors (p > .05).

148
Table 4-46. Regression Analysis for Real and Ideal
Self-Worth Distance Scores on Athletic Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
'' Tiy
.143
25.232
MT
Real Athletic Dist Worth
-.039
.010
-.393
-4.041
.000
Real Behavior Dist Worth
.000
.008
.000
-.003
.998
Real Friend Dist Worth
-.001
.010
-.013
-.109
.914
Real Job Dist Worth
-.017
.014
-.166
-1.246
.215
Real Physical Dist Worth
.002
.007
.023
.220
.826
Real Romantic Dist Worth
-.002
.010
-.029
-.237
.813
Real Scholastic Dist Worth
.007
.013
.064
.533
.595
Real Social Dist Worth
-.004
.013
-.034
-.307
.759
Ideal Athletic Dist Worth
.029
.012
.289
2.342
.021
Ideal Behavior Dist Worth
.005
.009
.060
.542
.589
Ideal Friend Dist Worth
-.010
.013
-.094
-.748
.456
Ideal Job Dist Worth
.004
.013
.043
.277
.782
Ideal Physical Dist Worth
.005
.010
.055
.497
.620
Ideal Romantic Dist Worth
-.010
.015
-.095
-.661
.509
Ideal Scholastic Dist Worth
-.005
.009
-.066
-.545
.587
Ideal Social Dist Worth
-.003
.013
-.026
-.216
.830
R = .434
R Square = .188
F = 1.988
Significance of F = .018.
The distance worth scores with significant relationships with athletic behaviors were real
athletic distance worth (Beta = -.39**) and ideal athletic distance worth (Beta = .28*,
NOTE: This relationship is the opposite direction as predicted). Note that the inclusion
of real distance worth scores affected the relationships between ideal distance worth
scores and athletic behaviors.
Social Communication
Real distance worth. No significant relationships were found in the regression analysis
for real distance worth scores on social communication (p > .05).

149
Table 4-47. Regression Analysis for Ideal Self-Worth
Distance Scores on Social Communication.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
rnr
1W
26.007
7J"
Ideal Athletic Dist Worth
-.005
.011
-.053
-.495
.622
Ideal Behavior Dist Worth
-.005
.008
-.054
-.560
.576
Ideal Friend Dist Worth
.013
.011
.120
1.148
.253
Ideal Job Dist Worth
.005
.010
.062
.493
.623
Ideal Physical Dist Worth
.004
.008
.045
.480
.632
Ideal Romantic Dist Worth
-.031
.012
-.302
-2.501
.013
Ideal Scholastic Dist Worth
-.007
.008
-.098
-.956
.340
Ideal Social Dist Worth
-.002
.012
-.020
-.181
.856
R = .318
R Square = .101
F = 2.042
Significance of F = .045.
The only distance worth score with a significant relationship with social communication
was ideal romantic distance worth (Beta = .-.30*).

150
Table 4-48. Regression Analysis for Real and Ideal Self-Worth
Distance Scores on Social Communication.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
3.240
.144
22.551
MT
Real Athletic Dist Worth
-.033
.010
-.324
-3.321
.001
Real Behavior Dist Worth
.005
.008
.064
.607
.545
Real Friend Dist Worth
.009
.010
.115
.931
.354
Real Job Dist Worth
.000
.014
-.004
-.027
.979
Real Physical Dist Worth
-.004
.007
-.061
-.580
.563
Real Romantic Dist Worth
-.002
.010
-.021
-.173
.863
Real Scholastic Dist Worth
-.002
.014
-.018
-.144
.886
Real Social Dist Worth
.010
.013
.082
.740
.461
Ideal Athletic Dist Worth
.011
.013
.112
.904
.367
Ideal Behavior Dist Worth
-.005
.009
-.059
-.531
.596
Ideal Friend Dist Worth
.007
.013
.064
.510
.611
Ideal Job Dist Worth
.006
.013
.076
.489
.626
Ideal Physical Dist Worth
.011
.010
.119
1.083
.281
Ideal Romantic Dist Worth
-.036
.015
-.354
-2.444
.016
Ideal Scholastic Dist Worth
-.008
.009
-.105
-.871
.385
Ideal Social Dist Worth
-.006
.013
-.055
-.452
.652
R = .318
R Square = .101
F = 2.042
Significance of F = .045.
The distance worth scores with significant relationships with social communication were
real athletic distance worth (Beta = -.32**) and ideal romantic distance worth (Beta =
-.35*). Note that the inclusion of ideal distance worth scores affected the relationships
between real distance worth scores and social communication.
Consumer Behaviors
Real distance worth. No significant relationships were found in the regression analysis
for real distance worth scores on consumer behaviors (p > .05).
Ideal distance worth. No significant relationships were found in the regression analysis
for ideal distance worth scores on consumer behaviors (p > .05).

151
Real and ideal distance worth. No significant relationships were found in the regression
analysis for real and ideal distance worth scores on consumer behaviors (p > .05).
Overall Behaviors
Table 4-49. Regression Analysis for Real Self-Worth
Distance Scores on Overall Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
nr
.104
28.843
MT
Real Athletic Dist Worth
-.020
.007
-.262
-3.102
.002
Real Behavior Dist Worth
-.003
.005
-.048
-.542
.588
Real Friend Dist Worth
.005
.006
.085
.880
.380
Real Job Dist Worth
-.012
.008
-.144
-1.485
.140
Real Physical Dist Worth
.009
.005
.166
1.839
.068
Real Romantic Dist Worth
-.006
.006
-.097
-.977
.330
Real Scholastic Dist Worth
-.003
.008
-.039
-.397
.692
Real Social Dist Worth
-.005
.009
-.056
-.569
.570
R = .372
R Square = .138
F = 2.947
Significance of F = .004.
The only distance worth score with a significant relationship with overall behaviors was
real athletic distance worth (Beta = -.28**).
Ideal distance worth. No significant relationships were found in the regression analysis
for ideal distance worth scores on overall behaviors (p > .05).

152
Table 4-50. Regression Analysis for Real and Ideal
Self-Worth Distance Scores on Overall Behaviors.
Unstandardized
Standardized
Coefficients
Coefficients
B
Std. Error
Beta
t
Sig.
(Constant)
JMT
TTfiT
28.073
MT
Real Athletic Dist Worth
-.028
.007
-.364
-3.777
.000
Real Behavior Dist Worth
.002
.006
.035
.336
.737
Real Friend Dist Worth
.005
.008
.080
.659
.511
Real Job Dist Worth
-.013
.011
-.157
-1.186
.238
Real Physical Dist Worth
.004
.005
.080
.775
.440
Real Romantic Dist Worth
-.002
.007
-.037
-.309
.758
Real Scholastic Dist Worth
.001
.010
.008
.065
.948
Real Social Dist Worth
-.004
.010
-.043
-.394
.694
Ideal Athletic Dist Worth
.013
.009
.173
1.418
.159
Ideal Behavior Dist Worth
-.004
.007
-.057
-.524
.601
Ideal Friend Dist Worth
.002
.010
.026
.214
.831
Ideal Job Dist Worth
.003
.010
.055
.353
.724
Ideal Physical Dist Worth
.005
.007
.073
.672
.503
Ideal Romantic Dist Worth
-.013
.011
-.166
-1.165
.246
Ideal Scholastic Dist Worth
-.007
.007
-.123
-1.027
.306
Ideal Social Dist Worth
-.002
.010
-.023
-.190
.850
R = .447
R Square = .199
F = 2.149
Significance of F = .009.
The only distance worth score with a significant relationship with overall behaviors was
real athletic distance worth (Beta = -.38**). Note that no change resulted from the
inclusion of ideal distance worth scores.
Tattoo or Earring Modeling Behaviors
Real distance worth. No significant relationships were found in the regression analysis
for real distance worth scores on tattoo or earring modeling behaviors (p > .05).

Table 4-51. Regression Analysis for Ideal Self-Worth
Distance Scores on Tattoo or Earring Modeling Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant]
rw
TT"
12.999
MT
Ideal Athletic Dist Worth
.018
.009
.203
1.919
.057
Ideal Behavior Dist Worth
-.002
.007
-.024
-.255
.799
Ideal Friend Dist Worth
.020
.010
.216
2.101
.037
Ideal Job Dist Worth
.010
.009
.134 _
1.078
.283
Ideal Physical Dist Worth
-.002
.007
-.026 <
-.286
.776
Ideal Romantic Dist Worth
-.009
.010
-.096
-.816
.416
Ideal Scholastic Dist Worth
-.005
.006
-.076
-.758
.450
Ideal Social Dist Worth
-.016
.010
-.169
-1.596
.113
R = .447
R Square = .199
F = 2.149
Significance of F = .009.
The only distance worth score with a significant relationship with tattoo or earring
modeling behaviors was ideal friend distance worth (Beta = .21 *, NOTE: This
relationship is the opposite direction as predicted).

154
Table 4-52. Regression Analysis for Real and Ideal Self-Worth
Distance Scores on Tattoo or Earring Modeling Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
mr
.124
10.794
MT
Real Athletic Dist Worth
-.012
.008
-.133
-1.388
.167
Real Behavior Dist Worth
.004
.007
.056
.548
.585
Real Friend Dist Worth
.000
.009
-.001
-.012
.990
Real Job Dist Worth
.006
.012
.063
.480
.632
Real Physical Dist Worth
-.016
.006
-.269
-2.617
.010
Real Romantic Dist Worth
.018
.008
.257
2.152
.033
Real Scholastic Dist Worth
.009
.011
.101
.850
.397
Real Social Dist Worth
-.006
.012
-.058
-.530
.597
Ideal Athletic Dist Worth
.025
.011
.287
2.359
.020
Ideal Behavior Dist Worth
-.003
.008
-.036
-.330
.742
Ideal Friend Dist Worth
.023
.012
.247
2.004
.047
Ideal Job Dist Worth
.006
.011
.087
.563
.574
Ideal Physical Dist Worth
.006
.008
.079
.735
.464
Ideal Romantic Dist Worth
-.026
.013
-.288
-2.030
.044
Ideal Scholastic Dist Worth
-.007
.008
-.106
-.890
.375
Ideal Social Dist Worth
-.012
.011
-.127
-1.060
.291
R = .455
R Square = .207
F = 2.247
Significance of F = .006.
The distance worth scores with significant relationships with tattoo or earring modeling
behaviors were real physical distance worth (Beta = -.26**), real romantic distance worth
(Beta = .25*, NOTE: This relationship is the opposite direction as predicted), ideal
athletic distance worth (Beta = .28*, NOTE: This relationship is the opposite direction as
predicted), ideal friend distance worth (Beta = .24*, NOTE: This relationship is the
opposite direction as predicted), and ideal romantic distance worth (Beta = .28*, NOTE:
This relationship is the opposite direction as predicted). Note that the inclusion of both
real and ideal distance worth scores affected the relationships between both real distance
worth scores and ideal distance worth scores with tattoo or earring modeling behaviors.

155
Using Andr Modeling Behaviors
Real distance worth. No significant relationships were found in the regression analysis
for real distance worth scores on using Andr modeling behaviors (p > .05).
ideal distance worth. No significant relationships were found in the regression analysis
for ideal distance worth scores on using Andr modeling behaviors (p > .05).
Real and ideal distance worth. No significant relationships were found in the regression
analysis for real and ideal distance worth scores on using Andr modeling behaviors
(p > .05).
Summary Chart of Significant Results of Regression Analyses for H3 and H4
The results of the above regression analyses are summarized in the following
table. A number in a cell indicates a significant relationship was found between that role
modeling index/variable and the corresponding distance worth score. The particular
regressions in which the significant relationships were found are explained by the key
beneath the table. In other words, the numbers indicate whether the relationship was
found in the regression including only real distance worth scores, the regression including
only ideal worth distance scores, or the regression including both real and ideal distance
worth scores.

156
Table 4-53. Significant Results of Regression Analyses for H3 and H4.
DISTANCE
WORTH SCORE
Athletic
Beh
Social
Comm
Prosocial
Beh
Consumer
Beh
Overall
Beh
Tattoo
&
Earring
Andr
&
Drugs
Real Athletic
1,3
3
1,3
1,3
Real Behavior
Real Friend
Real Job
Real Physical
3
Real Romantic
6
Real Scholastic
Real Social
Ideal Athletic
6
6
Ideal Behavior
Ideal Friend
5,6
Ideal Job
Ideal Physical
Ideal Romantic
2,3
6
Ideal Scholastic
Ideal Social
1 = Significant relationship found in regression for real distance worth score
2 = Significant relationship found in regression for ideal distance worth scores
3 = Significant relationship found in regression for real and ideal distance worth
scores
4 = Significant positive relationship found in regression for real distance worth
scores (opposite direction as predicted)
5 = Significant positive relationship found in regression for ideal distance worth
scores (opposite direction as predicted)
6 = Significant positive relationship found in regression for real and ideal distance
worth scores (opposite direction as predicted).
As indicated in the table above, the results for H3 and H4 are similar to those for
HI and H2. However, fewer significant negative relationships were found between role
modeling behaviors and distance worth scores (H3 and H4) than were found between role
modeling and distance scores (HI and H2). Because distance worth scores combined the
distance between self-concept and the image of the sports hero for a domain of
self-concept (distance scores) with the importance of that domain (domain importance
score), these results indicate the incorporation of domain importance scores actually
weakened the relationships between distance scores and role modeling. In other words,

157
just because a domain of self-concept was important did not make subjects more likely to
role model their sports heroes based on self-concept congruity with their sports hero in
that domain. Domain importance does not appear to be a predictor of role modeling a
sports hero, and the incorporation of distance scores did not add to the predictive nature
of the distance scores between self-concept and the image of the sports hero.
More real distance worth scores than ideal distance worth scores had significant
relationships with role modeling behaviors. As with the results for HI, real athletic
distance worth had the greatest number of significant relationships with role modeling.
Real physical distance worth had a significant relationship with the role modeling
variable for getting a tattoo or earring like ones hero, a non-athletic role modeling
behavior and an obvious link with the distance worth score for physical appearance.
Unlike HI, job distance worth was not significantly related with role modeling.
For H4, ideal romantic distance worth was the only ideal distance worth score
with a significant negative relationship with role modeling. This relationship came with
the index for role modeling social communication. Three domains of ideal distance
worth actually had positive relationships with some role modeling behaviors, the opposite
direction as predicted in by H4. As with the results for H2, which found little support for
significant negative relationships between role modeling and ideal distance scores, little
support was found for any significant negative relationship between role modeling and
ideal distance worth scores. Therefore, the incorporation of domain importance scores
did not increase the predictive nature of the congruity between the ideal self-concept and
the image of the sports hero. As indicated by the above chart, real distance worth scores
are more predictive of role modeling behaviors than ideal distance worth scores.

158
Summary of Results of H3 and H4
Again, several clarifications can be made about the data analysis related to these
hypotheses. First, the support for both of these hypotheses is limited. This is particularly
true for H4 (ideal distance worth scores). Of all the regression analyses, only ideal
romantic distance worth was found to have a significant negative relationship with any of
the role modeling behaviors (social communication), and this relationship is limited.
Additionally, three ideal distance worth scores had significant relationships in the
opposite direction as predicted. Therefore, it can be concluded that ideal distance worth
scores were not predictive of role modeling behaviors for the subjects in this study as
predicted.
Greater support was found for H3 (real distance worth scores); selected real
distance worth scores were significantly related to four role modeling indexes (prosocial
behaviors, athletic behaviors, social communication, and overall behaviors) and one
variable (getting a tattoo or earring like ones sports hero). However, of all eight domains
of real distance worth, only two real distance worth scores, real athletic distance worth
and real physical distance worth, had significant negative relationships with any of these
indexes/variable. Distance worth scores from the other domains were not significantly
related to role modeling behaviors. These results are similar to those for H1, which did
not incorporate domain importance. The most notable differences between the results for
the two hypotheses came in the domain of job competence. Unlike real job distance
scores, real job distance worth scores were not significantly related to any of the role
modeling indexes/behaviors. This would indicate that adding measures of the domain
importance of job competence did not help predict role modeling.
Placed into greater perspective, few significant relationships were found for all the
real distance worth scores. Not surprisingly, the limited support for H3 is found almost

159
exclusively with real athletic distance worth, a domain commonly associated with sports
heroes.
Perhaps more importantly, the use of domain importance did not add to the
predictive nature of the distance between self-concept and the image of the sports hero.
The results for H3 were similar to those for HI, indicating that domain importance did
not help to predict role modeling behaviors in this study. Therefore, study conclusions
will instead focus on the predictive nature of the congruity between self-concept and the
image of the sports hero.
Summary of Results
Limited support was found for HI and H3, indicating that congruity between real
self-concept and the image of the sports hero may be a factor in determining role
modeling of sports heroes for the participants of this study. This has been supported for
only a few domains of self-concept, and the relationships with the stated
indexes/variables of role modeling behaviors are fairly weak. Therefore, real self-concept
scores can be considered only limited factors in predicting role modeling behaviors of
sports heroes.
Almost no support was found for H2 and H4, indicating that congruity between
ideal self-concept and the image of the sports hero is not likely a factor in determining
role modeling of sports heroes for the participants of this study. While a limited number
of ideal distance scores and ideal distance worth scores had significant relationships with
a limited number of role modeling indexes/variables, there were not enough significant
relationships to consider these two constructs as statistically significant predictors of
behavioral role modeling of sports heroes.
Finally, the incorporation of domain importance scores did little to help predict
role modeling of sports heroes. While it was predicted that incorporating domain

160
importance scores with distance scores would create more accurate predictors of role
modeling, this did not appear true for this research, as evidenced by the lack of significant
results found for H3 and H4 as compared to HI and H2.
The final chapter, Conclusions, addresses the stated results, discusses the
importance and relevance of these results, and explains potential reasons these results
were found. Additionally, weaknesses of the study and recommendations for future
research are addressed.

CHAPTER 5
CONCLUSIONS
In drawing conclusions about the results of this research, the following topics are
addressed: inferences made from the descriptive statistics, including reported images of
sports heroes, adolescent choice of sports heroes, and reported role modeling statistics;
results of HI and H2 and the importance of real and ideal distance scores; results of H3
and H4 and the importance of real and ideal distance worth scores; and study limitations
and recommendations for future research.
Descriptive Statistics
Although the analyses of descriptive statistics do not pertain directly to the
hypotheses of this research, this information still should be reviewed to gain insight into
adolescents choices of sports heroes and the areas in which these adolescents are most
likely to model their sports heroes. This provides a greater understanding of American
sports heroes and their relationship to the adolescent audience.
Image of the Sports Hero
Due to the negative information available about many popular athletes in some
areas of their lives and the potential impact such information might have on fans views
of these popular athletes, it is particularly interesting to review the respondents reported
images of their sports heroes. While the rated mean image of sports heroes was fairly
high in all eight domains, ranging from 2.90 to 3.32 on a scale of 1 to 4, the ordered rank
of such scores may give greater insight into adolescent choices of sports heroes. The
161

Pages
are
misnumbered
following
this
insert

161
highest two rated domains were the social domain (3.32) and the athletic domain (3.31),
while the lowest two were the behavior domain (3.02) and the scholastic domain (2.90).
This is not surprising, given the frequent reports of negative behaviors of athletes,
collegiate athletic/academic scandals, and similar negative behaviors. It is interesting to
note, however, that the social domain of athletes rated as high as the athletic domain,
indicating that these subjects viewed sports heroes as more than merely supreme athletes.
The high domain scores in the romantic domain (3.19) and the friend domain (3.13)
reinforce this concept.
The adolescents from this study reported a fairly high opinion of their sports
heroes in a variety of areas beyond athletics. One potential explanation for this could be
the invasive nature of media coverage of popular athletes, allowing sports fans to learn
more about their sports heroes in a variety of different areas outside of actual athletic
performance. Because the mean scores for each of the eight domains were fairly close,
few general statements can be made about the relative rankings of the domain scores.
However, the fact that the behavior and scholastic domains had the lowest mean scores
does support a popular notion that many American athletes are portrayed as less than
exemplary students and often display negative behaviors. In other words, while these
adolescents rated their sports heroes as competent in all areas, they did seemingly indicate
that their sports heroes are not as competent in behavioral and scholastic areas as they are
in other areas.
Relationship between Hero Image and Self-Concept
A more informative means of viewing the rated image scores of the sports hero is
to examine the calculated distance scores, which examines the relative relationships with
individual self-concept. However, the calculated distance scores do not reflect negative
values, and therefore do not tell us whether the image of the sports hero is less than or
greater than the self-image in a particular domain. Distance scores only show the

162
absolute value of the distance. Therefore, it is first important to review whether the
sports heroes images were rated as higher or lower than individual self-concept.
For seven of the eight domains, the image of the sports hero fell in-between the
real self-concept and the ideal self-concept. In other words, these adolescents viewed
their respective sports heroes as more competent or having a better image than
themselves, but the heroes did not quite measure up to their ideal self. The hero did not
represent an ideal, but he did represent a step in that direction from the real self. Given
the high image of these athletes in juxtaposition to the real self-concepts, sports heroes do
represent logical choices as role models for these adolescents and a logical means for
moving their real self toward their ideal self.
The only domain in which the image of the sports hero fell below the real
self-concept is the scholastic domain. Obviously, these adolescents did not perceive their
sports heroes as ideal students or even of equal competence as themselves in the
classroom. This could be attributed to the strong educational background of the majority
of the subjects for this study, as most of the surveyed adolescents attended a private
college preparatory school or a selective public school. It would be expected that
participants would not look to their sports heroes as role models in the scholastic area.
Interestingly, this expectation was not fully met, since the role modeling variable STUDY
(studying harder because of ones sports hero) did receive a fairly high mean score (2.82
on a scale from 1 to 4), indicating subjects did consider their sports hero as a potential
role model for academic affairs. Conversely, the role modeling variable COL (going to
the same college as ones sports hero) received a low mean score (1.62), the third lowest
score of all role modeling variables.
It is clear that the image of the selected sports heroes typically fell in-between the
real self-concept and the ideal-self concept. Using the calculated distance scores, these
relationships can be viewed in terms of the actual distances for each domain. This
indicates exactly which particular domain of self-concept, for both real and ideal self, was

163
closest in distance to the image of the sports hero. In other words, this explains in which
domains the adolescent felt closest to his particular sports hero. From this information,
one can gain insight into the perceived similarity of todays sports heroes to adolescent
fans and perhaps even understand why certain types of popular athletes are chosen as
sports heroes.
Not surprisingly, the ideal athletic distance score had the lowest mean score
(3.70). In other words, these subjects viewed their sports heroes as closer to their ideal
self in athletics than for any other domain of self-concept, real or ideal. According to
these results, the sports hero, first and foremost, embodies ideal athletic abilities,
confirming the notion that sports heroes must be supreme athletes above all else (good
looks, good behavior, etc.) to be considered sports heroes. Although the media may
stress a variety of characteristics of popular athletes, these adolescents placed a premium
on athletic performance. This emphasis should help to further define the current state of
the evolving modem American sports hero.
The relative order of the 16 calculated distance scores (eight real distance, eight
ideal distance) gives insight into the qualities of the modem American sports hero. Many
theorists have suggested that the sports hero has lost his luster, largely due to invasive
media coverage, and this heroic figure no longer represents societys ideal. However,
such a sentiment is not wholly supported by this research. In fact, according to the
calculated distance scores, the sports heroes chosen for this research were more closely
aligned to the ideal self-concept than the real self-concept. Of the 16 distance scores, the
six scores with the smallest mean score (lower score equaling closer proximity between
hero image and self-concept) were ideal distance scores. The ideal distance scores for
athletic ability, physical appearance, and job competence, all domains related to athletic
and physical characteristics, were three of the four smallest distance scores. Such
proximity between the ideal self and the sports hero in these domains was expected due to
the athletic excellence and the physical appearance of the sports hero achieved from years

164
of training and competition. However, the low ideal distance scores in the social,
friendship, and romantic domains also indicate the ideal perceptions of sports heroes in
non-athletic domains. The media, with its focus on the personalities and lifestyles of
popular athletes, likely have contributed to such perceptions.
The two ideal distance scores that fell furthest down the list of the 16 distance
scores were the ideal behavior distance score, ranked tenth (6.24), and the ideal scholastic
distance score, ranked last (8.09). As addressed earlier, this indicates that these
adolescents did not necessarily choose sports heroes closely representing their ideal
academic selves or, to a slightly lesser extent, their ideal behavioral selves. As many
popular athletes are portrayed as poor students (dropping out of college, cheating on
exams, etc.) and often engage in questionable behaviors, such results reflect the modem
media portrayal of popular athletes. These adolescents appeared able to choose a sports
hero who epitomizes less than ideal behavioral and academic qualities yet still displays
athletic excellence and ideal social skills.
In general, the real distance scores were larger than the ideal distance scores,
indicating these adolescents perceived their sports heroes as closer to who they would
ideally like to be than who they actually perceive themselves to be. As with the ideal
distance scores, the sports hero was closest to the real self-concept in the athletic and
social domains (distance scores of 5.89 and 6.01, respectively), and, as with the ideal self,
close proximity from the real self to the sports hero in these two domains appear most
important in the selection of a sports hero. The real physical distance score and the real
romantic distance score ranked as the last two distance scores (the highest mean scores).
Upon investigation of the results of the tests for real self-concept, these high mean scores
are likely a reflection of a low reported real self-concept in these two domains by the
subjects of this study. These subjects reported relatively low opinions of their appearance
and romantic success compared to their opinions of themselves in other domains.
Because of a low reported self-concept for these subjects in the physical and romantic

165
domains, the distance scores for these domains were larger than the distance scores in
other domains.
In summary, the results have indicated that adolescents have a fairly high opinion
of their sports heroes, according to the Likert scale measurement, consistent among most
of the eight domains. While many athletes have been scrutinized for negative behaviors,
this was not reflected by the reported image scores for sports heroes in this study. These
reported images typically fell between the real self-concept and ideal self-concept for all
domains except the scholastic domain, in which the image of the hero fell below the real
self-concept. Therefore, in all but this one domain, these sports heroes represented a step
from the real self towards the ideal self. Further, when examining the calculated distance
scores between self-concept and the image of the sports hero, it was found that these
sports heroes were closer to the ideal self-concept than the real self-concept in all but two
domains of self-concept. Therefore, popular media images of sports heroes that stress
supreme athletic accomplishment and high social skills, but poor behavior and low
academic accomplishment, seem to have been mirrored in the variations in the image
scores and calculated distance scores from this study.
Role Modeling
A more drastic range of scores can be found amongst the individual role modeling
variable mean scores than for the mean scores for the domains of the image of the sports
hero. These differences in mean scores for different role modeling variables indicate that
these adolescents did choose the particular areas in which they plan to model their sports
heroes. Instead of simply modeling all behaviors of their sports heroes at the same rate,
adolescents may instead consider the individual behaviors, good and bad, and decide
which they intend to model. Additionally, a high variance for most of the role modeling
variables supports the notion that individual adolescents make individual choices about
their intentions to role model their particular sports hero, and not all adolescents will
model their sports heroes in the same areas to the same extent.

166
In reviewing the individual role modeling variables, several worthwhile
observations should be noted. First, the range on a 5-point Likert scale from the lowest
rated role modeling variable (getting a tattoo or earring like ones hero = 1.44) to the
highest rated role modeling variable (train harder because of ones hero = 3.84) is
considerable. While these adolescents appear very likely to model the training techniques
of their sports hero, they are very unlikely to get a tattoo like their sports hero. It is
similarly unlikely that these adolescents will get the same hair style as their sports hero
(HAIR = 1.46).
In more general terms, five of the six variables with the highest mean scores deal
with on-the-court behaviors (training, trash talking on the court, playing a particular
position, training techniques, and athletic maneuvers on the court). Besides the variable
dealing with trash talking (TRASH = 3.27), all of these variables deal specifically with
athletic behaviors and training. This indicates that these subjects value the athletic
actions of their heroes above other behaviors, and often they indicated intentions to model
these actions. Such would be consistent with the reported ratings of the image of their
sports heroes, where these subjects ranked the athletic and job performance domains as
two of the highest rated domains of this image. In other words, these adolescents seemed
more interested with the athletic components and qualities of their sports heroes than with
the many other facets of their sports heroes.
Conversely, many of the variables with the lowest mean scores deal either with
negative behaviors (getting a tattoo, hairstyle, taking Andr) or major life decisions
(choosing a college). It is predictable that these adolescents would not want to model
their sports heros college choice (COL = 1.62) given their relatively low opinion of the
scholastic image of their sports heroes. Somewhat contusing and contradictory, however,
is the relatively high mean score for the role modeling variable dealing with studying
harder (STUDY = 2.82). However, the choice to study harder is a much less intrusive
decision than choosing a college, and this distinction may account for the vast difference

167
between the two variables. Therefore, while the negative behaviors, hairstyles, tattoos,
etc., of sports heroes may earn heavy media attention, this coverage does not necessarily
influence adolescents to role model these behaviors and characteristics.
Perhaps most important to advertisers and athletic companies, the variable
measuring purchase intentions of the shoes worn by ones sports hero received a low
mean score (SNEAK = 1.70). The mean score was a bit higher for the variable measuring
whether adolescents would buy a product endorsed by their sports hero (ENDORSE =
2.34) and higher still for the variable measuring the likelihood of an adolescent wearing
the same shirt as their sports hero (SHIRT = 2.78). Still, the relatively low mean scores
for these variables, particularly for buying sneakers, indicates a potential weakness in
using popular sports heroes to promote and advertise products. Additionally, paying
professional athletes enormous sums of money to market products should be reevaluated
in terms of its effectiveness and corresponding value.
In summary, several generalizations can be suggested from the collected data on
reported role modeling of sports heroes. First, adolescents do not likely model all actions
of their sports heroes, but rather analyze and select those they intend to model. Second,
adolescents appear more likely to model athletic behaviors than those peripheral to
athletics. Besides on-court trash talking, this appears particularly true for negative
behaviors, potentially dangerous behaviors, consumer behaviors, and behaviors that
require major life decisions. Finally, there appears to be a high level of variance for each
of these role modeling variables, indicating that individual teenagers make independent
and individual decisions on role modeling intentions.
This leads to the discussion of the stated hypotheses, which address the potential
reasons for the variations of and motivations for role modeling behaviors.

168
Hypothesis I and Hypothesis II
As reviewed in the results section, limited support was found for HI, which
predicted a negative relationship between real distance scores and the role modeling
indexes and variables. Virtually no support was found for H2, which predicted a negative
relationship between ideal distance scores and the role modeling indexes and variables.
Several topics must be addressed in the analysis of these hypothesis. First, the significant
relationships should be addressed, particularly the pattern of relationships found for HI.
Second, the lack of support for H2 should be addressed, and focus should be given to the
relative importance of ideal self-concept versus real self-concept in determining role
modeling behaviors. Finally, the overall weakness of support for these hypotheses,
potential reasons for these results, and possible alternative explanations must be
examined.
The two real distance scores that produced multiple significant relationships with
role modeling indexes were real athletic distance and real job distance. These distance
scores were significantly related to the athletic behavior index and the overall behavior
index. Because these distance scores deal with athletic excellence, this relationship with
athletic behaviors would be expected. Generally speaking, as an adolescent feels more
similar in athletic competence to his sports hero, he is more likely to copy the athletic
behaviors of his sports hero.
These two distance scores, real athletic distance and real job distance, also had
significant relationships with prosocial behaviors, and athletic distance had a significant
relationship with the social communications index. Apparently, the proximity between
real self-concept and the image of the sports hero did have some influence on role
modeling in areas outside of athletic behaviors. However, it is the distance scores of the
athletic competence domain and the job competence domain, a domain inherently related
to athletics for this study, bearing that relationship. Therefore, although the perceived

169
proximity between real self-concept and the image of the sports hero did correlate with a
wide range of role modeling behaviors, this was only true for two domains of
self-concept (athletic competence and job competence). Distance scores from the other
domains of self-concept did not appear to influence role modeling decisions. The
perceived relationship between these adolescents and their sports heroes in domains such
as scholastic achievement and friendship had no measurable influence on their intentions
to role model their sports heroes. Again, it appears as though the athletic components of
the sports hero were the most crucial and influential with these adolescents, and this
further reinforces athletic performance as the key component of the American sports hero.
Regardless of the varied images seen on television about popular sports figures, it is the
most basic of functions, athletic accomplishment, which remains the most important and
influential component of the sports hero.
Only one other real distance score, real physical distance, bore any significant
relationship with role modeling behaviors. Real physical distance had a significant
relationship with the variable for getting a tattoo or an earring like ones hero. This
relationship is not surprising, as physical appearance is logically related to getting an
earring or tattoo. However, it should be noted that this variable of role modeling (TATT)
had a very low mean score (1.44), indicating that few of these subjects would consider
modeling such behaviors at all. Additionally, real physical distance actually had a
significant relationship in the opposite direction as predicted with two role modeling
indexes (prosocial behaviors and overall behaviors). Contrary to the hypotheses of this
research, it is possible that a close proximity between real physical self-concept and the
image of the sports hero indicated a high level of self-satisfaction in ones physical
appearance. As this satisfaction increased, these subjects may have been less likely to
model their role models in an attempt to increase their physical self-concept. In other
words, the better a subject felt about himself, the less he may have felt it necessary to
imitate his sports hero in order to increase his self-concept.

170
Another potential explanation for this relationship is that subjects who were
extremely concerned with their physical appearance may not have had great interest in
prosocial behaviors. Perhaps a fixation on physical appearance represented a level of
vanity that did not correlate highly with charity work and other prosocial behaviors. As a
subject placed more emphasis on his appearance, he may have put less effort into
prosocial behaviors. Although merely speculative and not supported in the literature
review of this study, this relationship is a potential explanation for these unexpected
results, one which presents a topic worthy of future research.
Therefore, while relatively few significant relationships were found between real
distance scores and role modeling, it appears that these adolescents were most influenced
by their perceived similarity to their sports hero in athletic areas and related job
competence. Athletics-related distance scores had the greatest impact on role modeling
of sports heroes, and this role modeling was more likely to occur in athletics-related
behaviors.
For Hypothesis II (ideal distance), the only relationship that supported this
hypothesis was that between ideal romantic distance and the index for social
communication and the variable for getting a tattoo or earring like ones hero. Again, this
is a logical relationship, since both these role modeling variables relate to socially
oriented, non-athletic behaviors, and such behaviors would more likely be correlated with
a non-athletic domain of self-concept such as romantic appeal. It is interesting to note the
positive romantic association with tattoos or earrings, an obvious reflection of the modem
image of todays athletes. It would be difficult to imagine past generations viewing such
behaviors in a positive light. Beyond the ideal romantic distance score, no other distance
scores had significant relationships supporting H2, and one distance score, ideal athletic
distance, actually had significant relationships in the opposite direction as was predicted.
Therefore, H2 is not supported by the results of this research.

