Title: Effects of sex and appearance on ratings of source credibility
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00102733/00001
 Material Information
Title: Effects of sex and appearance on ratings of source credibility
Physical Description: Book
Language: English
Creator: Engstrom, Erika, 1964-
Copyright Date: 1991
 Record Information
Bibliographic ID: UF00102733
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 25526605
ltuf - AJC1545

Full Text













EFFECTS OF SEX AND APPEARANCE
ON RATINGS OF SOURCE CREDIBILITY













By


ERIKA ENGSTROM


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


AUGUST 1991































For my parents














ACKNOWLEDGEMENTS


Many, many people helped me with this paper. I

sincerely thank each of them for giving me their time and

encouragement. First, I thank everyone who helped with

the production of the videotape used in this study. They

are: Drs. Julie Dodd, Paul Smeyak, John Wright, Kim

Walsh-Childers, Michael Weigold and Sherry Alexander;

fellow grad students and colleagues Matthew Power, Jane

Tolbert, Matthew Bunker, Darci Sirianni, Lisa Barr, and

Clay Conway; WUFT-TV production manager Frank Counts; and

WUFT-TV news director Rick Schneider, who so graciously

endured countless hours of taping and retakes.

I could not have conducted this study without the

following people: Ed Gilliland, Marge Pierce, Susan

Brown, Dr. Greg Neimeyer, Mark Mahon, Sunil Hewavitharana,

and my brother Bruce Engstrom, who all helped in one way

or another with the administration of the experiment; and

Simon DeYoung and Blake Milstead, who loaned me their

brand new video equipment without requiring collateral or

reimbursement other than St. Louis mineral water. I

especially thank my fellow doctoral inmate Barry

Hollander, who helped me practically every step of the


iii










way, not only with the experiment itself, but also with

the statistical analyses of the results.

I thank my committee members for their suggestions,

advice, and open doors: Drs. Leonard Tipton, John Wright,

Constance Shehan, and Jaber Gubrium.

Special thanks go to Dr. Julie Dodd, whose

understanding and support in so many ways throughout the

writing of this paper I appreciate with all my heart.

Very special thanks go to my committee chairperson,

Dr. Mickie N. Edwardson, for all the time, patience, and

wisdom she has so generously given me. She is, quite

simply, the best.

Finally, I thank my stockholders, Alex and Margaret

Engstrom. I hope I can make good on their investment.















TABLE OF CONTENTS


page

ACKNOWLEDGEMENTS . . . . . . . .. iii

LIST OF TABLES . . . . . . . . .. vii

ABSTRACT. . . . . . . . . . .. .viii

CHAPTERS

1 INTRODUCTION. . . . . . . . . 1

2 REVIEW OF LITERATURE. . . . . . 5

Nonverbal Aspects of Source Credibility . 5
Appearance and Social Judgments . . . 9
Appearance and Occupational Stereotypes . 16

3 HYPOTHESES AND RESEARCH QUESTIONS . . .. 21

Hypotheses . . . . . . . .. 21
Research Questions . . . . . .. 24

4 METHOD . . . . . . . . .. 28

Subjects. . . . . . . . . ... 28
Design . . . . . . . . .. 28
Materials . . . . . . . . . 31
Procedure . . . . . . . . .34
Pretest 1: Topic Interest . . . .. 34
Pretest 2: Topic Similarity . . .. 35
Pretest 3: Dress Conditions . . .. 37
Final Experiment . . . . . . 40
Instrument Validity and Reliability . .. 41
Factor Analysis . . . . . .. 42
Reliability . . . . . . .. 44
Analysis. . . . . . . . . ... 45

5 RESULTS . . . . . . . . . 50

Description of Subjects . . . . .. 50
Results for Hypotheses . . . . ... 52
Hypothesis One . . . . . . .. 52
Hypothesis Two . . . . . . .. 54
v










Hypothesis Three . . . . . .. 55
Summary of Hypothesis Findings . . .. 58
Results for Research Questions . . ... 59
Male Targets, Regardless of Dress. ... . 59
Female Targets, Regardless of Dress . 61
Conservatively Dressed Female Targets. . 63
Casually Dressed Female Targets. .. . 66
Conservatively Dressed Male Targets . 68
Casually Dressed Male Targets. . . ... 71
Summary of Research Question Findings . 71
Credibility of Dummy Sources ..... . . . 74
Male vs. Female Dummy Sources .. .. .. 75
Female Dummy Sources . . . . .. 75
Male Dummy Sources . . . . . .. 76
Summary of Dummy Source Findings ... 78
Additional Analyses . . . . . .. 80

6 DISCUSSION. . . . . . . . .. 84

General Implications . . . . . .. 85
Limitations and Considerations. . . ... 93
Internal Validity. . . . . . .. 94
External Validity. . . . .. . 95
Suggestions for Further Research. .. . . 96

APPENDICES

A QUESTIONNAIRE FOR PRETEST 1: TOPIC INTEREST 101

B QUESTIONNAIRE FOR PRETEST 2: TARGET SIMILARITY 118

C QUESTIONNAIRE FOR PRETEST 3: DRESS CONDITIONS 126

D QUESTIONNAIRE FOR FINAL EXPERIMENT. . ... 132

E DEMOGRAPHIC INFORMATION . . . . .. 141

REFERENCES. . . . . .. . . . . . 142

BIOGRAPHICAL SKETCH . . . . . . ... 148















LIST OF TABLES


page

Table 3-1. Order of Sources in Experimental
Conditions . . . . .... . . . 32

Table 3-2. Example of Three-Factor Solution
of Credibility Variables ... . . . . 46

Table 3-3. Example of Four-Factor Solution
of Credibility Variables . . . .. . 47

Table 4-1. Mean Credibility Dimension Scores
for Male and Female Targets, Regardless of
Dress Condition. . . .... . . . 55

Table 4-2. Mean Difference Scores for Conservative
and Casual Dress Conditions of Male and Female
Target Sources .. . . . .. . . 56

Table 4-3. Mean Credibility Dimension Scores
for Male and Female Targets by Dress Condition 58

Table 4-4. Mean Credibility Dimension Scores
for Male and Female Targets Regardless of
Dress Condition by Subject Age and Sex . .. 64

Table 4-5. Mean Credibility Dimension Scores
for Female Targets According to Dress Condition
by Subject Age and Sex . . . . . .. 69

Table 4-6. Mean Credibility Dimension Scores
for Male Targets According to Dress Condition
by Subject Age and Sex . . . . . ... .73

Table 4-7. Mean Credibility Dimension Scores
for Male and Female Dummy Sources .. . . 77

Table 4-8. Mean Credibility Dimension Scores
for Male and Female Dummy Sources by Subject Age
and Sex . . . . . . . . . . 81


vii















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

EFFECTS OF SEX AND APPEARANCE
ON RATINGS OF SOURCE CREDIBILITY

By

Erika Engstrom

August 1991

Chairperson: Dr. Mickie N. Edwardson
Major Department: Journalism and Communications

This study focused on the effects that a person's sex

and appearance have on ratings of his or her credibility.

Specifically, it is an examination of differences in

ratings of expertise, character, and dynamism of male and

female sources who appeared as experts in simulated

television interviews, and whether variations in clothing

affect perceptions of credibility of males and females

differently.

A stimulus videotape was created in which male and

female actors posing as college professors appeared in

three modes of dress: conservative, casual, and neutral.

Subjects drawn from a sample of college students and from

a civic organization comprised mostly of adults over age

35 viewed the interviews and rated the sources according

to characteristics which measured source expertise,

character, and dynamism.
viii











Contrary to expectations, no significant differences

were found between expertise and character ratings of male

and female sources. However, male sources were rated as

more dynamic. Credibility ratings of female sources were

expected to be significantly lower when they were dressed

casually than when dressed conservatively. No significant

differences in ratings between the two dress conditions

for females were found. It was also predicted that

subjects would perceive smaller differences in credibility

between male sources dressed casually and conservatively

than between females sources dressed casually and

conservatively. This was not supported. Results showed

that male sources dressed conservatively were regarded as

lower in character than when dressed casually.