171
Even though all of the regression models produced relatively low values,
looking at the relative relationships found for real and ideal distance scores with role
modeling may provide insight into the motivations for adolescent role modeling of sports
heroes. Interestingly, the relationships found for real distance scores were with domains
of athletic and job performance, while the few relationships found for ideal distance were
with the romantic domain. There is an obvious and distinct difference between these two
relationships. In the first hypothesis, it was found that the closer an adolescent was to his
sports hero in athletic areas or job performance related areas (which, for sports heroes,
would essentially refer to athletic accomplishments), the more likely he was to role model
that athlete, particularly in athletic and prosocial behaviors. However, the ideal athletic
distance scores and job distance scores did not have the same significant relationships
with role modeling. Therefore, while these adolescents tended to choose sports heroes
that were closer to their ideal self than their real self, particularly for the athletic domain,
it was conversely the close proximity with real self, not ideal self, that correlated with
role modeling behaviors, particularly for athletic behaviors. Generally speaking, the
adolescent might choose a sports hero who represents his athletic ideal, but if he does not
perceive himself as actually being similar to this sports hero, the task of role modeling an
athletes behaviors may appear too difficult or daunting. This could be particularly true
for athletic behaviors, which, due to the very nature of the sports hero, are extremely
difficult to model. As an adolescent perceives himself more similar to his sports hero as
an athlete, this adolescent would express more willingness to role model the athlete in
these difficult behaviors due to his own sense of ability and likelihood to succeed.
Conversely, although the ideal romantic distance scores did have significant
relationships with both social behaviors and getting a tattoo or earring, such social
behaviors are more easily accomplished than the athletic behaviors of a sports hero. It is
much easier to strive towards the ideal in these behaviors (social, tattoo or earring) than to
achieve ideal athletic behaviors. In other words, while it is fairly simple for an adolescent

172
to either copy social behaviors of an athlete or get a tattoo or earring like his hero in order
to strive towards the ideal romantic self, it is much more difficult to copy physically
demanding athletic behaviors of a favorite athlete when they do not perceive themselves
as having similar athletic abilities as their sports hero, even if this athlete is representative
of the ideal self. For difficult behaviors, real distance scores are more crucial in
determining role modeling intentions, indicating a component of the individuals sense of
competence in those difficult areas. In less difficult behaviors, ideal distance scores are
more crucial, as this represents an opportunity to strive toward the ideal without a fear of
low competency.
One questionable aspect of congruity theory addressed in the literature review was
the inability of this theoretical model to predict whether real self-concept or ideal
self-concept would predict purchase behaviors. Furthermore, the theory did not predict
when either of the two components of self-concept would be more predictive. This
research project can add to this understanding and provide some clarification to this
aspect of congruity theory. Congruity theory held true, albeit only to a limited degree, in
correlating real self-concept distance scores to modeling athletic behaviors and prosocial
behaviors, or behaviors that might be extremely difficult to copy without perceived
competence in these areas. If the real self-concept is far from the image of the sports
hero, it would likely seem extremely difficult to copy intricate athletic behaviors of the
sports hero. This is true for the job and athletic domains, domains that are related to the
athletic performance of the sports hero. Conversely, congruity theory held true, albeit to
an even lesser degree, to correlate ideal self-concept distance scores to social behaviors
and behaviors related to physical appearance, or behaviors not nearly as difficult to model
as athletic behaviors or even prosocial behaviors. This was found to be true for the
romantic domain of self-concept, an area that is not necessarily crucial to the sports hero.
To place this in the same language used in marketing studies, real self-concept
distance scores correlate with role modeling the more utilitarian behaviors of the sports

173
hero, those including athletics and prosocial behaviors. Ideal self-concept distance scores
correlate with role modeling the less utilitarian behaviors, behaviors not inherently a
component of the sports hero, but behaviors related to external appearance, social
interactions, and popularity.
This study expanded on previous marketing studies of congruity theory by
subdividing self-concept into eight distinct domains. Therefore, it is possible to better
understand not only which types of behaviors will correlate with real or ideal distance
scores, but also which specific domains of distance scores will correlate with such
behaviors. Congruity theory was supported for real self-concept in the athletic and job
domains, the more utilitarian or performance related components of self-concept. These
domains deal with ones perceptions of their ability to perform a function. For ideal
self-concept, congruity theory was supported for the romantic domain, a relatively
non-performance related or utilitarian domain of self-concept. Therefore, congruity
theory will more likely apply to real distance scores that are utilitarian in nature and to
ideal distance scores that are less utilitarian in nature.
However, despite the conclusions that can be drawn from the results of HI and
H2, it is just as important to note the overall lack of support for the predictions made for
these hypotheses. Although limited relationships were found between the calculated
distance scores and role modeling indexes/behaviors, the distance scores did not account
for the majority of these subjects decisions of whether or not to role model their sports
heroes. Other prevalent figures, including peers, parents, and teachers, likely had greater
influence in determining behaviors for these adolescents than sports heroes.
Some of the behaviors, such as taking Andr, had no significant relationships with
distance scores. This is particularly notable given the negative publicity given to popular
athletes using Andr and the potential influence this could have in encouraging
adolescent athletes to use this drug. Even if adolescents may learn about such
performance-enhancing drugs from mass-mediated sports figures, these adolescents likely

174
have other motivations for deciding whether or not to use the drug. Although it is
fortunate to learn that these adolescents do not simply model potentially negative
behaviors of their sports heroes, such results also illuminate the importance of coaches,
peers, and parents in guiding behaviors.
The index for consumer behaviors also had no significant relationships with any
of the distance scores. The perceived proximity between the real self or the ideal self and
the sports hero had no significant relationship on purchase behaviors of items
wom/endorsed by the sports heroes of these adolescents. Athletic and non-athletic
companies spend enormous sums of money for famous athletes to endorse and promote
their products, yet this research indicates that using such star athletes may not influence
purchase decisions. Many marketing campaigns attempt to portray the athlete as the ideal
and encourage the viewer to copy this athlete. For example, Gatorade promotes its
products by reminding viewers to Be Like Mike [Jordan] and therefore to drink the
same sports drink that Jordan drinks. However, marketing campaigns portraying popular
athletes as similar to the viewer or similar to the ideal of the viewer may have little or no
influence on the purchase decisions of adolescent viewers. Instead, there could be other
motivations for such purchase decisions. Given the amount of money paid to athletes to
market certain products, advertisers should seriously consider the effectiveness of using
these athletes to increase sales. More importantly, these advertisers should address the
motivations for using an athlete as a product spokesperson/endorser and the message of
the campaign, because this popular advertising strategy may have little or no influence on
purchase intentions.
For those behaviors that did have significant relationships with distance scores as
reported above, it is crucial to recall the relatively small amount of influence of distance
scores in predicting role modeling behaviors. Although the descriptive statistics did
indicate a relatively high level of role modeling in a variety of areas, these distance scores
accounted for a relatively small percentage of the variance in role modeling intentions. In

175
other words, these adolescents indicated that while they do role model their sports heroes,
the congruity between their self-concept and the image of their sports heroes was not the
primary motivation for their modeling intentions. This leaves researchers with much
more to learn about the potential reasons why adolescents choose to engage in role
modeling behaviors of sports heroes.
The next section, which examines the results for Hypothesis III and Hypothesis
IV, discusses domain importance and the calculated distance worth scores as additional
explanations for role modeling.
Hypothesis III and Hypothesis IV
It was predicted that the analysis of domain importance scores would explain
more about the potential motivations for role modeling a sports hero. It was expected
that distance worth scores, incorporating both the level of importance placed in a
particular domain and the congruity between self-concept and the image of the sports
hero in that domain, would be more predictive of role modeling behaviors. Increased
importance of a domain was predicted to increase the importance of the distance score in
determining role modeling intentions. However, as noted in the results section, no new
information was gained in these two hypotheses that was not already found in H1 and H2,
and distance worth scores were in fact less predictive of role modeling behaviors than the
distance scores used in HI and H2.
As with distance scores from HI and H2, real distance worth scores, particularly
real athletic distance worth scores, were the most significant in predicting role modeling
behaviors. Ideal distance worth provided little in terms of predicting role modeling
behaviors, similar to the results for H2 with ideal distance scores. Therefore, the
importance of a domain of self-concept and the creation of a distance worth score did not
help predict role modeling of sports heroes. It was predicted that increasing importance

176
of a domain of self-concept, particularly in conjunction with a closer proximity between
self-concept and the image of the sports hero for that domain, would correlate with role
modeling behaviors. This was not supported from the data collected in this study.
Interestingly, it is possible that incorporating domain importance actually
detracted from the predictive nature of distance scores, and the creation of distance worth
scores actually provided less information on role modeling of sports heroes than the
original distance scores used in HI and H2. Several distance worth scores had
relationships with role modeling in the opposite direction as predicted. For example,
several domains of self-concept distance worth scores, both real and ideal, were
negatively correlated with tattoo or earring modeling behaviors. These relationships were
the opposite as predicted in H3 and H4. Furthermore, these relationships were not found
in the analysis of HI and H2. It is possible that increasing importance in an area of
self-concept actually made these subjects less likely to look towards their sports heroes as
role models. As an area of self-concept became more important, these subjects likely
considered other motivations for deciding whether to model certain behaviors, such as
getting a tattoo or earring like a sports hero. Given the strong ties between domain
importance and self-worth (in contrast to self-concept, which does not incorporate
domain importance), perhaps a low sense of self-worth contributed to the decision to get
an earring or tattoo of a sports hero, while increasing self-worth makes this decision less
likely. As respondents felt better about themselves, they became less likely to get a tattoo
or earring like a sports hero.
Another explanation for this relationship between domain importance of
self-concept and role modeling is provided by the elaboration likelihood model of
persuasion (ELM), a theory that addresses the effectiveness of persuasive messages (Petty
& Cacioppo, 1986; Petty & Priester, 1994). According to ELM, messages can be
processed two ways. The more involved route, the central route, involves effortful
cognitive activity whereby the person draws upon prior experience and knowledge in

177
order to carefully scrutinize all of the information relevant to determining the central
merits of the position advocated (Petty & Priester, 1994, pp. 98-99). This is a
thoughtful and involved process of evaluation that can lead to lasting attitude change.
The less involved means of persuasion can occur through the peripheral route in which
simple cues in the persuasion context influence attitudes (Petty & Priester, 1994, p. 101).
The peripheral route is a more passive, less involved means to attitude change. These
changes are not typically as enduring as those made through the central route. Further,
the messages evaluated through the peripheral route are not evaluated as critically as
those evaluated thought the central route (Petty & Cacioppo, 1986; Petty & Priester,
1994).
A primary factor in determining the type of processing used in attitude change is
the personal relevance of the message (Petty & Priester, 1994). In other words, the more
important the message, the more likely an involved cognitive process will be used to
evaluate the message in considering attitude change. For this study, ELM might help to
explain the impact of domain importance on intentions to role model the sports hero. As
a domain of self-concept was considered more important (higher domain importance
scores), the central route more likely was used to evaluate the sports hero and the
congruity between self-concept and the image of the sports hero for that domain. When
the central route was used to carefully consider the implications of role modeling the
behaviors of a popular athlete, role modeling intentions likely decreased, particularly for
important, negative behaviors such as drug use or getting a tattoo or earring. In other
words, when this cognitive process was used to evaluate the messages of sports heroes,
the subjects for this study might have realized that popular athletes do not provide the
proper model to influence important decisions. Instead, these subjects might turn to other
sources to help determine behaviors, such as messages from parents and educators.
Conversely, decreasing importance of a domain (lower domain importance scores)
might have resulted in the use of the peripheral route to evaluate the sports hero and

178
self-concept congruity with the sports hero. The peripheral route, which does not involve
the same cognitive processing as the central route, might allow for greater role modeling
of sports heroes based on the perceived messages and images of the hero. In other words,
the less these subjects cared about a domain of self-concept, the less likely these subjects
were to carefully consider the congruity between their self-concept and their image of
their sports hero in that domain. Instead, they may have decided to role model their
sports hero without considering the importance of this decision or without considering
reasons why they should not copy the behaviors of their sports hero. This was most
likely true for behaviors that did not have lasting implications. Therefore, the ELM
provides one possible reason that not only did increases in domain importance not result
in greater role modeling intentions as predicted, but increasing domain importance may
have actually lead to reduced role modeling intentions. Although interesting and
important, this determination goes beyond the scope and limitations of this study. It
certainly is grounds for future research on role modeling of sports heroes, and researchers
should address ELM as a theoretical framework for determining role modeling intentions.
In summary of H3 and H4, this research indicates that incorporating domain
importance with distance scores does little to further predict role modeling intentions of
sports heroes. While domain importance may have an influence on role modeling, it was
not in the direction as predicted, and this should be studied independently of the distance
between self-concept and the image of the sports hero.
Limitations of the Study
This study adds to the understanding of several concepts, primarily the potential
motivations for adolescent role modeling of sports heroes. On a more theoretical level,
this study expands the boundaries and applications of congruity theory and examines the

179
relative influences of real distance scores and ideal distance scores. However, several
limitations that reduce the generalizability of the results should be examined.
First, subjects for this study were predominantly upper-middle-class white males,
typically from stable home environments. Most of these subjects were students in
rigorous, high achieving schools. Therefore, it is impossible to generalize the results of
this study to a larger, less homogenous population. Most of the subjects for this study
have had frequent interaction with many role modeling influences, including family,
peers, teachers, and religious leaders, who likely have a strong influence on their
behaviors. Therefore, the influences of televised sports heroes as role models for these
adolescents may be less pronounced than for adolescents who lack such a variety of role
models.
Second, for a variety of reasons, much of the recent debate on the role modeling
function of sports heroes has focused on African-American adolescents and
African-American sports figures. African-American adolescents may perceive their
sports heroes differently than the participants in this study and may have a different
perception of the similarity between themselves and popular African-American athletes.
While virtually all subjects for this study were white, slightly greater than half of these
adolescents chose African-American sports heroes. It is difficult to fully understand
whether African-American adolescents would perceive these sports figures differently
than their white counterparts and whether influences on role modeling would differ.
Certainly, this is an important area for future research.
Third, this study forced all subjects to chose one particular sports hero for the
study of role modeling and distance scores. However, it is likely that adolescents look to
a variety of sports figures as potential role models. Although limiting the selection to one
sports hero allowed for an examination of the relationships between the perceived
similarities to one particular hero and role modeling of that same hero, the study did not
capture all information about adolescent role modeling of sports heroes. For example,

180
some adolescents may look to one particular sports hero as an athletic role model, yet
look to another athlete altogether as a role model for social behaviors. Unfortunately, this
study could not capture the influences of different athletes, but rather had to examine the
influences of and relationships with one particular athlete, the athlete who best embodied
elements of the sports hero to each individual subject.
Fourth, obvious limitations exist in the use of quantitative methods to study
behaviors. While this research project attempted to capture the key components of role
modeling through the use of focus groups of adolescent males and a study of the
literature, it is unlikely that all components of role modeling were identified for the
creation of surveys. Additionally, because this study used survey research and
quantitative methods, there is the possibility that the subjects could not accurately assess
their role modeling behaviors from the surveys used in this research. Some types of role
modeling may have best been assessed through a more qualitative approach, allowing for
subjects to describe specific role modeling behaviors. Although this approach would not
have allowed for quantitative analysis of congruity theory, qualitative methods such as
interviews and observations would provide a less restrictive and more subject driven
means to understanding the range of behaviors that adolescents role model from their
sports heroes.
Finally, some participants may have given socially desirable responses, which
could have accounted for the low levels of reported intentions to role model negative
behaviors and consumer behaviors. Certainly, it would be preferable to find a more
accurate measure of role modeling, one that might involve a more qualitative assessment
of role modeling sports heroes. Again, this is an important area for future research.

181
Recommendations for Future Research
This dissertation has added to the academic knowledge base in two distinct areas.
First, this research has provided a better understanding of the role modeling function of
American sports heroes. Second, this study examined congruity theory outside the
context of consumer behaviors and provided a greater understanding of the relative
importance of real distance scores and ideal distance scores in determining behaviors.
From the results of this work, several recommendations can be made to further build the
knowledge base in these areas. Such recommendations have been divided into logical
categories.
Diversity of Subjects
As mentioned in the study limitations, this study examined a relatively
homogenous subject pool. Additional research should address a variety of different
subject populations. One such group is African-American adolescents. Due to the
prevalence of African-American athletes in professional and amateur athletics, cultural
critics have suggested that there is great potential for role modeling of sports stars by
African-American adolescents.
Similarly, with the remarkable growth of womens athletics at both the youth and
professional levels in the past two decades, researchers should examine role modeling
intentions of adolescent girls. More than 33 percent of high school girls participate in
school athletics, an increase from only 4 percent participation in 1972 (Womens Sports
Foundation, 1998). Female athletes now also have a much larger presence on television
and greater opportunities in collegiate and professional athletics. This increasing
involvement in athletics by adolescent girls has likely lead to greater opportunities for
girls to role model the new wave of popular female athletes. Therefore, female role
modeling of sports heroes has become an important yet largely tapped area for future
study. While this dissertation only examined predominantly white, adolescent male

182
athletes, future research should study role modeling of adolescents of all races and
genders, including African-Americans, females, and a variety of minority subgroups.
Future research also should address individual and socioeconomic characteristics
of adolescents and the potential influence these characteristics might have on role
modeling decisions. These characteristics could include single-parent vs. two-parent
home environments; financial conditions of the subjects family; academic achievement
of subjects; and the number of other, non-athletic role models identified by each
individual subject. Such research could provide insight into the relative influence of
mass-mediated sports heroes compared to other factors in the lives of adolescents.
Qualities of the Sports Hero
This dissertation allowed for individual selection of sports heroes. The only
stipulation was that each subject choose a popular athlete who was regularly covered by
television and other mass media outlets. In other words, sports heroes for this study
could not merely be personal acquaintances, and information about these heroes must
have been received through media channels. Future studies in this area could examine
different types of athletes to measure their potentially different impact on adolescent role
modeling. For example, researchers could examine whether role modeling of sports
heroes from any particular sport was different than role modeling of athletes from other
sports.
Additional research projects also could address the relative impact of positive vs.
negative athletes as role models. Most of the subjects from this research selected sports
heroes who generally act and behave in a positive, socially acceptable fashion. However,
a select few subjects chose athletes who have exhibited negative behaviors. These
athletes were all star performers, yet their non-athletic behaviors have resulted either in
frequent arrest or suspension from their respective leagues. Negative athletes who were
selected include boxer Mike Tyson and basketball players Charles Barkley, Dennis
Rodman, and Allen Iverson, all of whom have been arrested for off-the-court behaviors

183
and suspended for blatant rule infractions during athletic competition. Future research
should examine the potentially different influences of these positive and negative sports
heroes and identify the specific areas of role modeling in which differences exist.
Additionally, differences in congruity between self-concept of adolescents and the image
of the sports hero should be examined. This information could further an understanding
of the range of the sports hero construct, potential motivations for selecting positive vs.
negative heroes, and the impact of congruity on role modeling both positive and negative
behaviors.
Media. Influence
This study examined the influence only of popular, mass-mediated sports heroes,
emphasizing the importance of media in publicizing star athletes. While television plays
a crucial and dominant function in this process, a variety of different types of media also
provide information about popular star athletes. Such media include newspapers,
magazines, and a rapidly increasing number of Internet web sites. Each of these different
types of media can provide varying amount of information about individual athletes, from
peripheral coverage on television to in-depth analysis available in magazines and the
Internet. Additionally, some media types, such as magazines and the Internet, require an
active, information seeking process by the sports fan that may not be necessary for other
types of media. Future research on mass-mediated sports heroes should examine the
potential influence of these different types of media and the role modeling that may result
from the use of various media. For example, research could examine whether adolescents
who receive the majority of information about their sports heroes from television are
more or less likely to role model their heroes than those adolescents who seek
information from the Internet. Such research would help further an understanding both of
adolescent sports fans and the relative influence of various media types in influencing
behaviors of these fans.

184
Other future research could address the impact of mass mediated sports heroes, as
studied in this dissertation, vs. personally known sports heroes. This would give insight
into the impact of mass media in providing role models for adolescents. Similar research
projects should address the impact of mass mediated sports heroes vs. traditional,
personally known role models, including teachers, peers, and parents. Again,
comparisons of this nature would further examine the relative influence of mass media in
providing role models and could also speak to the importance of traditional role modeling
influences in the lives of adolescents.
Self-Concept
This dissertation examined two distinct types of self-concept, real self-concept
and ideal self-concept, across eight separate domains. In doing so, knowledge was gained
into the relative impact of real self-concept congruity vs. ideal self-concept congruity in
influencing role modeling of sports heroes. In other words, adolescent subjects for this
research expressed whether they role model sports heroes who more closely represented
their real self or their ideal self and in which areas such role modeling was likely to occur.
However, given the prevalence of negative actions of star athletes and the coverage this
negative activity receives from mass media, future research should address the impact of
negative athletes, athletes not typically seen as sports heroes, and the impact these
athletes have on the behaviors of adolescent viewers. Specifically, research should
address whether viewers model the negative behaviors observed through the media or if
they instead actively avoid the negative behaviors of these athletes. Negative, non-heroic
athletes would not provide a model for how to act, but rather for how not to act.
To conduct such research, another component of self-concept should be
addressed, the negative self-concept. While researchers have typically limited studies to
real and ideal self-concept, particularly when addressing congruity theory, future research
could examine a negative component of self-concept (myself as I do not want to be) and
how this construct impacts avoidance behaviors. Research should address whether

185
negative behaviors of athletes, such as shoving coaches, fighting, and spouse abuse, are
actively modeled or actively avoided and whether the congruity between negative
self-concept and the image of a negative sports hero is predictive of this process.
Because cultural critics, parents, and educators alike have criticized mass mediated sports
heroes as poor role models due to their negative behaviors, this direction of research
would present a logical step in further addressing these concerns.
Another area for future research involves the importance of self-worth or
self-esteem in impacting role modeling of sports heroes. For this research, domain
importance scores, a primary component of determining self-worth, did not contribute to
the importance of self-concept congruity. Conversely, increased domain importance
scores, which lead to higher self-worth, often reduced the impact of congruity in
determining role modeling. Therefore, it is important to consider whether adolescents
with a high self-worth are less likely to role model sports heroes. In other words,
researchers should study whether adolescents with high opinions of themselves are less
likely to copy the behaviors of mass mediated sports heroes. Similarly, adolescents with
a low sense of self-worth might role model their sports heroes more frequently in an
effort to feel better about themselves. Research on adolescent females has suggested that
low self-worth can serve as a behavioral influence to adopt behaviors such as weight loss
techniques in attempt to achieve a higher sense of self-worth (Orenstein, 1994). Similar
research on adolescent role modeling of sports heroes and self-worth would give new
insight into the actual motivations for role modeling decisions.
Methodological Alterations
The final area for additional research deals with potential methods to study
adolescent role modeling of sports heroes. This dissertation used surveys and
quantitative statistics to measure and analyze role modeling of sports heroes. However,
while this approach is both practical and allows for access to a large subject pool, these
methods are not necessarily the most effective means to study this topic. Subjects may be

186
inclined to provide socially desirable responses to survey questions, and surveys,
regardless of how well designed, may not fully address all role modeling behaviors.
Therefore, future research should use a variety of methods to study role modeling.
Qualitative research could allow researchers to learn more about the individual
role modeling behaviors of adolescents. Through the use of probing questions,
adolescents would be able to describe specific examples of role modeling intentions and
potential motivations for such behaviors. Specific questions pertaining to self-concept
congruity with the image of ones sports hero also could be asked. This research design
might reveal role modeling behaviors not addressed in survey research. Additionally,
through the interview process, adolescents could indicate individual motivations for their
specific role modeling intentions. These motivations could include self-concept
congruity, but they could also address a variety of motivational sources. Therefore, this
type of qualitative approach would allow for greater perspective on role modeling
intentions and would also place fewer restrictions on the range of topics to be examined.
Further, in-depth qualitative studies would help to shape a future quantitative research
agenda on this subject. This approach would allow for future quantitative research to
address the most important and most relevant areas unearthed through the more
naturalistic form of inquiry.
Another potential methodological approach for this topic is to utilize
observational research methods. By observing the behaviors of adolescents, particularly
over extended periods of time, researchers could see the actual behaviors of adolescents
that are modeled from popular sports stars. Some behaviors, such as wearing a sports
jersey or mocking a clothing style, could easily be identified and matched to specific
sports stars, while other, more generic behaviors, such as athletic maneuvers and
behaviors toward teammates, would be more difficult to match with one specific athlete.
Therefore, observational research would be most helpful in studying specific, easily
observed athletic and consumer behaviors. This approach would be most useful when

187
combined with interviews and focus groups, a qualitative approach, to allow subjects to
explain the observed behaviors. This research approach would reduce the potential for
socially desirable responses given for surveys and interview questions. Through
observation, researchers would be able to study role modeling behaviors that may not be
addressed through survey research, behaviors that may be overlooked during the creation
of survey instruments. Although quantitative research is the most appropriate approach
for theory testing, a more qualitative research method would give greater access to the
range of behaviors that are modeled and the conditions under which such role modeling
occurs.
The role modeling of televised sports heroes by adolescents, particularly in
conjunction with self-concept theory, is largely unexplored by academia. With the
impressive growth of sports media and the emphasis on sports and sports heroes in
modem society, future research should address a variety of topics and employ a range of
methodologies. Additionally, given the ongoing changes in the heroic construct and the
evolution in media technologies and coverage of sports by mass media, researchers
should continue to examine and reexamine the qualities of the mass mediated sports hero,
the relationships between these athletes and adolescent fans, and the various impacts of
these sports heroes on adolescents. Such a research agenda would help to provide insight
into an important area of modem society and help guide media critics, parents, educators,
and adolescents in better understanding the social impact of the modem American sports
hero in their individual lives.

APPENDIX A
SELF-CONCEPT/IMAGE OF THE SPORTS HERO QUESTIONNAIRE

189
What I Am Like
Name.
SAMPLE SENTENCE
Really
THie
lor Me
Sort o 1
TVue
(or Me
Sort ol
Thio
lor Me
Really
Thie
for Me
a)
m
0
Some teenagers like
lo go 10 movies in
their spare lime
BUT
Other teenagers would rather
go to sports events.
0 H
1.


Some teenagers (eel
that they are jusl
as smart as others
their age
BUT
Other teenagers arent so
sure and wonder il they are
as smart.


2.


Some teenagers find
it hard to make
friends
BUT
For other teenagers it's
pretty easy.


a


Some teenagers do
very well at all
kinds o! sports
BUT
Other teenagers dont feel
that they are very good when
it comes to sports.


4.


Some teenagers are
not happy with the
way they look
BUT
Other teenagers are happy with
the way they look.


5.


Some teenagers (eel that they
are ready to do well at a o
part-time job "
BUT
Other teenagers feel lhat they
are not quite ready to handle
a part-time job.




Some teenagers leel that il they
are romantically interested in
someone, that person will like
them back
BUT
Other leenagers worry that when
they like someone romantically,
that person wont like them
back.


* I
i

Some teenagers usually do
Ihe right thing
BUT
Other leenagers often dont do
what they know is right.


8- l
i

Some teenagers are
able to make really
close Iriends
BUT
Other teenagers find it hard
to make really close friends.


* I
i

Some teenagers are pretty
slow in finishing their
school work
BUT
Other teenagers can do
their school work more
quickly. 1


10. [
i

Some teenagers have a lot
ol friends
BUT
Other teenagers dont
have very many Iriends. L


, [
[

Some leenagers think they
could do well al just about any
new athletic activity
BUT
Other teenagers are afraid they ["
might not do well at a new
athletic activity. L



190
Really Sort of
TVue Thje
(or Me lor Mo
Sort of
Thie
for Me
Really
Thie
for Me
|

Soma teenagers wish
their body was different
BUT
Other teenagers Re their body
the way it is.


13. \

Some teenagers feel that they
don't have enough skills to
do well at a job
BUT
Other teenagers feel that they
do have enough skills to
do a job well.


*

Some teenagers are not
dating the people they
are really attracted to
BUT
Other teenagers are
dating those people
they are attracted la


,5. |

Some teenagers often get in
trouble for the things
they do
BUT
Other teenagers usually don't
do things that get them in
trouble


II.

Some teenagers do have a
close friend they can share
secrets with
BUT
Other teenagers do not
have a really close friend
they can share secrets with


n. |

Some teenagers do very well
at their dasswork
BUT
Other teenagers don't do very
well at their dasswork.


ii. |

Some teenagers are very
hard to like
BUT
Other teenagers are
really easy to like.


* 1

Some teenagers leel that 9
they are better than others "
their age at sports
BUT
Other teenagers don't
feel they can play as well.


HO.

Some teenagers wish their
physical appearance was
different
BUT
Other teenagers Ska
their physical appearance
the way it is. '


31. |

Some teenagers feel they are
old enough to get and keep a
paying job
BUT
Other teenagers do not feel
they are old enough, yet. to
really handle a job well 1


3X

Some teenagers leel that people
their age will be romantically
attracted to them
BUT
Other teenagers worry about
whether people Iheir age will
be attracted to them. I


2.3.

Some teenagers feel really
good about the way they act
BUT
Other teenagers don't feel that
good about the way they often 1
act



Some teenagers wish they had
a really close Iriend to share
things with
BUT
Other teenagers do have
a close friend to share
things with. >


25.

Some teenagers have trouble
figunng out the answers in school
BUT
Other teenagers almost always
can figure out the answers. 1



101
Really Sort of
Thi* Thie
for Mo for Mo
Sort of
Thie
for Me
Realty
True
for Me
u


Some teenagers are popular
with others their age
BUT
Other teenagers are not
very popular.
c

n.


Some teenagers don't do well
at new outdoor games
BUT
Other teenagers are good at
new games right away.
c

31


Some teenagers think that
they are good looking
BUT
Other teenagers think that they
are not very good looking.


3X


Some teenagers feel like they
could do better at work they
do lor pay
BUT
Other teenagers feel that they
are doing really wen at work
they do lor pay.


30.


Some teenagers feel that they
are fun and interesting on
a date
BUT
Other teenagers wonder about
how fun and interesting they
are on a date


S'-


Some teenagers do things
they know they shouldn't do
BUT
Other teenagers hardly ever
do things they know they
shouldn't da


32.


t
Some teenagers find it hard
to make friends they can
really trust
BUT
Other teenagers are able
to make dose triends they
can really trust.


33.


Some teenagers feel that
they are pretty intelligent
BUT
Other teenagers question
whether they are intelligent.


3*


Some teenagers feel that they
are socially accepted
BUT
Other teenagers wished
that more people their age
accepted them.


35.


Some teenagers do not feel
that they are very athletic
BUT
Other teenagers feel that they
are very athletic

!

3i.


Some teenagers really like
their looks
BUT
Other teenagers wish they
looked different.

1

31.


Some teenagers feel that they
are really able to handle
the work on a paying job
BUT
Other teenagers wonder if they
are really doing as good a job
at work as they should be doing

1

31


Some teenagers usually don t
go out with the people they
would really like to date
BUT
Other teenagers do go out
with the people they really
want to date.

1

39.


Some teenagers usually act
the way they know they are
supposed to
BUT
Other teenagers often don't
act the way they are
supposed ta

1

Ho.


Some teenagers don't have
a friend that is close enough
to share really personal
thoughts with
BUT
Other teenagers do have a
close friend that they can share
personal thoughts and
feelings with.
1


102
Name Age Grade Group
HOW IMPORTANT ARE EACH OF THESE THINGS TO YOU?
Really
True
lor Me
Sort of
True
lor Me
Sort of
True
for Me
Really
True
for Me
HI.

Some teenagers think it is
important to be intelligent
BUT
Other teenagers don't think
it is important to be intelligent


13.

Some teenagers don't think
its all that important to
have a lot ol friends
BUT
Other teenagers think that
having a lot of friends is
important


13.

Some teenagers think its
important to be good at
sports
BUT
Other teenagers don't care
much about being good at
sports


11.

Some teenagers don't really
think that their physical
appearance is all that
important
BUT
Other teenagers think that
their physical appearance is
important


15-

Some teenagers don t care
that much about how well
they do on a paying job
BUT
Other teenagers feel its
important that they do well
on a paying job


.

Some teenagers think its
important that the people they
are romantically interested in
like them back
BUT
Other teenagers don't really
care that much whether
someone they are interested in
likes them that much


17.

Some teenagers don't think
its that important to do the
right thing
BUT
Other teenagers think that
doing the right think is
important


.

Some teenagers think its
important to be able to make
really close friends
BUT
Other teenagers don t think
making close friends is all
that important


11.

Some teenagers don t think
that doing well in school is
really that important
BUT
Other teenagers think that
doing well in school is
important


5o.

Some teenagers think its
important to be popular
BUT
Other teenagers don't care
that much about whetner they
are popular


5/

Some teenagers don t think
that being athletic is that
important
BUT
Other teenagers think that
being athletic is important


53.

Some teenagers think that
how they look is important
BUT
Other teenagers don't care that
much about how they look


53.

Some teenagers think its
important to do their best
on a paying job
BUT
Other teenagers don't think
that doing their best on a
job is all that important


51.

Some teenagers don t care
that much whether they are
dating someone they are
romantically interested in
BUT
Other teenagers think its
important to be dating
someone they are
interested in


55.

Some teenagers think its
important to act the way
they are supposed to
BUT
Other teenagers don t care that
much whether they are acting
the way they are supposed to


51.

Some teenagers don t care
that much about having a
close friend they can trust
BUT
Other teenagers think its
important to have a really
close friend you can trust



APPENDIX B
ANSWER GUIDE FOR SELF-CONCEPT/IMAGE OF THE SPORTS HERO
QUESTIONNAIRE

194
Name:
My sports hero is
1)What I am like
What I want to be like
What my sports hero is like
Grade: Age:
ID t:
9)What 1 am like
What 1 want to be like
What my sports hero is like
2) What I am like
What I want to be like
What my sports hero is like _
3) What I am like
What I want to be like
What my sports hero is like _
4) What I am like
What 1 want to be like
What my sports hero is like _
5) What 1 am like
What 1 want to be like
What my sports hero is like _
6) What 1 am like
What 1 want to be like
What my sports hero is like .
7) What 1 am like
What 1 want to be like
What my sports hero is like
8) What I am like
What I want to be like
What my sports hero is like
10) What I am like
What I want to be like
What my sports hero is like _
11) What I am like
What I want to be like
What my sports hero is like _
12) What 1 am like
What 1 want to be like
What my sports hero is like _
13) What I am like
What I want to be like
What my sports hero is like .
14) What I am like
What I want to be like
What my sports hero is like .
15) What 1 am like
What I want to be like
What my sports hero is like
16) What I am like
What I want to be like
What my sports hero is like

105
17) What I am like
What 1 want to be like
What my sports hero is like _
18) What I am like
What I want to be like
What my sports hero is like .
19) What I am like
What I want to be like
What my sports hero is like .
20) What I am like
What 1 want to be like
What my sports hero is like
21) What 1 am like
What 1 want to be like
What my sports hero is like
25) What I am like
What I want to be like
What my sports hero is like _
26) What I am like
What 1 want to be like
What my sports hero is like _
27) What 1 am like
What I want to be like
What my sports hero is like .
28) What 1 am like
What I want to be like
What my sports hero is like
29) What 1 am like
What 1 want to be like
What my sports hero is like
22) What I am like .
What 1 want to be like
What my sports hero is like
23) What 1 am like
What 1 want to be like
What my sports hero is like
24) What 1 am like
What I want to be like
What my sports hero is like
30) What I am like
What 1 want to be like
What my sports hero is like .
31) What 1 am like
What 1 want to be like
What my sports hero is like
32) What 1 am like
What 1 want to be like
What my sports hero is like

\%
33)What I am like
What I want to be like
What my sports hero is like
41) What I think
42) What I think
34)What I am like
What I want to be like
What my sports hero is like
43) What I think
44) What I think
35)What I am like 45) what 1 think
What I want to be like
What my sports hero is like 46) What I think
36)What I am like 47) What 1 think
What I want to be like
What my sports hero is like 48) What I think
37) What 1 am like 49) What I think
What I want to be like
What my sports hero is like 50) What I think
38) What I am like 51) What I think
What I want to be like
What my sports hero is like 52) What I think
39)What I am like 53) What I think
What I want to be like
What my sports hero is like 54) What I think
40)What 1 am like 55) What 1 thmk
What I want to be like
What my sports hero is like 56) What 1 think

APPENDIX C
ROLE MODELING QUESTIONNAIRE
Name:
ID Number:
Your sports hero is .
I play on the following athletic teams in school .
I watch approximately hours (fill in the blank) of sports television per
week.
Instructions: Thinking about the sports hero that you have chosen, please read each
question carefully and circle the number that is most appropriate for you. Note that
not all questions have the same answer scale, so please pay close attention. There
are no correct or incorrect answers.
1) I plan to buy (or already own) the same sneakers that my sports hero wears.
1 2 3 4 5
(Unlikely) (Likely)
2) I plan to get (or already have) a hair style like my sports hero.
1 2 3 4 5
(Unlikely) (Likely)
3) I wear parts of my uniform the same way my sports hero wears his uniform (ex.
baggy shorts, chinstrap, etc.).
1 2 3 4 5
(Never true for me) (Always true for me)
4) I am planning to get (or already have) a tattoo or an earring like my sports hero.
1 2 3 4 5
(Unlikely) (Likely)
5) I want to play (or already play) the same position in sports that my sports hero plays.
1 2 3 4 5
(Never true for me) (Always true for me)
197

198
6) When Im doing sports, I try to copy the athletic moves of my sports hero.
1 2 3 4 5
(Never true for me) (Always true for me)
7) Watching my sports hero play sports has inspired me to train harder for sports.
1 2 3 4 5
(Never true for me) (Always true for me)
8) I learn new ways and techniques to train for sports from my sports hero.
1 2 3 4 5
(Never true for me) (Always true for me)
9) I will apply to the same college my sports hero attended.
1 2 3 4 5
(Unlikely) (Very Likely)
10) If my sports hero takes sports supplements (or were to start taking them) to improve
his performance, (ex. Andr, creatine, etc.) I would take (or already have taken) the same
supplements to improve my performance.
1 2 3 4 5
(Unlikely) (Very Likely)
11) I try to treat people in my family the same way my sports hero treats people in his
family.
1 2 3 4 5
(Never true for me) (Always true for me)
12) I try to support my friends and teammates the same way my sports hero supports his
teammates.
1 2 3 4 5
(Never true for me) (Always true for me)
13) I learn ways to communicate with my coaches by watching how my sports hero
treats (and acts around) his coaches.
1 2 3 4 5
(Never true for me) (Always true for me)
14) Watching my sports hero has taught me how to be more polite and honest.
1 2 3 4 5
(Not true for me) (True for me)
15) I have plans to get more involved with volunteer and charity work because I have
seen my sports hero donating his time to volunteer causes.
1 2 3 4 5
(Unlikely) (Very likely)

199
16) I try to be as popular with girls as my sports hero is with girls.
1 2 3 4 5
(Never true for me) (Always true for me)
17) I try to use the same kind of behavior while playing sports (ex: trash talk, silence on
the field, communication with teammates, etc.) that my sports hero uses when he plays.
1 2 3 4 5
(Never true for me) (Always true for me)
18) Watching my sports hero work hard in sports inspires me to study harder for school.
1 2 3 4 5
(Never true for me) (Always true for me)
19) I look to my sports hero as an example to help me overcome difficult obstacles in my
life.
1 2 3 4 5
(Never true for me) (Always true for me)
20) One way I will try to make more friends is to act more like my sports hero.
1 2 3 4 5
(Unlikely) (Likely)
21) I would be more likely to use a product if my sports hero endorsed it
1 2 3 4 5
(Unlikely) (Likely)
22) I will buy (or already own) a sports jersey of my sports hero.
1 2 3 4 5
(Unlikely) (Likely)
23) Whenever possible, I try to act like my sports hero.
1 2 3 4 5
(Never true for me) (Always true for me)

REFERENCES
Adams-Price, C. & Green, A. (1990). Secondary attachments and adolescent
self-concept. Sex Roles: A Journal of Research, 22, 187-198.
Agresti, A. & Finlay, B. (1986). Statistical methods for the social sciences.
San Francisco: Dellen.
Ajzen, I. (1989). Attitude structure and behavior. In A. Partakes, S. Breckler, &
A. Greenwood (Eds.), Attitude structure and function (pp. 241-274).
Hillsdale, NJ: Erlbaum.
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and
Human Decision Processes, 50, 179-211.
Almquist, M. & Angrist, S. (1971). Role model influence on college womens career
aspirations. Merrill-Palmer Quarterly, 17, 263-279.
Anderson, W. (1993, May/June). Have we outgrown the age of heroes? Utne Reader,
pp. 95-100.
Andreasen, M. (1990). Evolution of the familys use of television: Normative data from
industry and academe. In J. Bryant (Ed.), Television and the american family
(pp. 3-55). Hillsdale, NJ: Erlbaum.
Another hero. (1996, August 17). New York, p. 11.
Austin, E. (1993). Exploring the effects of active parental mediation of television
content. Journal of Broadcasting & Electronic Media, 37, 147-158.
Ball-Rokeach, S., & DeFleur, M. (1976). A dependency model of mass media
effects. Communication Research, 3, 3-21.
Bandura, A. (1992). Self-efficacy mechanism in psychobiological functioning.
In R. Schwarzer (Ed.), Self-efficacy: Thought control of action (pp. 355-394).
Washington, DC: Hemisphere.
Bandura, A. (1994). Social cognitive theory of mass communication. In J. Bryant &
D. Zillman (Eds.), Media effects: Advances and theory in research (pp. 61-90).
Hillsdale, NJ: Erlbaum.
200

201
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-
Hall.
Bandura, A., & Walters, R. (1963). Social learning and personality development.
New York: Holt, Rinehart, & Winston.
Bassow, S., & Howe, K. (1980). Role model influence: Effects of sex and sex-role
attitude in college students. Psychology of Women Quarterly, 4, 558-572.
Baughman, E., & Welsh, G. (1962). Personality: A behavioral science. Englewood
Cliffs, NJ: Prentice-Hall.
Becker, A., & Popenoe, R. (1996). Body, self, and society: The view from Fiji.
American Anthropologist, 98, 919.
Bell, A. (1970). Role modeling and interaction in adolescence and young adulthood.
Developmental Psychology, 2, 123-128.
Benedict, J., & Yaeger, J. (1998). Pros and cons: The criminals who play in the NFL.
New York: Warner Books.
Berger, L., & Luckman, P. (1966). The social construction of reality. New York:
Doubleday.
Betz, N., & Fitzgerald, K. (1987). The career psychology of women. Orlando:
Academic Press.
Birdwell, A. (1968). A study of influence of image congruence on consumer choice.
Journal of Business, 41, 76-88.
Birrell, S., & Cole, C. (1981). Women, sport, and culture. Champaign, IL: Human
Kinetics.
Boorstin, D. (1961). The image. London: Penguin.
Boorstin, D. (1980). A guide to pseudo events in America. New York: Athenaeum.
Botta, R. (1999). Television images and adolescent girls body image disturbance.
Journal of Communication, 49, 22.
Botvin, E., Botvin, G., Michela, J., Baker, E., & Filazzola, A. (1991). Adolescent
smoking behavior and the recognition of cigarette advertisements. Journal of
Applied Social Psychology, 21, 919-932.

202
Bowra, C. (1969). The hero. In V. Brombert (Ed.), Hero in literature (pp. 22-52).
New York: Fawcett.
Brady, E. (1999, December 15). Carruth now faces murder charges. USA Today,
Section C, p. 1.
Breckler, S., & Wiggins, E. (1989). On defining attitude and attitude theory:
Once more with feeling. In A. Partakes, S. Breckler, & A. Greenwood (Eds.),
Attitude structure and function (pp. 241-274). Hillsdale, NJ: Erlbaum.
Brim, O. (1968). Socialization through the life style. In C. Gordon & K. Gergen (Eds.),
The self in social interaction (pp. 227-240). New York: John Wiley & Sons.
Brombert, V. (1969). The idea of the hero. In V. Brombert (Eds.), Hero in literature
(pp. 11-21). New York: Fawcett.
Browne, R. (1983). Hero with 2000 faces. In R. Browne & M. Fishwick (Eds.),
The hero in transition (pp. 91-106). Bowling Green, OH: University Popular
Press.
Bumstein, E., Stotland, E., & Zander, A. (1961). Similarity to a model and self-
evaluation. Journal of Abnormal and Social Psychology, 62, 257-264.
Calhoun, D. (1981). Sport, culture, and personality. Champaign, IL: Human Kinetics.
Campbell, J. (1949). The hero with a thousand faces. New York: Pantheon.
Carlyle, T. (1840). On heroes and hero worship. London: Oxford University Press.
Caughley, J. (1984). Imaginary social worlds: A cultural approach. Lincoln, NE:
University of Nebraska Press.
Chandler, J. (1985). American televised sport: Business as usual. In W. Umphlett
(Ed.), American sports culture: The humanistic dimensions (pp. 83-97).
Lewisberg, PA: Bucknell University Press.
Charlton, A., & Blair, V. (1989). Predicting the onset of smoking in boys and girls.
Social Science and Medicine, 29, 813-818.
Chen, K. (1997, January 6). Paradise tossed: Chinese now lionize sports stars, tycoons,
singers not workers. The Wall Street Journal, p. 1 A.
Cialdini, R., Kallgren, R., & Reno, R. (1991). A focus theory of normative conduct: A
theoretical refinement and reevaluation of the role of norms in human behavior.
Advances in Experimental Social Psychology, 24, 201-234.

203
Clarkin, G. (1990, June). TV sports: Heres the score. Marketing and Media Decisions,
pp. 19-21.
Coakley, J. (1994). Sport in society. St. Louis: Mossy.
Cohen, S. (1993). Television in the lives of children and their families. Childhood
Education, 70, 103-104.
Cole, M. (1996). Cultural psychology: A once and future discipline. Cambridge, MA:
Belknap Press of Harvard University Press.
Coleman, J. (1961). The adolescent society. New York: Free Press.
Collins, D. (1999, October 19). Saints dock Ditka $20,000 for gestures. USA Today,
Section C, p. 12.
Condry, J. (1989). The psychology of television. Hillsdale, NJ: Erlbaum.
Coopersmith, S. (1967). The antecedents of self-esteem. San Francisco: Freeman.
Conrad, K. (1987). Why children start smoking cigarettes: Predictors of onset. British
Journal of Addiction, 87, 1711-1724.
Covington, M., & Beery, G. (1976). Self-worth and school learning. New York:
Holt, Rinehart, & Winston.
Crepeau, R. (1985). Where have you gone, Frank Merriwell? The decline of the
American sports hero. In W. Umphlett (Ed.), American sports culture: The
humanistic dimensions (pp. 76-82). Lewisberg, PA: Bucknell University Press.
Davies, R. (1994). Americas obsession: Sports and society since 1945. Forth Worth,
TX: Harcourt Brace.
Decker, P., & Nathan, B. (1985). Behavior model training: Principles and training.
New York: Praeger.
Deford, F. (1969, June 9). At what price heroes?. Sports Illustrated, pp. 33-40.
Delozier, W,. & Tillman, R. (1972). Self image concepts can they be used to design
marketing programs. Southern Journal of Business, 7, 9-15.
Denzin, N. (1996). More rare air: Michael Jordan on Michael Jordan. Sociology of
Sport Journal, 13, 319-324.

204
Diamond, E. (1993, December 4). Covering Powell: Is hero worship replacing
investigative reporting? National Journal, pp. 2902-2903.
Dolich, I. (1969). Congruence relationships between self images and product brands.
Journal of Marketing Research, 6, 80-84.
Donohew, L., Helm, D., & Hass, J. (1989). Drugs and the (Len) Bias on the sports page.
In L. Wenner (Ed.), Media, sports, and society (pp. 225-237). Newbury Park,
CA: Sage.
Dorr, A., Kovaric, P., & Doubleday, C. (1989). Parent-child coviewing of
television. Journal of Broadcasting and Electronic Media, 33, 35-51.
Douvan, E., & Gold, M. (1969). Adolescent development: Readings in research and
theory. Boston: Allyn & Bacon.
Duck, J. (1990). Children's ideals: The role of real-life versus media figures. Australian
Journal of Psychology, 42, 158-176.
Duncan, M., & Brummett, B. (1993). The mediation of spectator sport. In A.Yiannakis,
M. Melnick, & T. McIntyre (Eds.), Sport sociology: Contemporary themes
(4th ed.) (pp. 217-227). Dubuque, IA: Kendall/Hunt Publishing.
Duthie, J. (1971). Sport: A component of the counter-culture. Unpublished
paper. University of Windsor, Windsor, Ontario, Canada.
Edwards, H. (1994, November). The athlete as role model: Relic of Americas sports
past? Sport, p. 32.
Erikson, E. (1950). Childhood and society. New York: Norton.
Fairlie, H. (1978). The parties: Republicans and democrats in this century. New York:
St. Martins Press.
Farhi, P. (1997). A television trend: Audiences in black and white. In S. Biagi &
M. Kem-Foxworth (Eds.), Facing differences: Race, gender, and mass media
(pp. 23-28). Thousand Oaks, CA: Pine Forge Press.
Farhi, P. (1997, November 11). Deals may double NBAs TV fees. Washington Post,
Section C, p. 1.
Farrey, T. (1997, September 3). Why athletes go bad. World Wide Web site
http ://www. ESPNET. SportsZone.com/ editors/gen/ features/crime/index3

205
Fatsis, S. (1998, February 6). NBA bravely plans for post-Jordan era. Wall Street
Journal, Section B, p. 1.
Fehr, S. (1997, October 31). Pricey new sports venues help make Washington number 1
for high cost tickets. Washington Post, Section C, p. 1.
Felker, D. (1974). Building positive self-concepts. Minneapolis: Burgess.
Figler, S., & Whitaker, G. (1995). Sport and play in American life. Dubuque, IA:
Brown & Bechmark.
Fink, A. (1995). The survey handbook. Thousand Oaks, CA: Sage.
Fishwick, M. (1983). Introduction. In R. Browne & M. Fishwick (Eds.), The hero in
transition (pp. 5-13). Bowling Green, OH: Bowling Green University Popular
Press.
Fishwick, M. (1985). Seven pillars of popular culture. Westport, CT: Greenwood.
Fishwick, M. (1979). The hero in the context of social change. In J. Hague (Ed.),
American character and culture in a changing world (pp. 339- 348).
Westport, CT: Greenwood.
Franks, J. (1996, February 22). NFL Moon found not guilty of wife-beating.
Reuters News Service, BC cycle.
French, J., & Pena, S. (1991). Childrens hero play of the 20th century: Changes
resulting from televisions influence. Child Study Journal, 21, 79-94.
Frey, D., & Carlock, C. (1989). Enhancing self-esteem. Muncie, IN: Accelerated
Development.
Gecas, V., & Mortimer, J. (1987). Stability and change in the self-concept from
adolescence to adulthood. In T. Honess & K. Yardley (Eds.), Self and identity:
Perspectives across the lifespan (pp. 265-286). London: Routledge & Kegan
Paul.
Gergen, K. (1971). The concept of self. New York: Holt, Rinehart, & Winston.
Goldberg, K. (1995, November 10). Violence out of bounds: For some athletes,
women are the enemy. The Washington Times, Section B, p. 1.
Goldberg, D. (1998, January 22). Football owners officially approve TV deal.
Associated Press.

206
Golden, M. (1986, February). For single mothers: Choosing a male role model for your
child. Essence, pp. 134-135.
Goode, E. (1999, May 20). Study finds TV alters Fiji girlsview of body. New York
Times, Section A, p. 17.
Goodman, M. (1993, May/June). Where have you gone, Joe DiMaggio? Utne Reader,
pp. 103-104.
Graves, S. (1993). Television, the portrayal of African Americans, and the development
of childrens attitudes. In G. Berry & J. Asamen (Eds.), Children & television:
Images in a changing sociocultural world (pp. 179-190). Newbury Park, CA:
Sage.
Graziano, A., & Raulin, M. (1993). Research methods: A process of inquiry.
New York: Harper Collins.
Greenberg, B., & Atkin, C. (1982). Learning about minorities from television: A
research agenda. In G. Berry & C. Mitchell-Keman (Eds.), Television and the
socialization of the minority child (pp. 215-243). New York: Academic Press.
Greenberg, B., & Brand, J. (1994). Minorities and the mass media: 1970s to 1990s.
In J. Bryant & D. Zillman (Eds.), Media effects: Advances and theory in
research (pp. 273-314). Hillsdale, NJ: Erlbaum.
Groller, I. (1988, March). Who are kids heroes today? Parents Magzine, p.33.
Grubb, E., & Grathwohl, H. (1967). Consumer self-concept, symbolism, and market
behavior: A theoretical approach. Journal of Marketing, 31,22-27.
Gunter, B., McAleer, J., & Clifford, B. (1991). Childrens views about television.
Averbury, England: Aldershot.
Hackett, G., Esposito, D., & OHalloran, S. (1989). The relationship of role model
influences to the career salience and educational plans of college women.
Journal of Vocational Behavior, 35, 164-180.
Hall, E., & Post-Kammer, P. (1987). Black mathematics and science majors: Why so
few? The Career Development Quarterly, 35, 206-219.
Hargreaves, J. (1986). Sport, power, and culture: A social and historical analysis of
popular sports in Britain. London: St. Martins Press.
Harris, J. (1994). Athletes and the American hero dilemma. Champaign, IL: Human
Kinetics.

207
Harris, R. (1994). A cognitive psychology of mass communication. Hillsdale, NJ:
Erlbaum.
Harrison, K. (1998). The role of self-discrepancies in the relationship between
media exposure and eating disorders. Doctoral dissertation, University of
Wisconsin, Madison, Wisconsin.
Harter, S. (1982). The perceived competence scale for children. Child Development,
53, 87-97.
Harter, S. (1987). The determinants and functional role of self-worth. In N. Eisenberg
(Ed.), Contemporary topics in developmental psychology. New York: Wiley.
Harter, S. (1988). Manual for the self-perception profile for adolescents. Denver:
University of Denver.
Hattie, J. (1992). Self-concept. Hillsdale, NJ: Erlbaum.
Heath, T., & Solomon, G. (1997, March 30). For new arena, expensive salaries, fans
must pay the ticket. Washington Post, Section D, p. 1.
Hendricks, J. (1992). Learning to act old: Heroes, villians, or old fools. Journal of
Aging Studies, 6, 1-11.
Heroes of young America: 1980-1989. (1990). The world almanac and book of facts
1990. Newspaper Enterprise Association.
Higgs, R. (1978). Sports. In M. Inge (Ed.), American popular culture (pp. 225-291).
Westport, CT: Greenwood.
Higgs, R. (1981). Laurel & thorn: The athlete in American literature. Lexington, KY:
University Press of Kentucky.
Hilliard, D. (1984). Media images of male and female professional athletes: An
interpretive analysis of magazine articles. Sociology of Sport Journal, 1,
251-262.
Hoagland, E. (1974, March 10). Where have all the heroes gone? New York Times,
Section 6, p. 20.
Hook, S. (1957). The hero in history: A study in limitation and possibility. Boston:
Beacon.
How to be great. (1995). Psychology Today, p. 46.

208
Howard, J. (1995, May 15). Sending the wrong message. Sports Illustrated, p. 84.
Hunter, M. (1983). A need for heroes. Horn Book Magazine, 59, 146-154.
Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from childhood to
adolescence. New York: Basic.
Jensen, G., & Rojek, D. (1992). Delinquency and youth crime. Prospect Heights, IL:
Waveland.
Johnson, R. (1995, September). Ebony interview with Mike Tyson. Ebony, pp. 78-81.
Jung, J. (1986). How useful is the concept of role model?: A critical analysis. Journal fr-
of Social Behavior and Personality, 1, 525-536.
Katona-Sallay, H. (1993). The characteristics of adolescentsself-concept. In
L.Oppenheimer (Ed.), The self-concept: European perspectives on its
development, aspects, and applications (pp. 95-105). New York:
Springer-Verlag.
Katz, D. (1994). Just do it: The Nike spirit in the corporate world. Holbrook, MA:
Adams.
Katz, M. (1996, May 13). Cable networks carve sports niches. Broadcasting &
Cable, p. 40.
Kem, E. (1969). The modem hero: Phoenix or ashes? In V. Brombert (Ed.), Hero in
literature (pp. 266-277). New York: Fawcett.
Kellner, D. (1996). Sports, media culture, and race -- some reflections on Michael >
Jordan. Sociology of Sport Journal, 13, 458-467.
Kelly, C. (1993, May/June). Why do we need celebrities? Utne Reader, pp. 100-101.
Keteyian, A. (1998, October). Crime season. Sport, pp. 32-33.
King, L. (1996, February 28). The Larry King Show. CNN Network, Transcript 1682.
Kinikema, K., & Harris, J. (1992). Sport and the mass media. Exercise and Sport
Science Reviews, 20, 127-159.
Klapp, O. (1962). Heroes, villains, and fools: The changing American character.
Englewood Cliffs, NJ: Prentice-Hall.

209
Klein, A. (1991). Sugarball: The American game, the Dominican dream. New Haven,
CT: Yale University Press.
Klein, F. (1997, October 10). Student athlete raises questions about education. Wall
Street Journal, Section B, p. 10.
Kleinbaum, D., & Kupper, L. (1978). Applied regression analysis and other
multivariable methods. North Scituate, MA: Duxbury.
Koppett, L. (1981). Sports illusion, sports reality. Boston: Houghton Miffflin.
Krech, D., Cruchfield, R., & Ballachey, E. (1962). Individual in society. New York:
McGraw-Hill.
Kubey, R., & Csikszentmihalyi, M. (1990). Television as escape: Subjective experience
before an evening of heavy viewing. Communication Reports, 3, 92-100.
Lapchick, R. & Stuckey, D. (1993). Professional sports: The racial report card.
In D. Eitzen (Ed.), Sport in contemporary society: An anthology (pp. 164-168).
New York: St. Martins.
Larsen, O. (1968). Violence and the mass media. New York: Harper & Row.
Larson, M. (1993). Family communication on prime time television. Journal of
Broadcasting and Electronic Media, 37, 349-357.
Lehman, B. (1928). Carlyles theory of the hero. Durham, NC: Duke University
Press.
Lemer, M. (1957). America as a civilization. New York: Simon and Schuster.
Levin, H. (1957). Contexts of criticism. Cambridge, MA: Harvard University Press.
Licht, M. (1995). Multiple regression and correlation. In L. Grimm & P. Yamold
(Eds.), Reading and understanding multivariate statistics (pp. 19-58).
New York: Longman.
Lipsky, R. (1981). How we play the game: Why sports dominate American life.
Boston: Beacon.
Long, G. (1991). Social perceptions of sports figures: Dumb jocks, flawed
heroes, and superstars. In L. Diamant (Eds.), Psychology of sports, exercise, and
fitness (pp. 227-236). New York: Hemisphere.

210
Lord, L. (1999, March 22). DiMaggios America: The way we were. U. S. News &
World Report, pp. 18-21.
Loundon, D., & Bitta, A. (1979). Consumer behaviors. New York: McGraw-Hill.
Lowry, B. (1995, September 25). Networks cheered by TV violence study. Variety,
p. 34.
Lowery, S., & DeFleur, M. (1983). Milestones in mass communication research.
New York: Longman.
Loy, R., & Hesketh, B. (1984). The Aegon motif. In K. Olin (Ed.), The contribution of
sociology to the study of sport (pp. 18-35). Jyvaskyla: University of Jyvaskyla.
Lubin, H. (1968). The entertainer as hero. In H. Lubin (Ed.), Heroes and anti-heroes
(pp. 305-309). San Francisco: Chandler.
Lupica, M. (1994, November). Fall of the legends. Esquire, p. 62.
MacMechan, P. (1840). Heroes and hero worship. London: Oxford Press.
Mannheim, K. (1936). Ideology & utopia. New York: Harvest.
Malgady, R., Rogler, L., & Constantino, G. (1990). Hero/heroine modeling for Puerto
Rican adolescents: A preventive model. Journal of Consulting and Clinical
Psychology, 54, 469-474.
Malone, K. (1993, June 14). One role model to another. Sports Illustrated, p. 84.
Maslow, A. (1954). Motivation and personality. New York: Harper.
McEvoy A., & Erikson, E. (1981). Heroes and villians: A conceptual strategy for
assessing their influence. Sociological Focus, 14, 111-122.
McGuire, W., & McGuire, C. (1981). The spontaneous self-concept as affected by
personal distinctiveness. In M. Lynch, A. Norem-Hebeisen, & K. Gergen (Eds.),
Self-concept: Advances in theory and research (pp. 145-167). Cambridge,
MA: Ballinger.
McLuhan, M. (1964). Understanding media: The extensions of man. New York:
McGraw Hill.
McPherson, B., Curtis, J., & Loy, J. (1989). The social significance of sport.
Champaign, IL: Human Kinetics.

211
Meichenbaum, D. (1984). Teaching thinking: A cognitive-behavioral perspective. In
R. Glasser, S. Chipman, & J. Segal (Eds.), Thinking and learning skills (Vol. 2)
(pp. 407-426). Hillsdale, NJ: Erlbaum.
Messner, M., & Solomon, W. (1994). Sin & redemption: The Sugar Ray Leonard
wife-abuse story. In M. Messner & D. Sabo (Eds.), Sex, violence & power in
sports: Rethinking masculinity (pp. 53-65). Freedom, CA: Crossing Press.
Michael Jordans magical powers (1991, June 1). The Economist, p. 28.
Midgley, M. (1976). A simple mathematical theory of innovative behavior. Journal of
Consumer Research, 5,31-41.
Miller, T. (1983). Identification process and sensory impact of childrens television
programming on the preschool child. Child Study Journal, 13, 203-207.
Monaco, J. (1978). Celebration. In J. Monaco (Ed.), Celebrity: The media as image
makers (pp. 3-14). New York: Dell.
Muuss, R. (1975). Theories of adolescence. New York: Random House.
Nack, W., & Munson, L. (1995, July 31). Sportsdirty secret. Sports Illustrated,
pp. 62-74.
Nance, R. (1995, February 16). Hecklers irritate some players, amuse others.
USA Today, Section C, p. 1.
Newcomb, T. (1956). Social psychology. New York: The Dryden Press.
Nixon, H. (1984). Sport and the American dream. New York: Leisure Press.
Noble, G. (1975). Children in front of the small screen. Beverly, CA: Sage.
Novak, M. (1976). The joy of sports. New York: Basic.
OFaloain, J. (1956). The vanishing hero: Studies in novelists of the twenties.
Boston: Little & Brown.
Oppenheimer, L., & Oosterwegel, A. (1993). The self-system: Developmental changes
between and within self-concept. Hillsdale, NJ: Erlbaum.
Orenstein, P. (1994). Schoolgirls: Young women, self-esteem, and the confidence gap.
New York: Doubleday.
Oriard, M. (1982). Dreaming of heroes. Chicago: Nelson-Hall.

212
Osgood, C., Suci, G., & Tannenbaum, P. (1957). The measurement of meaning.
Urbana, IL: University of Illinois Press.
Ostlund, L. (1974). Perceived innovation attributes as predictors of innovativeness.
Journal of Consumer Research, 1, 23-29.
Owens, C., & Ascione, F. (1991). Effects of the models age, perceived similarity and
familiarity on childrens donating. The Journal of Genetic Psychology, 151,
341-357.
Parker, W., & Lord, S. (1993). Characteristics of role models for young
African-American men: An exploratory survey. Journal of Multicultural
Counseling and Development, 21, 97-105.
Paterno, S. (1994, July 23). Covering sports heroes: Sports editors discuss why
the 1989 wife arrest of O.J. Simpson was not taken seriously by the media.
Editor & Publisher, pp. 11-12.
Peart, K. (1996, April 12). Money + fame = hero? Scholastic Update, pp. 10-11.
Pekrun, R. (1993). Social support, achievement, evaluations, and self-concepts in
adolescence. In L. Oppenheimer (Ed.), The self-concept: European perspectives
on its development, aspects, and applications (pp. 108-119). New York:
Springer-Verlag.
Petty R., & Cacioppo, J. (1986). The elaboration likelihood model of persuasion.
In L. Berkowitz (Ed.), Advances in experimental and social psychology (Vol. 19)
(pp. 123-205). New York: Academic.
Petty, R., & Priester, J. (1994). Mass media attitude change: Implications of the
elaboration likelihood model of persuasion. In J. Bryant & D. Zillmann (Eds.),
Media effects: Advances in theory and research (pp. 91-122). Hillsdale, NJ:
Erlbaum.
Piers, E., & Harris, D. (1969). Piers-Harris childrens self concept scale. Los Angeles:
Western Psychological Services.
Pleiss, M., & Feldhusen, J. (1995). Mentors, role models, and heroes in the lives of
gifted children. Educational Psychologist, 30, 159-169.
Poindexter, P., & Stroman, C. (1981). Blacks and television: A review of the
research literature. Journal of Broadcasting, 25, 103-122.

213
Porter, D. (1983). Americas greatest sports figures. In R. Browne & M. Fishwick
(Eds.), The hero in transition (pp. 248-259). Bowling Green, OH: Bowling
Green University Popular Press.
Prebish, C. (1984). Heavenly father, divine goalie: Sport and religion. The Antioch
Review, 42, 312.
Rada, J. (1997). Colorblind-sided: Racial bias in network televisions coverage of
professional football games. In S. Biagi & M. Kem-Foxworth (Eds.), Facing
differences: Race, gender, and mass media (pp. 23-28). Thousand Oaks, CA:
Pine Forge Press.
Raglan, L. (1956). The hero: A study in tradition, myth, and drama. New York:
Vintage.
Raine, H. (1997, March 24). The glory and cost of childhood dreams. U.S. News and
World Report, p. 6.
Rao, B. (1979). The American fictional hero. New Delhi: Bahri Publications.
Reed, L. (1995). Secrets in the bedroom: Adolescentsprivate use of media. Journal of
Youth and Adolescence, 24, 535-550.
Reynolds, W., Ramirez, M., Magrina, A., & Allen, J. (1980). Initial development and
validation of the academic self-concept scale. Educational and Psychological
Measurement, 40, 1013-1016.
Richards, B. (1998, October 2). Nike says it plans to slash SI00 million on spending
for star endorsements. Wall Street Journal, Section B, p. 6.
Rogers, E., & Shoemaker, F. (1971). Communication on innovations: A cross-cultural
approach (2nd ed.). New York: Free Press.
Role models, sports and youth. (1985). Report from the National School Safety Center,
Malibu, CA.
Rollin, R. (1983). The Lone Ranger and Larry Skutnik: The hero as popular culture.
In R. Browne & M. Fishwick (Eds.), The hero in transition (pp. 14-45).
Bowling Green, OH: Bowling Green University Popular Press.
Rodin, J. (1992, January/February). Body mania. Psychology Today, pp. 56-60.
Rodman signs endorsement deal with Converse. (1997, January 26). Associated Press.

214
Rodman sued for sexual assault. (1999, September 29). World Wide Web site
http://www.mrshowbiz.go.com/news/Todays_Stories/990929/rodman092999
Rogers, C. (1951). Client-centered therapy: Its current practice, implications, and
theory. Boston : Houghton Mifflin.
Rosekrans, M. (1967). Imitation in children as a function of perceived similarity to a
social model and vicarious reinforcement. Journal of Personality and Social
Psychology, 7, 307-315.
Ross, I. (1971). Self-concept and brand preference. The Journal of Business, 44, 38-50.
Rummel, R. (1970). Applied factor analysis. Evanston, IL: Northwestern University
Press.
Runyon, K. (1977). Consumer behavior and the practice of marketing. Columbus, OH:
Charles E. Merrill.
Ryan, J. (1995, September). Why sports heroes abuse their wives. Redbook, pp. 83-89.
Ryan, T. (1929). The literature of American school and college athletics. New York:
The Carnegie Foundation.
Sabo, D., & Jensen, S. (1992). Images of men in sports media. In S. Craig (Ed.), Men,
masculinity, and the media (pp. 169-184). Newbury Park, CA: Sage.
Sage, G. (1993). Sport and religion. In D. Eitzen (Ed.), Sport in contemporary society:
An anthology (pp. 164-168). New York: St. Martins.
Schaefer, C. (1978). How to influence children. New York: Van Nostrand Reinhold.
Schafer, W. (1971). Sport socialization and the school. Paper presented at the
Third International Symposium on the Sociolgy of Sport, Waterloo, Ontario.
Schlesinger, A. (1968). The decline of heroes. In H. Lubin (Ed.), Heroes and
anti-heroes (pp. 341-365). San Francisco: Chandler.
Schneider, K. (1996, June 3). Mission Impossible. People Weekly, pp. 64-68.
Schwartz, J. (1971). Causes and effects of spectator sports. Paper presented
at the Third International Symposium on the Sociology of Sport, Waterloo,
Ontario.
Secord, P., & Backman, C. (1964). Social psychology. New York: McGraw-Hill.

215
Seiko, E. (1994, June 23). Rockets nix Knicks 90-84, win NBA title. Houston
Chronicle, Section B, p. 1.
Severin, W., & Tankard, W. (1992). Communication theories: Origins, methods,
and uses in the mass media. New York: Longman.
Seggar, J., Hafen, J., & Hannonen-Gladden, H. (1981). Televisions portrayal of
minorities and women in drama and comedy dramas, 1971-1980. Journal of
Broadcasting, 25, 277-288.
Shapiro, L. (1999, January 13). Networks prepare for lost air. Washington Post,
Section D, p. 5.
Shugar, C. (1988). In search of the modem hero. Childhood Education, 65, 202-205.
Siegler, R. (1991). Childrens thinking. New York: Prentice Hall.
Singer, M. (1991, May 14). Heroines and role models. Science, p. 249.
Sirgy, M. (1983). Social cognition and consumer behavior. New York: Praeger.
Simons, J. (1997, March 24). Improbable dreams. U.S. News and World Report,
pp. 46-52.
Smith, G. (1973). The sport hero: An endangered species. Quest, 19, 59-70.
Smith, L. (1985). Media images and ideal body shapes: A perspective on women with
emphasis on anorexics. Paper presented at the annual meeting of the Association
for Education in Journalism and Mass Communication.
Spring, J. (1993). Seven days of play. American Demographics, pp. 50-53.
SPSS base 9.0 applications guide. (1999). Chicago: SPSS Inc.
Starr, M. (1999, January 25). Basketball looks at life after 23. Newsweek, pp. 56-57.
Starr, M., & Samuels, A. (1997, December 15). Hoop nightmare. Newsweek, pp. 26-29.
Stotland, E., & Patchen, M. (1961). Identification and change in prejudice and in
authoritarianism. Journal of Abnormal and Social Psychology, 62, 265-214.
Stotland, E., Zander, A., & Natsoulas, T. (1962). Generalization of interpersonal
similarity. Journal of Abnormal and Social Psychology, 61, 250-256.

216
Suci, G., & Tannenbaum, P. (1957). The measurement of meaning. Urbana, IL:
University of Illinois Press.
T.V. sports. (1996, May 13). Broadcasting and Cable, p. 34.
Tapin, I. (1988). Why we need heroes to be heroic. Journal of Popular Culture, 22,
133-142.
Taylor, P. (1991, November 25). A model role model. Sports Illustrated, pp. 78-81.
Television and growing up; The impact of televised violence. (1972). Report to the
Surgeon General, United States Public Health Service. Rockville, MD: National
Institute of Mental Health.
Temple, K. (1992). Interview. In J. Coakley (1994), Sport in society, St. Louis: Mosby.
Tener, R. (1987). The significance of hero choice as related to self-concept and
ego strength in adolescence. Doctoral dissertation, University of Detroit.
Thomson, R. (1993). Sport and ideology in contemporary society. In A. Yiannakis,
M. Melnick, & T. McIntyre (Eds.), Sport sociology: Contemporary themes
(4th ed.) (pp. 90-96). Dubuque, IA: Kendall/Hunt.
Tiggerman, M., & Pickering, A. (1996). Role of television in adolescent womens body
dissatisfaction and drive for thinness. The International Journal of Eating
Disorders, 20, 199.
TKO: Tyson gets year in jail for assault. (1999, February 6). Associated Press.
Tollefson, T. (1993, May/June). Is a hero really nothing but a sandwich? Utne Reader,
pp. 102-103.
Vesser, L. (1998, January 24). [Interview with Charles Groden]. The Charles Grodin
Show. CNBC Network.
Videbeck, R. (1958). Properties of the concept role. Midwest Sociologist, 20, 104-108.
Voigt, D. (1978). Myths after baseball: Notes on myths in sports. Quest: Literature
Myth, 30, 46-57.
Vogt, W. (1993). Dictionary of statistics and methodology. Newbury Park, CA: Sage.
Walden, D. (1986, January). Where have all our heroes gone? USA Today Magazine,
pp. 20-25.

217
Wecter, D. (1941). The hero in America: A chronicle of hero-worship. Arm Arbor, MI:
University of Michigan Press.
Weisberg, H., Krosnick, J., & Bowen, B. (1989). An introduction to survey research
and data analysis. Thousand Oaks, CA: Sage.
Weisman, J. (1993). Big-buck basketball: Acolytes in the temple of Nike. InD. Eitzen
(Ed.), Sport in contemporary society: An anthology (pp. 164-168). New York:
St. Martins.
Wells, L., & Marwell, G. (1976). Self-esteem: Its conceptualization and measurement.
Beverly Hills, CA: Sage.
Whalen, J., & Jensen, J. (1993, October 11). Is Shaquille the air apparent to endorser
crown? Advertising Age, p. 49.
Whitman, C. (1969). The matrix of heroism: Ajax. In V. Brombert (Ed.), The hero in
literature (pp. 61-72). Greenwich, CT: Fawcett.
Wilson, B., & Sparks, R. (1996). Its gotta be the shoes: Youth, race, and sneaker
commercials. Sociology of Sport Journal, 13, 398-427.
Wilson, J. (1994). Playing by the rules: Sport, society, and the state. Detroit: Wayne
State University Press.
Womens Sports Foundation. (1998). The future of womens sports and fitness.
Retrieved January 29, 2000 from the World Wide Web:
http://www.womenssportsfoundation.org
Wylie, R. (1961). The self concept. Lincoln: University of Nebraska Press.
Young, B. (1990). Television advertising and children. New York: Oxford Press.
Zahn-Waxler, C., & Smith, K. (1992). The development of prosocial behavior.
In V. Hasslet & M. Hersen (Eds.), Handbook of social development
(pp. 229-256). New York: Plenum.

BIOGRAPHICAL SKETCH
Keith Strudler received his bachelor of arts degree in English from Cornell
University in 1992. Upon graduation, Keith returned to his hometown of Houston,
Texas, to complete his master of education at the University of St. Thomas.
Concurrently, Keith worked as a junior high school English teacher at the Village School,
where he also founded the cross country and track and field programs for the school and
served as head coach for these programs. In 1994, Keith enrolled in the College of
Journalism and Communications at the University of Florida to work toward his Ph.D. in
mass communication. Keiths primary areas of study include children and media, sports
communication, and the use of new media in secondary education. Keiths other interests
include teaching and coaching children and adolescents, competing in triathlon races, and
playing guitar and singing for the rock band X Butlers.
218

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
shn Sutherland, Chair
Professor of Journalism and
Communications
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Kim Walsh-Childers, Cochair
Professor of Journalism and
Communications
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosoj
RutlVAlexander
Professor of Exercise and Sports Science
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Professor of Journalism and
Communications

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy^.
>.uri Kent
Professor of Journalism and
Communications
This dissertation was submitted to the Graduate Faculty of the College of
Journalism and Communications and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
May 2000
Dean, College of Journalism and
Communications
Winfred M. Phillips
Dean, Graduate School



170
Another potential explanation for this relationship is that subjects who were
extremely concerned with their physical appearance may not have had great interest in
prosocial behaviors. Perhaps a fixation on physical appearance represented a level of
vanity that did not correlate highly with charity work and other prosocial behaviors. As a
subject placed more emphasis on his appearance, he may have put less effort into
prosocial behaviors. Although merely speculative and not supported in the literature
review of this study, this relationship is a potential explanation for these unexpected
results, one which presents a topic worthy of future research.
Therefore, while relatively few significant relationships were found between real
distance scores and role modeling, it appears that these adolescents were most influenced
by their perceived similarity to their sports hero in athletic areas and related job
competence. Athletics-related distance scores had the greatest impact on role modeling
of sports heroes, and this role modeling was more likely to occur in athletics-related
behaviors.
For Hypothesis II (ideal distance), the only relationship that supported this
hypothesis was that between ideal romantic distance and the index for social
communication and the variable for getting a tattoo or earring like ones hero. Again, this
is a logical relationship, since both these role modeling variables relate to socially
oriented, non-athletic behaviors, and such behaviors would more likely be correlated with
a non-athletic domain of self-concept such as romantic appeal. It is interesting to note the
positive romantic association with tattoos or earrings, an obvious reflection of the modem
image of todays athletes. It would be difficult to imagine past generations viewing such
behaviors in a positive light. Beyond the ideal romantic distance score, no other distance
scores had significant relationships supporting H2, and one distance score, ideal athletic
distance, actually had significant relationships in the opposite direction as was predicted.
Therefore, H2 is not supported by the results of this research.