Subject variables of age and sex were also

investigated in terms of their influence on source

ratings. Younger female subjects tended to give both male

and female sources higher credibility ratings than did

younger males, older males, and older females. Older

females tended to give the lowest credibility ratings to

all sources. Most significant differences among subjects

were found between older and younger women. Older men and

younger men did not significantly differ in their ratings

of sources.















CHAPTER 1
INTRODUCTION


The importance of one's appearance in the

establishment of one's identity and credibility as a

source of information can be traced to ancient Greece.

The concept of ethos, the totality of characteristic

traits of a speaker, includes not only those aspects of

contemporary concepts of credibility, such as competence

and character, but also physical appearance (Aristotle,

1932).

The purpose of this paper is to investigate the link

between perceptions of a person's competence in terms of

his or her qualifications as an expert source and that

person's appearance. Specifically, the study will focus

on the effects that variations of a communicator's

appearance have on evaluative opinions by an audience.

The present study is an attempt to explore the subject of

credibility in terms of how different people, in terms of

age and sex, view men and women who are presented as

"experts." Such sources often appear as guests or news

sources on television programming, such as network news

programs and magazine talk shows. Often, panels of

experts may appear on these programs. Experts of this

1











type come from a variety of backgrounds, especially the

professions. A common type of expert is the college

professor, whose specialty is studying the particular

issue at hand. Also, attorneys, medical personnel, such

as medical doctors and psychologists, and government

spokespersons, may appear. These sources may serve as

stereotypes for their professions. It is the intent of

this study to examine the extent to which variations in

the appearance of such sources, both male and female, may

result in different evaluations of the perceived expertise

of these types of sources.

Of particular interest will be how gender of an

expert source is related to different evaluative opinions

by perceivers.1 Gender and credibility have been studied

in terms of societal perceptions of women's competence as

professionals (Lott, 1985) and trust between the genders,

both interpersonal and in general (Carocci, 1988).

Lott (1985) reviewed gender studies in which

comparisons of how people evaluated male and female target

persons on a variety of personal traits, including

competence, were investigated. In general, studies

examining such gender differences support the notion that

there is a tendency to devalue competent women, even in

situations in which both men and women targets perform

equally well on a given task. Thus, she concluded, equal












performance by men and women does not necessarily yield

equal evaluations.

Carocci's (1988) study of subjects' trust of same-sex

and opposite-sex members as friends and as people in

general found that, overall, women's ratings of both male

and female others' expertness, character, and dynamism--

all components of credibility--were higher than men's

ratings. Also, men rated other men as higher in

expertness than they rated other females, while, contrary

to her expectations, women did not rate other women as

significantly higher in expertness than other men.

Attitude objects in that study were either subjects'

"close" friends (male or female) or males and females "in

general" (p. 73). This may indicate that females, at

least in Carocci's sample, viewed men and women as equally

competent, while males may still be biased against women

in terms of their competence.

It is hoped that this investigation of the

interrelationship between gender and appearance of a

source and the gender of the receivers of that source's

message will lead to a better understanding of appearance

as a component of communicator credibility. The

Aristotelian concept of credibility, ethos, is composed of

a speaker's good sense, good moral character, and good

will. According to this concept, good will includes those











characteristics about a speaker which do not have to do

with his morals but nevertheless win approval from

observers. Among these traits is appearance: a speaker

who is "clean and neat in person" and in dress will

enhance the good will others may see in him (Aristotle,

1932, p. 104). This study is intended to provide

empirical evidence for the influence appearance may have

on an audience's regard for a communicator.

Note

1As Unger (1979) points out, the term "sex" implies
biological mechanisms, such as chromosomes, genes,
hormones, and other physical characteristics which serve
as the foundation for social distinctions between males
and females. The term "gender," on the other hand, is
more concerned with the nonphysical, sociocultural
characteristics and behaviors which are considered
appropriate for males and females, such as those which
denote masculinity and femininity. While the researcher
recognizes the different meanings implied by both terms,
"sex" and "gender" will be used interchangeably in this
paper to refer to the obvious physical distinctions
between men and women rather than behavioral traits. The
sex of sources and subjects will be used as the basis for
the independent experimental and subject variables in this
study.















CHAPTER 2
REVIEW OF LITERATURE


The present investigation draws upon several

theoretical perspectives, including source credibility,

nonverbal communication, the symbolic interaction theory

of communication, the social cognition of stereotypes, and

the sociology of gender differences. The following is a

review of relevant findings which provide the bases for

the main research question posed: How does the perceived

credibility of male and female experts differ as a

function of differences in the appearance of such sources,

and to what extent do differences in receivers' age and

sex affect their ratings of those experts?

Nonverbal Aspects of Source Credibility

One of the most well-documented and widely accepted

generalizations in persuasion research is that the

credibility of a source influences the effectiveness of

persuasive messages (Miller, 1987). Early studies of

credibility involved experiments in which print messages

were used. Manipulation focused mainly on the pre-

established trustworthiness (defined as credibility) of

familiar or fictional organizational and individual

sources and how it affected intended attitude change

5











(Hovland & Weiss, 1951). In the time since then, two of

the main areas of research into the processes and elements

associated with the construct of source credibility have

been the isolation of the dimensions of a source's

perceived credibility and how manipulation of certain non-

and paraverbal source characteristics, such as speaking

style and appearance, affect an audience's perceptions of

those dimensions.

Two broad dimensions which emerge consistently from

factor-analytic research of source credibility are the

concepts of "competence," that is, a speaker's perceived

expertise and regard as an authority on a particular

subject, and "trustworthiness" (or "character"), that

component which embodies a source's personal integrity and

likelihood of telling the truth about a subject (O'Keefe,

1990; McCroskey & Young, 1981). Other dimensions which

have been found to comprise overall credibility are

dynamism, that is, a speaker's assertiveness or energy

(Berlo, Lemert, & Mertz, 1969), and his or her sociability

or friendliness (Miller, 1987).

Of more importance to the present study, however, is

the other approach--studying credibility formation: the

"process by which communicators use symbolic inducements

to persuade others to perceive them more positively"

(Miller, 1987, p. 467). The purpose for this line of











research is to discover which characteristics about a

source and/or his or her message can be manipulated so as

to affect how a person regards that source in terms of

various credibility dimensions. These characteristics

include information about the source (such as education

level and experience), liking for the source (including

similarity), position advocated by the source, use of

humor in the message, citation of evidence, and topic

sequencing (O'Keefe, 1990; Miller, 1987).

A more global construct with which to study how

characteristics of a source may influence perceptions of

credibility is that of communicator style. Included in

this area of research are paraverbal elements such as

speaking rate, articulation, and speech errors or

nonfluencies. In general, studies of this type have found

that faulty articulation and nonfluent speech tend to

result in lower ratings of credibility, especially on

dimensions such as competence and dynamism (Addington,

1971; Burgoon, 1978; Miller & Hewgill, 1964; Sereno &

Hawkins, 1967). Less consistent findings have been found

concerning the effects of speaking rate on perceptions of

sources (Burgoon, 1978; Miller, Maruyama, Beaber, &

Valone, 1976; Bowers, 1978).

Nonverbal characteristics of sources have been found

to be related to how people regard sources in terms of











various credibility dimensions. Studies of this type

typically compare ratings of sources who are judged to be

physically attractive in terms of facial features with

those of sources judged to be less physically attractive.

Generally, it has been found that physically attractive

communicators receive higher ratings of likability than

less attractive communicators (Joseph, 1982; Patzer, 1983)

and rate higher on expertise and trustworthiness (Patzer,

1983).

While the relationship between somatic features such

as facial appearance and perceived credibility has been

for the most part supported (O'Keefe, 1990; Joseph, 1982),

other physical attributes of a person also may play a role

in how he or she is viewed by others. Perhaps the most

obvious of these is clothing. One's personal artifacts,

such as clothing and insignia, contribute to perceptions

of social power (Emmert & Donaghy, 1981). Malandro and

Barker (1983) acknowledge the importance of clothing as a

nonverbal cue to impression formation by pointing out that

the extent to which we consider another person to be

credible may be in part revealed by his or her clothing.

Experimental evidence supporting the notion that

clothing can serve as a cue to perceptions of credibility

is provided by Harp, Harp, and Stretch's (1985) study

which investigated how ratings of credibility of a











television anchorwoman were affected by the different

styles of clothing she wore. Credibility ratings, as

determined by the sum of scores of seven characteristics,

including competence, honesty, and viewer preference, were

higher when the anchorwoman was dressed in a conservative,

professional manner (suit jacket and blouse) than when

dressed in a simple dress. The authors concluded that

their results support prior research findings that dress

and appearance are important to newscaster success in

terms of credibility, and that people do have perceptions

of what type of clothing is suitable work apparel.

Appearance and Social Judgments

Appearance often is the most basic perceptual cue

available in order for us to determine the status of

another person. The symbolism which clothing provides

acts as a cue which can be used to classify a person

according to his or her character, occupation, or other

traits. The importance of physical appearance and

clothing is acknowledged in symbolic interaction theory.

Goffman (1959), in his dramaturgical analysis of everyday

social interaction, emphasized the nonverbal communication

accomplished through our "personal front"--that part of us

which we present to society. Personal fronts include all

aspects of our physical selves, including the clothing we

wear. Our appearance, then, may be taken to refer to











those stimuli which function to tell us of a performer's

social status. Indeed, status symbols such as clothing

are the specialized means of displaying one's position or

role in societal exchange (Goffman, 1951). Insignia on

uniforms and certain standard modes of dress, such as the

business suit and tie, are automatically used to infer who

a person is and his or her rank in an organization or in

society in general.

The importance of nonverbal symbolism has been

reiterated by several theorists who have recognized the

effectiveness of appearance as a medium for social

meaning. Stone (1962) recognized the lack of attention

given to this dimension of communication. Appearance

establishes the identification of persons and facilitates

or hinders interaction. Indeed, asserted Stone,

"appearance means identification of one another" (p. 91).

The neglect of empirical study of clothing and

artifactual symbolism is pointed out by Davis (1982), who

explored the lack of attention given to the "vast realm of

human communication lying beyond the essentially

denotative sphere of spoken and written language" (p.

116). Despite this discursive bias evident in the

symbolic interactionist perspective, "we know that through

clothing people communicate some things about their

persons and at the collective level this results typically









11

in locating them symbolically in some structured universe

of status claims and lifestyle attachments" (Davis, 1985,

p. 16).

The extent to which the symbolic can be subjected to

systematic and sustained analysis has been underestimated

in the symbolic interactionist perspective. Indeed, a

proliferation of empirical studies supports several major

conclusions which can be made about clothing, dress, and

appearance, and how they influence social judgments.

In general, research of this type is concerned with the

impressions made of strangers on the basis of their

appearance. The majority of these types of studies, based

in a theory of social perception which assumes that

clothing acts as a nonverbal cue in communicating

information, has demonstrated that clothing does affect

the social impressions made of stimulus persons (Davis &

Lennon, 1988).

A good many of these studies have originated from a

more practical perspective; their main focus has been on

clothing, in terms of the impact dress style, color, and

form have on perceptions of personality traits, social

status, or attractiveness assigned by subjects to stimulus

persons depicted in drawings, photographs, or in person

(Hoult, 1954; Compton, 1962; Rosencranz, 1962; Douty,

1963; Taylor & Compton, 1968; Hamid, 1972; DeLong, 1978;












Pinaire-Reed, 1979; Harris et al., 1983; Fiore & DeLong,

1984, Sweat & Zentner, 1985; Francis & Evans, 1987; Hewitt

& German, 1987).

Reactions to different modes of dress have also been

measured by subject behavior toward a stimulus person.

When approached to sign a petition, subjects were more

likely to comply with a request made by a similarly

dressed other (Bryant, 1975) or a stimulus person dressed

neatly than by one dressed in an untidy manner (Lambert,

1972). In the latter study, it was found that for all

subjects, compliance was higher for a "smartly dressed"

female experimenter posing as a market researcher.

Compliance rates were significantly higher among older

subjects than among younger ones. Lambert (1972)

concluded that this suggests older people have more

conventional attitudes toward appearance.

Gender differences have been the focus of several

clothing and behavior studies. Rosenwasser, Adams, and

Tansil (1983) compared the amount of time males and

females looked at slides of differently dressed male and

female stimulus persons. The researchers' premise was

that in American society, physical appearance has been

more important for and to women than men. Stimulus

persons appeared either fully clothed or in bathing suits.

It was found that male subjects looked significantly











longer at clothed women than at women in bathing suits.

Female subjects behaved similarly. Though their sample

was small, the authors suggest that while men's reactions

can be explained as a function of their interest in the

opposite sex, women's interest in other women's clothing

may provide some evidence for the notion that women are

interested in other women's clothing and artifacts, such

as jewelry.

Gender differences and clothing styles were studied

by Stead and Zinkhan (1986) in their investigation of

salespeople's reactions to similarly dressed males and

females. Male and female confederates were both dressed

in business or casual attire and arrived at a checkout

counter in a department store simultaneously. It was

found that men received service from both male and female

salesclerks sooner than did women. Both male and female

stimulus persons received service more promptly when

dressed in business attire (suit and jacket) than in

casual attire (jeans and shirt). This difference was more

pronounced for male "customers."

Recently, the importance of appearance, especially

for professional women, has been recognized by researchers

examining appearance and its relationship to judgments of

job-skill ratings and competence in the workplace. The

effect of colors and layering (wearing of a suit jacket)










14

on perceptions of a person's capabilities in the business

world was examined by Scherbaum and Shepherd (1987).

Subjects were master of business administration students

who viewed drawings of male and female stimulus persons

dressed in blue or red business suits with or without

jackets. Highest ratings of self-assuredness, likelihood

of success, and appropriateness of attire were given to

male and female targets who wore blue suits with jackets.

Differences between dress conditions were especially

pronounced for male targets, while differences between

attire conditions for female targets were less varied.

The authors concluded that their findings suggest attire

may be more important for men in the business world, and

deviance from the standard business suit may not be as

tolerated for men as for women. Women may have less

stringent restrictions on what is proper or improper

business attire, but the authors admit their findings may

have been a function of social desirability on the part of

subjects in that they may not have wanted to appear as

discriminating against females based on their appearance.

The effect of grooming style on hiring decisions was

studied by Cash (1985) in an experiment in which corporate

personnel served as subjects. Female stimulus persons'

appearance was operationalized as being "managerial" or

nonmanageriall." Managerial appearance consisted of a












tailored suit, little jewelry, and short, simple

hairstyles which were away from the face and lacked

adornment. Nonmanagerial appearance included more

"feminine" clothing, dangling earrings, and hairstyles

which were long, concealed the face or appeared to require

considerable maintenance, and were adorned with ribbons or

barrettes. Subjects were asked to make hiring evaluations

based on targets' photos as well as interview evaluations

and resumes. Male subjects were found to rate targets

with managerial appearance more favorably, while female

subjects' evaluations were dependent upon geographic

region. Subjects from New York rated managerial and

nonmanagerial styles similarly, while in Chicago the

nonmanagerial style was actually preferred. Managerial

appearance was rated more favorably than nonmanagerial

appearance by female subjects in Texas. Overall, both

male and female subjects recommended higher salaries for

targets in the managerial appearance condition and rated
9
managerial appearance as more likable.