162
absolute value of the distance. Therefore, it is first important to review whether the
sports heroes images were rated as higher or lower than individual self-concept.
For seven of the eight domains, the image of the sports hero fell in-between the
real self-concept and the ideal self-concept. In other words, these adolescents viewed
their respective sports heroes as more competent or having a better image than
themselves, but the heroes did not quite measure up to their ideal self. The hero did not
represent an ideal, but he did represent a step in that direction from the real self. Given
the high image of these athletes in juxtaposition to the real self-concepts, sports heroes do
represent logical choices as role models for these adolescents and a logical means for
moving their real self toward their ideal self.
The only domain in which the image of the sports hero fell below the real
self-concept is the scholastic domain. Obviously, these adolescents did not perceive their
sports heroes as ideal students or even of equal competence as themselves in the
classroom. This could be attributed to the strong educational background of the majority
of the subjects for this study, as most of the surveyed adolescents attended a private
college preparatory school or a selective public school. It would be expected that
participants would not look to their sports heroes as role models in the scholastic area.
Interestingly, this expectation was not fully met, since the role modeling variable STUDY
(studying harder because of ones sports hero) did receive a fairly high mean score (2.82
on a scale from 1 to 4), indicating subjects did consider their sports hero as a potential
role model for academic affairs. Conversely, the role modeling variable COL (going to
the same college as ones sports hero) received a low mean score (1.62), the third lowest
score of all role modeling variables.
It is clear that the image of the selected sports heroes typically fell in-between the
real self-concept and the ideal-self concept. Using the calculated distance scores, these
relationships can be viewed in terms of the actual distances for each domain. This
indicates exactly which particular domain of self-concept, for both real and ideal self, was


26
accomplishment, but instead due to mass exposure and visibility. Boorstin provided the
following description of the American celebrity as:
a person who is known for his well-knownness. His qualities or lack of
qualities illustrate our particular problems. He is neither good nor bad,
great nor petty. He is the human pseudo-event. He has been fabricated on
purpose to satisfy our exaggerated expectations of human greatness.
(1961, pp.57-58)
Unlike heroes, celebrities do not display a model of greatness. They are a product of the
media as much as they are a product of their accomplishments. Celebrities do not
consistently display excellence, and they exhibit ordinary behaviors and accomplishments
(Klapp, 1962; Kelly, 1993; Monaco, 1978).
Therefore, many theorists of the 1960s and 1970s believed, largely due to
Americas increased reliance on mass media and television, that the celebrity had
replaced the hero in American society, and the heroic figure no longer existed in America
(Schlesinger, 1968; Monaco, 1978; Hoagland, 1974; Kelly, 1993; Harris, 1994; How
to, 1995). This cynical view blamed television for initiating the extinction of the heroic
figure. This viewpoint also expressed a static view of the hero. Accordingly, this heroic
figure did evolve over time and was eventually replaced by the celebrity. However, many
theorists of the 1980s and 1990s have refuted these arguments, writing instead about the
ongoing evolution of the hero into what can be called the modem American hero
(Fishwick, 1979; Fishwick, 1983; Fishwick, 1985; Browne, 1983; Walden, 1986;
Harris, 1994). These authors argued that heroes have not disappeared, but instead heroic
qualities have changed with the growing influence of mass media. In particular, because
mass media have made heroes more available to the public, Americans are given a larger
number of heroic choices, and these heroes cannot maintain the longevity of heroes of
past generations.
Mass media exposure has become a critical component of modem heroic stature,
and therefore the modem hero does share characteristics with the less substantial


172
to either copy social behaviors of an athlete or get a tattoo or earring like his hero in order
to strive towards the ideal romantic self, it is much more difficult to copy physically
demanding athletic behaviors of a favorite athlete when they do not perceive themselves
as having similar athletic abilities as their sports hero, even if this athlete is representative
of the ideal self. For difficult behaviors, real distance scores are more crucial in
determining role modeling intentions, indicating a component of the individuals sense of
competence in those difficult areas. In less difficult behaviors, ideal distance scores are
more crucial, as this represents an opportunity to strive toward the ideal without a fear of
low competency.
One questionable aspect of congruity theory addressed in the literature review was
the inability of this theoretical model to predict whether real self-concept or ideal
self-concept would predict purchase behaviors. Furthermore, the theory did not predict
when either of the two components of self-concept would be more predictive. This
research project can add to this understanding and provide some clarification to this
aspect of congruity theory. Congruity theory held true, albeit only to a limited degree, in
correlating real self-concept distance scores to modeling athletic behaviors and prosocial
behaviors, or behaviors that might be extremely difficult to copy without perceived
competence in these areas. If the real self-concept is far from the image of the sports
hero, it would likely seem extremely difficult to copy intricate athletic behaviors of the
sports hero. This is true for the job and athletic domains, domains that are related to the
athletic performance of the sports hero. Conversely, congruity theory held true, albeit to
an even lesser degree, to correlate ideal self-concept distance scores to social behaviors
and behaviors related to physical appearance, or behaviors not nearly as difficult to model
as athletic behaviors or even prosocial behaviors. This was found to be true for the
romantic domain of self-concept, an area that is not necessarily crucial to the sports hero.
To place this in the same language used in marketing studies, real self-concept
distance scores correlate with role modeling the more utilitarian behaviors of the sports


77
4) Concurrent Validity: Concurrent validity is maximized when a new
measurement tool compares favorably with an existing measurement tool
that is already considered valid (Fink, 1991). Concurrent validity
increases with high correlation between the new survey instrument and the
existing measurement tool. For this research, the survey instruments used
to assess self-concept and the image of the sports hero are adaptations of
Harters self-concept test, and therefore new tests have not been created to
replace existing, more valid measures. For the created measurement tool
of role modeling intentions, no known existing measurement tool was
found. Therefore, correlation between this new tool and an existing tool
cannot be calculated, and concurrent validity could not be established for
this measure.
5) Construct Validity: Construct validity refers to how well the studys
results support the theory or constructs behind the research and asks
whether the theory supported by the findings provides the best available
theoretical explanation of the results (Graziano & Raulin, 1993, p. 171).
Therefore, a focal component is to eliminate rival hypotheses and other
theoretical explanations for the research results. Threats to construct
validity are reduced by clearly stating definitions and building hypotheses
on well-validated constructs (Graziano & Raulin, 1993). For this study,
threats to construct validity have been reduced through building the
hypotheses on a sound theoretical foundation (congruity theory) that has
been used successfully in marketing research. Construct validity also
refers to the ability of a survey to distinguish between persons who have
certain characteristics and those who do not (Fink, 1991). One means of
establishing construct validity of a survey instrument is to check for
discriminant validity with another measure. In other words, construct


171
Even though all of the regression models produced relatively low values,
looking at the relative relationships found for real and ideal distance scores with role
modeling may provide insight into the motivations for adolescent role modeling of sports
heroes. Interestingly, the relationships found for real distance scores were with domains
of athletic and job performance, while the few relationships found for ideal distance were
with the romantic domain. There is an obvious and distinct difference between these two
relationships. In the first hypothesis, it was found that the closer an adolescent was to his
sports hero in athletic areas or job performance related areas (which, for sports heroes,
would essentially refer to athletic accomplishments), the more likely he was to role model
that athlete, particularly in athletic and prosocial behaviors. However, the ideal athletic
distance scores and job distance scores did not have the same significant relationships
with role modeling. Therefore, while these adolescents tended to choose sports heroes
that were closer to their ideal self than their real self, particularly for the athletic domain,
it was conversely the close proximity with real self, not ideal self, that correlated with
role modeling behaviors, particularly for athletic behaviors. Generally speaking, the
adolescent might choose a sports hero who represents his athletic ideal, but if he does not
perceive himself as actually being similar to this sports hero, the task of role modeling an
athletes behaviors may appear too difficult or daunting. This could be particularly true
for athletic behaviors, which, due to the very nature of the sports hero, are extremely
difficult to model. As an adolescent perceives himself more similar to his sports hero as
an athlete, this adolescent would express more willingness to role model the athlete in
these difficult behaviors due to his own sense of ability and likelihood to succeed.
Conversely, although the ideal romantic distance scores did have significant
relationships with both social behaviors and getting a tattoo or earring, such social
behaviors are more easily accomplished than the athletic behaviors of a sports hero. It is
much easier to strive towards the ideal in these behaviors (social, tattoo or earring) than to
achieve ideal athletic behaviors. In other words, while it is fairly simple for an adolescent


38
Self-Concept
Introduction and Definitions
Much of the literature and debate dealing with the choice of sports heroes has
focused on the qualities of the sports figure, not the individual qualities of the sports fans.
For example, theorists have debated whether the sports hero is determined solely through
athletic feats, or whether he must also personify outstanding qualities in his life outside of
sports (Harris, 1994; Walden, 1986; Smith, 1973). The limited discussion of the
qualities of sports fans typically has focused on group affiliation and characterization of
sports fans, including race, nationality, socio-economic status, and gender, and how these
general characteristics might influence the choice of a sports hero. (Harris, 1994a;
Wilson & Sparks, 1996; Lapchick & Stuckey, 1993; Klein, 1991). However, these
perspectives ignore the fact that selection of a sports hero is an individual choice based on
the individual characteristics and needs of each person. One potential means of
addressing such individual characteristics in a standardized and measurable method is to
examine self-concept in relation to the choice of a sports hero.
Self-concept is an extension of the term self that has been defined as the
individuals attitudes toward self (Frey & Carlock, 1989, p. 2; Secord & Backman,
1964). The self serves as the reference point for each individuals view of the world.
From self, theorists have established self-concept. Self-concept has been defined with
several slight variations, as explained below.
The simplest definition of self-concept is myself as I see myself (Loundon &
Bitta, 1979, p. 373; Dolich, 1969; Landon, 1974; Delozier & Tillman, 1972). This is a
summation of the process by which persons evaluate their own selves to form their
self-concept. Newcomb (1956) elaborated, defining self-concept as the individual as
perceived by that individual in a socially determined frame of reference. This definition
allows for the context of environment to influence the formation of self-concept, and it


145
affect the results of regression analysis, it was hoped that including all the distance worth
scores (real and ideal) would increase the prospect of achieving statistically significant
results (Licht, 1995).
The results of all of the regression analyses for these two hypotheses are detailed
in the following tables. The tables are organized by dependent variable (role modeling
index/variable), not by hypotheses. Therefore, all three regression analyses for each of
the seven dependent variables are grouped together. A more general discussion of the
individual hypotheses follows these results. For each of the significant regression
analyses that produced a significant F-value (p less than .05), one table details both the
overall significance of the regression (including the R and values and the F-value) and
the significance of the separate independent variables (distance worth scores). Following
each table is a description of the significant relationship(s) between the independent
variable(s) and the role modeling index/variable. This includes the Beta coefficient for
each relationship. (For each relationship, indicates significance at p less than .05, and
** indicates significance at p less than .01.)
For those regressions whose F Statistic is not significant at the .05 level, only a
brief text description of the analysis is given. In the proceeding section, for each role
modeling index/variable examined in this research, the regression analysis with only real
distance worth scores is explained first, followed by the regression analysis with only
ideal distance worth scores. The regression analysis with both real and ideal distance
worth scores is described last for each index/behavior.
Following the tables of the separate regression analyses, one overall table
summarizes the results of these analyses. This table indicates which role modeling
indexes/variables were significantly related to the distance worth scores and which
particular distance worth scores had significant relationships with each of these
indexes/variables.


79
of the subjects, attrition, instrumentation, etc.) did not affect this study. It
is possible that the sequence of the administration of exams could
potentially affect the answers given on the questionnaires. This is often
controlled by using more than one order of administration of
questionnaires (Graziano & Raulin, 1993). However, for this research, it
was important that subjects used the questionnaire of real self-concept to
anchor the other tests, and therefore one order of administration, with the
real self-concept questionnaire completed first, was used.
8) Subject and Experimenter Effects: Subject effects can threaten the
validity of a study when subjects give socially desirable answers or
unintentionally (or intentionally) try to satisfy the demands of the
researcher through their answers (Graziano & Raulin, 1993). For this
research, it is possible that participants might report higher or lower levels
of behavioral intentions to model a sports hero depending on what they
deem more socially desirable. Additionally, they may name a sports hero
who may not actually be their sports hero, but rather a person they feel
would be a more socially desirable sports hero. These threats to validity
have been reduced by assuring participants that there are no correct or
incorrect answers and by limiting the questions of role modeling to those
that were more likely to be answered honestly. Therefore, questions
concerning anti-social behaviors such as domestic abuse and recreational
drug use were not included. Experimenter effects could occur in this study
if the experimenter tried to influence the answers of the subjects (Graziano
& Raulin, 1993). This was reduced by having teachers administer the
actual surveys and by closely monitoring the procedure. Because there are
no observational data (as in experimental research), the threats to validity
from experimenter effects are minimal in this study.


36
between players; perhaps the most notable of these occurred when boxer Mike Tyson bit
Evander Holifields ear during a title fight. These behaviors also include athlete
confrontations with fans, coaches, and officials (TKO: Tyson, 1999). For example, in
a much publicized case, NBA star Latrell Sprewell, then of the Golden State Warriors,
attacked his coach, P. J. Carlisimo, twice during a practice session (Starr and Samuels,
1997). In recent years, several players have been suspended in recent from league play
for physically attacking a referee. These include Nick Van Exel and Dennis Rodman of
the NBA, Orlando Brown of the NFL, and Roberto Alomar of Major League Baseball,
who spit on an umpire after a disputed strike call (Collins, 1999). Rodman also has
incurred fines and a suspension for kicking a TV cameraman during a game (Starr and
Samuels, 1997).
Star athletes also have had conflicts with fans during games. Houston Rockets
guard Vernon Maxwell punched a heckling fan in Portland during a 1995 game, costing
Maxwell $20,000 and a ten-game suspension (Nance, 1995). Even professional coaches,
who themselves have become highly paid, popular sports figures, have displayed
extremely inappropriate behaviors during games. NFL coach Mike Ditka made an
obscene gesture to home fans during a home New Orleans football game (Collins, 1999).
Such negative on-court behaviors, including fighting, yelling at other players, shoving
and arguing with officials, kicking cameramen, and, in the most extreme of cases, even
physically harming a coach or fan, have become a somewhat regular occurrence for many
popular sports figures (Wilson & Sparks, 1996; Weisman, 1993; Katz, 1994; Harris,
1994; Harris, 1994b; Starr & Samuels, 1997; Smith, 1973; Walden, 1986; Coakley,
1994).
Negative off-the-court behaviors of popular athletes have become even more
troubling than such on-court behaviors, and many of these behaviors have resulted in
much more serious consequences. One such example is the case of Ray Carruth, wide
receiver for the NFL Carolina Panthers, who was indicted in December 1999 for


47
describing the process of self-actualization, which refers to the process of achieving ones
potential. According to this process, people are motivated to behave and act in order to
achieve their potential. However, according to Maslows theory, this need for
self-actualization is the last in a long list of human needs. Therefore, self-actualization
will be the last factor to impact behavioral decisions. Since Maslows theory of
connecting self-concept and behaviors is deeply embedded within a larger theory of
human needs and existence, it is somewhat difficult to isolate this concept within
Maslows work.
Much of the subsequent research relating self-concept to behaviors involved
marketing and product selection. According to such research, consumers choose products
which are consistent with their own image, or self-concept, and this can lead to a variety
of purchase decisions. Kretch et. al (1962) explained this relationship by describing how
self-concept motivated middle and upper-class Europeans to choose a lower quality of
bread due to the lofty image of this product.
the very foods that an individual chooses to eat must not only satisfy his
hunger but also be congruent with his conception of himself as a certain
kind of person. Thus the less nutritious white bread came to be preferred
to the more nutritious dark bread because the dark bread was reminiscent
of the European peasants black bread. The color of the bread had
become a status symbol and therefore, a determinant of how well it would
assuage the status seekers hunger, (pp. 83-84)
The choice of bread was motivated by congruence between the white bread and
self-concept. Because white bread was perceived as a high status product, it probably
was indicative of the ideal self-concept of these Europeans. Food was selected not only
to satisfy hunger, but also due to its congruence with oneself as a person.
Several studies have found consumers to choose products consistent with their
self-concepts (Landon, 1974; Sirgy, 1983; Loundon & Bitta, 1979; Runyon, 1977;
Hattie, 1992; Delozier & Tillman, 1972; Birdwell, 1964; Ross, 1971; Felker, 1974;
Dolich, 1961; Grubb & Grathwohl, 1967; Krechetal., 1962). Loundon and Bitta (1979)


206
Golden, M. (1986, February). For single mothers: Choosing a male role model for your
child. Essence, pp. 134-135.
Goode, E. (1999, May 20). Study finds TV alters Fiji girlsview of body. New York
Times, Section A, p. 17.
Goodman, M. (1993, May/June). Where have you gone, Joe DiMaggio? Utne Reader,
pp. 103-104.
Graves, S. (1993). Television, the portrayal of African Americans, and the development
of childrens attitudes. In G. Berry & J. Asamen (Eds.), Children & television:
Images in a changing sociocultural world (pp. 179-190). Newbury Park, CA:
Sage.
Graziano, A., & Raulin, M. (1993). Research methods: A process of inquiry.
New York: Harper Collins.
Greenberg, B., & Atkin, C. (1982). Learning about minorities from television: A
research agenda. In G. Berry & C. Mitchell-Keman (Eds.), Television and the
socialization of the minority child (pp. 215-243). New York: Academic Press.
Greenberg, B., & Brand, J. (1994). Minorities and the mass media: 1970s to 1990s.
In J. Bryant & D. Zillman (Eds.), Media effects: Advances and theory in
research (pp. 273-314). Hillsdale, NJ: Erlbaum.
Groller, I. (1988, March). Who are kids heroes today? Parents Magzine, p.33.
Grubb, E., & Grathwohl, H. (1967). Consumer self-concept, symbolism, and market
behavior: A theoretical approach. Journal of Marketing, 31,22-27.
Gunter, B., McAleer, J., & Clifford, B. (1991). Childrens views about television.
Averbury, England: Aldershot.
Hackett, G., Esposito, D., & OHalloran, S. (1989). The relationship of role model
influences to the career salience and educational plans of college women.
Journal of Vocational Behavior, 35, 164-180.
Hall, E., & Post-Kammer, P. (1987). Black mathematics and science majors: Why so
few? The Career Development Quarterly, 35, 206-219.
Hargreaves, J. (1986). Sport, power, and culture: A social and historical analysis of
popular sports in Britain. London: St. Martins Press.
Harris, J. (1994). Athletes and the American hero dilemma. Champaign, IL: Human
Kinetics.


9
lead some theorists to question whether the sports hero still truly exists in American
society (Rollin, 1983; Crepeau, 1985; Walden, 1986; Smith, 1973; Fairlie, 1978;
Nixon, 1984; Harris, 1994a; Farrey, 1997; Ryan, 1995; Coakley, 1994; Nack&
Munson, 1995; Goldberg, 1995), sports figures have frequently been chosen as heroes in
surveys of America teens (Heroes of, 1990; Simons, 1997; Harris, 1994a; Harris,
1994b). Such survey results refute the idea of a recent extinction of the American sports
hero and instead infer that the hero exists as a constantly evolving construct (Fishwick,
1983; Fishwick, 1985; Rollins, 1983; Harris, 1994a). In other words, the qualities once
considered essential for the sports hero may be very different than those of the modem
sports hero.
Sports fans now seem willing to overlook negative, or perhaps unheroic, activity
in certain areas of the lives of their sports heroes. This is likely due to the impressive
physical skills of the popular sports figure. A star athlete, while performing spectacular,
seemingly impossible feats on the playing field or in the arena, transcends typical mortal
athletic potential and appears, if only for that brief period of time, almost god-like
(Nixon, 1984; Oriard, 1982; Goodman, 1993; Walden, 1986; Birrell & Cole, 1981;
Loy & Hesketh, 1984). Sports may be one of the few remaining areas of American
society in which one consistently can find greatness, and fans may be willing to overlook
negative activity to embrace this greatness in their sports heroes. This, coupled with the
astonishing growth of sports television, has made sports a very popular and frequent area
from which American society finds its heroes. As Joan Ryan wrote (1995), Every
culture has its gods, and ours hit baseballs, make baskets, and score touchdowns (p. 85).
Regardless of the views one takes of the heroic condition in American society, the fact
remains that the popular sports figure has continued to grow as a popular choice of hero
for American society, and television has contributed greatly to this growth.


22
individual becomes more confident in his or her ability to perform a behavior, he or she
will be more likely to model such behaviors.
Another factor influencing modeling is the perceived benefits the role model gains
from his or her behaviors (Ostlund, 1974; Midgley, 1976). Role models who are
rewarded for their actions are more likely to be imitated than those who are not rewarded.
This has lead to the belief that popular, mass-mediated role models are often very
appealing to observers because they are well compensated with fame and wealth (Simons,
1997; Wilson & Sparks, 1996). Viewers, therefore, will attempt to role model those
behaviors perceived as instrumental in achieving such fame and wealth. Simons (1997)
explained this appeal of wealthy, mass-mediated figures as potential role models to
certain groups of adolescents, using basketball star Michael Jordan as an example.
Simons wrote,
a lot of them (low-income, African-American adolescents) will tell you
they want to be like Mike.. In this context, being like Mike doesnt mean
becoming an entrepreneur, a corporate spokesman, or a college graduate.
It means being a high-flying, windmill slammer of a ballplayer, (p. 50)
In other words, the most glamorous aspects of Michael Jordan (athletic fame, financial
riches from athletic accomplishment) made Jordan appealing as a role model to these
adolescents.
Self-Concept as a Condition for Role Modeling
Most research on the impact of role models, and specifically athletes as role
models, has focused on the characteristics of the potential role model; the societal role of
the role model; and easily observable characteristics of the observer, such as race,
socio-economic status, and perceived positive or negative value to society (Simons, 1997;
Malone, 1993; Role models, 1989; Wilson & Sparks, 1996; Chen, 1997; Golden,
1996; Walden, 1986; Pleiss & Feldhusen, 1995). This dissertation addresses the
relationship between role model and observer from a slightly different perspective.
Instead of measuring the aforementioned characteristics, this research examines


92
(less importance for a domain), the more the distance worth scores will increase from the
distance score. Therefore, higher domain importance scores result in lower relative
distance worth scores. In other words, if two distance scores are equal but the domain
importance scores for that domain are different, the lower relative distance worth score
will be the one with the higher domain importance score.
The following hypothetical examples of the creation of two distance worth scores
further explain this concept.
Example of the Creation of Distance Worth Scores
In Example A, Frank and John are two hypothetical subjects. Both subjects have
identical real distance scores in the athletic domain (distance score = 5). In other words,
both Frank and John have the same congruity between their real athletic self-concept and
their athletic image of their individual sports hero. However, Frank has a higher domain
importance score than John for the athletic domain. Franks domain importance score is
three (3), while Johns domain importance score is two (2). In other words, Frank places
greater value on the athletic domain of self-concept than John does. Using the equation
presented in the previous section, distance worth scores have been calculated for the two
hypothetical subjects, and these scores are represented in the following table.
Table 3-1. Example A of the Calculation of Distance Worth Scores.
Real Athletic
Athletic Domain
Reduced Domain
Real Athletic Dist
Subject
Distance Score
Importance Score
Importance Score
Worth Score
Frank
5
3
0.75
6.67
John
5
2
0.5
10
In this example, while both subjects had identical distance scores, the subject who
placed greater importance on the athletic domain (Frank) had a lower relative distance
worth score for the athletic domain. John placed less value on this domain and had a
lower domain importance score, so his resulting distance worth score was higher.


45
self-evaluations that can occur with traditional, non-mass mediated models (Bumstein et
al., 1962). Therefore, personal interaction is not necessary to impact self-concept, and
media figures can help to shape the self-concept of a viewer. This may be particularly
important to adolescents who, in an effort to establish their own individual identity, often
turn from more traditional sources of persons, such as parents, and look toward new
sources, such as mass media figures, to define their self-concepts (French, 1991;
Katona-Sallay, 1993). Furthermore, most people form their self-concepts during
childhood and adolescence, and self-concept becomes more resilient to change during
adult years (Krech et al., 1962; Orenstein, 1994; Hattie, 1992; Katona-Sallay, 1993;
Frey & Carlock, 1994). Therefore, research into adolescent self-concept formation is
crucial, because this represents a period when self-concept is still volatile and changing
and is a time when one seeks new sources and people, including media figures, who may
help form ones self-concept.
Self-Concept Maintenance
A primary human objective is to maintain and enhance real self-concept, and this
goal influences human behaviors and interactions (Hattie, 1992; Runyon, 1977; Grubb
& Grathwohl, 1967; Felker, 1974; Frey & Carlock, 1989; Delozier & Tillman, 1972;
Sirgy, 1983; Wells & Marwell, 1976; Loundon & Bitta, 1979; Orenstein, 1994; Krech
et al., 1962). One prominent means of achieving this goal is to bring ones real
self-concept, or ones view of oneself, closer to ones ideal self-concept. Peoples
behavioral intentions and choices of friends and role models are often formed in an effort
to maintain or improve their real self-concepts. Therefore, people act in such a way as to
gain higher opinions of themselves or to preserve the opinion they have of themselves.
Such self-maintenance and self-improvement occurs because it can result in a higher
sense of self-worth (Orenstein, 1994; Harter, 1987; Harter, 1982; Hattie, 1992; Felker,
1974). In other words, improving ones self-concept can be a primary behavioral
influence.


133
Tattoo or Earring Modeling Behaviors
Real distance scores. No significant relationships were found in the regression analysis
for real distance scores on tattoo or earring modeling behaviors (p > .05).
Table 4-38. Regression for Ideal Self-Concept Distance Scores
on Tattoo or Earring Modeling Behaviors.
Unstandardized
Coefficients
Standardized
Coefficients
t
Sig.
B
Std. Error
Beta
(Constant)
OTT
.112
11.905'
IT
Ideal athletic distance
.043
.016
.247
2.662
.009
Ideal behavior distance
-.006
.012
-.045
-.467
.642
Ideal friend distance
.019
.014
.145
1.348
.180
Ideal job distance
.022
.016
.173
1.413
.160
Ideal physical distance
.000
.014
-.003
-.029
.977
Ideal romantic distance
-.036
.019
-.255
-1.937
.055
Ideal scholastic distance
-.004
.012
-.040
-.355
.723
Ideal social distance
-.007
.018
-.048
-.402
.688
R = .337
R Square = .114
F = 2.338
Significance of F = .022.
The only distance score with a significant relationship with tattoo or earring modeling
behaviors was ideal athletic distance (Beta = .24**, NOTE: This relationship is the
opposite direction as predicted).


124
sports hero is closer to the real self-concept than the ideal self-concept, reflecting both a
high ideal academic self-concept and a correspondingly low image of the sports heros
competence in academic areas. For six of the areas of self-concept, however, it can be
stated that the sports hero more closely approximates the ideal self, or who these subjects
want to be, than the real self, or who these subjects currently perceive themselves to be.
Results for Hypothesis I and Hypothesis.!!
The first hypothesis (HI) predicted a negative relationship between real distance
and role modeling intentions. Therefore, it was expected that there would be a negative
and significant relationship between the individual indexes of role modeling and the real
distance scores. To test this, a separate regression analysis was done for each role
modeling index, the overall role modeling index, and the individual variables TATT
(getting a tattoo or earring like ones hero) and ANDRO (using Andr or drugs like
ones hero). For each regression analysis, the role modeling variable/index served as the
dependent variable, while each of the eight real distance scores for the eight domains
served as the independent variables. Therefore, seven separate regression analyses were
done to test H1.
The second hypothesis (H2) predicted a negative relationship between ideal
distance and role modeling intentions. The regression analyses for this hypothesis were
identical to those done for HI, except that ideal distance scores were used instead of real
distance scores as the independent variables.
A third group of regression analyses were done that included both real distance
scores and ideal distance scores. These analyses, which addressed both HI and H2, were
the same as those done for real distance and for ideal distance, except that all 16 distance
scores (eight real distance, eight ideal distance) were used as the independent variables
for each of the seven regression equations. For each equation, one of the four role


165
domains, the distance scores for these domains were larger than the distance scores in
other domains.
In summary, the results have indicated that adolescents have a fairly high opinion
of their sports heroes, according to the Likert scale measurement, consistent among most
of the eight domains. While many athletes have been scrutinized for negative behaviors,
this was not reflected by the reported image scores for sports heroes in this study. These
reported images typically fell between the real self-concept and ideal self-concept for all
domains except the scholastic domain, in which the image of the hero fell below the real
self-concept. Therefore, in all but this one domain, these sports heroes represented a step
from the real self towards the ideal self. Further, when examining the calculated distance
scores between self-concept and the image of the sports hero, it was found that these
sports heroes were closer to the ideal self-concept than the real self-concept in all but two
domains of self-concept. Therefore, popular media images of sports heroes that stress
supreme athletic accomplishment and high social skills, but poor behavior and low
academic accomplishment, seem to have been mirrored in the variations in the image
scores and calculated distance scores from this study.
Role Modeling
A more drastic range of scores can be found amongst the individual role modeling
variable mean scores than for the mean scores for the domains of the image of the sports
hero. These differences in mean scores for different role modeling variables indicate that
these adolescents did choose the particular areas in which they plan to model their sports
heroes. Instead of simply modeling all behaviors of their sports heroes at the same rate,
adolescents may instead consider the individual behaviors, good and bad, and decide
which they intend to model. Additionally, a high variance for most of the role modeling
variables supports the notion that individual adolescents make individual choices about
their intentions to role model their particular sports hero, and not all adolescents will
model their sports heroes in the same areas to the same extent.



PAGE 1

1 THE MASS MEDIATED SPORTS HERO AS A ROLE MODEL FOR ADOLESCENT MALES: MEASURING SELF-COITCEPT CONGRUTTY WITH THE PERCEIVED IMAGE OF THE HERO TO PREDICT ROLE MODELING BEHAVIORS By KEITH ANDREW STRUDLER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2000

PAGE 2

For Seth, CJ, and X Butlers evaywhere.

PAGE 3

ACKNOWLEDGMENTS Thanks to my paraits and family for always believing tJiat I could make it and to my friends for not holding it against me if I did not. Thanks to all the professors who made me think and all the staff who made me remember. Finally, a special thanks to Dr. John Sutherland and Dr. Kim Walsh-Childers who, both in their own way, provided the guidance, knowledge, and patience necessary to write ftas dissertidon. iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGMENTS iii ABSTRACT -vi CHAPTERS 1 INTRODUCTION 1 Mass Media Research -3 Self-Concept Theory 1 1 Problran Statement 13 2 REVIEW OF LITERATURE 16 Role Models and Behavioral Modeling 16 The Sports Hero as a Role Model Figure 24 Self-Concept 38 Congruity Theory ^4 Hypotheses -58 3 METHODS .62 Subjects for Study 62 Tests and Te^ AdministratiQn .65 Issues of Validity for this Study .75 Data Analysis .80 4 RESULTS AND DISCUSSION .96 Descriptive Statistics 97 Self Concept, Image of Sports Hero, and Distance Scores 1 10 Results for Hypothesis I and Hypothesis II 124 Summary Chart of Significant Results of Regression Analyses for HI and H2 135 Summary of Results of HI and H2 138 Domain Importance and Distaike Worth 140 Results for Hypothesis III and Hypothesis IV 143 Summary Chart of Significant Results of Regression Analyses for H3 and H4 155 iv

PAGE 5

Summary of Results of H3 and H4 158 Summaiy of Results 159 5 CONCLUSIONS 161 Descriptive Statistics 161 Hypothesis I and Hypothesis II 168 Hypothesis III and Hypothesis IV .• 175 Limitations of the Study 178 Recommendations for Future Research 181 APPENDICES A SELF-CONCEPT/IMAGE OF THE SPORTS HERO QUESTIONNAIRE 1 88 B ANSWER GUIDE FOR SELF-CONCEPT/IMAGE OF THE SPCmTS HERO 193 C ROLE MODELING QUESTIONNAIRE 197 REFERENCES 200 BIOGRAPHICAL SKETCH 218

PAGE 6

Abstract of Dissertation Presented to the Graduate Sdiool of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE MASS MEDIATED SPORTS HERO AS A ROLE MODEL FOR ADOLESCENT MALES: MEASURING SELF-CONCEPT CONGRUITY WITH THE PERCEIVED IMAGE OF THE HERO TO PREDICT ROLE MODELING BEHAVIORS By Keith Andrew Strudler May 2000 Chair: John Sutherland Major Department: Journalism and Conmiunications The sports hero has become a popular and influential figure in modem American society. The growth of this social phenomenon can be attributed largely to heavy mass media coverage of sports, with a particular emphasis placed on the star athlete. This emphasis by a variety of media outlets, most notably television, allows fans to learn more about their favorite athletes, both positive and negative, than possible for previous generations of fans. Such coverage has created concern over the potential role modeling influence of star athletes, particularly such behavioral influoice on adolescents. Two primary areas of interest are the actual behaviors that are most likely to be role modeled and the conditions under which such behavioral role modeling will occur. A theoretical foundation to understanding human behaviors is provided by congniity theory. According to congruity theory, the closer one's self-concept is to his or her image of a consumer product, the more likely this person is to purchase this product. vi

PAGE 7

This study adapts congruity theory to examine behavioral role modeling of mass mediated sports heroes by American adolescent msles. Surveys were given to 172 9th and 10th grade male student athletes. These subjects first identified their individual sports hero. Subjects then used Barter's Self-Perception Profile for Adolescents to rate their real and ideal self-concepts and the image of the titeir sports heroes. Finally, subjects rated role modeling of their sports heroes for a variety of behaviors. Regression analyses were completed to examine whether the proximity between self-concept and the image of a sports hero, or congruity, correlated with role modeling bdiaviors of &e sports hero. According to the statistical analysis, behavioral role modeling correlated with congruity between real self-concept and the image of the sports hero for a limited number of behaviors. This correlation came primarily with sports-related behaviors. Ahnost no correlation was found for the congruity betweai ideal self-concept and the image of the sports hero. Additionally, the increasing importance of an area of self-concept did not correlate with increased role modeling as predicted. vii

PAGE 8

CHAPTER 1 INTRODUCTION Star athletes for years have assumed a heroic position in American society. Historically, these star athletes were assumed to r^%sent the best of Amoican society. As written by Lewis Lord in U.S. News & World Report as an dbituaiy to the recently deceased baseball hero Joe DiMaggio, the recoit death of baseball legend Joe DiMaggio recalls an cats wfaoi a sports hero csnbo^ed America's hopes and (faemas. Stcxies al>out and pictWK dTDiMf^giO are extremely evocative, representing as fliey do the way America was — cetdX least wished it was. (1999, p. 18) DiMaggio played his career in an era when television and media did not heavily influence and potentially overexpose star athletes and when such athletes did not command the enormous salaries relative to the society at large. Howev, mass media, particularly television, has had a profound infiuoice in recoit years on popular ^XHts and the fans' perspectives of star athletes. Certainly, fans now have greater opportunity to watch their favorite athletes perform. Fans also have greater insi^ into the personal lives of star athletes, an insight that often has exposed a less thas lieimc side of th^ stars. The money generated through media coverage has allowed star athletes to command yeariy salaries of millions of dollars, not including the incmne they doive from advising and oidorsemoits. Obviously, the pa-^)e(^ve of today's mass mediated sports stars differs greatly from that of the sports heroes of past generations. For a variety of reasons, many journalists and social aitics have argued that tiie sports hero no longer exists in modem American society. Popular journalist Nfike Liqnca wrote in 1994, 1

PAGE 9

2 America's sports heroes were once larger than life, but today evm tfie idea of the sports hero is dead in America. A great nimiber of sports heroes • have fallen off their pedestals in the public eye. Sports heroes have destroyed themselves, but the fans and the media have helped them along. The fans want to know too much about the players, and ibc me(fia are only too willing to give them what they want. (p. 62) Whether the sports hero still exists in America or whether this figure has merely changed with changing sodal conditions is d^atable. However, it is obvious that the star aMeto assumes a visible and lofty position in American society. Of greater importance than the labeling of star athletes is the potential influence of these highly conspicuous persons. It is often assiuned that popular star athletes, such as basketball players Michael Jordan and Charles Barkl^, are role models to many views, particularly children and adolescents (Peart, 1996). If young sports fans model positive behaviors of athletes, thai this relationship has a positive influaice on Amoican society. Positive behaviors of star athletes might include athletic skill, good sportsmanship, and conmiunity involvement. However, if adolescents model the highly publicized negative activities and charactedstics of many star athletes, including drug use, greed, and violent behavior, tfien this status is highly problematic for society at large (Benedict & Yaeger, 1998; Keteyian, 1998). Therefore, the more important debate about sports heroes focuses not only on the characteristics of these athletes, but rather whether these star athletes serve as role models to America's youth. This debate is comphcated by the increasing diversity of American society, fhe diversity of star athletes, and the huge number of sports stars accessible through various media outlets. Some journalists and social critics have written that athletes are no longer role models for American adolescoits. Harry Edwards expressed that opinion in Sport magazine (1994), writing, athletes have long been expected to be role models, but this notion is becoming outdated. The role model expectation is based on the assumption that star athletes and their fans share a uniform group of values, but this is no longer true. Indeed, the racial, ethnic, gender, and

PAGE 10

3 lifestyle diversity of the United States has shattered all but the jnetraise of consensus upon which this notion is founded. Athletes are sports entertainment celebrities with no more special calling, capability, or obligation to be role models than anyone else in the mass-oriented entertainment business, (p. 32) According to this viewpoint, not only have star athletes lost their heroic credential, but due to the complexity of modon American society, these athletes are no longo: role models. The question of whether sports heroes still serve as role models for young American sports fans is of great interest to parents, educators, and society in general. Further, it has become increasingly important to understand the specific bdiaviors that are most likely to be modeled and the motivations for individual fans to model their own favorite sports stars. This research project will address these topics to help gain a better understanding of the function of the modem star athlete in the lives of American adolescents. One area of study crucial to understanding the modem American sports hero is mass media theory and research. TTiis topic will be reviewed extoisively tluoughout this dissertation. Mass Media Research Years of acadonic research has focused on the vast impact mass media, and specifically television, historically have had upon society. This sdiool of research has produced theories predicting a range of effects, varying from theories examimng general societal impact of media to tfiose dealing with a narrow range of media types and populations. Obvious reasons exist for the ongoing interest in this research agenda. By 1980, televisions were ivesent in 98% of American households, a po-coitage that has held steady since that time (Harris, 1994b; Andreasen, 1990). Television viewing comprised an inordinate amount of American leisure time, more than four times more than any other

PAGE 11

4 single activity (Cohen, 1993; Spring, 1993). Americans watch, on average, four hours of television a day, and the television remains on for seven hours a day, on average, in the American household (Harris, 1994b; Kubey & Csikszentmihalyi, 1990). Statistics on television in the lives of children are even more staggering. American youths, on avaage, watch approximately 1,500 hours of television a year, fat greater than the 900 hours the American youfli will spend in school annually (Harris, 1994b). With so much of America's time invested in television, it is natural and vital that researchers attempt to understand the various in:q>acts of television contrait on society. Television and Society America's interest in television has produced a society somewhat dependent upon television for its daily existoice. Ball-Rokeach and DeFl^ (1976) explained such a relationship with their "dependency theory," which suggests that, "particularly in a modon iffban-industrial society, audiences have a high level of dQ)endence on mass media information" (Severin & Tankard, 1 992, p. 262). America, a highly industrial and increasingly urban society, exemplifies such a reliance on mass media information. This dependent relationship creates a need for continuous research on the ramifications of such dependence on American society. One area of particular interest for this research project is the potential for television viewers to copy the behaviors and actions of characters they see on television and to use television figures as role models. Much of the prior research has focused on violent behaviors and models, including such notable research works as the Surgeon Genotd's Report on Television Violraice, Television and Growing Up (1972), whidi established tentative relationships between viewing violrat television contoit and modeling of those behaviors (Lowry & Defleur, 1983). Other examples of research concerning modeling of television content have shown that viewers may model anti-social behaviors such as smoking and drinking (Charlton & Blair, 1989; Botvin et al., 1991; Conrad et al., 1987), prosocial behaviors (Cohen, 1993; Bandura, 1994), styles

PAGE 12

5 of dress and fashion (Muuss, 1975; Wilson & Sparks, 1996; Hams, 1994b), verbal expression (Muuss, 1975), and career selection (Muuss, 1975; Harris, 1994b). Research has indicated that many television viewers look to television for role models, and these televised role models have replaced traditipoal models such as family, peers, and teachers (Jung, 1986; Hendricks, 1992; Harris, 1994a; Pleiss & Fddhusoi, 1995; "Role models", 1989; Duck, 1990). This may be particularly true during adolescence, an age when children often reject traditional role models, such as parents, and seek new role nM)dels to help establish thdr own identity (Jung, 1986; Qrenstein, 1994; Krechetal., 1962; Hendricks, 1992; Felker, 1974). While one would hope television role models would display positive behaviors, this may not always be the case, and critics have suggested that much of television content pcntrays potential role models engaged in negative activity (Harris, 1994a). These potentially negative role models on television have drawn growing attention, and they are paramount in the study of the questionable behaviors of popular athletes as role models. Sports Television Sports television, central to this research project, has received growing attrition by networks and viewers alike in recent years. While the first televised sporting event was a college baseball game between Columbia and Princeton in 1939, sports did not gain much attention fix)m television networks until the 1960s (McPherson ei al., 1989; Davies, 1994; Coakley, 1994). Sports has been an inCTeasingJy vital part of the television industry since that time. Growing interest in sports television developed fi:om a vari^ of &ctors, including a gradual exodus from downtown living areas, where stadiums are located, to the suburbs; realization by networks that sports offered an opportunity to boost ratings during traditionally slow times, such as Sunday aftemoons; and the ability for advertises to consistently attract young male viewers (Davies, 1994; Nixon, 1984; Harris, 1994a; Harris, 1994b; Katz, 1996; McPherson et al., 1989; Coakley, 1994; Clarkin, 1990).