The authors concluded that the managerial style of

grooming conveyed a more credible and effective corporate

image. Again, according to the authors, social

desirability may have played a role in some of the female

subjects' ratings; they may have rated target women












similarly because they may not have wanted to appear to

discriminate on the basis of appearance.

Thysjcal. appearan-e, in terms: of dress and grooming,

then, has beecn shown to influence social judgments not

only in terms of attribution of personality traits, but

also of behavior toward stimulus persons. Studies of this

latter type have also found differences in the way males

and females behave toward target persons (Rosenwasser,

Adams, & Tansil, 1983; Cash, 1985; Stead & Zinkhan, 1986;

Scherbaum & Shepherd, 1987).

Appearance and Occupational Stereotypes

A review of stereotype research reveals a number of

findings and general conclusions concerning the role

physical appearance plays in the formation of social

categories. These studies have dealt mainly with somatic

or facial features. The greatest amount of research has

been conducted to verify the existence of a physical

attractiveness stereotype and how perceptions of

attractiveness influence social judgments, expectations,

and interaction (Dion,'Berscheid, & Walster, 1972; Derner

& Thiel, 1975; Bar-Tal & Saxe, 1976; Adams, 1977; Chaiken,

1986; Dion, 1986).

Though most research of stereotypes has concentrated

on specific traits associated with stereotype categories,

the popular conception of stereotypes, especially those












involving gender, includes more than just personality

traits. It also includes physical appearance, role

behaviors, and occupations (Ashmore & Del Boca, 1979;

Deaux, 1984). Deaux and Lewis (1984) investigated the

saliency of physical descriptors as a major component of

gender stereotypes. This type of information described

features such as "tall, strong, sturdy, and broad-

shouldered" as "masculine" physical qualities, while "soft

voice, dainty, graceful, and soft" were "feminine"

qualities. They found evidence that once information

about physical appearance was provided, subjects seemed to

rely heavily on it to make inferences about other aspects.

Deaux and Lewis concluded their study by calling for

further research investigating the extent to which people

hold stereotypes in terms of appearance, adding, "it is

surprising that investigators have given so little thought

to this aspect of gender stereotypes" (p. 1003).

Physical appearance as a component of stereotypes was

the focus of Freeman's (1987) study of gender stereotype

structure. Subjects were asked to rate stimulus persons,

based on written descriptions, on their masculine and

feminine traits and behaviors. It was found that when

stimulus females were described as unattractive, they were

more likely to be rated as engaging in traditional

feminine behaviors (such as housework), and less likely to









18

possess "desired" masculine traits (such as independence)

than females described as being of high or medium

attractiveness. The results "indicate an unfavorable

image of the woman who is unattractive" (p. 66).

Furthermore, concluded the author, physical appearance

served as the most potent source of stereotyping compared

to other information provided about the target persons.

As far as professional image is concerned, there

seems to be a kind of "professional" stereotype, in that

business attire and professional-type grooming seem to

reflect societal norms of what is appropriate for certain

types of occupations. Studies have been done concerning

the appearance of persons in certain occupations and the

importance that physical appearance may have in terms of

prediction of success. For example, Croxton, Van

Rensselaer, Dutton, & Ellis (1989), in their study of

gender-typing of occupations, found that subjects

considered physical attractiveness an important predictor

of success for two occupations: "mayor" and "TV

newsperson." Several studies concerning the appearance of

television newscasters have found that viewers do have

preferences concerning how an anchorman should "look"

(Sanders & Pritchett, 1971), and that the apparel of a

television newswoman does influence her perceived

credibility, in that a conservative dress style which











included suit and jacket resulted in higher ratings of

credibility than a simple dress (Harp et al., 1985).

Because of the entertainment aspects of the television

news industry, it is not surprising that appearance plays

an important role in images of professionalism for

television news anchors, and female anchors in particular.

Indeed, this overemphasis on physical appearance has been

cited as a barrier to equal employment opportunities for

women in broadcasting (Ferri & Keller, 1985).

Mode of dress as a component of gender stereotyping

of occupations was examined by Davis (1982). Subjects

viewed slides of males and females dressed in "masculine"

or "feminine" styles, as determined by protests. The

"masculine" dress for males consisted of a gray, tailored

three-piece suit, white shirt, striped tie, and briefcase.

The "feminine" dress for males consisted of pleated

slacks, a pastel cotton sweater, mesh shoes, and a gold

hoop earring. The "masculine" dress for females consisted

of a navy skirted suit, white blouse, tie, and briefcase,

while the "feminine" dress consisted of a pink cotton

dress, sandals, and white beaded necklace. Male and

female targets were assigned to one of two gender-related

dress conditions. They were also assigned to one of three

occupations: engineering technician (masculine),

administrative secretary (feminine), or educational











counselor (neutral). Subjects rated targets on gender-

related personality traits, attractiveness, interpersonal

attraction, occupational success, and psychological well-

being and adjustment.

It was found that the masculine clothing conditions

were associated with perceptions of greater occupational

success and higher ratings of masculine traits for both

men and women. Though the study does have limitations,

such as generalizability to real-world settings, the

results suggest that masculine traits, occupations, and

modes of dress are more highly valued than feminine ones

(Davis, 1982). The preference for masculine-typed

occupations and traits, and the prestige accorded to

masculine occupations have been supported by several

studies regarding gender and occupations (Shaffer &

Wegley, 1974; Shaffer & Johnson, 1980; Shepelak, Ogden, &

Tobin-Bennett, 1984; Beyard-Tyler & Haring, 1984).
















CHAPTER 3
HYPOTHESES AND RESEARCH QUESTIONS


Hypotheses

The present study is designed to examine the

relationship between appearance and credibility formation

in a mass communication setting: the television interview

in which an individual serves as an expert source of

information. Based upon the findings of previous studies

which have demonstrated the effects of visual cues on

impressions and subsequent behavior and gender differences

in credibility, the following hypothesis is proposed:

H1. Overall, ratings of expertise, character,
dynamism, and total credibility of male experts
will be significantly higher than ratings of
female experts with identical credentials.

"Credibility" in this context will refer to overall

ratings of competence and character as defined by

McCroskey and Young (1981). These include ratings of

expertness and trust. It is believed that because there

seems to be a society-wide bias against women (Richardson, i

1988), especially in terms of their competence as

professionals, a comparison of male and female experts of

similar background will again provide evidence that men

and women are not evaluated equally.










22

Lott's (1985) review of research concerning people's

evaluations of men and women in a variety of contexts

provides extensive evidence for a general societal

tendency to devalue a competent woman. A competent woman

is more likely to be devalued in serious, believable, and

realistic contexts and when there are consequences for the

evaluator, such as in a hiring situation. By extending

this type of inquiry to a source credibility context in a

mass communication setting, we can more fully determine

under what circumstances women professionals might be

rated differently than men in terms of how competent and

trustworthy they appear to be as sources of information.

The following hypotheses are offered:

H2. Overall, ratings of expertise, character,
dynamism, and total credibility will be
significantly different for the different
styles of appearance of female target persons.
Specifically, ratings for female experts will
be significantly lower in the casual appearance
condition than in the conservative condition.

H3. Subjects will perceive smaller differences in
expertise, character, dynamism, and total
credibility between men dressed conservatively
and men dressed casually than between women
dressed conservatively and casually.

In these last two hypotheses, "conservative"

appearance refers to dress styles which reflect

managerial, "masculine" business attire. "Casual"

appearance will reflect nonmanagerial, less formal attire.

Operationalization of these two conditions will be similar












to that developed by Cash (1985) and Scherbaum and

Shepherd (1987).

Hypothesis Two is based on findings of several

studies which have shown that people do hold certain

appearance expectations for professional, managerial-type

occupations (Davis, 1985; Cash, 1985). Harp et al. (1985)

demonstrated that people perceive marked differences in

modes of dress and what is appropriate work attire for one

type of professional. In that study, female news anchors

dressed in a conservative manner received higher ratings

of credibility than when dressed in less conservative

attire.

It is believed that because males hold a higher

initial aura of credibility as experts than females in our

society (Lott, 1985) and physical appearance is a more

important factor in the evaluations of females than of

males (Bar-Tal & Saxe, 1976), alterations in appearance

will be expected to affect the credibility of males less

than the credibility of women. That is, variations in

appearance are not expected to affect a male expert's

credibility ratings greatly. On the other hand, Bar-Tal

and Saxe (1976) point out that appearance plays a more

important part in the role fulfillment of women than of

men, and that our society expects women to emphasize their

physical appearance. That is, women are expected to care












about the way they look because they are aware of the

importance appearance has on their evaluations by others.

Thus, it is expected that the way in which a female target

is dressed would have a greater impact on how others would

rate her as being a competent, expert source than would be

the case with male experts.

In summary, the three hypotheses are intended to test

the notion that appearance can influence perceptions of

competence and trustworthiness. It is expected that

differences in appearance will significantly affect

credibility ratings more for female sources than for male

sources.

Research Questions

To better understand the effects visual cues such as

physical appearance have on perceptions of credibility,

subject factors must also be considered. The way in which

one interprets appearance as an indicator of credibility

would depend on the salience of this type of nonverbal cue

in forming an impression of the target. For instance,

Lambert (1972) found that age may affect how a person will

react to differences in the appearance of a target

individual. When approached by a neatly dressed female,

older subjects (over 35 years of age) were more willing to

comply with her request to interview them than when the

same target was less neatly dressed. This suggests that












older people may have more conventional attitudes toward

dress and may be less tolerant toward "sloppy" appearance.

Younger people, on the other hand, may not consider

"deviant" or unconventional appearance as inappropriate in

situations such as the one in which Lambert conducted his

study. However, in their study of female news anchors,

Harp et al. (1985) did not investigate differences between

age groups and their assessments of newscaster

credibility.

Another important variable to consider is subjects'

sex. Findings concerning how male and female subjects

evaluate different dress styles of male and female targets

have not been entirely consistent. For instance, in

Cash's (1985) study, male subjects preferred a female

target's managerial style over a nonmanagerial look.

However, other studies have found no significant

differences between men and women's ratings of target

persons according to clothing styles, regardless of

targets' sex (Hewitt & German, 1987).

Regarding how men and women rate other men and women

in general on credibility dimensions such as expertness

and trust, Carocci (1988) found that the college men in

her study tended to rate men higher on expertness than

women. The college women in her study, however, rated

both men and women similarly on the same dimensions. It











was also found that women rated men and women higher on

expertness, character, and dynamism than men did. In

other words, it seemed that women were more generous in

their responses than were men. Though Carocci's

definition of the study's targets was vague (men and women

"in general"), her findings do provide a direction of

study which would help us to understand further the

dynamics of person perception and how men and women differ

in their assessment of male and female sources in terms of

credibility.

The main purpose of this study is to investigate how

appearance can affect perceptions of credibility. In

addition to the hypotheses to be investigated, the

following research questions are posed to examine if and

how subjects' age and sex influence ratings of a source's

credibility depending upon that source's appearance style:

Q1. Do older and younger subjects differ in
their ratings of expertise, character,
dynamism, and total credibility of male and
female experts of similar backgrounds who
appear in different clothing?, and

Q2. Do male and female subjects differ in their
ratings of expertise, character, dynamism,
and total credibility of male and female
experts who appear in different clothing?

Results of this study may also add to the

understanding of the composition of stereotypes and the

role that physical appearance plays in their formation.

In addition, the study may also demonstrate under what











conditions and communicative situations differences in

gender may be found.

This study will be limited to investigation of

credibility of expert sources who appear on television

informational programming, such as news and talk shows.

"Expert" in this context is defined as an individual who

has some expertise in a particular area and who appears on

such programming.

In terms of practical applicability, it is believed

that research of this type would help producers of

informational programming to enhance the credibility of

their programs by having credible and credible-looking

expert guests appear on their programs. Professionals who

appear on such shows who want to enhance their own

credibility might also benefit, as well as professionals

in general. Perhaps more important, it is the

professional woman who may benefit more from research

which provides evidence of ways in which appearance

influences perceptions of credibility.















CHAPTER 4
METHOD


Subjects

Subjects were 76 undergraduates at the University of

Florida enrolled in classes during the spring semester of

1991, and 89 members of a civic organization in the

Orlando, Florida area. The researcher was familiar with

the civic organization and solicited the group's

participation during one of its monthly meetings. Data

were collected during the months of April and May, 1991.

There was a total of 67 males and 96 females in the final

experiment sample. The average number of subjects per

experimental condition was 41. A power analysis showed

that the sample size and experimental condition mean met

the requirements for achieving a .05 level of significance

for the overall F-tests in analyses of variance (Winer,

1962).

Design

A 2 X 2 X 2 X 2 factorial design was used.

Independent variables were: (1) appearance of target

source (conservative or casual), (2) sex of target source

(male or female), (3) sex of subject (male or female), and











(4) age group of subject (18 to 34 years old or 35 years

old and above).

There were eight sources appearing in the stimulus:

(1) two male targets dressed conservatively; (2) two male

targets dressed casually; (3) two female targets dressed

conservatively; and (4) two female targets dressed

casually. A total of four target sources was used. Each

was dressed in both casual and conservative attire.

However, subjects saw each target dressed either

conservatively or casually, but not both. Thus, each

subject saw one male target dressed conservatively, the

other male target dressed casually, one female target

dressed conservatively, and the other female target

dressed casually.

"Conservative" appearance for the male targets

consisted of a navy blue suit (jacket and slacks), white

shirt, and blue and red striped tie. Conservative

appearance for the female targets consisted of a navy blue

suit jacket, white blouse, red polka-dotted bow tie, and

navy blue skirt. The video was shot so that subjects did

not see the feet of any of the sources. The conservative

mode of dress for the male and female targets, then, was

designed to convey a professional, managerial image.

"Casual" appearance, on the other hand, was designed

to convey a less professional image. Casual appearance












for the male targets consisted of a short-sleeved, pink-

and-gray knit shirt and blue jeans. Casual appearance for

the female targets consisted of a pink, short-sleeved knit

top and dark blue slacks.

Hairstyles in both dress conditions were identical.

Male targets had short, above-the-collar hairstyles.

Female targets had shorter than shoulder-length hair.

Longer hair was coiffed into an upswept hairstyle.

In order to counteract any order effects, a "dummy"

source appeared before each target. A different dummy

appeared before each of the four targets used, for a total

of four dummies (two male and two female) and four target

sources. These were rotated so that subjects saw all four

targets, each in a different mode of dress.

The "dummy" source appeared in a "neutral" mode of

dress, neither conservative nor casual. Male dummy

sources were dressed in a white long-sleeved shirt, dark

blue slacks and either a gray or brown cardigan sweater.

Female dummy sources appeared either in a light-blue,

long-sleeved blouse and dark blue skirt or a white long-

sleeved blouse and a brown skirt.

None of the targets or dummies wore glasses. Male

targets and dummies had no beards or moustaches. All were

judged by the researcher to appear to be between 30 and 45

years of age.












There were four experimental conditions. In each

appeared two male targets, one conservative, one casual;

two female targets, one conservative, one casual; and four

dummy sources, two males and two females dressed

neutrally. Each target was preceded by a dummy; the same

dummy appeared before the same target.

Materials

Four videotaped versions of the stimulus were

produced. Each consisted of a videotape containing eight

simulated television interviews. The source was

identified by words below the face consisting of his or

her name and occupation.