PAGE 13

6 Sports programming has increased through to the 1980s, when, on average, sports programming accoimted for 15% of total programming on commercial television, a percentage that has held steady through the mid-1990s (Harris, 1994a; Coakley, 1994). Several television networks, including ESPN and Fox Sports, are now devoted solely to sports contoit, 24 hours a day, making sports extremely accessible to television viewors. Along with the increase of sports media coverage has come a change in the contoit of televised sports. Much of this can be attributed to the enormous amounts of money networks pay for the rights to televise amateur mi professional sports. For example, in 1998 four networks (ABC, CBS, ESPN, and FOX) agreed to pay the National Football League (NFL) $17.6 billion over eight years for the right to broadcast professional football (Goldbo:^, 1998). This amount far surpasses the amount teams receive each year in ticket revenues. The National Basketball Association (NBA) has signed a similarly lucrative contract with two networks (NBC and Turner), receiving $2.4 billion for four years (Farhi, 1997). This does not include revalues received from lesser contracts signed with local networks for regional sports coverage. Given the dependence of major sports leagues on television for financial success, popuiBi sports are influenced greatly by television networics and orchestrated for television viewing evrai more so than for live spectating. This coverage may cause the television audience to believe that they are watohing the real game, not a media representation of die game (Tonple, 1992; Coakley, 1994; Kinkema & Harris, 1992; Nixon, 1994; Harris, 1994b). In oftcr words, viewers may be obUvious to the potential influence of networks to shape and influence the sporting events regularly seen on television. Additionally, the exoibitant prices of tickets to professional sporting contests (an average NBA ticket cost $36.32 in 19%; an average NFL ticket cost $40.93 in 1998) has made live attendance of such professional sporting contests virtually impossible for most Americans (Fdir, 1997; Heath & Solomon, 1997). Therefore, professional sports, an American institution rivaling religion in popularity and societal influence (Prebish, 1985; Novak. 1976; Sage, 1993), has

PAGE 14

7 become a phenomenon Americans see almost exclusively on television (Temple, 1992; Chandler, 1985). Coakley (1994) explained, the emphasis comes in a form that makes it seem as if the "real" game is the one on television, not the one seen in the stadium. In other words, television has constructed sports and spectator experiences in important ways. It's happened so smoothly that most people tiiink fliat vtbai Ihey watch a game on television they are experioicing sports in a "natural" form! (p. 332) What most Amoicans see and know about professional sports, they see and know from television. Considering the potential extent to which television can impact society, sports media should receive considerable attention from academics. Television S tars and Sports Hemes A focal point of televised sports and the focal component for this research project is the star athletic p^ormer. Through the process of mediation — using nairaticm, visuals, intimacy widi particular athletes, and constant fixture on star performers ~ television has placed emphasis on the best individual performers and their greamess (IXincan & Brummett, 1987; Weisman, 1993; Katz, 1993; Harris, 1994b; Kiiikana& Harris, 1992; Hargreaves, 1986; Figler & Whitaker, 1995; Shugar, 1988). Harris (1994b) wrote, often the "star mentality" of television and the mbextBintamt bunaess in general affects the presentation of sports in the media. The superstiurs C exalted and glorified far more than, for example, a show of fine teamwork on the field. Stories about sports tend to focus on individual performers, much like stories about religion focus on the Pope or Jerry Falwell or stories about government tend to focus on the pre^deiit (p. 1 29) This emphasis on the star has been accomplished not only through the actual coverage of regular sporting events, but also through other sports media, iodudu^ sports highli^ts programs, preand post-game programs, and entertainment-based programs that highlight individual athletes over team performance (Kinkema & Harris, 1992; Figler& Whitaker, 1995; Harris, 1994b). Networks have promoted individuals m their

PAGE 15

ongoing effort to increase the popularity of their programming, and the respective sports leagues hope to prosper from this coverage through the potoitial sale of moduoidise from these star performos and overall increased interest in their sports product. This emphasis on the individual has created a society in which television sports viewers can learn more about popular athletes than in prior times, and the star sports figure has become one of the most prominent and well-known persons in American society (McPherson, 1989; Nixon, 1984; Harris, 1994a; Harris, 1994b; Coakley, 1994; Duncan & BnmimeM, 1987; Kinkona & Harris, 1992). This focus on individual star athletes in sports television has promoted the growing societal interest in the sports hero. As previously mentioned, the sports hero has bei a part of American society for years before the existence of television, but television has contributed to changes in this construct (Porter, 1983; Smith, 1973; Voigt, 1978). Before the growth of sports television, sports heroes such as baseball legend Babe Ruth were accessible local figures, and many fans had the opportunity to meet their sports heroes in person (Walden, 1986; Katz, 1993). However, the lack of heavy coverage and scrutiny from mass media kept fans from knowing all the details, both good and bad, of the lives of thefr sports heroes. Due to television's current ubiquitous presence, while the sports fan today will not personally know his or her sports heroes or even have the opportunity to watch them play in person, these fans likely will know much more about their heroes (Walden, 1986; Nixon, 1984; Harris, 1994a; Harris, 1994b; McPherson et al., 1989; Malone, 1993; Patemo, 1994). Television has allowed greater access to sports heroes than ever before, as Caugliley (1984) explained: "Evoi an avmige fan knows more than a baseball player's name. He also knows his team, uniform number, position, batting average, salary, and something about his appearance, personality, medical history, and off-the-field conduct" (p. 32). Despite the increasingly visible information about the negative actions of popular sports stars, including rape, drug abuse, and a variety of criminal activities, which has

PAGE 16

9 lead some theorists to question whether the sports hero still truly exists in American society (RoUin, 1983; Crepeau, 1985; Walden, 1986; Smith, 1973; Fairlie, 1978; Nixon, 1984; Harris, 1994a; Farrey, 1997; Ryan, 1995; Coakley, 1994; Nack& Munson, 1995; Goldberg, 1995), sports figures have frequently been chosen as heroes in surveys ofAmerica teens ("Heroes of, 1990; Simons, 1997; Harris, 1994a; Harris, 1 994b). Such survey results refute the idea of a recent extinction of the Ammcan sports hero and instead infer that the hero exists as a constantly evolving construct (Fishwick, 1983; Fishwick, 1985; Rollins, 1983; Harris, 1994a). In other words, the qualities once considered essential for the sports hero may be very different than tiiose of the modem sports hero. Sports fans now seem willing to overlook negative, or peAaps unheroic, activity in certain areas of the lives of their sports heroes. This is likely due to the impressive physical skills of the popular sports figure. A star athlete, while performing spectacular, seemingly impossible feats on the playing field or in Ae arena, transcoids typical mortal athletic potential and appears, if only for that brief period of time, almost god-like (Nixon, 1984; Oriard, 1982; Goodman, 1993; Walden, 1986; Birrell & Cole, 1981; Loy & Hesketh, 1 984). Sports may be one of the few remaining areas of Amoican society in which one consistently can find greatness, and fans may be willing to overlook negative activity to embrace this greatness in their sports hm)es. This, coupled with the astonishing ^wtb of sports television, has made sports a vy p(^ular and frequent area from which American society finds its heroes. As Joan Ryan wrote (1995), "Every culture has its gods, and ours hit baseballs, make baskets, and score toudidowns" (p. 85). Regardless of flie views one takes of the heroic condition in American society, the feet remains that the popular sports figure has continued to grow as a popular choice of hero for Ammcan society, and television has contributed greatly to this growth.

PAGE 17

Societal Functions of Sp orts Heroes A more clouded but peili£q}s more important topic is die popular bdief that athletes inherently serve as role models, so much so that the terms "hero" and '*role moderareoftenusedinterchangeably ("Role models", 1989; Malone, 1993; Harris, 1994b; Simons, 1997; Wilson & Spaiks, 1996; Parker & Lord, 1993; Taylor, 1991; "Michael Jordan's", 1991; Goodman, 1993). In a column authored for Sports Illustrated magazine, NBA basketball star Karl Malone (1993) wrote, "We don't choose to be role models, we are chosen. Our only choice is whether to be a good role model or a bad one" (p. 84). A lesser held, contrary opinion is that while popular mass media figures such as sports heroes may be idolized, they often do not serve as role models because they are not perceived as similar enough to fans of the hero (GroUer, 1995; Goodman, 1993). In other words, while fans may enjoy watching their sports heroes, these fans cannot actually view these heroes as role models due to the vast percdved differraces betwem thCTiselves and these athletes. Yet another suggestion is that^torts hax>es can serve as role models to certain people for certain areas of their life (Simons, 1997; Wilson & Sparks, 1996; Harris, 1994a; Harris, 1994b; Dononhew, Helm, & Haas, 1989; Nixon, 1 984). Research about sports heroes should address whether tiie sports hero is a role model and the specific conditions under which a sports hero is most likely to serve as a role model. Authors have addressed flie potential for mass media figures to serve as role models. Caughley (1984) explained the potential relationship between media figures and their fans, writing, relationships are those in which the media figure becomes the object of intense admiration. Such relationships approximate the general stereotypic concq>tualization of the fen, but much more is involved than esthetic appreciation. Characteristically, the admired figure comes to represent some combination of idol, hero, altO' ego, mentor, and role model, (p. 53)

PAGE 18

As popular sports heroes have become mass mediated figures, they too are subject to the same types of relationships with their fans. Many athletes have become aware of their potential influence on tilie viewing public. Wdsanan (1993) wrote oti fliis topic, athletes' bdmvior, whether it involves drug use, the shoes they wear or their views on society, will be emulated. They are taught, very early in their careers, to be the best example of American manhood, or at the very least to escape with all their defects undetected, (p. 1 66) In OTd&[ for such a definitive statement to be made about the influence of televised star athletes, detailed r^earch should address the specific impact of these mass mediated figures. Critical insight and research in this area would help society to further understand the impact of the positive and negative actions of sports heroes. Self-Concept Theory Importance of Self-Concept One theoretical framework used to study motivations for human behavior is self-concept theory (Grubb&Grathwohl, 1967; Ross, 1971; Dolich, 1969; Hattie, 1992; Baughman& Welsh, 1962; Sirgy, 1983; Loundon& Bitta, 1979; Krechetal., 1962). Humans typically develop their wants and goals around maintaining and improving their self-concq)t, which can be defined as "myself as I see mysdf (Kretch et al., 1962; Loundon & Bitta, 1979; Sirgy, 1983; Hattie, 1992). This desire to maintain and improve self-concept is a strong predictor for purchase intentions (Grubb & Grathwohl, 1967; Ross, 1971; Dolich, 1969; Hattie, 1992; Baughman & Welsh, 1962; Su-gy, 1983; Loundon & Bitta, 1979; Krech et al., 1962). Self-concept theory also has been linked to behaviors such as choosing friends and academic efforts in school (Felker, 1974; Hattie, 1992). While most of the research on self-concept and behavior has focused on consumer behaviors, the link between self-concept and behavior has beai strongly estabhshed and

PAGE 19

12 is a valuable tool for future investigation into human behavior, Grubb and Grathwohl (1967) explained the importance of this topic, writing, the self represents a totality which becomes a principal value around which life revolves, something to be safeguarded and, if possible, to be made still more valuable. An individxial's evaluation of himself will greatly influence his behavior, (p. 204) Felker (1974) provided a more detailed explanation for the relirtion^p betwem self-concept and behaviors, writing, the role of self-concept is three-fold. The sdf-concept operates as a mechanism for maintaining inner consistaicy; &e sdf-concq)t determines how experiences are interpreted; and the self-conc^t provides a set of expectancies. Each of these three factors is a powerful determiner of behavior, (p. 7) Although self-concept theory has not previously been used to examine role modeling behaviors of sports heroes, use of this theory, which predicts that people continuously try to maintain and improve self-concept, provides a promising new framework to examine the motivations for such role modeling. Role Modeling Sports Heroes and Self-Concept Theory This research project will look at the behaviors involved in role modeling sports heroes as related to self-concept. Similar types of studies have been done for consumo: behaviors, determining tiiat similarity between self-concept and product image correlates with purchase behaviors (Dolich, 1969; Grubb & Grathwohl, 1967; Ross, 1971; Hattie, 1992; Baughman& Welsh, 1962; Sirgy, 1983; Loundon & Bitta, 1979; Kredietal., 1962). Similarly, researchers have suggested that mass media figures can serve as role models,particularly for adolescents (French &Pina, 1991; Bandura,.1992; Harris, 1994b; Caughl^, 1984). This research project incorp(ates fliese two concepts while examining a increasingly popular, criticized, and critical aspect of American society, the sports hero.

PAGE 20

13 Prnhlem Statement Americans, and particularly American adolescents, historically have held up popular sports figures as hm>ic figures. This is not surprising, as dieorists have asserted that the need for hopoes is inherent in human societies (Fishwick, 1983; Walden, 1986; Harris, 1994a; Schlesinger, 1968; Browne, 1983), and sports has become one of America's most celebrated pastimes and po^nifications of American dreams and values (Nixon, 1984; Porter, 1983; Goodman, 1993; Oriad, 1982; Ryan, 1995; Walden, 1986; Smith, 1973; Coakley, 1994; Crepeau, 1985; Harris, 1994a; "Role models", 1989; "Michael Jordan's", 1991). Given the growing mass media interest in professional and major collegiate sports, Americans now have more access to watch star athletes than American sports fans of generations past (*T^. aports", 1996; McPherson, Curtis, & Loy, 1989; Coakley, 1994; Davies, 1994; aaridn, 1990; Harris, 1994a; Hanisi tmbi KdSktBt, 1996; Caughley, 1984; Hargreaves, 1986; Kinkema & Harris, 1992; Denzin, 1996). Accordingly, the oj^rtunity to find heroes in sports appears aidless. In a sodety where popular athletes have become incredibly wealthy figures about whidi spectators know more than ever before, the issue of choosing a particular sports hero is a complex phenomenon. While theorists have suggested that heroes no longer truly exist in Amoica (Boorstin, 1961; Klapp, 1962; Monaco, 1978; Lubin, 1968; Kelly, 1993; McLuhan, 1 964), and many of today's popular athletes have displayed behaviors that certainly appeal "uidim>ic," athletes are still popular choices as heroes among Amoican youths and adolescents C*Heroes of, 1990; "Role models", 1989; Wilson & Sparks, 1996; Simons, 1997). Television has only aided this process, bringing heroes and hero worshippers togetfao* more easily (Higgs, 1978). However, one is left to wonder exactly what the impact of these sports heroes is on American adolescents.

PAGE 21

14 Critics and fans of sports heroes alike have suggested that popular American sports heroes serve as role models for American adolescents (Harris, 1994f^ Harris, 1994b; "Role models", 1989; Malone, 1993; Simons, 1997; Wilson & Sparks, 1996; Parker & Lord, 1993; Taylor, 1991; "Michael Jordan's", 1991; Goodman, 1993). Little scientific research has addressed tiie motivations for such role modding or the conditions under which such role modeling is most likely to occur for ibt individual adolescent. Socioeconomic status, race, and parental and peer influences have been suggested as potaitial factors affecting such role modeling (Harris, 1994a; *'Role models", 1989; Sunons, 1997; Wilson & Sparks, 1996; "Michael Jordan's", 1991; Goodman, 1993). However, these factors do not address the individual role modeling decisions of each adolescent and the motivations for the individual to model ihe behaviors of lus or her own particular sports hero. One area of research which has examined motivations for individual behavioral intentions is the study of individual self-concept and how this self-concept relates to the images of other objects (DoUch, 1969; Grubb & Gratfiwohl, 1967; Ross, 1971; Hattie, 1992; Baughman & Welsh, 1962; Sirgy, 1983; Loundon& Bitta, 1979; Krech et al., 1962). This research on self-concept has primarily been used to better understand consumer behaviors and individual purchase decisions. However, self-concept research also presents a potential means to better understand a variety of behaviors, including the emulation of sports heroes addressed in this research. This Aeoretical approach to studying adolescent role modeling of sports heroes has lead to the following research questions: 1) Among adolescent males, how do the relationships between self-concept and perceived images of a sports hero relate to behavioral intentions to role model a sports hero? In other words, how do the perceived similarities and difiB^ences between one's individual self-concept and the image of a sports hero relate to one's intentions to emulate that sports hero?

PAGE 22

15 2) What specific role modeling behaviors are most strongly related to these relationships between self-concept and the image of a sports hoo? 3) Which areas of self-concept are most influential in predicting adolwit role modeling of a sports hero? Chapter 2, the Literature Review, addresses prior research related to tfiese questions. Specifically, each construct used in this study is defined and discussed. Prior research on related subject matter is reviewed. Finally, Cbapta 2 addresses areas of research that have not been studied adequately as well as the ways that this research project adds to the base of knowledge and contributes meaningfiil information to the academic community. Three additional chapters are included in this dissertation. Chapter 3, Methods, addresses the specific methods and statistical analysis used in this dissotation. Cluq)ta' 4, Results and Discussion, details the results of the stated analysis. Finally, Chapter 5, Conclusions, addresses the results, discusses the importance of these results, and explains potential reasons these results were found.

PAGE 23

CHAPTER 2 LITERATURE REVIEW This chapter reviews relevant prior research. Additionally, important terms are defined for use in this research. Finally, the contribution of tfiitasesrdi project is discussed. The topics reviewed in this chapter, in order, are as follows: role modeling, characteristics of the sports hero, the sports hero as a role model, self-concq>t, and self-concept congruity theory. Role Models and Behavioral Modeling Sports heroes are widely assumed to be role models, especially for children and adolescents, although such assumptions are often made based on pem>nal opimon and lack concrete support (Malone, 1993; Harris, 1994a; Harris, 1994b; Wilson & Spaiks, 1996; Nack&Munson, 1995; Taylor, 1991; Walden, 1986; "MichaelJordan's", 1991; Simons, 1997; "Role Models", 1989). One primary purpose of this research is to determine how tl^ relationships between self-concept and tiie ima^ of sports hero determine role modeling behaviors of this sports hero. Therefore, an adequate understanding of the role model and its ftmction and infiuoice is necessary. This section of the literature review defines role models, explains the inocess of imitating role models, and outlines the conditions under which such modeling is most likely to occur. Finally, issues specific to role models and adolescrats are dismissed. The term "role model," like self-concept, has been used vaguely in reference to a variety of different constructs, so it is necessary to define this term specifically. A role 16

PAGE 24

17 model has been described as "someone whose hfe and activities influenced the respondent in specific life decisions" (Bassoon & Owe, 1980, p. 559). This description does not address actual copying of behaviors or intentions to copy behaviors, but rather presents the role model as one influence in such intentions. Other definitions assert that role models are "adults who are worthy of imitation in some areaof fhdr life" ^lass & Feldhusen, 1 995, p. 1 63) or "someone who demonstrates the appropriate bdiavlor for a specific role or relationship with another person" (Jung, 1986, p. 528). Role models generally exert influence ova* one of three areas of life: educali
PAGE 25

18 can be replicated in laboratory settings. For example, learning the technical processes involved in juggling fix)m a model in a laboratory setting would be an example of modeling. The person dononstrating the juggling behaviors would be the model. However, role modeling, while it does involve the technical processes of modeling, also involves an motional attachment between the observer and role model. Jung (1986) explained distinct differences in the two constructs: "Modeling does not imply that there is any emotional tie between the model and observer. In contrast, many instances of role modeling seem to require some affectional bond for the model to be effective in influencing the observer" (p. 527). Role modeling involves a relationship between role model and observer, and such relationships can influence behavior through learned roles in specific situations (Hendricks, 1992; Brim, 1968; Bell, 1970; Jung, 1986; Milla-, 1 983). Therefore, the process of role modeling involves adopting and modeling behaviors based on an emotional tie between observer and model, and this process of modeling involves context and relationships with the role model. In orda for role modeling to occur, the role model and the observer must interact and the observer must identify the role model as such (Bell, 1970; Basow & Howe, 1980). One popular view of role model influence is that everyone has role models, and these figures serve as guides for our own actions and possible templates for our own selves (Hendricks, 1992; Gecas & Mortrimer, 1987). Such influence can be bdrnvicnral and attitudinal. However, behtivioral influence has been studied and ffiscussed more frequently than attitudinal influence by researchers and theorists (Basow & Howe, 1980; Ahnquist & Angrist, 1971 ; Hendricks, 1992; Jung, 1986), and these influences are the focal point of this research project. Observers are not limited to one role model, and it is quite normal for observers to have several role models for different aspects of their lives (Jung, 1 986). Additionally, some role models influence more aspects of observers' lives than other role models (Gecas & Mortimer, 1987; Basow & Howe, 1980). Hendricks (1992) explained this.

PAGE 26

19 writing, "Some models are relevant in specific realms only; that of the older, wiser colleague at woik, or the sports figure we seek to emulate. Others cut across specific roles and have a more profound influence on our lives" (p. 7). Most descriptions of role modeling refer to the process of copying the behaviors of the role model, although they also can, in some circumstances, ref^ to avoiding behaviors of the role model and actively seeking alternative behavioral techniques to those displayed by the role model (Pleiss&Feldhusen, 1995; Jung, 1986). While including such avoidance techniques broadens the definition of the role model, this research specifically examines the intention to model the behavior of a role model, not the intention to avoid certain behaviors. Many available sources of role models exist Persons who have fi'equent interaction with and are well known by the obsCTver, including p^ffoits, teachea^, peers, co-workers, and siblings, typically are popular role models for adolescents. (Schaefer, 1978; Muuss, 1975; Basow&Howe, 1980; Malgady et al., 1990; Hackett et al., 1989; Almquist & Angrist, 1971; Betz & Fitzgerald, 1987). In addition to p^nal acquaintances, mass media figures also can serve as role models (Caughley, 1985; Schaefer, 1978; "Role models", 1989; Jung, 1986; Pleiss&Feldhusen, 1995; Duck, 1 990). In contrast to personally known role models such as paroits and peers, mass media figures, including famous athletes, actors, and politicians, have the potential to serve as role models for large numbers of people due to their iiss exposure. Researchers and critics have takoi varying positions on the potential influence of mass-mediated figures such as sports hero. One popular belief is that popular, mass-mediated figures, particularly those portrayed as positive or hat>ic in tiieir bdiaviors, inherently serve as role models due to tiieir stature and position in society. In fact, the terms "public hero" and "role model" are often used interchangeably without regard to a distinction between the two (Pldss & Feldhusen, 1995, p. 163; Malgady et al., 1990; Hunter, 1983). Conversely, some have concluded that a public hero cannot truly be a role model because there is almost no potential for interaction between the role

PAGE 27

model and the observer (Singer, 1991; Hall & Post-Kammer, 1987). Both of these contradictory opinions are too inflexible to fully sununarize the potential influoace of the mass media figure, and the true influence of this figure lies somewhere inbetween these viewpoints. Research has indicated that mass-mediated figures, particularly those who are considered hooes, can serve as role models (Pleiss & Feldhusen, 1995; '*Role models", 1989; Schaefer, 1978; Bandura, 1981; "Michael Jordan's", 1991; Malone, 1993; Kelhier, 1996; Weisman, 1993; Wilson & Sp^, 1996). Parker and Lord (1993) have echoed this sentimrat in research with African-Amedcan mal^, noting that "faig^y visible and successful individuals (such as sports personalities or entertainers) make excellent role models for young Black males" (p. 102). Therefore, while popular media figures are not inherently role models, they certainly have the potential to be role models. Media users often develop intense imaginary relationships with their favorite mass-media figures, and these relationships contribute to the role modeling potential of such figures (Caughl^, 1985; McEvoy & Erikson, 1981). McEvoy and Erikson (1981) called these public heroes "reference idols." Reference idols, although not personally known by the observer, become part of the referoice group of the observer, "a personification of values and behaviors perceived as unportant to tiie individual" (p. 1 14). The authors explained the potential influence of these mass media heroes, writing, "Reference idols, real or fictitious, can affect the formation of attitudes, belief, idoitities, and behaviors" (p. 114). In other words, mass media heroes provide example attitudes, beliefs, identities, and behaviors that individuals use in creating, modifying, and/or maintaining/reinforcing their self-concept. It is these potential influences of mass media figures that will be the focus of research for this dissertation. In summary, it is necessary to define the term "role model" for use in this dissertatioiL For this study, the role model is a mass mediated figure (in this instance, a sports figure). The processes of role modeling involve modeling of specific behaviors of an admired figure. Finally, an affectionate bond must exist between the observer and the

PAGE 28

21 role model, unlike laboratory models. Therefore, role modeling will occur as jwrt of daily life, making the study of role modeling one of naturalistic inquiry, not laboratory observation. Conditions of Role Modeling Many researchers have studied tiie conditions or factors which make observers more likely to adopt the behaviors of a role model. These conditions are reviewed in this section. One primary factor is poveived personal similarity b^eoi observer and role model (Hackett et al., 1989; Jung, 1986; Decker & Nathan, 1985; Owens & Ascione, 1991; Wilson & Sparks, 1996; Muuss, 1975; Bandura, 1994; "Role models", 1989; Malgady et al., 1990; Bumstein et al., 1961; Stotland & Patchen, 1961; Stotland et al., 1 962; Rosekrans, 1 967). People are more likely to adopt tiie behaviors of a role model whom they perceive as having similar attributes to thanselves. Specific attributes include sex, race, and cultural identification. Wilson and Spades (1996) provided examples of this through interviews with AfiicanAmerican teens in reference to their sports heroes. They specifically studied AfricanAmerican teenagers and their role modeling of Afiican-American basketball stars. They wrote, in reference to potential modeling of clothing styles, "Many of the Black adolescents gave responses acknowledging the celebrity Black athletes' definitive influoice on popular style through athletic s^arel and the athletic-{q>parel commercial. This influence was manifested most often in the respondents' desire to look like the celebrity athlete" (p. 414). These adolescents were more likely to use Afiican-American athletes as their role models due to perceptions of racial similarity, cultural idmtity, and masculine idoitity. Tho^re, the perceived similarities between popular male, Afiican-American athletes and male, Afiican-American adolescent fans contributed greatly to this selection of role models. Perceived self-cflScaqr of the observer also can affect the dioice of a role model (Simons, 1997; Weisman, 1991; "Role models", 1989; Bandura, 1994). Self-efficacy refCTs to the a belief that one C£di successfiilly model an actiwi (Bandura, 1994). As the

PAGE 29

22 individual becomes more confident in his or her ability to perform a behavior, he or she will be more likely to model such behaviors. Another factor influencing modeling is the perceived braefits the role model ^das firom his or her behaviors (Ostlund, 1974; Midgley, 1976). Role models who are rewarded for their actions are more likely to be imitated than those who are not rewarded. This has lead to the belief that popular, mass-mediated role models are often very appealing to observers because they are well compensated with fame and wealth (Simons, 1997; Wilson & Spaiks, 1996). Viewers, therefore, will aitmspt to role modd tiiose behaviors perceived as instrumental in achieving such fame and wealth. Simons (1997) explained this appeal of wealthy, mass-mediated figures as potential role models to certain groups of adolescents, using basketball star Michael Jordan as an sample. Simons wrote, a lot of them (low-income, Afiican-Ammcan adde:(aits) wifi teU you they want to be like Mike. In this context, being like Mike doesn't mean becoming an entrepreneur, a corporate spokesman, or a college graduate. It means being a high-flying, windmill slammer of a ballplayer, (p. 50) In other words, the most glamorous aspects of Michael Jordan (athl^c fame, financial riches &om athletic accomplishment) made Jordan appesiHag as a role moi^ to tiiese adolescents. Self-Concept as a Condition for Role Modeling Most research on the impact of role models, and specifically atti]et sit< models, has focused on the characteristics of the potential role model; the societal role of the role model; and easily obso^able charactoistics of the observer, such as race, socio-economic status, and perceived positive or negative v^e to society (Snnons, 1997; Malone, 1993; "Role models", 1989; Wilson & Sparks, 1996; Chen, 1997; Golden, 1996; Walden,1986; Pldss&Feldhusen, 1995). This dissertation addresses the relationship between role model and observer from a slightly different perspective. Instead of measurmg the aforementioned characteristics, this research examines

PAGE 30

23 individual perceptions of the self, or self-concept, and the influence self-concept has in determimng role modeling bdiaviors. Of primary importance for this research are the relationships between self-concept and the image one has of his role model (for this research, the sports hero). In other words, this research attempts to better understand whether one's bdiavioral intentions to model the behaviors of a role modd toe infiuoiced by how similar or dissimilar one's self-concept is to his or her perceptions of his or her sports hero. This approach of examining self-concq)t has not bera widely used to investigate bdiavioral intentions of role modeling. However, this research q>proach has been used quite frequently in marketing research, and this dissertation will borrow investigative strategies fix>m such marketing studies. (See later section on "self-concq)t" for a detailed description of this construct and a review of research and theory in this area.) Marketing researchers have provided a foundation for the use of self-concept in studies of behaviors and behavioral intentions (specifically, purchase intoitions). First, th^r have established that self-concept maintenance and enhancement is a factor in determining behavior (Sirgy, 1983; Landon, 1974; Birdwell, 1968; Dolidi, 1969; Grubb & Grathwohl, 1967; Ross, 1971; Delozier & Tilhnan, 1972; Loundon & Bitta, 1979; Krechetal., 1962; Wells & Marwell, 1976; Baughman & Welsh, 1962; Felker, 1 974; Prey & Cariodc, 1 989). Several studies bave established flsrt omaimas choose to purchase certain products based on the similarity between fhdr reported sdf-eoncqrt and their perceived image of a product (Dolich, 1969; Ross, 1971; Delozier & TiUman, 1972; Grubb & Grathwohl, 1967; Hattie, 1992; Baughman & Welsh, 1962). These studies have shown that individual perceptions of self (self-concept) can influence behavioral intentions when viewed in reference to the individual perceptions of a consumer product. Therefore, each individual has perceptions that influence his or her behaviors, and these perceptions may be completely different than the perceptions of

PAGE 31

24 another observer. This is the most important justification for the use of self-concept to examine behavioral intentions for this dissertation. In the aforementioned consumer studies, subjects rated tfieir perceptions of consumer products, and these ratings were compared to the individual self-concepts of these subjects. From tiiese comparisons, researchers examined the role of self-concq>t in influencing consumer behaviors toward selected products. However, this dissertation looks at perceptions and influence of sports heroes, not consumer products. Therefore, it is necessary to review the characteristics of popular, mass-mediated sports heroes and their potential to serve as role models for adolescents. This is the focus of the next section. The Sport s Hero as a Role Model Figure The term "hero" is a heavily cited yet frequently changing construct in American society. However, the hero has been a vital part of societies throughout history, and a brief review of this evolution will assist in better und^-standing the qualities of the modem hero and, more specifically, the modem sports hero. The term "hero" dates back to ancient Greece and originally was defined as "a superior man, embodimoit of composite ideals" (Fishwick, 1979, p. 340). Qualities that the Greeks associated with the hero included an iutimacy with gods; superiority to otho" humans; and an abundance of the best of human characteristics, sudi as stroigtfa aiMl courage (Bowra, 1969; Brombert, 1969; Whitman, 1969; Rshwick, 1983). The ancient hero was a pre-destined character, and he was inherently greater than the normal man (Macmechan, 1 840; Ldmian, 1 928). Another historical name for the heroic figure is the Great Man (Hook, 1957; Wecter, 1941).