A total of eight segment topics was used. All dealt

with a similar subject--marine animals. This subject was

chosen because of its noncontroversial and informative

nature. Each segment dealt with a different aspect of the

topic, and each appeared to be an excerpt from a full-

length interview in a studio setting.

In each segment a male program "host" appeared with

the source. The host was dressed similarly to the dummy

sources. He wore a beige suit jacket, cream-colored

shirt, yellow patterned tie, and tan slacks. The host did

not have glasses, a beard or moustache, and his hairstyle

was the same as the other male targets and dummies.












Table 3-1
Order of Sources in Experimental Conditions


Condition 1

Source


Condition 2

Dress Source


Dummy 1

Male A

Dummy 2

Female A

Dummy 3

Male B

Dummy 4

Female B

Condition 3

Source


Dummy 4

Female B

Dummy 1

Male A

Dummy 2

Female A

Dummy 3

Male B


Neutral

Casual

Neutral

Casual

Neutral

Conservative

Neutral

Conservative



Dress


Neutral

Casual

Neutral

Conservative

Neutral

Conservative

Neutral

Casual


Dummy 1

Male A

Dummy 2

Female A

Dummy 3

Male B

Dummy 4

Female B

Condition 4

Source


Dummy 4

Female B

Dummy 1

Male A

Dummy 2

Female A

Dummy 3

Male B


Neutral

Conservative

Neutral

Conservative

Neutral

Casual

Neutral

Casual



Dress


Neutral

Conservative

Neutral

Casual

Neutral

Casual

Neutral

Conservative


and 4 are


Dress


Note: Dummies 1 and 3 are female; Dummies 2
male.












The same program host appeared in the same clothing

in all eight segments. The host asked the source (target

or dummy) a question five to 10 seconds in length

concerning marine biology, with the source responding with

an answer that did not exceed 50 seconds in length. The

segment began with a wide-shot in which both the host, who

was not identified with a graphic, and source appeared.

After the host posed a question, the source was shown in a

medium-shot with an identification across the chest area

which consisted of two lines. On the first line was a

fictitious name, and on the second line were the words

"associate professor of zoology." All sources, both

target and dummy, were identified in the same manner. The

graphic remained on screen for an average of seven seconds

while the source was shown responding to the host's

question. It appeared again during the last seven seconds

of the source's response.

All targets and dummies were assigned a separate

topic. Targets spoke about the same topic in both dress

conditions. Each interview, with both dummy and target

sources, lasted approximately 45 to 55 seconds in length.

The interviews were professionally produced in the studios

of a local public television station. The host was the

station's news director who also hosted authentic

informational programming produced at the station. All












interview scripts were displayed on teleprompters.

Segments were thoroughly rehearsed so that gestures and

animation by the target sources were equivalent in both

dress conditions. The interviews were recorded onto 1/2"

videotape. Total tape time was approximately eight

minutes.

Procedure

Several protests were conducted. These were to

ensure that all segment topics were rated similarly in

interest, that target and dummy sources were perceived to

be of similar physical attractiveness and rated similarly

on several credibility variables when dressed in identical

costumes, and that the casual, conservative, and neutral

dress conditions were rated as being significantly

different.

Pretest 1: Topic interest

A group of 19 students, 11 males and 8 females,

enrolled in an introductory mass-media writing class

served as subjects. All were between the ages of 18 and

34.

Subjects read 16 segments of purported scripts.

After reading each segment, subjects filled out a seven-

point semantic differential scale indicating how

interesting or boring they found each segment topic (see

Appendix A). Mean scores for each segment were used to











assess the similarity in interest ratings of each topic.

Mean ratings ranged from 4.37 to 2.89, with higher ratings

indicating higher interest. The two segments that rated

highest and the two that rated lowest in interest were

discarded. The remaining 12 topics were randomly assigned

to the actors appearing in the stimulus videotape used in

Pretest 2.

Pretest 2: Target similarity

A group of 42 students, 21 males and 21 females,

enrolled in an undergraduate telecommunication survey

course for non-majors served as subjects. Forty-one

subjects were between the ages of 18 and 24, and one was

between 25 and 34.

Subjects viewed a videotape of interview segments

either with or without audio of 12 persons (six male and

six female) dressed in conservative attire. Subjects were

randomly divided into two groups: 23 viewed the stimulus

videotape without audio; 19 viewed it with audio.

The purpose of this pretest was mainly to see that

the four people who were to serve as target sources in the

final experiment were perceived as similarly attractive

and competent. The 12 topics which were pretested as

being of similar interest were used; each source talked

about one topic. After viewing each segment, subjects

rated each target on the following seven-point semantic-











differential scales: attractive-unattractive,

intelligent-unintelligent, trained-untrained, expert-

inexpert, and competent-incompetent (see Appendix B). A

reliability analysis of the latter four semantic-

differential scales used to assess competence similarity

for the 12 sources resulted in a Cronbach's alpha of .92.

Ratings on all four scales were then totaled and treated

as composite credibility scores. The highest possible

credibility score was 28, the lowest was 4.

Credibility scores for each of the 12 actors were

then submitted to t-tests to check for significant

differences between the audio and no-audio conditions. No

differences were found at the .05 significance level based

on two-tailed probability. Not hearing the actors had no

significant effect on how subjects rated their

credibility.

The credibility scores from the audio condition were

then used to determine which eight actors would appear in

the stimulus to be used in Pretest 3 and the final

experiment. Mean credibility scores ranged from 20.1 to

22.8. The four actors (two male and two female) receiving

the lowest credibility scores were eliminated from the

stimulus. The four sources with the highest competence

scores served as the dress-condition targets in the final











experiment. Those four sources with the next highest

competence scores served as dummies.

All 12 actors were rated similarly on attractiveness,

with mean ratings on this item ranging from 3.2 to 4.5 (on

a seven-point scale). In addition, t-tests between male

and female actors' credibility scores showed no

significant differences in either audio or no-audio

condition. Male and female actors dressed conservatively

were rated similarly in credibility.

Pretest 3: Dress conditions

Subjects in this final pretest were 27 students (10

males, 17 females) enrolled in an undergraduate sociology

course. Twenty-four subjects were between the ages of 18

and 24, and three were between 25 and 44.

Subjects were randomly assigned to one of two groups:

one group contained 11 subjects, the other contained 16.

One group viewed segments of Male Target A dressed

casually, Male Target B dressed conservatively, Female

Target A dressed casually, Female Target B dressed

conservatively, and all four dummies dressed neutrally.

The other group viewed segments of Male Target A dressed

conservatively, Male Target B dressed casually, Female

Target A dressed conservatively, Female Target B dressed

casually, and all four dummies.








38

After viewing the targets, subjects rated them on the

following seven-point semantic-differential scales:

conservative clothing--casual clothing, professional-

looking--unprofessional-looking, and masculine clothing-

feminine clothing (see Appendix C). Subjects also rated

the dummies on the same scales to determine that dummy

dress styles were perceived to be neither very

conservative nor very casual and neither very professional

nor very unprofessional.

For each of the four actors who were designated to be

targets, individual t-tests run between dress conditions

showed significant differences at the .05 level (based on

two-tailed probability) on the conservative-casual

clothing and professional-unprofessional looking scales.

When actors appeared in conservative attire, they received

significantly higher ratings on both scales than when they

appeared in casual clothing. The overall mean rating on

the conservative clothing--casual clothing scale for

targets in the conservative clothing condition was 6.73,

with seven indicating very conservative clothing. The

mean rating for targets in the casual clothing condition

was 1.92. The overall mean rating on the professional-

looking--unprofessional-looking scale for targets in the

conservative condition was 6.42, with seven indicating











very professional-looking. The mean rating on the same

scale for targets in the casual condition was 2.65.