PAGE 32

25 According to Carlyle's Theory of the Hero, the hero is one who is destined for greatness from a divine source (Carlyle, 1841; Lehman, 1928; Smith, 1973; Wecter, 1941 ; Fishwick, 1 985). Following this theoretical perspective, heroes were predestiiwd by God as leaders to move the world forward (Carlyle, 1941; Lehman, 1928; Campbell, 1949; Wecter, 1941). Examples of Great Men in history include Moses, Frederick Ae Great, and Napoleon. The evolution of the hero has been quite pronounced in the 20th century, particularly in a relatively youthfiil American society. The youtii of this society allowed Americans to define the characteristics of their own horoic types (Hook, 1 957; Lubin, 1968; Wecter, 1941; Lemer, 1957). Theorists of this period asserted the hero was not pre-determined, but instead was a product of his society and those who followed him (Hook, 1957; Klapp, 1962). Additionally, an increasingly complex society has allowed for a variety of different heroic types and characteristics. In other words, 20th centuiy America experiraiced a fragmentation of heroic types, allowing for greater ha:oic diversity and more heroic choices than in ancient Greek society. Instead of possessing God-like qualities, the American hero of the mid-20th coituiy had a closer relationship with the common man (Wect^, 1941; Hook, 1957). This presented a strong contrast with European Great Men of that time period, including Hitler and Mussolini. Amoicans, increasingly fearful of stidi powerful and destructive heroic figures, did not embrace Great Men as their heroic dfEMoes. Th^ instead dhose more familiar, less intimidating heroic types. (Hook, 1957; Fishwick, 1985). The increasing influence of mass media in American society in die late 20tli century further changed the dynamic of the American hero. Theorists of the 1960s and 1970s described a new type of figure, the celebrity, which had become prominent in American society (Boorstin, 1961; Klapp, 1962; Lubin, 1968; Bowra, 1969; Monaco, 1978; Fishwick, 1979; Browne, 1983; Rollin, 1983; Hoagland, 1974; Hams, 1994). The celebrity, first described by Boorstin (1961), does not gain fame through greatness or

PAGE 33

26 accomplishment, but instead due to mass exposure and visibility. Boorstin provided the following description of the Ammcan celebrity as: a person who is known for his well-knownness. His qualities — or lack of qualities ~ illustrate our particular problems. He is neither good nor bad, great nor petty. He is the human pseudo-event He has beoi fiibricated on purpose to satisfy our exaggerated expectations of human greatness. (1961, pp. 57-58) Unlike heroes, celdnities do not display a model of greatness. They are a product of the media as much as they are a product of their accomplishments. Celebrities do not consistoitly di^lay excelloice, and they exhibit ordinary bdiaviors and accomplishm a its (Klapp, 1962; Kelly, 1993; Monaco, 1978). Therefore, many theorists of the 1960s and 1970s believed, largely due to America's increased reliance on mass media and television, that the celebrity hai replaced the hero in American society, and the heroic figure no longer existed in Amorica (Schlesinger, 1968; Monaco, 1978; Hoagland, 1974; Kelly, 1993; Harris, 1994; "How to", 1995). This cynical view blamed television for initiating the extinction of the hennc figure. This viewpoint also expressed a static view of the hrao. Accordingly, this heroic figure did evolve over time and was eventually replaced by the celebrity. However, many theorists of the 1980s and 1990s have refuted these argumraits, writing instead about Hbc ongoing evolution of the hero into what can be called the modem American hero (Fishwick, 1979; Fishwick, 1983; Fishwick, 1985; Browne, 1983; Walden, 1986; Harris, 1994). These authors argued that heroes have not dis^>peared, but instead heroic qualities have changed with the growing influence of mass media. In particular, because mass media have made ho-oes more available to the public, Americans are given a larger number of heroic choices, and these heroes cannot maintain the longevity of heroes of past generations. Mass media exposure has become a critical component of modrni heroic stature, and therefore the modem hero does share characteristics with the less substantial

PAGE 34

27 celebrity. Rollin (1983) clarified the influence of mass media on the modem hero, writing, "Publicizing the heroic act and celebrating the performer, tiierefore, are audal. All heroes then are celebrities; no hero is not a celebrity" (p. 14). Unlike the traditional hero, the modem hero roast receive pubHcity by mass media in order to gain heroic consideration. Publicity is cmdal in acceptance as a modon hero, regardless of the characteristics of the individual. However, unlike the celebrity figure, publicity in itself is not enough to create the modem hero, and this figure must also exhibit certain heroic qualities. This dissertation takes the perspective of these modem theorists, asserting that American heroes still do exist and that mass media have sped the evolutionary process of these heroic figures. While the modmi American hero can take several different forms, theorists have attempted to define some parameters of this modem heroic figure. These include attractiveness, wiiming, style, risk-taking behavior, individualism, and skill in a particular field (Browne, 1983; Rollin, 1983; Fishwick, 1985; Crepeau, 1985; Harris, 1994; "How to", 1995). Americans have become less likely to require moral greataess fi-om their heroes than was required by previous generations. Additionally, due to an increasingly Augmented Amoican society and an increasing numbo* of hooic dioices available through television, Americans of today are less likely than Americans of formCT genoations to unite bdiind one particular hero (WMea, 1986; Fishwidc, 1983; Rollin, 1983; Crepeau, 1985). This research deals with one specific type of hero, the sports hero, and this is the focus of the following section. The Sports Hem Sports has become a popular and vital area in which Americans now find their heroes, and this trend has grown through the 20th centary ("Heroes of, 1990; Ryan, 1995; Nixon, 1984; Oriard, 1982; Simons, 1997; "Role models", 1989). One potential reason for this is that sports remains one area where true greatness and sxtperior beauty

PAGE 35

28 still can be found in a complex society (Goodman, 1993). Regardless of the negative perceptions Americans may have of politicians, actors, and other frequently tarnished public figures, a star athlete, xmlike other mass mediated figures, will have rare moments during which they appear to surpass typical mortal limitations through spectacular, seemingly impossible athletic feats. He will appear, if for only Aat brief poiod of time, almost "god-like" (Nixon, 1984; Oriard, 1982). While the sports hero has been part of American culture since well before the turn of the 20th coitury, a fantastic growth and emphasis on the Amoican sports hao occurred beginning in the 1960s with the rapid growth of sports television (McPherson, 1989; Davies, 1994; Harris, 1994; Nixon, 1984). By the mid-1980s, sports programming accounted for 15 percent of total programming on commercial television, including networks, such as ESPN, devoted solely to sports programming (Harris, 1994b; Davies, 1994; Nixon, 1984; Katz, 1996). This growth in sports media allowed for inoieased exposure and popularity of the sports hero to the gmanl public. In addition to publicizing the modem sports hero, the current focus of the content of sports television has contributed also to the evolution of the American sports hero. To increase ftn interest and provide more exciting television matoial, Amoicaii ^rts programs are piesoited to emphasize heroic actions of the individual and individual emotions and personalities of the star athletes (Kinkema & Hanis, 1992; Coakley, 1994; Haii^Yes, 1986; Hilliard, 1 984; Sabo & Jensen, 1 992). TMs focus on the individual has helped to create a strong and \mique relationship between the viewer and the individual star athletes. Dimcan and Brummett (1987) explained media's ability to create such relationshif^>xitng, during sporting events, television tends to focus more on the individual than the group, thus establishing a kind of intimacy between the audience and the playo*. This intimacy is created by close-iq)S of partiodar adiletes before, during, and after the competition and shots capturing the athletes' expressions and reactions at critical moments, by the announcers' creation of imaginary conversations, but their use of the players' first names, and by personal interviews with the athletes or coaches. Television, therefore

PAGE 36

29 turns spectator sports into observance of intimates, albeit a controlled intimacy within flie contepl^ ft 19-indi screen, (p. 171) The individual has become so paramount in televised sports ibat television ratings often hinge upon whether a star athlete, such as basketball star Michael Jordan, plays in a particular contest (Fatsis, 1998; Shapiro, 1999; Starr, 1999). Professional baskefl)all (NBA) in particular has employed a marketing strategy of promoting star players ahead of the actual game (Starr, 1999). For the 1997-98 basketball season, NBC's ratings for nationally televised Chicago Bulls games, featuring Michad Jordan, were 75% higher than for the network's other nationally televised NBA contests (Fatsis, 1998). With Jordan's recent retiremoit ftom basketball, both league and network executives have openly worried over the potential loss of league popularity and ratings of televised games. Peter Roby, vice president of marketing for Reebok Athletics, asserted that the league would never again reach the status attained through Jordan's star q>peal, saying, "I don't think the NBA will ever again attain the glory it did with Midiael" (Starr, 1999, p. 56). To combat this potential loss, networks have embraced the opportunity to create and publicize new athletic stars. NBC Sports chairman Dick Eberson said in referoice to the potential for deoreased ratings upon Jordan's retiremait, "We do have an opportunity to reintroduce a whole new generation of stars" (Shapiro, 1999, p. D5). In other words, the individual star remains the prime emphasis for television sports, and ihc retironent of one star leads to the creation of other stars. Nike corporation CEO Phil Knight, wlK>se company airs numerous conmiercials featuring star athletes, has stated that even his television advertising campaigns alone are powerful enough to create popular sports figures. For instance, in reference to tiien-NBA newcomer Al
PAGE 37

30 distinction between sports hero and sports celebrity spears to be a confusing and blurred line. However, this research follows the perspective that mass media, specifically television, are now necessary to publicize sports heroes. Additionally, the fact that sports figures have primarily become mass-mediated figures does not mean all popular athletes must be celebrities instead of heroes, but instead reflects the evolution of tiie modem sports hero. Due to tiie growth of media coverage of sports figures, Americans now know more about popular sports figures than ever before, including bofli then* on-the-field and oflP-the-field activities. As Caughley (1984) wrote, "Even an average fan knows more than a baseball player's name. He also knows his team, uniform number, position, batting avmige, salary, and something about his appearance, personality, medical history, and ofif-the-field conduct" (p. 32). The popular sports hero has been demystified, and fans now see the impofections as well as the greatness (Hargreaves, 1986; Harris, 1994a; Hoagland, 1974; Coakley, 1994; McPherson et al., 1989). Such imperfections include wife/girlfriend abuse, drug and alcohol use, sports gambling, fighting, and fatiieringofillegitimate children (Harris, 1994; Patemo, 1992; Johnson, 1995; Mesaia & Solomon, 1994; Long, 1991; "Michael Jordan's", 1991; Nack&Munson, 1995; Starr & Samuels, 1997; Wilson & Sparks, 1996; Farhi, 1997; Franks, 1996; Howard, 1995). These imperfections must be considered in addressing die potoitial influmce of popular sports figures, particularly their potential influence as role models for adolescents. Despite these imperfections, many theorists still believe the modon, mass-mediated sports figure can be a hero, and they have identified several characteristics that are commonly associated with this modem sports hero. These qualities include the following: suprone athleticism on the field or court, high winning percentages and the potential to win important games and championship matches, statistical records and greatness throughout a career, flair and charisma in style of play and personality, and

PAGE 38

31 confidence in one's abilities (Nixon, 1984; Harris, 1994a; Harris, 1994b; Crepeau, 1985; Goodman, 1993; Smith, 1973; Porter, 1983; Starr & Samuels, 1997). According to surveys of American teenagers, particular sports-related behaviors, such as sportananship and fair play, are also important to some sports fans (Harris, 1994a). While not necessarily related to the athletic performance, financial success and lucrative commercial endorsement deals are commonly identified qualities of the sports hero, particulariy to adolescent boys who aspire to reach similar financial heights through professional athletics (Katz, 1994; "Michael Jordan's", 1991; Wdsman, 1993; Wilson & Sparks, 1996; Simons, 1997). Finally, theorists also have identified several non-performance-related characteristics, including civic and community involvement, academic accomplishment, strong family ties, and avoiding illegal and iimnoni behaviors, as components of the modem heroic qualification (Walden, 1986; Smith, 1973; Harris, 1994a; Harris, 1994b; Hoagland, 1974; Nixon, 1984; Coaldey, 1994). While the off-lhe-court actions Of the sports star may have some impact on heroic classification, on-the-court excellence has been identified as more instrumental in determining heroic status. Nixon explained this, writiog, ''Wayward athl^es may be excused by fans. in their lifestyle off the field as long as they work hard and produce on the field and. their behavior on or off the field does not dqyart too much firom conventional standards" (1984, p. 174). This dissertation accepts the perspective that the sports hero is a non-static, evolving construct. Additionally, each individual may have his or her own views of the characteristics that are most important in being a sports hero. For Ms tcaemh, however, the sports hero will be defined as adhering to the following guidelines: 1) The sports hero must be well known. Therefore, he must be shown prominently on television. This perspective acknowledges the impact of celebrity on the heroic condition in America.

PAGE 39

32 2) The sports hero must display athletic excellence on the field or coiirt. This can be displayed through record or notable performances (scoring, defoise, etc.), flamboyant or outstanding style of play, or other feats that set an athlete apart &om his or her peers on the field or court. 3) The sports hero must display the ability to be a wimier on the field or court 4) The sports hero may display the ability to overcome difficult odds against success in sports, such as injury, physical limitatioiis, w undeqffiyileged &mily life as a youth. The sports hero may also display txtimtp^acvermce in reaching athletic goals. 5) The sports hero may display exemplary behavior
PAGE 40

33 evidence (Jimg, 1986). The limited research that has been d9ne on this topic is addressed below. The popular sports figure can serve as a role model in several different areas of adolescent life. Some of these roles are positive, while otheiis are negative. One identified arcsL of role modeling is style of play and athletic perfomupce. Obsorvos may pattern aspects of their athletic participation after the athletic accomplishments of their sports role models (Wilson & Sparks, 1996; Kellner, 1996; Simons, 1997; Harris, 1 994b). For example, golfers rqwrted a noticeable slowing of play at their local courses when golf tournaments were first televised (Harris, 1994b). This was because amateurs began lining up their putts as the pros did on television, despite the fact that many of than had no idea why this was done or the effectiveness of this procedure. Similarly, observers model the athletic moves of NBA basketball stars (Harris, 1994; Wilson & Sparks, 1996; Simons, 1997; Nixon, 1984; "Michael Jordan's", 1991). ObsCTvers may attempt to adopt aspects of their athletic role model's play style, whether it be the flamboyant dribbling style of Julius Erving or the physical play of Karl Malone. The mfluCTce of professional athletes often extends fi-om one gmeration of athlete to the next, as many current stars list former athletic stars as their role mo^ls in devdoping playing styles (Nixon, 1984; Wilson & Sparks, 1996). Non-sports-related on-(x>urt behaviors also can be modeled by observes. These behaviors can be positive, sudi as exhibiting grace during defeat or helping players fixMn other teams. They can also be negative, including throwing temper tantrums during competition, fighting with other players and officials, and abusing equipmoit (Wdanan, 1993; Wilson & Sparks, 1996; Starr & Samuels, 1997; Harris, 1994). Several popular athletes, including Charles Barkley, Joe Namath, and Deon Sanders, have bec(ne well known for their brash and arrogant behavior, and adolescent observers may copy sudi behaviors (Harris, 1994b). Atiiletic role models also can display positive behaviors

PAGE 41

34 during play, and observers can imitate these behaviors as well (Coakley, 1994; "Role models", 1989; Goodman, 1993; Harris, 1994b). ^ Theorists have suggested that athletes can serve as positive role models when observers imitate positive work behaviors of success^ sports heroes. These behaviors include hard work toward reaching a goal, practice, desire to win, and prosodal behaviors such as charitable work and sportsmanship (Kellner, 1996; Wilson & Sparics, 1996; Coakley, 1994; Malone, 1993; Weisman, 1993; "Michael Jordan's", 1991; Nixon, 1984; 'Tlole models", 1989; Harris, 1994a). ~^ One prominent potential area of role modeling of popular mass-mediated athletes involves glotbifig. style and athletic footwear, including athlete-endorsed shoes costing in excess of $100 per pair. While obsovers may not be able to perform the athletic feats of their role models, they can wear or hope to wear the clothing endorsed by role models (Weisman, 1993; Katz, 1994; "Michael Jordan's", 1991; Kelhier, 1996; Simons, 1997). Mass media advertising campaigns have contributed to the marketing power of Ithe athletic role model, with companies telling the consumer that shoes are a key component to being more like their athletic hero, whether it be Michael Jordan or Bo Jackson (Weisman, 1993; Wilson & Sparks, 1996; Simons, 1997). As Wilson and Sparks (1996) wrote about young African American males and their favorite NBA players, **Many of the Black adolescents gave responses adax>wledging the cdebrity Black alMetes* definitive influence on popular style through atiiletic appsoKi and the athletic-apparel commercial. This influence was manifest most oilai in the respondoits' desire to look like the celebrity athlete" (p. 414). The coinciding occurrence of imitation of athletic play style and choice of athletic shoes points to the influence of the athletic role model. Katz (1994) described this in his chronicle of flie Nike corporation, writing.

PAGE 42

35 by 1991, a visitOT to any playground in fhe world could find hefiy yoin^ Charles Barkleys and towering David Robinsons playing Charles-style or Mr. (David) Robinson-style in their Air Force shoes. Lean and athletic Scottie Pippens flew in Air Flights; little Bo Jacksons pumped iron in thousands of gyms, and, by then, miniature Andre Agassis lit up tainis courts in their peacock-colored Nike clothe mti'iftioes. (p. 148) Fans could couple their role modeling of athletic behaviors of their sports hero with purchases of equipment of &eir ^kmIs heto, certainly an inviting concept to athl^c companies marketing such oidorsed sports products. Popular athletes also may serve as role models of ^^ggmicsuccess. Professional athletes receive exorbitant salaries; NBA salaries avoage mcife lhan $2 million a season, not including endorsement income (Simons, 1997). Saliffies of top aretes far exceed this figure, and such wealth is often accomplished at a very young age. Theorists have suggested that Amoican youths use famous athletes as rote taodds of economic success and mobility (Simons, 1997; Weisman, 1993; Wilson & Spaiffe 1^; Kelteer, 1996; Harris, 1994; "MichaelJordan's", 1991; Nixon, 1984; Starr & Samuels, 1997). Tho^fore, such youtiis may desire to copy this financial success instead of imitating more realistic role models, such as teachers and local business persons. This occurs despite the fad that only one of every 50,000 high school basketball players actually reaches the NBA (Simons, 1997). In other words, an adolescent's chances of ever achieving fame and fortune through professional athletics are quite slim. Additionally, many popular athletes have exhibited negative bdiaviors, both on and off-the-court, that possibly could be role modeled by fens. It is this type of role modeling that is most worrisome to critics of the modem sports hero. Although certainly not all popular sports stars display sudi behavioral tendencies, the list of athletes have done so is long and covers a wide range of sports and teams. A few specific examples of such well pubhcized negative behaviors are described below. Many of these negative behaviors occur during competition, refored to as on-court behaviors. These behaviors oftoi involve argumoits and confijontations

PAGE 43

36 between players; pofaaps the most notable of these occurred whoi boxer Mike Tyson bit Evander Holifield's ear during a title fi^t. These behaviors also indude athlete confrontations with fans, coaches, and officials ("TKO: Tyson", 1999). For example, in a much pubUcized case, NBA star Latrell Sprewell, then of the Golden State Warriors, attacked his coach, P. J. Carlisimo, twice during a practice session (Starr and Samuels, 1997). In recent years, several players have been suspended in recent from league play for physically attadcing a referee. These include Nick Van Exel and Dennis Rodman of the NBA, Oilando Brown of the NFL, and Roberto Alomar of Major League Baseball, who spit on an umpire after a disputed strike call (Collins, 1999). Rodman also has incurred fines and a suspoision for kicking a TV camonman during a game (Starr and Samuels, 1997). Star athletes also have had conflicts with fans during games. Houston Rockets guard Vemon Maxwell punched a heckling fan in Portland during a 1995 game, costing Maxwell $20,000 and a ten-game suspension (Nance, 1995). Even iofessional coaches, who themselves have become highly paid, popular sports figures, have displayed extremely inappropriate behaviors during games. NFL coadi Mike Ditka made an obscene gesture to home fans during a home New Orleans foo^l game (Collins, 1 999). Such negative on-court behaviors, including fighting, yelling at other players, shoving and arguing with ofifidals, kicking cameramen, and, in the most extreme of cases, even physically harming a coach or fen, have become a somewhat regular occurrence for many popular sports figures (Wilson & Sparks, 1996; Weisman, 1993; Katz, 1994; Harris, 1994; Harris, 1994b; Starr & Samuels, 1997; Smith, 1973; Walden, 1986; Coakley, 1994). Negative ofif-the-court behavi
PAGE 44

V 37 conspiring to murder his wife (Brady, 1999). Unlike many other athletes who have committed serious criminal offenses, Carrudi did not have a prior record of n^ative and illegal behaviors. Other athletes, such as Mike Tyson and Dennis Rodman, became well known as regular law breakers. Among other offenses, Tyson has been soitoiced to jail twice, once for r^ and once for assault, and Rodman has been arrested for sexual assault, domestic abuse, and drug use ("TKO: Tyson", 1999; "Rodman sued", 1999). Althou^ this papa cannot effectively address the many examples of the different varieties of negative off-Ae-court behaviors, some bdiaviofs of popldar athletes that are frequently reported by mass media include wife and girlfriend abuse, drug abuse, alcohol abuse, and assault (Howard, 1995; Taylor, 1991; Star & Samuels, 1997; Donohew, Hehn, & Haas, 1989; Patemo, 1994; Messner & Solomon, 1994; Goldberg, 1995). Given the invasive media coverage of the off-the-field behaviors of athletes, these negative behaviors hold smous ramifications for any study of role modeling of popular athletes. The previous section addressed the sports hero as a potential role model and some specific types of role modeling that could result fijom the actions of mass-mediated ^rts figures. This study, however, addresses not only different types of role modeling, but also the motivations for such role modeling and conditions under which role modeling might occur. Specifically, this study examines adolescent self-omoqit and Hm proximity of such self-concept to perceived images of sports heroes as a factor in determining a variety of role modeling b^viois of ^rts heroes. Theref^ &e mxt section addresses self-concept and its relationship wi^ bdiavior and role modding.

PAGE 45

Self-Concept Introducti on and Definitions Much of the literature and debate dealing with the (^ice of spoits heioes has focused on the qualities of the sports figure, not the individual qualities of die ^ofts fims. For example, theorists have debated whether the sports hero is determined solely through athletic feats, or whetho* he must also personify outstanding qpaUties in bisMs outside of sports (Hanis, 1994; Waldrai, 1986; Smith, 1973). The limited dtelbfi^ of Ae qualities of sports fans typically has focused on group affiliation and characterization of sports fans, incltiding race, nationality, socio-economic status, and gender, ai^ how these goieral characteristics might influ^ce the choice of a sports hero. (Harris; 1994a; Wilson & Sparks, 1996; Lapchick & Stuckey, 1993; Klein, 1991). However, these per^)ectives ignore the fact that selection of a sports hero is an indlvidNiid db^SQ on the individual characteristics and needs of each person. One potentitd^ m^ms of addressing such individual characteristics in a standardized and measurable method is to examine self-concept in relation to the choice of a sports hsxo. Self-concept is an action of the term "self Hiatlias been defined as "the individual's attitudes toward self (Frey & Carlock, 1989, p. Z; §f^rd & Backman, 1964). The self serves as the Tef&mce point for each individual's view of the w
PAGE 46

39 also hints at the importance of the external environment in the detomination of self-concept. Several researchers have noted the difference between one's individual self-concept and the self as observed by other people. In oihet words, an individiud's view of himself or herself may not be the same as how otho* people view that person. One reason for this is because people do not always behave consistently with their self-concepts. One's overt behavior is often linked with one's self-concept, but inconsistencies may exist between the behavioral component of self and one's actual self-concept. For example, a person might think that he is an excellent student, yet this same person may not study diligently and complete his acadonic assignments. Therefore, his self-concept, or perceptions of himself, would not be consistent with his behaviors. This could lead to observers viewing a person differently than that person views himself or herself (Hattie, 1992; Baughman & Walsh, 1962). While the self-concept is likely instrumental in determining future behaviors, the self-concept is still a cognitive measure, and it measures one's own perception of reality. This may be inconsistent both with an objective reality and Avith the ways in whidi oHi&cs perceive someone (Loundon&Bitta, 1979; Frey & Carlock, 1989; Felker, 1974). Felker(1974) explained this in his definition of the self-concept, as follows, self-concept is the sum total of the view which an individual has of himself. Self-c(mcq)t is a unique set of perceptions, ideas, and attitudes which an individual has about himself. The view wlddi an individual has about himself is unique and to varying degrees is different &om the view that anyone else has about him. (p. 2) All the aforementioned definitions of self-concept refer to a single self-construct known as the actual self-concept, or the real self-concept (Sirgy, 1983; Dolich, 1%9). However, the self-concept actually has multiple components. The nimiber of measurable components of self-concept has varied greatly among researchos (Siigy, 1983; Hattie, 1992; Loudon & Bitta, 1979; Runyon, 1977). For this research project, two constructs

PAGE 47

40 of self-concept are addressed. They are as follows, with the corresponding definitions for die purposes of this research: 1) The Real Self: An individual's perception of how he actually is (Dolich, • 1969; Biidwell, 1964; Ross, 1971; Runyon, 1977; Loundon & Bitta, 1979). This refers to the most common self-concept, the self-concept which has been described in the previous paragraphs. This construct has also been called the "actual self-concept" because it refers to the individual's self-perceptions as he or she actually sees them (Sigry, 1983). For this research, this construct does not take into account a person's perceptions of how otfaos view him or her, known as the "social self-concept" (Sirgy, 1983). Instead, this refers only to the individual's own perceptions of himself or herself 2) The Ideal Self: An individual's percq>tion of himself as he would like to be (Delozier & Tilhnan, 1972; Loundon & Bitta, 1979; Runyon, 1977; Baughman & Welsh, 1962; Ross, 1971). These two components are used as separate measures of self-concq)t and togetfao: make up the construct of "self-concept" for this research. Domflins n f Self-Concqtt The construct "self-concept," whether it be real or ided sdf^^oncept, is often regarded as a singular measurement, combining self-perceptions on a variety of diffoent areas togeflier into one construct According to this classification, individual percqntions toward oneself as an athlete would not be differentiated from perceptions toward oneself as an academic. In oth^ words, measuronents of self-concq)t would not allow the individual to differentiate betweoi distinct areas of himself or ho^lf. Althou^ widely used, this means of assessing self-concept is limited and somewhat incomplete (Harter, 1 982; Runyon, 1977). Therefore, this research examines self-concqH not as one

PAGE 48

41 measuranent, but rather as a measurement of several distinct domains, all of which are part of self-concept Harter (1982) pointed out distinct problems with using a single measure of self-concept in a raitique of Coopersmith's (1967) self-esteon inventory, a test of 58 questions which make up one global evaluation of sdf-worth, and the Piers-Hmris Self-Concept Scale (1969), a similar test of 80 questions used to evaluate overall self-concept. Harter wrote, items on such popular scales as the Coopersmith Self-Esteem Inventory and the Pia:s-Harris Self-Concept Scale t^ a range of diverse ctmtoit including cognitive competencies, physical skills, popularity, acceptance by parent, morality, personality traits, physical characteristics, and affective reactions. Responses to these heterogeneous items are then summed, and the total score is interpreted as an index of global selfregard. In employing such a procedure, Coopemnitii has a^umed diat children do not make distinctions among die dfmiains in tfadr lives. This assumption was seriously questioned in the scale-OHistruction efforts reported here. (1982, p. 87) Early research on self-concept did not break down this construct into specific domains. However, theorists did find that self-concept could differ in varying situations, such as a work environment as opposed to a social or an athletic environment (Loundon & Bitta, 1979; Newcomb, 1956). This was a precursor to the idea that there may actually be several different components of self-concept, and these different components accounted for a diffa&A self-concept in the context of dififorent oiviromnaits. Researchers subsequently determined that the general measuremoit of self-concept consisted of several components, including one's perception of one's athletic ability, morality, academic accomplishment, and physical appearance (Dolich, 1961 ; Krech et al., 1962; Loundon & Bitta, 1979; Reynolds et. al, 1980). Furthermore, these components of self-concept could change with cognitive development and aging (Frey & Carlock, 1989; Felker, 1974; Hattie, 1992; Katona-Sallay, 1993; Harter, 1982; Harter, 1988). Similarly, Felker (1974) identified separate categories of self-concq>t Aat ^lied

PAGE 49

42 specifically to adolescents. These categories were behavior, anxiety, intellectual or school status, h^piness, and satisfaction (Felker, 1974). Although self-concept was still measured as a single construct, this understanding of the separate components of self-concept lead to the evoitual measurement of separate domains of self-concept Susan Harter (1 982) conducted research on the sq>arate domains of self-concept and subsequently created tests to measure these separate domains of self-concept. This research provided the foundation for the collection of data on self-concept for this study. Harter created the Perceived Competency Scale for Childr^ a self-concqit test measuring four separate subscales of self-concept. These subscales are cognitive, social, physical, and general self-worth (Harter, 1982). Harter later developed the Self-Perception Profile for Adolescents, a similar self-concept measure for adolescents that assesses the following eight domains: scholastic competence, social accq)tance, athletic competoice, physical appearance, job competence, romantic appeal^ behavioral conduct, and close fiiendship (Harter, 1988). Because this research examines self-concept and behaviors of adolescents, these are the domains used as the subscales of self-concept for this research. The use of domains for research on self-concept and intentions to role model sports heroes is crucial because different domains of self-concept may relate quite differently to role modeling intuitions of sports hax)es. For example, the atiiletic domain of self-concept may have a vastly different relationship with the decision to role model a sports figure than behavioral conduct or romantic appeal has with sudi role modeling. Additionally, as discussed earlier, modem sports heroes are divo^ figures, and different domains of self-concept may relate in different ways to these different qualities of sports heroes. Popular athletes such as NBA stars Charles Barkley and Latrell Sprewell, both whom are excelloit athletes but who, on numerous occasions, have participated in illegal and unethical activity, can display seemingly contradictory behaviors and characteristics in different areas of their life (Starr & Samuels, 1997; Farrey, 1997; Malone, 1993). I

PAGE 50

43 The ability to separate these different aspects of the individual is paramount to this research, and this is why the use of Harter's research on the separate domains of self-concept for adolescents guides this examination of self-concept. While Harter and others have identified specific domains for self-concept, this does not mean that everyone places equal importance on each of the specific d(nains. In fact, each individual likely holds different opinions of the importance and salience of the various facets of self-concq>t(Hattie, 1992; Harter, 1987; Gergen, 1971). For example, Hattie (1992) pointed out, "A minister is more likely to have a more clearly delineated and richer moral self-concept than an athlete. The athlete may have a clearer physical self-concept" (p. 55). Individuals may deem particular domains important for a variety of reasons, including the amount of familiarity or learning witii a particular domain, the presence of a need or motivation toward a particular domain of self-concept (ex. peers may give approval to those who are proficioit in sports, yet may not give the same approval to high marks in academics), competence in a domain, and i^ticular social situations which may cause particular domains to become increasingly important (ex. fi-equently attoiding parties where sodal interaction is important (Hat^, 1992; Gergoi, 1 97 1 ; Videbeck, 1 958). While any of these factors may help dictate which domains are • perceived as more important than others for the individual, the importance of any particular domain may be differrait for each individual. Ther^re, because domain importance can vary both by domain and for each individual, it is extremely important to measure self-concept as a fimction of domains instead of an overall measure. The importance of each domain should also be measured, as is done in ^s research project Further explanation of the importance of domain specific self-concept evaluation will be given during the following section of self-concept formation and enhancement. Development and Maintenance of Self-rnnr.ftpf Self-concept is formed through life experiences and interactions. No one is bom with a self-concept, but rather it is learned (Felker, 1974; Frey & Carlock, 1989;

PAGE 51

44 Baughman& Welsh, 1962; Wells & Marwell, 1976; Orenstein, 1994; Krechetal., 1962). Self-concept is developed through self-appraisal, reflecting upon one's behaviors; reflected appraisal, reflecting upon cues from others; social comparison, comparing oneself to others; and biased scanning, seeking out information to confirm one's opinions while filtering out contradictor/ information (Loundon & Bitta, 1979; Runyon, 1977; Baughman & Welsh, 1962; Krechetal., 1962; Bumstein et al., 1962). One may look to a variety of people and sources in the formation of self-concq>t These include paroits, peers, teachers, siblings, and other persons deoned significant to the individual (Loundon & Bitta, 1979; Hattie, 1992; Frey & Carlock, 1989; Covington & Beery, 1976; Orenstein, 1994; Baughman & Welsh, 1962; Krechetal., 1962; Felker, 1974; Katona-Sallay, 1993). Although many influences on self-concept are personally known acquaintances, mass media figures, and specifically sports heroes, can also influoice the formation and maintenance of self-concept (French, 1991; Katona-Sallay, 1993; Caughley, 1984). A fantasy relationship with a media figure can serve to confirm one's sense of his or her own idaitity. Caughley (1984) wrote in reference to such a perceived relationship between fan and media star, "Landing such a widely adored celebrity figure confirms your self worth; it makes you somebody" (p. 50). To create such a mythical relationship, a fan may actually fantasize fliat flie media star chooses the individual ova millions of other viewers, picking that person to be a special fiiend or even a lovar or wife. Caughley (1984) explained the allure of these fantasy relationships, explaining that "through ^tasy, a media love relationship is exquisitely turned not to the needs of the celdmty, but to the needs of the self (p. 5 1 ). The relationships with mass-mediated, celebrity figures can be extremely powerfiil in shaping personal identity and self-concept. Media fans also can make social comparisons with popular mass media figures, and such comparisons, both negative and positive, can lead to fiirther development of self-concept (Oppenheimer & Oosterwegel, 1 993). This process is similar to I

PAGE 52

45 self-evaluations that can occur with traditional, non-mass mediated models (Bumstein et al., 1962). Tha-efore, personal interaction is not necessary to impact self-concq)t, and media figures can help to shape the self-concept of a viewer. This may be particularly important to adolescents who, in an effort to establish their o^ individual identity, often tum fiom more traditional sources of p:sons, such as paroits, and look toward new sources, such as mass media figures, to define their self-concepts (French, 1991 ; Katona-Sallay, 1993). FurthCTmore, most people form their self-concepts during childhood and adolescence, and self-concept becomes more resilient to change during adult years (Krech et al., 1962; Orenstein, 1994; Hattie, 1992; Katona-Sallay, 1993; Frey & Carlock, 1994). Therefore, research into adolescent self-concept formation is crucial, because this represents a period when self-concept is still volatile and changing and is a time when one seeks new sources and people, including media figures, who may help form one's self-concq)t. Self-Concept Maintenanr^ A primary human objective is to maintain and enhance real self-concept, and this goal influences human behaviors and interactions (Hattie, 1992; Runyon, 1977; Grubb & Grathwohl, 1967; Felker, 1974; Frey & Carlock, 1989; Delozier & TiUman, 1972; Sirgy, 1983; Wells & Marwell, 1976; Loundon & Bitta, 1979; Orenstein, 1994; Krech et al., 1962). One prominent means of achieving this goal is to bring one's real self-concq)t, or one's view of oneself, closer to one's ideal self-concept. People's behavioral intentions and choices of fiiends and role models are often formed in an effort to maintain or improve their real self-concepts. Therefore, people act in sudi a way as to gain higher opinions of ftemselves or to preserve the opinion they have of tfaonselves. Such self-maintenance and self-improvement occurs because it can result in a higher sense of self-worth (Orenstein, 1994; Barter, 1987; Barter, 1982; Hattie, 1992; Fdkcr, 1974). In othowords, improving one's self-concept can be a primary behavioral influence.

PAGE 53

46 However, this view of self-concept enhancement is likely a simplification of the process. Instead of viewing self-concept as a singular construct, self-concept maintenance and enhancement should be studied in terms of the many distinct domains of self-concept. This is because each person places a different level of importance on each particular domain, or, as described by Hattie (1992), "The saliraicy of various dimensions of self-concept differs across individuals" (p. 169). This difference in saliency among individuals leads to each specific domain contributing differently to one's sense of self-worth (Hart-, 1982). While people likely attempt to maintain and improve real self-concept in all areas, people are more likely to do so in their most important domains ofself-concept(Harter, 1987; Harter, 1982; Hattie, 1992; Runyon, 1977). Hattie (1992) explained, "An individual's sense of worth may be in a variety of areas: the body (beautiftil) is of the highest importance to some, but for others it may be more important to be academically able, to have a happy family fife, to gain respect for each other, or to have a desirable personality" (p. 55). Therefore, it is these particular areas for which the individual will have the most interest in maintaining and improving the real self-concept, because these areas/domains will contribute the most to one's sense of value. Self-Conce pt and Behaviors The focus area for this research project is the potential link between self-concept and behavior. The desire to maintain real-self concept and reach towards one's ideal self-ooncq)t by behavioral intention has been examined primarily Aro^igji studies of consumer behavior and product selection. While this niay differ from the topic researched in this study (role modeling of sports heroes), prior maiketing research should help build a foundation of knowledge and guide the r^earch for this study. Two of the earliest examinations of self-concept as a behavioral influence were done by Rogers (1951) and Maslow (1954). Rogers explained that each person will strive to actualize, maintain, and enhance himself or herself, and these needs will guide people's actions (Rogers, 1951; Felker, 1974). Maslow (1954) introduced similar ideas in

PAGE 54

47 describing the process of self-actualization, which refers to the process of achieving one's potoitial. According to this process, people are motivated to bdiave and act in order to achieve their potential. However, according to Maslow*s theory, this need for self-actualization is the last in a long Ust of human needs. Therefore, self-actualization will be the last factor to impact behavioral decisions. Since Madow's tiieoiy of connecting self-concept and behaviors is deeply embedded within a larger theory of human needs and existence, it is somewhat difficult to isolate this concept within Maslow's woik. Much of the subsequent research relating self-concept to behaviors involved mariceting and product selection. According to such research, consumes choose products which are consistent with their own image, or self-concept, and diis can lead to a variety of purchase decisions. Kretch et. al (1962) explained this relationship by describing how self-concept motivated middle and upper-class Europeans to choose a lower quality of bread due to the lofty image of this product. the very foods that an individual chooses to eat must not only satisfy his hunger but also be congruent with his conception of himself as a certain kind of person. Thus the less nutritious white bread came to be preferred to the more nutritious dark bread because the dark bread was raniniscoit of the European peasant's black bread. The color of the bread had become a status symbol and therefore, a determinant of how well it would assuage the status seeker's hunger, (pp. 83-84) The choice of bread was motivated by congruence between fbe white bread and self-concept. Because white bread was perceived as a high status product, it probably was indicative of the ideal self-concept of these Europeans. Food was selected not only to satisfy hunger, but also due to its congruence with oneself as a po^n. Several studies have found consumers to choose products consistent with their self-concepts (Landon, 1974; Sirgy, 1983; Loundon & Bitta, 1979; Runyon, 1977; Hattie, 1992; Delozier & Tillman, 1972; Birdwell, 1964; Ross, 1971; Felker, 1974; Dolich, 1961; Grubb & Grathwohl, 1967; Krech et al., 1962). Loundon and Bitta (1979)

PAGE 55

48 explained, "Products and brands are considered as objects that consiuners purchase either to maintain or to enhance their self-images. The dioice of which brand to buy dq)cads on how similar (or consistent) the consumer perceives the brand to be wifli his or her self-image" (p. 376). This theory has been examined and tested thoroughly, and it helps explain &e motivations for certain behaviors and behavioral intoitiras. Theorists have not definitively established whether red self^ooncept or ideal self-concept is more instrumental in guiding consumechoices (Loimdon & Bitta, 1979; Hattie, 1992; Delozier & TiUman, 1972; Sirgy, 1983; Runyon, 1977; Landon, 1974; Grubb & Grathwohl, 1967; Ross, 1971). It has been suggested that visible products, such as formal clothing or expulsive wines, might reflect the ideal self-concept, while privately used products, such as toothpaste and soap, might reflect flie real self-concept Additionally, researchers have suggested certain products, such as a luxury/sports car, are purchased more for status appeal and will be chosen for congruoice with the ideal self-concept. In contrast, other products, such as a station wagon, are purdiased for their utility value and will be chosen for congruence with the real self-concept. However, researchers have yet to establish exacfly \mder which conditions real self-concept is used or when ideal self-conc^t is used. Grubb and Grathwohl (1967) explained, further research is needed in terms of specific consumer decision situations to determine to what extent self-enhancement involves a conformity concept or an ideal self-image concept. For example, are consumers, through their consuming behavior and the interaction process, seddng siq>pQrt for their self-concept as they now perceive dionselves, or are they seeking reactions that will promote the attainment of a more ideal self? (p. 208) While both self-concepts have been identified as important in determining consumo* behaviors, the specific instances and motivations for particiilar behaviors is not yet fully understood. The importance of the individual domains, or domain importance, of self-ccmcqrt may also help determine consiraier behavior. In other words, it is possible that choice of

PAGE 56

49 consumer products also depends upon which domains of self-concept are most important to the individual and how the selected consumer product relates to this particular domain. For example, the purchase of a calculator may be more related to academic self-concept, and the selection of athletic shoes may be more related to the physical self-concept. While not specifically addressed in any known research, researchCTS hinted at this potoitial relationship when they explained that diffwent types of people (ex. housewives, tennis players, aspiring executives) may seek different types of consimier products (ex, linens, white shorts, leathobrief cases) in reference to their individual self-concq)ts (Hattie, 1992; Grubb & Grathwohl, 1967; Loundon & Bitta, 1979). Therefore, the areas of self-concept that are most important to the individual will guide the selection of consumer products. The selection of a brief case may not correlate at all to the self-concept of a housewife because the related domains (professional accomplisliment, etc.) may be unimportant to this individual. Because this area of study is largely unstudied, domain importance and its correlation with behaviors is further examined in this research project. Self-Concept and Mass. Media Figures Little research has addressed the selection and adoption of behaviors of mass media figures with respect to self-concept or other related constructs. Caughley (1984) addressed the perceived relationship between a viewer and an admired media figure, writing, *The appeal is often complex, but the admired figure is typically felt to have qualities that the person senses in himself but desires to develop further. The admired figure represents an ideal self-image" (p. 54). In other words, the obsovachooses a media figure who represents^an ideal Self-concept, particularly in areas that are important to the observer. Accordingly, domain importance is a crucial factor in the relationship between media heroes and fans. Through this complex relationship, the fan can attmpt to develop the most appealing characteristics of his or her favorite celebrity figure, allowing the fan to strive towards the most valued areas of his or her ideal self-concq)t.

PAGE 57

50 Evai less is known about the relationship between the sports hero and the observer in reference to the fan's individual self-concq)t Tener's (1987) diss^tion work found tiiat the hero choices of an adolescent can be used to predict self-concq)t, suggesting a consistency between the hero and the observer. However, this work did not d^ wiih the real and ideal self-oncepts, but rather just the strengfli of an adolescoit's real self-concept. Additionally, this research focused on the importance of choosing a personally known vs. an unknown h&o, and it did not address the different characteristics of chosen heroes. While not addressing the specific construct of self-concept, several authors have suggested that fans may dhoose and role model their favorite s|x>rts figures based on their perceived similarities between themselves and the athlete (Wilson & Sparks, 1996; Cole, 1996; Kellner, 1996; Harris, 1994a; Simons, 1997; "Role models", 1989). The work in this area has not addressed the individual's perc^tion of himself, or self-concept, but radio: studied obsovable characteristics of the individual, including race, playing style, and gender. Therefore, one primary goal of this research is to examine modeling of sports heroes in relation to individual self-concept, not geaasi group characteristics or qualities. Adolescent Self-Concq>t Hiis study deals q)ecifically with self-concept of adoles(aits. Adolescoice represCTts the beginning of drastic change to one's self-concqst, mariced by an increased awareness of self-concept (Orenstein, 1994; Harter, 1988; Frey & Carlock, 1989; Hattie, 1992); Felker (1974) explained, 'Thiring a^kaGems^ lkax is an increased drive for change in the individual. The increased change and corresponding lack of stability brings with it a need for reorganization of the self-concept" (p. 104). Such reorganization has beai labeled a "crisis of identity," a process of drastic change and instability in self-concept diffing adolescraice (Hattie, 1992, p. 132; Erikson, 1950).

PAGE 58

51 Frey and Carlock (1989) described adolescence as a movement from a period of dq>enden(^ to a period maiked by increasing separateness and indq>eiidaice. They wrote, "The movement away from dependency during adolescence is characterized by such questions as 'Who am I? How do I want to be?'" (p. 285). Additionally, adolescents typically become aware of a variety of expectations and standards, including standards for academics, social standing, and athletic competence, against whidi the self can be measured (Hattie, 1992; Frey and Carlock, 1989; Orenstein, 1994). Therefore, this ongoing process of measuremoit creates a greater interest in and awaroiess of one's own perceived competence and ability. However, most authors have acknowledged that while adolescoice marks a period of change for the self-concept, it is gaierally a gradual, less rebellious period during which adolescents "make the bid for autonomy gradually and these requests meet reasonable consideration and deference from parents who ally themselv^ with the child's need to grow" (Douvan & Gold, 1969, p. 485; Hattie, 1992; Wylie, 1961). In other words, parents and children typically engage in a give-and-take as adolescents attempt to form their own identity. It is reasoned that adolescents often shift from using family as their {nimary reference and instead rely more on peers, teachers, and mass media to establish their own identity (Hattie, 1992; Felker, 1974; Katona-Sallay, 1993; Pekrun, 1993; Frey & Carlock, 1974). This can create an instability in sdf-concq)t that is only increased by the ongoing physical changes throughout adolescence ((^istein, 1994; Frey & Carlock, 1974; Felker, 1974). A study by McGuire and McGuire (1981) examined the types of people adolescents named whoi asked to describe themselves in relative terms. Compared to younger children, adolescents were more likely to shift fixun parents to teachers and from siblings to friends when defining themsdves. In other wordst adolescoits woe more likely to seek external sources to define themselves during this period of self-concept awareness and development. Research done by Pekrun (1993) indicated that while the

PAGE 59

52 self-concepts of adolescents are still influenced by family support, adolescents also look to other sources, particularly teachers, to formulate self-concept Pdcrun further added that different sources may impact different areas, or domains, of self-concept. Pekrun wrote, "School factors have an impact on domain-specific, school related self-concept development. Furthermore, it may be that peers exert primarily influences on the development of domain-specific self-concepts of social competencies" (1993, p. 118). This research not only indicates the variety of influoices on adolescent self-concept, but also stresses the importance of using separate domains of self-concept for the study of adolescent self-concept. Adolescents also may begin to look toward mass media figures and popular figures in reference to tiieir own self-concept (Orenstein, 1994; Frey and Cariock, 1989; Hattie, 1992). Unfortunately, despite the prevalence of media figures in the lives of adolescents, relatively littie research addresses the influoice of such figures on adolescent self-concept. Adams-Price and Green ( 1 990) found that a crush on or an attadimait to a celebrity figure is an important aspect in self-concept development during adolescence. This process of develoinnent often occurs during solitary use of media, aUowing an adolescent to discover his or her new sense of self (Reed, 1 995). Such solitary involvement with media reflects the increased awareness and individjiality of self-concept expected during adolescence. : The majority of the recent literature on television and development of self-concept has focused on adolescent girls and their percq)tions of their bodies. Due to the prevalence of thin, attractive females in mass media, adolescent fonale media usos often develop lower self-concepts of their own physical appearance, seeing themselves as ovoweight and imattractive in comparison to these media figures (Schneider, 1996; Rodin, 1992; Harrison, 1998; Tiggermann and Pickering, 19%; Goode, 1999; Botta, 1999; Becker and Popenoe, 1998). These images also have caused many female adolescent viewers to develop much higher, and likely unadiievable, ideal sdf-concepts

PAGE 60

53 for their physical appearance (Smith, 1985; Becker and Popenoe, 1998; Goode, 1999; Harrison, 1998). According to dissertation work by Harrison (1998), viewing of television with conspicuously thin charactCTS resulted in hi^erbody dissatisfaction, while viewing of television with overweight characters did not have the same effect on physical self-concept. Similarly, Beckraand Popenoe (1998) found lAmt the recoit introduction of television and its images of attractive, thin lead characters to a remote South Pacific island led to a decrease in individual self-image of the female adolescent viewers. Additionally, these adolescent viewers described^l^^^teir kteal body image from that of a robust, rounded body, typical of the adults of the island, to that of a thin body, typical of the televised figures. The changed ioq^iesulting fpm watching media figures has lead to chmiged behaviors of adolescrat fmales, nKMrt ixQldrfy eating disorders (Goode, 1999; Becker and Popenoe, 1998; Schneider, 1996; Harrison, 1998; Smith, 1 985). In one study, 1 5 percent of surveyed teenagj^ j^lf admitted tp dieting or exercising to look more like an image seen on television (Schnddtf 19%). While these studies have focused primarily on adolescent females and the physical component of self-concept, they do show examples of adolescents developing their self-concqpt based on television figures. Furthermore, some of these studio detailed behaviors reftdting from such social comparison to media images. This behavioral component of mass media social comparison is a focal point of this study, and diese studies itieidme provide sound foundation for the thecmtical base of this research. The following section describes congruity theory, a theory j^edicting tbc relationships between self-concept congruity and bdiavior. Congjuity ftnetxy builds
PAGE 61

54 Congruity Theory The reviewed literature has established a link betwerai self-concqrt and behaviors. In order to logically establish hypotheses for such a link in this research, a theoretical framework mxjst be established. For this research, congruity theory provides such a framework. Congruity theory, jwoposed by Sirgy (1 980), thus fer has beoi used only in referraice to self-concq)t and consumer behaviors. However, Sx this research, the framework of congruity theory has been extended to study self-concept and role modeling of sports heroes. According to congruity theory, individuals assign images to consumer products, much like they assign self-concq>ts to themselves. Ccm^iitty, or congruence, refers to the relationship between one's self image and the image one assigns to a consumer product. Specifically, congruity increases as the distance between self-coocept and product image decreases. Conversely, an increase in Has distance results in less congruity. This distance is also referred to as a "distance score." (Distance scores have an inverse relationship with congruity. Therefore, as congruity increases, the distance score decreases.) According to congruity theory, the subsequent intention to purchase a product is based on the congruity between one's self-concept and one's individually assigned product image. Sirgy (1982) has idoitified two conditions that predict one's intration to vise a product. The first involves real (actual) self-concept and product image congruity. Sirgy wrote, (1982), "The relationship between actual self-image/fHtnliict-image congruity (self-congruity) and brand preference has been supported by numerous studies" (p. 1 65). According to Sirgy (1982), as the two images become more similar, or more congruous (lower distance score), the resulting affect will be cognitive consistency, ^ch will lead to subsequent purchase of that product. Conversely, a growing distance between these two images, or decreasing congruity (higher distance score), will lead to cognitive

PAGE 62

55 dissonance (consistency affect), which will likely result in the individual avoiding such consumer products. Numerous marketing studies have supported this relationship. These include studies by Dolich (1969), Delozier and Tilhnan (1972), Grubb and Grathwohl (1967), and Ross (1971). In each of these studies, researchers found that the real self-conc^t was more closely matched to the most preferred brands and less closely matched to the least preferred brands. These studies siq)port the first component of Sirgy's (1982) congruity theory. The second condition involving self-concept and brand image as a predictor for purchase intention deals with the ideal self-concept. Sirgy (1982) wrote, "The relationship between ideal self-image/product-image congruity (ideal congruity) has beoi generally supported. Ideal congruity will lead to positive product evaluation, whereas ideal incongruity may be responsible for negative product evaluation" (p. 167). In other words, a highocongruity between ideal self-concept and an individual's imi^ of a product will lead to a greater likelihood of purchase intentions. (The distance between ideal self-concq)t and &e image of a product is also called an 'Ideal distance so(h%." As congruity increases, or as the ideal self-concq)t is closer to the image of a product, the ideal distance score decreases.) While theorists have siq^rted this aspect of con^uity theory, not as many studies have addressed ideal self-concept as have addressed real self-concept. Dolich (1969) concluded that both real self-image and ideal self-image are predictive of brand preferoice, and purchase intentions for some products tend to be more closely tdated to real self-image, while purchase intentions for other products tend to be more closely related to ideal self-image (Dolich, 1969; Ross, 1971; Loundon & Bitta, 1979). Loundon and Bitta (1979) provided a goieral assessment of the rdation^ps for both real and ideal self-concept predicted by congruity theory, writing, "The consumer behavior of an individual will be directed towards the furthering and enhancing of his self-coDcqpt

PAGE 63

56 through the consumption of goods as symbols" (p. 377). In other words, people will use products both to maintain self-concq)t, fiirther establishing title red self, and to enhance self-concept, moving closer toward the ideal self Therefore, bofli the real and ideal self-concepts are useful as predictors of consumer behavior, and both should be considered in the design of academic research. A pplication of Theoretical Model to this Study Congruity theory has significantly added to the understanding of human behavior and the importance of self-concept in detmnining behavior. According to congruity theory, self-concept has an impact upon consumer bdiaviors. ConsumCTs use objects that are congruent in image with either their real self-concept, their real ideal self-concept, or both self-concepts, as this will satisfy the inherent need to &&i&t maintain or improve self-concept. Congruity theory, therefore, explains the relationship between self-concq)t and some bdiaviors. However, while theorists have indicated that self-concq)t has a major influence upon many human behaviors, including the choice of friends and effort placed into academic work, little (if any) research exists outside of the study of consumer bdiaviors (Hattie, 1992; Orenstein, 1994). Because marketing studies have been successful in identifying behavioral intentions, academics should attempt to apply congruity theory and the study of self-concept to other important areas of behavioral iaflueoce. This is the onphasis of this research project. It is hoped that the appliodoD of sound tiiecxry to the study of important societal behaviors, in this case role modeling of sports heroes, will be a valuable addition to the academic coniinunity and will fiaiiid: die unda-standing of the impact of sports heroes on American adolescents. Prior research studies have hcea used to provide the frazhework for this research. However, because subsequent studies examined consumer behaviors, and this researdi project examines role modeling of sports heroes, obvious alterations have berai made. Most obviously, the behavioral component of prior studies, purchasing consumer

PAGE 64

57 products, has been replaced with role modeling behaviors of the sports hero. Popular, mass-mediated sports figures have become so entrmched with marketing products and mass-media promotions, often for products named after them, it is not such a extension to view sports figures as products themselves (Katz, 1994; "Michael Jordan's", 1991). Perfa^s the most important justification for this study design is Hic lack of scientific research in the area of self-concept and role modeling. Few studies of role modeling have examined the influence of self-concq>t. Because congruity theory provides a sound researdi fi'amework, a somewhat liberal alteraticm of marketing studies using this theory provides the most logical approach for this dissertation. Self-Concept Domains as Part of Research Design Prior Studies (Dolich, 1969; Ross, 1971) ofcongruity theory have used matching questionnaires to measure the individual self-concept, both real and ideal, and the product image. FrcMn these matched surveys, researchers searched for statistical congruity between the answers fitnn the two surveys. In oiher words, distance scores between self-concept and product image were calculated. Researchers in these studies then used differoit questionnaires to measure the behavioral componoit, purdiase intaitioa. To test congruity theory, distance scores (or congruity between sdH-^tacitept and product image) were compared to intentions to buy the product. According to congruity theory, as distance scores deoease (higher congruity), purchase intentions 9b;>vld incret^. This dissertation follows a similar design. However, in this study, questi
PAGE 65

athletic competence, physical appearance, job competence, romantic appeal, behavioral conduct, and close fiioidship. This use of domains in this research allows for more specific results. As suggested previously, fans of a sports hero may attempt to copy certain aspects of the athlete's behavior or personality and not other aspects (Simons, 1997; ''Role models**, 1989). Therefore, addressing role modeling across the eight separate domains further clarifies the particular areas in which the role model will be copied. For example, an individual may role model his s^rts hero much more in the domain of atUetic competence than physical appearance. The congruence between one*s self-concept and the image of the sports hero should be analyzed across those particular domains instead of using a more general, overall measure. This provides a more accurate and usefiil analysis of the importance of congruity in role modeling sports heroes. Additionally, as addressed in the literatiu-e review, certain domains of self-concept may be more important to an individual than other domains (ex. physical appearance may be more important than job competence to an individual). The literature review has suggested that individuals will copy role models in areas that are most important to than. Therefore, this research measures the importance of each domain to see if increasing domain importance leads to increased role modeling of one's sports hero. Hypotheses The review of prior research has lead to die formation of four hypotheses, all four of which are listed and explained in this section. These four hypotfieses address the research questions presented at the end of Chapter 1 and are examined through the collection of survey data. The desaiption of this analysis is outlined in C3iapter 3, tiie Mediods section.

PAGE 66

59 Hypothesis I. For eadi domain of self-concept, role modeling of the sports hero will be positively related to the congruity between real self-c(mcq>t and the image of the sports hero. The reader is reminded of tiie positive relationship theorized throng congruity theory; as the distance (difTeroice) between self-concept and the image of the sports hero decreases, congruity increases. For this study, distance scores for each domain of self-coiu^ept are calculated, and these distance scores reflect the distance between self-concept and the image of the sports hero in each domain of self-concept. As a distance score increases, congruity between self-concq)t and image of a sports hero decreases. As a distance score decreases, congruity between self-concept and the image of the sports hero increases. Therefore, a positive relationship between role modeling and &e congruity of self-concept/image of the spofte hm> would be realized iifbi^ a negative relationship between role modeling and distance scores. HI deals specifically with real distance scores, or the distance bet^n^en real self-concept and the image of the sports hem. This hypothesis, along with the othathree hypotheses, will be writtrai to reflect the n^ative relationship with distance scores and role modeling described above, as this ne^^ative relationship, and not the inverse positive relationship with congnuty and role modding, will be Die fiscal pdnt oil^ examng statistical analysis. Tlus is rejected in tiie following restatement of HI. Hypothesis I For each domain of self-concept, role modeling of the sports hero will be negatively related to the distance between real self-concept and the image of the sports hero, also know as the real distance scores. The second hypothesis also addresses the relationship betweoi role modeling and die distance between self-concept and the image of the sports hero. However, H2 deals specifically with ideal self-concept, and the distances betwei ideal self-concq>t and flie image of tiie sports hao. Th^ distaoices, calculated for each domain of self-ccmcc^ are

PAGE 67

60 know as the ideal distance scores. Similar to HI, a positive relationship between role modeling mi the congruity of ideal s#l^:||p6pt the image of the sports hero is predicted. This is reflected in a predicted neg^ve relationship betwem icie mudtX^ and ideal distance scores, as stated in H2. Hypothesis II For each domain of self-concept, role modeling of the sports hero will be negatively related to the distance between ideal self-concept and the image of the sports hero, also know as the ideal ^riiftcie Scsores; Hypotheses III and IV build upon HI and H2 by inccoporating dcnsakkiportance into the independent variables (distance scores), creating new dependent variables known as distance worth scores. Distance worth incorporates both the distance betweoi self-concept and the image of the sports hero (distance score) in a domain of self-concept and the importance of that domain of self-concept. Therefore, this new variable is created by combining two other variables. A detailed explanation of distance worth scores and a description of the calculation of distance worth scores can be found in tiie {nxKeeding Methods section. A brief description of this construct follows below. For this study, distance worth scores are created by incorporating the importance of a domain (domain importance score) with the distance betweoi self-concept and the image of the sports hero (distance score). Distance worth scores reflect the distance score for a domain of self-concept multiplied by the importance of that domain. Howcvct, domain importance scores have an inverse relationship with distance worth scores. In other words, distance scores are actually divided by domain importance scores to create distance worth scores. Therefore, the lower the domain importance score is for a domain (domain deemed unimportant), the greater the distance worth score will be. Higgler domain importance scores (domains deemed important) will res\ilt in smaller distance worth scores. In oth: words, the less important a domain of self-concept is, the more tiie distance worth score for that domain will increase. Domain importance serves to further

PAGE 68

61 qualify the relationship betweai distance scores, or the congruity between self-concept and the image of the sports hero, and role modeling. By creating distance worth scores through the incorporation of domain importance and distance scores, it is predicted that the less important a domain of self-concept is to an individual, the less likely itat tiie congruity between self-conc^t and the image of a sports hm> in that domain will have a significant impact upon role modeling that sports hero. As a domain becomes more important to the individual, increased congruity (decreasing distance scores) for that domain will be more likdy to have a positive relationship with role modeling of a sports hero. Hypotheses III deals specifically with real distance worth scores, a set of variables that incorporate real distance scores in each domain and domain importance scores for the corresponding domain. Hypothesis HI. For each domain of self-concept, role naodding of the ^rts horo will be negatively related to real distance worth. The fourth hypothesis also addresses the relationship between role modeling and distance worth scores. However, H4 deals specifically widi ideal distance wordi, which is abated in the same manner as real distance worth, except that ideal distance scores, not real distance scores, are divided by domain importance scores for each corresponding domain. As with H3, a negative relationship betweoi role modeling and ideal distance wordi is predicted. This predicted relationship is reflected in Hypothesis IV, as listed below. Hypothesis TV For each domain of self-concept, role modeling of the sports hero will be negatively related to ideal distance woifli. The following ch^ter. Methods, addresses the specific methods and statistical analyds used in Has dissotation.

PAGE 69

CHAPTER 3 METHODS The methods chapter covers the following topics crucial to the collection of data and testing of stated hypotiieses: the subjects for this study, the measures administered for self-concept and the image of the sports hero; creation of the role modeling questionnaire; pretesting the measures used in this study; the collection of data (the administratian of q^stionnaires); statistical analysis of collected data, ^diessdng each of the stated hypotheses; and potential limitations of this analysis. Validity and idiability are discussed throughout this chapter as necessary. Subjects for Study This dissertation was designed to analyze the adolescait ^-^a^sxpt, fhe relationship between this self-concept and perceptions of sports heroes, and the subsequent likelihood of adolescents role modeling the |)^viors of sports hooes. Therefore, all subjects were hi^ school studoits in grades nine and te^ hpf^jdmatcly aged 14-16. Additionally, this dissertation only examined the behavioral intentions of male adolescoits, so no females were included in the pool of subjects. Tliere are tiuree reasons for this restricti(Hi. First, such a research design prevents gender from being a confounding variable in data analysis. Second, researchers have suggested that male adolescents are more likely to look to mass media figures, including athletic hm)es, as role models than are their female counterparts, who are more inclined than males to continue to view parents and personal acquaintances as role models (McEvoy & Erikson, 62

PAGE 70

63 1 98 1 ). Third, as the vast majority of popular mass media sports figures are male, adolescent males have many more available sports hero choices of die same gmdo'. Because perceived similarity has been suggested as a potential factor for role modeling, adolescent males may be more likely to find role models in spotts hooes than fenaie adolescents. Such an approach is consistent with research in the area of spotts hm>es as role models, where researchers have suggested that popular sports figures, such as NBA basketball players, are more likely to serve as role models for male adolescents ("Role models", 1989; Wilson & Sparks, 1996; Simons, 1997). The particular subjects for this dissertation consisted of ninth-grade and tendi-grade males fnm Houston, Texas. Of the 1 72 valid subject used Ibr ikii malysis in this study, 120 subjects were students in a suburban private ^ool. All of those 120 students were participants in school athletics. The decision to use only student-athletes was made due to the increased likelihood that student-atiiletes have strong ties and potential perceived similarities betwera themselVes aind meir sports herdes. The decision to use adolesc^ts firom this particular sdiool as subjects was largely a pragmatic one. First and foremost, die principal of the hig^ school willingly agreed to allow his students to participate in this study. Second, because the school is private, the researcher was able to avoid much of the difBculty involved with using subjects fmm the public school system. Finally, because all students are male, the research process could avoid the perception of being exclusionary to female students in a coed school, as the female students would not be able to participate as male students did. The students fiiom this school were predominantly yMte, a small petcmtage of minorities (Asian, Hispanic, Afiican-American). Most students live in stable uj^)er-middle-class family environments. Because die student body is not as large as diat of many public schools in the greater Houston area, a relatively large percentage of students participate in school athletics, and such participation is encouraged and

PAGE 71

64 promoted by school faculty and leadership. Therefore, athletics, at least in the participatory sense, is a part of the daily lives of many of this school's students. The remaining 52 subjects were participants in a sports tournament in Houston run through a local community crater. These siibjects, of the same grade and age parameters as the first 120 subjects, also were participants in school athletics. The majority of these subjects participated in baseball, but other selected sports included soccer, basketball, and track. These subjects share similar donogr^hics traits with the first 120 students surveyed (white, upper-middle-class, supportive fionily environment), and in fact the data collected from the two groups were virtually identical. Again this group of subjects was selected largely for pragmatic reasons (availability, access, age ^propriateness). Using two groups of subjects and receiving similar results hdped to strengths the validity of this research. Therefore, the data analysis for this study and the subsequent analyses of the stated hypotheses were based on the responses of 172 valid subjects. While more than 200 subjects participated in the study, surveys fix)m approximately 30 subjects could not be used due to subject error in completing the surveys. The majority of these errors involved subjects not completing all parts of the sjffvey questionnaires. One criticism of this study may be that the students do not represent a diverse sample, decreasing external validity. Similarly, because most of these students come fixMn stebie femily environments with many available role models, these students may not use sports heroes as role models as much as adolescaits wjio do ^ot have many other available role models. However, this study will not claim to be geno-alizable to all populations, nor is it able to address issues of race, socioeconomic status, and family/home environment. Additionally, because this study addresses each parson's individual self-concq>t and allows each person to identify his individual choice of sports hero, the focus of this research is not on societal patterns, but rather individual choices based on individual relationships between self-concqrt and the image of their ^rts hero.

PAGE 72

65 In other words, this research is designed to enable better understanding of how self-concq)t affects individual choices to role model a sports hem, not flie impact of external sources on role modeling, and therefore demographic factors (race, family, etc.), while important in the study of role modeling sports heroes, are not paramount to this research. Teste and Test Admini.stration The following sections detail the survey instruments used for data collection in this study. This includes the selection of existing instrumoits, the altmtions of such instrumoite, and the construction of new instruments. Self-Concept and Ima ge of the Sports Hero For the measuronait of real and ideal self-concept across all domains, Ibe Adolescent Self-Perception Profile, created by Susan Harter (1988), was used. TWs test, an upward extension of Barter's Self-Perception Profile for Children (1982), measures self-concept in eight specific domains that are most important and relevant to adolescoits. Barter's rationale behind using such a domain-specific format, as opposed to a singular score of self-concq)t, is as follows: "An instrument providing separate measures of perceived competence or adequacy in different domains. would provide a ridioand more differentiated picture than those instruments providing only a single score" (Harter, 1 988, p. 2). This test was designed for students in approximately grades nine through twelve. The test measures self-concept in eight specific domains. Each domain is described below, as reported in Barter's manual for this test (1988):

PAGE 73

66 1) Scholastic Competrace: The adolescent's perception of his/ho: coiiq)etaice in scholastic perfonnance, achievement in classwork, perception of one's intelligence. 2) Social Acceptance: The degree to which one is accq>ted by peers, popularity, number of friends, feels that he/she is easy to like. 3) Athletic Competence: Adolescent's perceptions ofhis/her athletic ability, feelings ihat one is good at sports. 4) Physical Appearance: Degree to which the adolescxnt iis happy with how he/she looks, likes one's body, feels he/she is good looking. 5) Job Compet^ce: Extent to which the adolescmt feels he/she has job skills, prepared to succeed at part-time jobs, feels he/she is doing well at current jobs. 6) Romantic Appeal: Adolescent's percq)tions that he/she is romantically attractive to those in whom he/she is intnested, dating ptople he/she waits to date, feels he/she is fun and interesting on a date. 7) Behavioral Conduct: The degree to which an adolescent likes die way he/she bdiaves, does the right thing, acts in appropriate manno", and avoids trouble. 8) Close Frioidship: Adolescent's perceptions ofhis/her ability to make close friends to share personal thoughts and secrets Avith. These eight domains make up the test of self-concept. Five questions address each domain of self-concept (40 questions ovoall). The original test has 45 questions because it includes a ninth scale iiat imasures overall self-concept. This ninth scale was not used in this study, so these questions were eliminated. Each question is rated on a 1 to 4 scale, and approximately half of all questions for each domain are counter-balanced such that some have the most adequate description on the lett (scored 4 to 1) and some have the most adequate description on the right (scored 1 to 4). (Please see Appendix A and Appendix B for actual tests used.)

PAGE 74

67 The reliability of Hatter's test of sdf-ooncq>t has beoi detennined tiuough repeated iise and examination of this instrument. Reliability refers to tiie reproducibility of a measure (Graziano & Raulin, 1993). Of primary importance in this study, particularly for the tests of self-concept, is internal consistoicy rdiability. bxtemal consistency reliability is relevant when several observations are made to obtain a score for each subject (Graziano & Raulin, 1993). Intemai consistaicy reliability is high when several items measuring the same thing (in this case, each of the dg^t d(nains of self-concq)t) correlate highly witfi eacli ottier. Several questions (5) are t^cd for each domain of self-concept, and therefore an intemai consistency reliabiUty score, reported as Cronbach's Alpha, is reported for each. Harris (1988) reported flie subscale internal consistency reliability for each sid>scale, as based on Cronbach's Alpha. They are as follows (for each domain): 1) Scholastic: .81 2) Athletic: .92 3) Acceptance: .78 4) Close Friend: .83 5) Romance: .80 6) Appearance: .86 7) Conduct: .78 8) Job Competence: .74 The high correlation of the items of each domain indicates a hi^ intonal consistency reliability (Graziano & Raulin, 1993). Additionally, Harter's (1988) fnetests reported mean scores for each domain ranging from 2.5 to 3.3, depending on the sex and age of the subjects. The standard deviations for each domain ranged fix>m {^proximately .41 to .8, again dq)oiding on the subjects.

PAGE 75

68 Multiple Use/ Administration of Harter's Self-Concept Test As explained in the literature review, five sq)arate measuremoits were made for this research (real self-concept, ideal self-concept, image of sports hero, domain importance, and behavioral intoition to role model one's sports hero). All but the last two constnicts, domain importance and behaviors, were measured using Harter's (1988) test of self-concq)t, but slight variations of the test were made for measurement purposes. In other words, Harter's test was altered so that the same questions addressing the real self-concept would instead address either the ideal self-concq>t or tiie perceived image of the participant's sports ho. Such a procedure for altering an existing test in this manner is derived fix)m the previously reviewed marketing studies that examine self-concept, product image, and purchase intentions (Dolich, 1969; Delozier & Tillman, 1972; Gnibb & Grathwohl, 1967; Ross, 1971; Landon, 1974). In these studies, the same (or extremely similar) tests of self-concept were used to measure flie real self-concqpt, the ideal self-concept, and the perceived image of the brand/product. For example, Dolich (1969) used the same semantic differential scale to measure all three constructs, as explained: "The semantic differratial was used to measure evaluations by respondents of real-self image, ideal-self image, and brand images. It is described as a combination of controlled association and scaling procedures and satisfies the requirement of measuring sevoal concepts with the same instrument" (p. 8 1 ). While these examples use such a scale to rate a brand image as opposed to the image of a sports hero, the process is quite similar and rqnresents a logical and sound research design. In order to use such a design, Harter's (1988) test was administered three times. (See Appendix A and Appendix B for actual test alterations.) The first administration measured the real self-concept. This was done with the original format of the questionnaire (subtitled "What I Am Like") (Hdoter, 1988). The second administration measured the ideal self-concept. In measuring this, references to self (me) were changed to the ideal self (How I Would Like to Be). The third administration measured the

PAGE 76

69 perceived image of the sports hero. In measuring this, references to self (me) were changed to eadi subject's individual choice of mass mediated sports hero. For each of the three variations of this test used in this research, the scaling, question order, and wording of the surv^ questions were idoitical. The only dififoence was that the first administration addressed the real self, the second administration addressed the ideal self, and the third administration addressed the sports hero. Mea5njrement nf TVhiiiiwi Importance In order to measure domain importance, this study used Harter's (1988) test of domain importance (subtitled "How Important Are Each of These Things to You?"), as included in Harter's adolescent self-perception profile. (See Appmdix A and Appendix B for actual test.) This questionnaire follows a similar format to the test of self-concept, with a rating scale of 1 to 4, with half of the questions keyed in reverse. This test contains 16 questions, 2 questions for each of the 8 domains. Data collected with this questionnaire were used for ihe cteati(m of distance worth scoks for analysis in Hypothesis III and Hypothesis IV. Mea.sMrement of Role Modeling The final questionnaire measured the adolescent's intention to role model his sports hero. In many of the questions, behavioral intentions were used instead of bdiavioral measures, as discussed earlio* in the literature review. Sudi a process of measurement also has been copied fi*om marketing studies, which often measure purchase intentions instead of actual purchase behaviors. For example, in a study of real self-concept, ideal self-concept, and consumer purchase intoitions, Landon (1974) measured purchase intentions with a five-point scale, with the fifth (most extreme answer) representing an intention never to purchase the product. (See Appendix C for actual questionnaire for role modeling one's sports hero.) This same structure has been used in this dissertation for the measurement of role modeling one's sports hero. A

PAGE 77

70 five-point scale, with one (1) r^resenting a strong negative response and five (S) iqHresoiting a strong positive response, was used to measure behavi(ral intentions. Two processes were used in creating the actual questionnaire to measure role modeling of one's sports hexo. First, specific potential types of role modeling were identified flirough literature review and examination of popular sports media (magazines, television, newspapers, etc.). Included in this examination was the purchase of many types of athletic goods and commercial products associated with popular sports figures, particularly those geared toward adolescent males. Second, the researcher asked sevoal adolescent males to identify role modeling behaviors typical of themselves and their peer group. Both a focus group of seven teenage males and opai-eadsd vmttea surveys (sent to approximately ten adolescent males) were used to identify aspects of role modeling that would be relevant for this study. From the areas of role modeling one's sports heso identified throu^ these procedures, a Likert scale survey insthmient was formulated. This instrument included the examples of role modeling most frequently identified by adolescoits and relevant literature. The instrumoit consisted of 23 Likert scale questions, 22 of which addressed specific behaviors of role modeling a sports hero and one ovoall role modeling question, used primarily to examine the validity of the questionnaire. Pretesting of Questionnaires The instrummt for role modeling and the ofhsx previously described measuronoit instruments (self-concept, image of sports hero) were further examined through pretesting. This pretest was conducted to allow for possible adjustmoits to the surv^ instruments before the actual test administration for this study. The pretest was conducted with 20 eighth-grade males from a private junior high school in Houston, Texas. These studaits completed all components of this study, just as was to be done in the actual research. This particular pretest population was chosen due to its similarity to the actual population that woiild be examined in this study.

PAGE 78

Students completed all questionnaires in the same order and manner as would be done for the actual study. Upon completion, students were given flie opportunity to provide both written and oral critiques of the survey instruments. Two issues were of particular interest in pretesting diese instruments. The first issue dealt vnHtk the age appropriateness of tiie language used in the questioimaires. In other words, did the pretest subjects fully understand the wording of each question, and was this wording appropriate for this age and demographic group? The second issue addressed whether all relevant areas of adolescent role modeling had been included in the questionnaire of role modeling. In addition to addressing these questions, the pretest allowed the researcher to practice survey administration and to observe the amount of time required by subjects to finish all questionnaires. While no major adjustments were necessary based on the pretest, minOT alterations were made to the wording of a few questions aboik iofc^ modeling. All of the role modeling questions can be found in Appendix C. Test Administration Five tests (two tests of self-concept, a test of domain iaq wii te ace. a test of tfie image of the sports hero, and a test of role modeling of one's sports hero) were administered to each subject. Tests were administered in either sdiool classrooms or gymnasiums. The total testing time, mcluding instructions mA time for qaesticms, lasted approximately one hour. Each student completed all five tests in one session. Tests were administo^ to small groiq)s of q)proximately 10-20 students for eadi sessi(Mi, and flie research^was {nesoit at all testing sessions, along with eithoa teadto* or coach. Aftof giving instructions about how the questionnaires should be completed and showing a brief example, the researcher provided an instruction pack^ for studoits instnicti(is beyond those provided verbally by the researcher. The researcher was available to answer questions during the entire testing period. No incentives were given to subjects for participation in the study.

PAGE 79

72 After tiie instructional session, subjects were instructed to pick the one athlete they most considered to be their sports hero. As discussed in the litorature review, while each subject was allowed to choose his own sports hero, individual choices were required to meet the following two cnteria: the athlete must be, or mi^ have been, covoed heavily by mass media (particularly television), and the atfilete could not be a personal acquaintance of the subject. This allowed for a wide range of hero choices but kept the focus of this study on mass mediated figures i potentiai vple models, flg^9||)osed to personal acquaintances, stich as family, friends, md coaches as role models. Subjects then were told that their selected sports hero would be the basis for all subsequent questions concerning sports hemes for the study (image, role modeling). In odier words, subjects were instructed to use their individual choice of sports hero as replacement for the generic "sports hero" of the questionnaires. After choosing a sports hero, subjects completed the tests for real sdf-concq^ ideal self-concept, and image of the sports hero at the same time. Having subjects complete the three similar tests together was done so students would consider their real self in juxt£qx>sition to their ideal self and the image of their qxts hero, hi other words, the real self-concept was meant to be used as an "anchor" or "standard" by which the ideal self-concq>t and the image of the sports hero would be considered. These Aree tests woe given using one questi
PAGE 80

73 and the image of their sports hoc. The sequoice of test adimmstrati(m was identical for all subjects. Because the same order was used to answer each question (real sdf answered first, fhea ideal sel^ and finally image of the hero), it is possible that this order affected the way in which each question was answered. In othen;9or& ed ^ ^i i^ aHBg fbe real self-concept first for each question might have impacted the scores giv^ for the ideal self-concq)t and, siibsequmtly, the image of the sports hero. The completed questionnaires were examined to ensure that such order effect did not occur. In particular, irregular answer patterns were looked for, such as a series of ascending answers for each question. The literature review has suggested ^it tiie image of the sports hero likely would fall between the real self-concept and the ideal self-concept, so inappropriate variations from this pattern would suggest potential order effect Howevo*, upon review of the completed questionnaires, order does appem: to have impacted the answers given for each question. Conducting all three similar tests during the same testing period should have eliminated the slight possibility of "history" as a confounding variable (Oraziano & Raulin, 1 993). If a week were to transpire between the administration of these three tests, it is possible, although unlikely, that an experience could alto: the real or ideal self-concept of an individual or an individual's image of his sports hoc, and this would impact the ensuing analysis based on these three constructs. Following the tests of self-concq)t and image of the sports heso, subjects answered flie 16 questions addressing domain importance. These were answered on the same question guide as the self-concept/image questions, and, upon completion, the entire answer page and question guide were gy/ea to the test administrator. Subjects then were given the role modeling instrumait. Before answering the 23 Likert scale questions on role modeling, students again wrote the name of their chosen sports hero at the top of the questioimaire and were reminded to focus answers on thdr

PAGE 81

74 particular choice of sports hero. This t^ took much less time to complete than the previous tests of self-concept/hero image, and subjects w&rc finished with their participation in the study when they completed and turned in this questionnaire. The process of analyzing role modeling through survey questionnaires posed an obvious threat to the internal validity of this study, specifically that subjects may have been inclined to give socially desirable responses for such questions as the behavioral intuitions to role model tiieir sports hero. In other words, subjects may not have been willing to admit, or evi be fiilly aware, that they would model such behaviors (ex. style of dress, dating behaviors, etc.). However, these threats to validity were reduced by allowing subjects to choose their own personal sports hm) (as of^sed to forcii^ any particular hero on all students) and by reminding students no question had either a correct or incorrect answer. Additionally, by using Likert scale que^ticHis for subject response (similar to the marketing studies that provide the firework for this research), this threat to internal validity was reduced, as subjects were allowed to consider such behaviors in degrees instead of merely being offered a "yes" or "no" option. Because of the length of the testing session and the large mimbar of questiiHis presented in the questionnaires, subject fatigue was a potential confounding variable for this study. Specifically, there was concern that subjects miglit not carefully consider answers given in the latter parts of Hie study questionnaires. To help reduce this risk, subjects were only given one of the two questionnaires at a time, which made the survey process less overwhehning for participants. Subjects first were given only die self-concept/image of the sports hero questionnaire. After completing and turning in this questionnaire, subjects were allowed to take a brief break before taking the role modeling questi(Hmaire, which was the less time consuming of the two survey instruments. Additionally, completed surveys were examined by the researcher to look for signs of subject fatigue as rq)resented by irregular or inconsistoit answer patterns in the latter

PAGE 82

75 parts of the questionnaires. Upon review of the completed questionnaires, subject fatigue does not appear to have impacted the answers given for this study. Upon completion of all questionnaires, the collected data for all 172 valid subjects were input into SPSS for statistical analysis. The procedures for data analysis are discussed in a later section of this chapter. Issues of Validity for this Study The following section addresses several types of validity, threats to validity and confounding variables, the relevance of different types of validity to this study, and the means by which this study attonpted to maximize validity. While several different types of validity and organizations of these concepts have been oiitlined by various theorists (Graziano & Raulin, 1993), the most frequently described types of validity and those most relevant to this study ate listed and outlined below. 1) Content Validity: "Contait validity refers to the extent to \n^ch a measure thoroughly and appropriately assesses the skills or characteristics it is intoided to measure" (Fink, 1991, p. 50). Content validity is maximized by assuring that the measurements used in the study actually address the ccmstructs in question. For this particular study, the constructs include self-concept and role modeling of sports heroes. Contoit validity is often maximized by adequately defining the variables in question and by using questioimaires that address the construct in question (Fink, 1991). This study has maximized content validity by using valid instruments designed to assess the particular constructs in question. For the questionnaire to measure role modeling, contoit validity was inoreased by

PAGE 83

fully defining role modeling and by the use of adolescent focus groups, who provided information as to which aspects of the b^avior of adolescents and role modeling were most appropriate to include. 2) Face Validity: Face validity refois to the sur&ce diaractedstics of a measure, and, unlike content validity, does not rely on established theofy (Fink, 1991). Face validity is increased by asking all appropriate questions and using language that is q)inx>piiate for die subject population. The face validity for this study was increased by pretesting questionnaires on adolescents similar to those in the study population and asking for appropriate feedback on the language of the questionnaires and li^ether any aspect of role modeling bdiaviors had been omitted. 3) Statistical Validity: Statistical validity refers to whetho" one's results are due to a systematic factor (the independoit variable) or merely to chance variations (Graziano & Raulin, 1993). This is particularly important in experimental research with a manipulation of the indepoident variable. Statistical validity has been increased for ^s study by using reliable measures, such as Harter's test of self-concept. Violation of the assumptions of a statistical test can decrease statistical validity. Because regression analysis was a primary statistical test used in this research, violations of the following assumptions could have decreased statistical validity: normal distribution, a linear relationship between depeodeoit and independoit variables, an assumption of homoscedastidty, and assumptions that measurements of Y values are independent of each other (Licht, 1996; Kleinbaum & Kupper, 1978). Statistical validity was increased by examining a scatter plot of responses before statistical analysis to assure that statistical procedures such as the potential use of cross products to account for interaction effects was not required.