T-tests conducted on ratings for the masculine-

feminine clothing scales for actors in the two dress

conditions revealed significant differences for only two

of the four targets, one male and one female. In both

cases significantly higher ratings on the masculine-

looking--feminine-looking scale (seven indicating very

masculine-clothing) were given to these targets when they

were dressed in conservative clothing than when dressed in

casual clothing.1 In short, conservative clothing was

rated as being more masculine for these two targets.

A reliability analysis conducted on mean ratings for

the four dummy sources appearing in the neutral dress

condition showed that subjects rated them similarly on the

conservative-casual and professional-unprofessional

looking scales (Cronbach's alpha= .71 and .62,

respectively). Ratings for both these scales were between

4.72 (highest) and 2.13 (lowest). A reliability analysis

conducted on mean ratings for the masculine-feminine

clothing scale showed that subjects rated dummy sources

similarly on this scale (Cronbach's alpha= .56). Mean

masculine clothing-feminine clothing ratings for dummies

were between 5.45 (highest) and 2.9 (lowest).












It was concluded that the conservative and casual

dress condition manipulation was successful. Actors

appearing in the videotape used for this pretest were

included in the final experimental stimulus.

Final Experiment

In the actual experiment, subjects were randomly

assigned to one of four groups. Each group viewed one of

four versions of the stimulus videotape. After each

segment, the videotape machine was stopped and subjects

filled out credibility scales measuring each source's

competence, character, and dynamism as defined by

McCroskey and Young (1981) and Miller and Hewgill (1964).

Each characteristic was presented as a series of

seven-point semantic differential scales. The competence

dimension was measured by the following scales:

intelligent-unintelligent, trained-untrained, expert-

inexpert, informed-uninformed, competent-incompetent, and

stupid-bright. The scales used for the character

dimension were: virtuous-sinful, dishonest-honest,

selfish-unselfish, sympathetic-unsympathetic, high

character-low character, and trustworthy-untrustworthy.

Dynamism was measured by the following scales:

aggressive-meek, bold-timid, energetic-tired, and

extroverted-introverted (see Appendix D).












After rating the last source in each condition,

subjects completed a questionnaire containing a series of

items concerning demographic information: age, sex,

education level, income level, and occupation (or, if

retired, former occupation) (see Appendix E).

Instrument Validity and Reliability

The 16 semantic-differential scales of items used in

the final experiment have previously been found to be

correlated with three principal dimensions used to measure

source credibility: competence (called expertise in this

experiment), character, and dynamism. The competence and

character variables used in the present study's instrument

are borrowed from factor-analytic research by McCroskey

and Young (1981). The four variables used to measure the

dynamism dimension were previously used by Miller and

Hewgill (1964).

Before analysis of results pertaining to the

hypotheses and research questions could be conducted,

validity of the experimental instrument was verified to

show, in terms of empirical validity, that the instrument

actually taps the attributes of expertise, character, and

dynamism that it purports to measure, and that the

instrument extracts only those attributes that are of

interest to the researcher. By doing this, the researcher











can be assured that the instrument also has construct

validity (Stamm, 1981).

The instrument used should also demonstrate some

degree of reliability in terms of its stability and

consistency. That is, the instrument should be able to

produce similar results when administered over time.

Reliable measures should be able to detect relationships

between variables and give the researcher confidence that

the instrument will yield the same results when used

repeatedly (Wimmer and Dominick, 1983).

Empirical and construct validity of the variables

used in the present study were demonstrated by conducting

a confirmatory factor analysis on the 16 scales used to

gauge how subjects perceived the actors appearing in the

stimulus video. A reliability analysis using Cronbach's

alpha was used to assess the general reliability of the

credibility measures. These two procedures were used to

verify the correlations between the variables contained in

each credibility dimension.

Factor Analysis

Confirmatory factor analysis begins with a

preconceived idea about the possible structure of a

particular area. Variables are then selected which might

fit that structure (Child, 1990). In this case, expertise

was measured by the following seven-point semantic









43

differential scales: intelligent-unintelligent, trained-

untrained, expert-inexpert, informed-uninformed,

competent-incompetent, and stupid-bright. Sources'

character was represented by the following seven-point

scales: virtuous-sinful, honest-dishonest, selfish-

unselfish, sympathetic-unsympathetic, high character-low

character, and trustworthy-untrustworthy. Dynamism was

measured on the following four scales: aggressive-meek,

bold-timid, energetic-tired, and extroverted-introverted.

The factor analysis was conducted after all data from

the final experiment were collected. Due to the unwieldy

nature of attempting to analyze all 16 variables for all

eight sources, a total of 128 variables, eight separate

factor analyses were conducted, one for each source

subjects saw, regardless of experimental condition. Thus,

only the 16 variables used to rate each source were

included in the factor extraction.

A principal factor design utilizing varimax rotation

was used, with missing values replaced by variable means.

Factor loadings produced in the rotated factor matrix were

used to determine how well variable groupings matched

their relevant indices.

The desired three-factor extraction was obtained for

two of the eight analyses. Five of the analyses resulted

in a four-factor solution, and one of the analyses









44

resulted in a five-factor solution (for examples of three-

and four-factor solutions, see Tables 3-2 and 3-3). Among

these, there were no consistent patterns of aberrant

variable groupings. In general, those variables which

were expected to be correlated, were. Variables which

were extracted as components of separate factors but had a

minimum loading of .30 on their predetermined index were

considered as part of the appropriate factor. Thus, for

purposes of this study, it was concluded that the

variables used to determine each credibility index were

sufficiently correlated.

Reliability

Reliability analyses were conducted by creating

separate scales for each of the three credibility

dimensions. Variable ratings for all eight sources viewed

by subjects, regardless of condition, were included in the

analyses. Results are as follows: expertise dimension

Cronbach's alpha = .98 (six variables, a total of 48 items

analyzed); character dimension Cronbach's alpha = .99 (six

variables, a total of 48 items analyzed); and dynamism

dimension Cronbach's alpha = .98 (four variables, a total

of 32 items analyzed) (n = 165 for all three analyses,

including missing values).

High reliability coefficients among the variables led

to the conclusion that they could be summed into their









45

appropriate indices for use in testing the hypotheses and

research questions.

Analysis

The scales for the credibility dimensions were

scored by assigning a value of seven to the responses

indicating the highest degree of perceived intelligence,

training, expertise, and so on. A value of one was given

to responses for the lowest ratings.

Scores for expertise, character, and dynamism were

obtained by summing the values of each dimension's various

factors. Accordingly, the possible range of scores for

each dimension was: competence--6 to 42, character--6 to

42, and dynamism--4 to 28. The maximum possible total

credibility score was 112.

Hypothesis One was tested by creating mean dimension

scores according to target sources' sex, regardless of

dress condition. Expertise, character, dynamism, and

total credibility scores for the two male target sources

were averaged, as were the scores for the two female

target sources. Thus, there were four scores for each

sex. These were submitted to t-tests to determine if mean

scores for male and female targets were significantly

different.

For Hypothesis Two, scores for both female target

sources were averaged according to dress condition. These












Table 3-2
Example of Three-Factor Solution of Credibility Variables



Factor Loadings


Variable Expertise Character Dynamism


Expert-Inexpert .86890

Intelligent-Unintelligent .72823

Informed-Uninformed .72127

Bright-Stupid .63599

Competent-Incompetent .58738

Trained-Untrained .57734

Honest-Dishonest .75405

Trustworthy-Untrustworthy .74913

High Character-Low Character .73580

Sympathetic-Unsympathetic .66107

Virtuous-Sinful .66021

Selfish-Unselfish .64408

Aggressive-Meek .83165

Energetic-Tired .68346

Bold-Timid .57408

Extroverted-Introverted .45554












Table 3-3
Example of Four-Factor Solution of Credibility Variables



Factor Loadings


Variable Expertise Character Dynamism


Expert-Inexpert .73516

Intelligent-Unintelligent .73308

Informed-Uninformed .78259

Bright-Stupid .53441

*Competent-Incompetent .45798

*Trained-Untrained .46763

Honest-Dishonest .63889

Trustworthy-Untrustworthy .69396

High Character-Low Character .62931

Sympathetic-Unsympathetic .55329

Virtuous-Sinful .77853

Selfish-Unselfish .67965

Aggressive-Meek .77490

Energetic-Tired .53406

Bold-Timid .56675

Extroverted-Introverted .46779



*These two variables were loaded onto a fourth factor.
Factor loadings on this separate factor were:
Competent-Incompetent .79621
Trained-Untrained .55140












means were then submitted to t-tests to check for

significant differences between casual and conservative

dress conditions.