PAGE 84

77 4) Concurrent Validity: Concurrent validity is maximized when a new measurement tool compares favorably with an existing measuronent tool that is already considered valid (Fink, 1991). Conciurent validity increases with high correlation between the new survey instrument aikl tiie existing measurement tool. For this research, the survey instrumoits used to assess self-concept and the image of the sports hero are adaptations of Harto-'s self-concqit test, and therefore new tests have not been oeated to replace existing, more valid measures. For the created measuremoit tool of role modeling intentions, no known existing measwement tool was foimd. Therefore, correlation between this new tool and an existing tool cannot be calculated, and concurrrait validity could iK>t be establidied for this measure. 5) Construct Validity: "Construct validity refers to how well the study's results support the theory or constructs behind the research and asks wheflio: the theory supported by the findings provides the best available theoretical explanation of the results" (Graziano & Raulin, 1993, p. 171). Therefore, a focal component is to eliminate rival hypotheses and other theoretical explanations for the research results. Threats to construct validity are reduced by clearly stating definitions and building hypotheses on well-validated constructs (Graziano & Raulin, 1993). For this study, threats to construct validity have been reduced tiuough building the hypoflieses on a sound theoretical foundaticm (ocmgruity theory) tiiat has been used successfully in marketing research. Construct validity also refers to the ability of a survey to distmguish between persons who have certain charactwistics and those who do not (Fink, 1 99 1 ). One means of estabhshing construct validity of a survey instrument is to check for discriminant validity with another measure. In other words, construct

PAGE 85

78 validity is inoeased by insuring that one measuremoit tool does not correlate highly with another measuremrat tool tiiat is used to measure a different, dissimilar construct. For this study, content validity has been establi^ed by diecking that real self-concept and idesi self-concqit do not have a high correlation (discriminant validity). A statistical compariscm of the collected data for real self-concept and ideal self-concept can be found in the results section of Has dissertation. 6) External Validity: External validity, often know as generalizability, refers to the extent to which the results of a study can be generalized to different populations and conditions (Graziano & Raulin, 1993). Hireats to extonal validity are reduced through careful selection of participants, including random sampling and attempting to choose a representative sample. However, for this study, subjects wceenot raadomly sheeted. Instead, this study used convenience samples, and therefore results must be geno'alized very tentatively, likely only to those who have characteristics similar to those in the stutfy. It is more likely that the results cannot truly be generalized at all to other populations (Graziano & Raiilin, 1 993). However, it should be noted fbat t}us study examined relationships between individual self-concqits, MiiE*t individual choice of a sports hero, and one's individual intentions to role model. Therefore, while this study cannot be truly genoalized to tbe adolescent male population at-large, the results present relationships that cm be studied and researched in other, broader adolescent populations. 7) Internal Validity: Intemal validity deals primarily wi& causality, particularly wheflier the independoit variable is iu;tually responsible for the observed changes in the dependent variable (Graziano & Raulin, 1993). For this study, most of the threats to intonal validity (maturation

PAGE 86

79 of the subjects, attritioii, ustramentation, etc.) did not affect this study. It is possible that the sequence of the administration of exams could potentially affect the answers given on the questioimaires. This is often controlled by using more than one order of administration of questionnaires (Graziano & Raulin, 1993). However, for this research, it was important that subjects used the questionnaire of real self-concept to anchor die other tests, and therefore one ordor of administration, with the real self-concept questionnaire completed first, was used. 8) Subject and Experimenter Effects: Subject effects can threaten the validity of a study when subjects give socially desirable answers en* unintentionally (or intentionally) try to satisfy ttie demands of the researcher through their answers (Graziano & Raulin, 1993). For this researdi, it is posable that partici|Kints ad^it vepbrt hi^aer or lower levels of behavioral intentions to model a sports hero depending on what they dean more socially desirable. Additionally, th^ may name a sports hero who may not actually be their sports hero, but rather a person ibgy feel would be a more socially desirable sports hero. These threats to validity have been reduced by assuring participants that there are no correct or incorrect answers and by limiting the questions of role modeling to iboac that were more likely to be answered honestly. Therefore, questions concerning anti-social behaviors such as domestic abuse and recreational drug use were not included. Experimenter effects could occur in this study if the experimenter tried to influence the answers of the subjects (Graziano & Raulin, 1993). This was reduced by having teachors administothe actual surveys and by closely monitoring the procedure. Because there are no observational data (as in experimental research), the threats to vahdity from experimenter effects are minimal in this study.

PAGE 87

so Data Analysis This section reviews the data analysis for this study and the statistical analysis of the four hypotheses of this study. All valid collected data were ii^)ut into SPSS for analysis. The topics to be discussed include the following: role modeling statistics, self-concept and calculation of distance scores, and analysis of stated hypotheses. All results of this data analysis and further explanations are found in the following "Results" chapter. Role Mode ling Data In order to further utilize the results of the role modeling questionnaire as the dependent variables for this study, reduction of the role modeling data was necessary. To accomplish this, a factor analysis of the 22 individual role modeling behavior variables (not including the general ovCTall role modeling variable) was conducted. Factor analysis, which identifies patterns among the variations in the values of large groups of variables, is effective in identifying patterns of interrelationships among these variables and, subsequeitly, reducing this large number of variables into a smaller number of variables known as factors (Agresti & Finlay, 1986; Rummel, 1970; Vogt, 1993). Seva-al rotations of this factor analysis were conducted to find the best fit for the role modeling variables used in this factor analysis. Rotations are typically done to identify factors that are more clearly named and distinct in their identity (SPSS, 1 999; Vogt, 1 993). In other words, by rotating this factor analysis, more distinct and specific areas of role modeling were idraitified as factors. For the variables of lole modeling for this study, a varimax rotation was used, and four significant factors of role modeling were identified. A significant factor was defined as having an eigenvalue greato* flian one (1) and at least three variables with significant factor loadings on that factor. For each of the significant factors of the factor analysis, an index was created. An index is a group of individual measures that, when combined, represent a more goieral

PAGE 88

81 characteristic (Vogt, 1993). For this study, each of the significant factors was used to create an index of role modeling bdiaviors. Indexes were created instead of merely using the factors as the more general variables of role modeling because factors place different weights on each variable in the factor and include components of o&na variables that are not strongly related to the targeted role modeling behavior of that particular factor. Each index, however, only includes the chosen variables that specifically relate to the core role modeling behavior of the more graieral index. An index was created for each factor by summing and then averaging across each subject's responses on the variables shown to load significantly on that factor. Significant variables were identified as those with a fasibor loading greats than .4 on that factor only. This was done for two reasons. First, this factor loading (.4) represoited a high level of correlation between the variables within an index. Second, upon examination of the significant Actors, this vaiae served as a logical break point for tilie selection of variables to each index. An overall index also was created from all variables that were included in each of the four indexes. Finally, two variid)les not included in any of the role modeling indexes (getting a tattoo or earring like a sports hero or taking Andro like a sports hero) were used in the test of the stated hypotheses. While these two variables did have a significant factor loading (greater than .4) on fiictors with eigenvalues greatathan one (1), both of these factors had fewer than tinee agnificant variables and therefore were not used to create indexes. These sevai variables (four role modeling indexes, one overall role modding index, and two indepoident role modelmg variables) were the depeadait variables for role modeling used in the four hypotheses of this study. Self-Concept and Tma gp of the Sports Hftro Data for real self-concept were collected using Harter's Self-Perception Profile for Adolescents (1988). The data for the ideal self-concept and the image of the sports hero WCTe collected using the altered forms of Harter's test, as described earlier in this

PAGE 89

82 chapto-. For each of these three constructs, a mean score was calculated in each of the eight domains of self-concept. Each of these 24 domains had five questions. Before a mean score could be calculated, the results of all questions asked in a negative direction had to be invoted sudi that the results fix>m each question ranged fix>m lowest rating of self-concept/image of sports hero (1) to highest rating (4). (See-thb J^|)peQiUx for questionnaires and directions for individual questions.) i": It should be noted that while these mean scfix^ bq^ saif-fExms^ and the image of the sports hero were reported in the results for informative purposes, these mean scores were not used in the calculation of the distance scores between self-concept and the image of the sports h&o. Instead, distance scores woe calculated fiom fhe individual questions, not the mean scores, to provide a more accurate determination of the specific distances between the constructs. An explanation of this calculation follows in the next section. Distance Scores Distance scores are a measure of the distance between two concepts, and they can be used to measure the similarity between two such concepts (Osgood et al,, 1957) For this study, distance scores were calculated to measure the distance between self-concept, both real and ideal, and the image of the sports hero. In other words, th^ distance scores indicate the similarity between self-concq)t (real and ideal) and tfie image of the sports hero. These two separate sets of distance scores, real distance and ideal distance, have been calculated for each of the eight domains. This resulted in 16 distance scores (8 real, 8 ideal) for use in this study. Several equations can be used to calculate distance scores. For this study, the difference squared model, which squares the dififCTence betwerai each paired set of questions and sums these differences, has been used (Sirgy, 1983; Osgood et al., 1957). The difference squared model assured a positive value for the distance betweai each set of values, regardless of which value, the self-concept or the image of Ae spoits hero, is

PAGE 90

83 greato:. This formula for calculating distance scores between the image of the sports hao and eitho" the real or the ideal self-concept in any of the eight domains can be represoited as follows: Distance (in each domain) = (Ql Sp Hero Ql SC)^ + (Q2, Sp hero Q2, SC)^ + (Q3, Sp hero Q3, SC)2 + (Q4, Sp hero Q4, SC)2 + (Q5, Sp hero Q5, SC)^. Qn, Sp Hero = Question n firom the test of the image of the sp>orts hero. Qn, SC = Question n from the test of self-concept This same formula was used in each domain for real distance and ideal distance. The sum of squared differences for each question was used instead of the siimmed difference of the overall construct scores as this provides a more accurate account of tiie descriptions between the sports hero image and the respective self-concept scores. In other words, this formula assures that differences on specific paired questions from each domain woe not lost or ignored when combined for an overall score for that domain (Osgood et al., 1957). The lower the distance score for each domain, the closer the particular domain of self-concept is to the image of the sports hero in that domain. Lower distance scores indicate closo* proximity between self-concept and sports hao image and a higher level of congruity between the two constructs. The higher the distance score, the less congruity that exists between self-concept and the image of the sports hero. These distance scores woe used as the independent variables for HI and H2 of this study. Hypothesi.s I and Hypothftsis TT HI and H2, which predict a negative relationship betwera role modeling and distance scores, were statistically examined with multiple regression analysis. Multiple

PAGE 91

1 84 regression is useful for the following types of analysis important to the testing of HI and H2: 1) Multiple regression analysis can "characterize the rdationship between the dependent and independent variables in the sense of detmnining the extent, direction, and strength of the association among these variables" (Licht, 1995, p. 34). 2) Multiple regression analysis can determine whidi of several variables are important for predicting a dependent variable, and which of these variables are tmimportant in predicting this dependent variable (Licht, 1995; Weisberg et al., 1989; Kleinbaum&Kupper, 1978). 3) Multiple regression analysis can detemune the best model for describing the relationships between various predictor variables and a dependent variable (Licht, 1995; Weisberg etal., 1989; Kleinbaum&Kupper, 1978; Vogt, 1993). For this study, regression analysis was used to detomine.^ relationships between the predictor variables (independent variables), distance scores, and the dependent variable, role modeling. The dependent and predictor variables were assumed to have a linear relationship, and an examination of the pretest data and actual data fiom this study confirmed this relationship. Therefore, linear regression analysis was used in this study. Although regression does not necessarily imply causality, regression can give a better understanding of the nature of a phenomenon by identifying factors with which it co-occurs (Licht, 1995; Weisberg et al., 1989). In other words, while this regression analysis cannot necessarily detomine whether congruence betweoi sdf-concept and flie sports hero causes role modeling of a sports hero, regression can determine whether

PAGE 92

85 congruoice co-occins with such role modeling, providing a better understanding of role modeling of sports heroes and the possible conditions undowhich it is likely to occur. For each of the seven role modeling variables (four indexes, one overall index, and two role modeling variables), three regression analyses woe completed. The first set of analysis used all eight real distance scores as the indqien^nl^predictor variables. The second set of analyses used all eight ideal distance scores as independent/predictor variables. The third set of analyses used all 16 distance scores (real and ideal) as indq)endent/predictor variables. One particular function of regression is to detomine tfie best model for describing a relationship between a dependent variable and independent variables (Licht, 1995). Therefore, this third set of regression analyses was done in attempt to create a more predictive regression equation for each of the role modeling indexes and variables than may be found in the regression analyses using only real distance scores or only ideal distance scores, and to account for a largop^catt^ of tilie variance of each of fliese dependent variables (in other words, to increase flie amount of variance explained for each role modeling index or variable). The exclusion of important predictor variables can drastically affect the results of regression analysis, so it was hoped that including all the distance scores (real and ideal) woidd increase the prospect of achieving statistically significant results (Licht, 1995). Finally, placing all 16 distance scores in one regression allows for a simple comparison of the impact of all distmce scores by examining the standardized beta weights of each of these 16 dependent variables (Kleinbaum & Kiq>per, 1978; Licht, 1995). One common fimction of regression analysis is to better understand which predictor vmiables Me more inqxntant than others, and this set of regression analyses has enabled that understanding. However, tiieorists do warn of using an excessively large number of predictcnvariables in a regression analysis, and, in fact, one potential goal of regression analysis is to find the best and most predictive model while minimizing the number of predictor variables (Kleinbaum & Kupper, 1978; Licht, 1995; Weisberg et al., 1989; Fink, 1995).

PAGE 93

86 This therefore presents the challenge of minimizing the number of {nredictor variables while still recognizing the importance of including important potential predictor variables, even those which do not make significant partial contributions to the dependent variable, as suggested by the theoretical firameworic of the study (Agresti & Finlay, 1986). For this study, while theory requires all distance scores to be included as potential predictor variables, even those that do not significantly contribute to role modeling, practice indicates that such a process will weaken the stability of these regression analyses. This inclusive set of regression analyses, using all 16 distance scores as predictor variables, does create difficulties in accurate assessment of the relationships between role modeling and distance scores. First, using all 16 distance scores does not tell the variance (R^) of role modeling accounted for by either the real distance scores or the ideal distance scores indq)endently, but rather gives the variance fi-om all distance scor^. Therefore, these regressions do not indicate whether real distance scores or ideal distance scores independently serve as significant predictors of role modeling behaviors. Because that question is a focus of this study, particularly in attempting to better understand congruity theory and the differing impact of real vs. ideal distance scores, these inclusive regression analyses do not fully answer all k^ questions raised in this study. Additionally, using all 16 distance scores inaeases the possibility for multicoUinearity among these independent variables. Multicollinearity refers to the intercorrelation between the independent/predictor variables in a regression analysis (Licht,1995; Kleinbaum&Kupper, 1978; Weisberg et al., 1989). As muliticolliniarity increases, regression analysis becomes less stable, and it becomes more difficult to assess the statistical significance of individual predictor variables whoi sudi variables are highly intercorrelated with othw predictor variables in the regression. For this study, high correlation could exist between several of the ideal and real distance scores, particulariy between the two scores from the same domain. Such correlation would result

PAGE 94

in hig^ multicollinearity, making the regression analyses with all 16 distance scores as predictor variables less stable than analyses using fewer predictor variables. For these reasons, the regression analyses using only the eight real distance scores (or only the eight ideal distance scores) as predicts variables are more crucial to the study of the first hypothesis. These less complex analyses likely give a more accurate and stable assessment of the impact of distance scores on role modeling behaviors. The seven regressions using all 16 distance scores serve to supplonent this assessment and add information for further discussion. Therefore, 21 separate regression analyses were completed to test HI and H2. A stq>wise regression procedure was used for these analyses. Stqiwise regressimi, unlike forward selection and backward elimination regressions, finds die best fit fin* a regression equation by entering the predictor variables in a variety of ordea:s (Licht, 1995; Vogt, 1993; iSPS'S, 1999). This type of regression drops predictor variables fix)m a r^ression model if they lose their significance when other variables are added to the regression equation, and this makes stepwise regression the most effective method when coirelation among the independent variables is strong (Agresti & Finlay, 1986; Weisbog et al., 1989; SPSS, 1999). Given the potential for strong correlation (or multicollinearity) between the distance scores used as the predictor variables for HI and H2, the stepwise method was the logical choice for the regression analyses far this study. The F value of each regression model was examined, and only those regressions withaplessthan.05 were deemed to support HI or H2. As sevoal regressi(ni& addressed each of the two hypotheses, an insignificant regression equatimi does isot reject either HI or H2, but rather indicates that a particular role modeling behavior does not have a significant relationship with the distance scores (real or ideal) used in that particular regression equation. For either HI or H2 to be folly rejected, none of flic regressions for either of the two hypotheses could be significant. By using several regression analyses for each hypothesis, this study identifies whidi of the particular role

PAGE 95

,88 modeling behaviors have significant relationships with distance scores, and whidi of the role modeling behaviors do not have such significant rel^cm^ps. The regressions with significant F statistics indicate which indexes/variables of role modeling support HI and H2. For each of the significant regressions, the t statistic, the dgnificance of this statistic, and the Beta statistic for each of the individual predictor/independent variables (distance scores) were examined. These statistics clarified vMch of tl^ distance scotcs had significant relationships with specific role modeling indexesArarii^les md the relative strength of these relationships. To determine significance of each of the predictor variables, the test statistic (t) was used. For any iredictor variable to be deoned to have a significant relationship with a role modeling index/variable, the p vahie of the test statistic must be less than .05 (p less than .05). The beta weights (B), or standardized regression coefiGcioits, of each significant relationship were examined to explain tb.c strength of the relationship between individual distance scores and role modeling indexes/variables. This statistic not only explained the extent to which any of the individual distance scores related to role modeling, but it also explained wbidi of die distance scores had the strongest relative relationships with role modeling. An examination of the results of the 2 1 completed regression analyses provide an explanation of HI and H2 and selective support (or lade of support) for these hypodieses, which predict significant relationships between role modeling and distance scores. Domain Importance and r)i.stance Wnrtb SnnrRg Domain importance for each of the eigiht domains of sdf-concq>t was odculated fi-om Harter's test of domain importance as part the Self-Perception Profile for Adolescents (19SS). (See Aiq)aidix A for the actual test questions.) Two questions addressed each of the d^t domains of self-concept, and a mean score was calculi^ for each domain. Domain importance was measured on a scale from one (1) (least important) to four (4) (most important). These domain importance scores were used, along wifli the

PAGE 96

89 calculated distance scores used for HI and H2, to create distance worQi scores. These scores were used as the predictor/independent variables for H3 and H4. Before the calculation of this new variable can adequately be described, it is necessary to explain fiilly the proposed relationship betweoi domain importance scores and role modeling of a sports hero. As a domain of self-concept becomes more important (domain importance scores increase), one should be more likely to role model bdiaviors of one's ha:o, particulariy in behaviors that are related to that domain of self-concept. Additionally, higher domain importance should increase the likelihood that one considers the congruity between one's self-concept and the image of one's sports hero in influencing role modeling. TherefOTC, it is predicted that as domain importance scores increase, role modeling of the sports hot) also will increase. In other words, domain importance and role modeling should be positively correlated. The following examples illustrate this predicted relationship. Two hypothetical persons, Frank and Mark, are used for illustration purposes in these examples. Frank and Mark each has his own individual sports hero. Frank places high importance on his athletic self-concept. Therefore, his domain importance score for the athletic self-concept is high (for this example, Frank's domain importance score is four). Mark places less importance on his athletic self-concq)t (domain importance score is two). According to the predicted relationship for this research, Frank should be more likely to role model his sports hero than Mark, especially for behaviors that are related to the athletic self-concept. Additionally, Frank's self-concept congruity witfi his sports hero in the athletic domain should be more likely to influence his decision to role model his hero. Therefore, as domain importance scores ino-ease, role modeling also should increase. For ease of further analysis and to make distance worth scores smalls and more easily read, the domain importance scores were reduced by dividing the scores from each domain by four (4). The new, reduced domain importance scores ranged from .25 (least

PAGE 97

90 important) to one (1) (most important). In other words, a domain importance score of"!" became a score of ".25", while a score of "4" became a score of"!". These reduced scores represent the same values and ratios as the original scores, and this data alteration did not change the nature of this construct nor the relative value of domain importance from one domain to any of the other seven domains. These reduced distance scores (d^t real, eight ideal) were used in the creation of the 16 distance worth scores used as indq)aident variables for H3 and H4. EKstance worth scores incorporate both the distance sccs used as the independent variables for HI and H2 and domain importance scores. Tho-efore, this variable represents both (a) the congruity betwerai individual self-concept and the image of a sports hero in a domain and (b) the importance of that domain to an individual. However, the statistical calculation of this new variable requires specific data alterations due to the nature of distance scores and domain importance scores. As predicted for this research, distance scores should have a negative correlation with role modeling. This is because increased congruity between self-concept and the image of the sports hero in a domain results in decreased distance scores for that domain. Convasely, it has been predicted that domain importance scores have a positive relationship with role modeling. ThCTefore, to create distance worth scores, it is necessary to make statistical alterations to one of the two variables sudi that they both have a relation^p witil role modeling miSm same direction. In other words, because distance worth has been created as a variable which combines congruity and domain importance to inedicst role modeling, Hae relation^ps of congruity (represented by distance scores) aHd domain importance with the dependent variable role modeling must be in the same direction for the creation of this new variable. The creation of distance worth scores requires distance scores to be multiplied by domain importance scores. However, to satisfy the above conditions, domain importance scores (the reduced domain importance scores have beoi used) have been inverted before

PAGE 98

91 being multiplied. For this calculation (see fonnula below), each of the distance scores was multiplied by Ae inverse of the reduced domain importance score for that particular domain. In other words, distance scores were actually divided by domain importance scores to create distance worth scores. This was done 1 6 times, resulting in etgjit real distance worth scores and ei^t ideal distance worth scores. (Each rechiced domain importance score was used two times ~ once for the real distance worth score for that domain, and once for the ideal distance score). These 16 distance wrarth scoies (d|^t real, eight ideal) were used as tiie predictor/independent variables Ibr ID and H4. To calculate distance worth scores, the following fonnula was used: Distance Worth = Distance Score x (1 / Reduced Domain Importance Score) According to this formula, higher distance scores result in higher distance worth scores, while lower distance scores result in lower distance worth scores. In this respect, distance worth scores are identical to distance scores in their proposed correlation with role modeling behaviors. Convwsely, due to the inversion of domain importance scores, higher domain importance scores result in lower distance worth scores, while lower domain importance scores result in higher distance worth scotcs. As a domain is rated more important (domain importance score increases), the distance wortii score for that domain will decrease for both real and ideal distance worth. Therefore, this formula has allowed for the negative correlation between role modeling and distance scores and the positive correlation between role modeling and domain importance scores to be represented in one variable (distance worth). Due to the reduction and subsequent inversion of domain importance scores, virtually all distance worth scores will be greater than thencorresponding distance scores. The greater the domain importance score, the less the distance worth score will increase from its corresponding distance score. The lower the domain importance score

PAGE 99

92 (less importance for a domain), the more Hie distance worfli scores will inoease from llie distance score. Therefore, higher domain importance scores result in lower rdative distance worth scores. In other words, if two distance scores are equal but the domain importance scores for tibat domain are differmt, the lower xdAtivedistance worth score will be tile one with the higho* domain importance score. The following hypothetical examples of the creation of two distance worth scores further explain this concept. Example of the Creati on of Distance Worth Scores In Example A, Frank and John are two hypothetical subjects. Both subjects have idoitical real distance scores in the athletic domain (distance sccse = 5). In other words, both Frank and John have the same congruity between their real athletic self-concept and their athletic image of their individual sports hero. However, Frank has a higher domain importance score than John for the athletic domain. Frank's domain imptntance score is three (3), while John's domain importance score is two (2). In other words, Frank places greater value on the athletic domain of self-concept than John does. Using the equation presented in the previous section, distance worth scores have been calculated for the two hypothetical subjects, and these scores are represented in the following table. Table 3-1 Example A of the Calculation of Distance Worth Scores. Keal Athletic Athletic Uomam Reduced Uomam Real Athletic DM Subject Distance Score Importance Score Inqwrtance Score Worth Score Fnmk 5 3 0.75 6.67 John 5 2 0.5 10 In this example, while both subjects had identical distance scenes, the subject who placed greater importance on the athletic domain (Frank) had a lower relative distance worth score for the athletic domain. John placed less value on this domain and had a lower domain importance score, so his resulting distance worth score was higher.

PAGE 100

93 Because H3 and H4 predict that distance wortti scores will have a negative correlation with role modeling, Ons research predicts that Frank would role model his sports hm> more than John, who had equal congruity yet a lower domain importance score. Therefore, distance worth scores incorporate both the importance of a domain and die congruity of that domain in an effort to more accurately predict and assess role modeling of the sports hero. It should be noted that, as explained in the previous section, both Frank's and John's distance worth scores are greater than their corresponding distance scores. This is due to the fact that domain importance scores are reduced before being inverted in the process of calculation. However, this numerical increase fix>m distance scores to distance worth scores does not afTect the statistical analysis of the hypotheses for this research. Instead, it is the relative value between distance worth scores, not the relative value between distance worth scores and corresponding distance scores, that guides the analyses of H3 and H4. In Example B, real distance worth scores are again calculated for Frank and John, this time for the social domain. In this example, the subjects have identical domain importance scores for the social domain (domain importance score = 3), but Frank has a lower real social distance score (distance score = 4) than John (distance score = 7). In other words, while both subjects place equal importance on the social domain, Frank has greater real self-concept congraity with the unage of his sp(ts hoo in tiiis dooEHdniluBi does John. The following table details the creation of distance worth scores for this scenario. Table 3-2. Example B of the Calculation of Distance Worth Scores. Keal Social Distance Social Domain Reduced Domam Keal Social JJist Subject Score Iiiqx)rtaiice Score Iiqportance Score Worth Score Frank 4 3 0.75 5.33 John 7 3 0.75 9.33

PAGE 101

94 Aldiough the domain impratanoe sebies are identical, the difforaices in die distance scores result in differences in the calculated distance worth scores. Again, bodi distance worth scores are greater than their corresponding distance scores. While Example A showed how dififeroices in domain importaoce le^ to differaices in distance worth somes, Utas example shows how distance worth scmts reflect diffooices in subjects' distance scores, or congruity with one's sports hero. This calculation of distance worth scores has allowed inr two c(KXpts, (a) tiie distance between self-concept and ihe image of the sports hero in a domain and (b) the importance of that domain to be statistically represented in one variable. This variable, distance worth (for both real and ideal self-concq>t), has a {seffiiBted zgative rdationship with role modeling, and this was examined through statistical analysis of hypothesis III and hypothesis IV. Hypothesis in and Hypothesis TV H3 and H4, which predict a negative relationship between role modeling and distance worth scores, were statistically examined Avith multiple r^ression anafysis. Tins process of analysis was identical to that used to test HI and H2, except that die dktanoe scores used for HI and H2 were replaced with distance worth scores. Twaity-one separate regression analyses were completed. Three regre^
PAGE 102

95 either H3 or H4, but ratho: indicates a particular role modeling behavior does not have a significant relationship with the real or ideal distance worth scores used in that particular regression equation. The regressions with significant F statistics indicate which indexes/variables of role modeling had significant relationships with distance worth scores, and th^fore which such indexes/variables would ^qipcxrt H3 and H4. For each of the significant regressions, the t statistic and the Beta statistic for each of the individual predictor/independrait variables (distance worth scores) were examined. To determine significance of each of the predictor variables, the test statistic (t) was used. For any predictor variable to have a significant relationship with a role modeling index/variable, the p value of die test statistic must be less than .05 (p less than .05). The beta weights (B), or standardized regression coefficients, of each significant relaticm^p were examined to explain the strength of the relationship between individual distance worth scores and role modeling indexes/variables. This statistic not only explained the extent to which any of the individual distance worth scores related to role modeling, but it also explained which of die distance worth scores had the stiongest relative relationships with role modeling. An examination of the results of the 21 completed regression analyses provided explanation of H3 and H4 and displayed selective support (or lack of support) of tiiese hypotheses, which predict significant relationships between role modeling and distance worth scores. The following chapter. Results, details the results of the described analyas. These results address the four stated hypotheses and the potential support for these hypothesis based on flie collected data for this study.

PAGE 103

CHAPTER 4 RESULTS AND DISCUSSION This cluster details the results of administered surveys and ensuing data analysis. In addition to a statistical review of the four stated hypotheses, fbs ch^toincludes a statistical review of the collected descriptive statistics of the study. These statistics were used as a foundation for analysis of the hypotheses. The following areas are addressed: 1) A detailed list of the athletes chosen as sports heroes, the sports played by these athletes, and the frequency of selection of these aretes. 2) Results of the participants' responses to the questionnaires on role modeling and intentions to role model sports heroes' behaviors; factor analysis of the results of the role modeling questionnaires; and the creation of role modeling indexes, which were used as the dependent variahles of role modeling for tiie tests of the hypotheses. 3) Results of the tests of self-concept, both real and ideal, and the test of the image of the sports hero. These resiilts are reported by the eight domains of self-concept 4) Results of &e calculated distance scores. These re^ts include botii real distance scores (distance between real self-concept and the image of the sports hero) and ideal distance scores (distance between ideal self-concq>t and the image of the sports hero), rqwrted for each of the eig^t domains of self-concept. These distance scores were used as the independent variables for Hypothesis I and Hypothesis II. 96

PAGE 104

97 5) Results of the 21 regression analyses completed to test HI and H2, predicting a negative relationship between the dependent variable (role modeling) and the independent variables (real and ideal distance scores). These results include detailed descriptions of which particular distance scores had negative relationships with lole modeling and for which of the pa^eulor rdie^Be^ing indexes/variables these significant relationships exist. In other words, the discussion desoibes the particular areas of role modeling and the particular distance scores that bore significant negative relationships to each otiior. 6) Results of the domain importance questionnaire, reported for each of the eight domains of self-concept, and the data reduction conducted on the results of these questionnaires. 7) Analysis of the creation of distance worth scores, both real and ideal, which incorporate both the distance scores and the corresponding domain importance score for that domain. These distance scores are reported for each of the eight domains of self-concept, and they were used as the indepoident variables for Hypothesis III and Hypothesis IV, 8) Results of the 2 1 regression analyses completed to test H3 and H4, predicting a negative relationship between the dependent variable (role iodeling) and the independent variables (real and ideal distance worth scores). These results woe reported in the same fashion as was done for HI and H2. 7) Summary of the results of the hypotheses. Descriptive Statistics Much of the data collected for this study could not be used to test tilie hypotheses for this study until undergoing statistical transformations. However, this data provides important information about the sports hm>es chosen in this study and the actual role

PAGE 105

98 modeling intentions of the subjects in this study. Specifically, this section will detail the aretes chosoi as sports heroes for this study, the results of the role modeling questionnaires, and the creation of role modeling indexes through factor analysis. Chosen Sports Heroes The actual athletes chosm as sports heroes in this study and the feeqamay widi which they were chosen are detailed in the following table (Table 4-1). All athletes who were selected by more than one subject are named in the table. The data for athletes selected only once are summarized at the bottom of tibe table. Also included are the sports played by the selected sports heroes. It should be remembered that each subject could only choose one athlete. Further, this athlete had to be well known through television and other mass media outlets, eliminating local athletes or athletes only known through personal acquaintance.

PAGE 106

99 Table 4-1. Athletes Chosen as Sports Hero, Listed in Descending Order by Frequency. Frequency \j c^-fr^ Ant r crccni l^UIllUlltllVC X ClL^CUt Micti^LCl Jordan (basketball) 9.3 Craig Biggio (baseball) 6 3.5 12.8 Maik M cOuiiv (baseball) 6 3.5 16.3 Ken Griffey Jr (baseball) 6 3.5 19.8 Jeff Bagwell (baseball) 5 2.9 22.7 Steve Austin (pro wrestling) 4 2.3 25.0 Bill Goldberg (pro wrestling) 4 2.3 27.3 Ken Caminetti (baseball) 4 2.3 29.7 Ricky Williams (football) 4 2.3 32.0 Tiger Woods (golf) 3 1.7 33.7 Barry Sanders (football) 3 1.7 35.5 Jason Williams (badEcfiMitt) 3 1.7 37.2 Charles Barkley (basketball) 3 1.7 39.0 Mike Tyson (boxing) 3 1.7 40.7 Sammv Sosa fbaseball^ 3 1.7 42.4 Piste Smmras/tmiieiis)' 3 1.7 44.2 Allen Tverson fbasketbain 2 1.2 45.3 Tonv Hawk rextretne snorts^ X X XiX VY Iv \ U will Vi' OL/V/l VO 1 2 1.2 46.5 Cal Rinkiti Jr Thasehain 2 1.2 47.7 Tim Duncan (baske0>all) 2 1.2 48.8 Patrick Roy (hockey) 2 1.2 50.0 Greg Maddux (baseball) 2 1.2 51.2 Jerry Rice (tootball) 2 1.2 52.3 jurgen jumsmaim ^soccery Michael Owen (soccer) 2 1.2 54.7 Cobi Jones (soccer) 2 1.2 55.8 Kevin Gamett (basketball) 2 1.2 57.0 Jdm Elway (football) 2 1.2 58.1 Emmitt Smidi (food>all) 2 1.2 59.3 70 Athletes wifli 1 Vote 70 40.7 100.0 Each; Total 172 100.0 Athletes with 1 vote: 23 baseball, 15 basketball, 8 football, 4 hockej soccer, 4 track and field, 3 golf, 8 other ^KHts.

PAGE 107

Ninety-nine different athletes were chosen as sports heroes for the 172 subjects, demonstrating a divCTsity of selections. Of the athletes, Michael Jordan was selected by the most subjects by an overwhelming margin, with 16 subjects choosing the basketball star. This is not surprising given his athletic excellence, extremely heavy media covonge and mariceting associations, and tiie generally positive conoqstiicm of Jordan both on and off the court. Broken down by sport, however, baseball players were chosen by the largest number of subjects (57), followed by basketball (43) and football playos (21). Because many of the subjects for this study were participants in a baseball tournament, the large number of subjects selecting baseball players is not surprising. These three major sports had far more sports heroes selected than all olim S|X>rts in this study. A distant fourth in number of hwoes chosen was professional wrestling, with eight subjects identifying wrestlers as their sports hero. Given the high television ratings for professional wrestling and the emphasis placed on the charactar development of wresdo^ throughout the staged productions, it is not surprising that this "fake" sport had more hero selections than established sports such as hockey, tennis, track and field, and golf This is true despite the fact that many subjects participated in these ^rts at school. Howevo', these sports do not receive the same media coverage as the three major sports in America (football, baseball, and basketball), which accounted for the majority of athletes chosoi in this study, providing support for the contention that televisiim mi oiher mass media are instruments in providing hero choices for adolescents. Only one of the 172 male subjects chose a fonale sports hero, soccer player Mia Hamm, stressing tiie iiBftoitance of po-cdved similarity in the selection of a sports hero. Addttioaally, bmmse male athletics typically receives far greater media coverage than female athletics, these subjects likely have had much more exposure to star male athletes than their fonale counterparts. This also may have influenced tiie dominant selection of male sports heroes by subjects in this study.

PAGE 108

101 Although it may be difficult to find commonalties in the hoo selections, virtually all athletes chosen currently are or once were stars in their sports. A select few may not have been stars, but did overcome overwhelming odds to achieve success. One such example is Rudy Rudiger (selected by one subject), an undersized fonner Notre Dame football player who achieved his dream of playing fbolban for Notre Dame despite his obvious lack of physical talent and size. Additionally, most of the identified athletes have displayed positive behaviors on and off the court throughout ibdr careos. There were a few notable exceptions, such as basketball player Chiiates Baikley (chosen by three subjects) and boxer Mike Tyson (chosen by three subjects), both of whom have di^layed negative behaviors and encoun