For Hypothesis Three, differences between expertise,

character, dynamism, and total credibility scores of

conservatively dressed male targets and casually dressed

male targets were compared to differences between the same

scores of conservatively dressed female targets and

casually dressed female targets. A separate difference

score was calculated for each dimension and for total

credibility by subtracting scores in the casual condition

from their corresponding scores in the conservative

condition. Mean difference scores for each dimension and

total credibility were then submitted to t-tests to see if

differences between dress conditions for male sources were

smaller than differences in dress conditions for female

sources.

In sum, the statistical procedures used to test the

three hypotheses were intended to analyze the differences

between subjects, since target sources were not seen in

both modes of dress. In testing Hypothesis One, scores

for male sources were compared to scores for female

sources, regardless of dress. Hypothesis Two was tested

by comparing scores of female sources when they were

dressed conservatively to their scores when they were











dressed casually. Hypothesis Three was tested by

comparing differences between clothing conditions of male

sources with differences between clothing conditions of

female sources.

In order to examine differences in how older and

younger subjects and male and female subjects rated the

four targets, dimension and total credibility scores for

each of the four target sources were submitted to two-way

analyses of variance tests (ANOVA), with subjects' age and

sex serving as independent variables.

Note

iMale Target A received significantly higher ratings
on the masculine clothing-feminine clothing scale (with
higher scores indicating higher masculinity) when dressed
conservatively (M = 7.0) than when dressed casually (M =
5.9), t(24) = -4.08, p < .001. Subjects rated Female
Target A's conservative clothing to be significantly more
masculine (M = 3.6) than her casual clothing (M = 2.36,
t(22.3) = -2.35, E = .028. No significant differences
were found for mean scores on the masculine clothing-
feminine clothing scale between conservative and casual
conditions for Male Target B (conservative: M = 6.27;
casual: M = 5.31) and Female Target B (conservative: M =
2.55; casual: M = 2.87).















CHAPTER 5
RESULTS


The present study was designed to examine the

relationship between appearance and subsequent perceptions

of credibility of male and female sources appearing in a

television interview. The hypotheses were intended to

investigate the differences between credibility attributed

to male and female experts, and to see if the way a source

dresses makes a difference in how others perceive his or

her credibility. The research questions were posed to

investigate the effects a receiver's age and sex have on

how he or she perceived a source based on that source's

appearance. Findings are presented first in terms of the

three hypotheses, and then in terms of the interactions,

if any, between the age and sex of message receivers and

their ratings of target sources' credibility.

Description of Subjects

The final experiment was conducted with a total of

165 subjects: 76 were students enrolled in various

undergraduate classes at the University of Florida; 89

were members of a civic organization based in Orlando,

Florida.












The student sample contained students enrolled in

undergraduate classes in psychology (n = 46), mass media

writing (n = 19), and art (n = 11). There were 26 males

and 50 females. Seventy-one were between the ages of 18

and 34, and five were between the ages of 35 and 70.

Most, 80%, were undergraduates, and 20% reported having a

bachelor's degree or some formal education beyond college.

Mean family income for this group was between $40,000 and

$60,000 per year.

The civic organization sample of 89 subjects

contained 41 males and 46 females. Two subjects did not

indicate their sex. Seventy-eight of the subjects from

this group were between the ages of 35 and 70 or above;

only eight were between the ages of 18 and 34. Three

subjects did not indicate their age. As for education

level, 24% had completed high school, 35% had some college

education or bachelor's degree, 8% had a vocational or

trade school education, and 31% had some formal education

beyond college or some type of graduate degree. Two

percent of the subjects did not indicate education level.

Mean family income for this group was between $40,000 and

$60,000 per year.

Of the 165 total number of subjects, 79 were between

the ages of 18 and 35, and 83 were 35 years old or older











(three subjects did not indicate age). Males totaled 67,

females totaled 96 (two subjects did not indicate sex).

The number of subjects assigned to each experimental

condition were as follows: Condition One--42, Condition

Two--40, Condition Three--38, Condition Four--45. There

was a mean of 17 males and 24 females per condition.

There was a mean of 19 subjects between the ages of 18 and

34 per condition, and a mean of 21 subjects 35 or older

per condition.

Results for Hypotheses

Hypothesis One

The first hypothesis predicted that male experts

would receive higher ratings of competence, character,

dynamism, and overall credibility than female experts with

identical credentials, regardless of dress condition.

This hypothesis was supported only in part. Results of

t-tests conducted between male and female target experts'

credibility dimensions show that significant differences

were found only for the dimension of dynamism. Subjects

rated male targets higher in dynamism (M = 19.72) than

they did female targets (M = 18.53), t(152) = 4.74,

E < .001.

Closer examination of this finding revealed that one

of the female targets (Female Target A) received

significantly lower dynamism ratings than the two male











targets when comparisons were made within both dress

conditions. In the casual dress condition, Female Target

A received significantly lower scores in dynamism (M =

18.26) than Male Target A (M = 19.19), t(77) = 2.18, E =

.03, and Male Target B (M = 20.6), t(145.8) = 3.48, R =

.001. In the conservative dress condition, Female Target

A also was rated significantly lower in dynamism (M =

17.32) than Male Target A (M = 19.69), t(67) = 4.13, p <

.001, and Male Target B (M = 19.96), t(147.6) = 3.86, E <

.001. Female Target A was also rated significantly lower

in dynamism (M = 17.32) than the other female target (B)

(M = 19.39) when the two were dressed in conservative

clothing, t(148.2) = 3.12, E = .002.

In addition, Female Target B was rated significantly

lower in dynamism (M = 19.38) than Male Target B (M =

20.6) when both were dressed casually, t(72) = 2.16, R =

.03. Though Female Target B was rated to be slightly

higher in dynamism (M = 19.38) than Male Target A (M =

18.89) when both were dressed casually, this difference

was not significant. As for comparisons made within the

conservative dress condition, Male Target A (M = 19.57)

and Male Target B (M = 19.88) were rated slightly higher

in dynamism than Female Target A (M = 19.39), but these

differences were not significant.












Taken together, these additional findings led the

researcher to conclude that the dynamism finding mainly

could be attributed to the lower ratings given to Female

Target A. With one exception, however, both female

targets were rated lower in dynamism than both male

targets, though these differences were not always

significant.

Means for male and female target credibility ratings

are presented in Table 4-1. Male targets received only

slightly higher expertise and total credibility scores.

Female targets, on the other hand, received higher scores

in the character dimension. However, none of these

differences was significant.

Hypothesis Two

Hypothesis Two predicted that credibility ratings for

female experts would be significantly lower when they

appeared in casual clothing rather than in conservative

clothing. This hypothesis was not supported. T-tests run

on each credibility dimension found no significant

differences between the two dress conditions, although it

seemed that when female targets were dressed

conservatively, they received slightly higher ratings in

expertise, dynamism, and overall credibility (see Table 4-

2). Casual dress of female targets did not seem to

detract significantly from their credibility.












Table 4-1
Mean Credibility Dimension Scores for Male and Female
Targets, Regardless of Dress Condition



Target


Credibility Dimension Males Females




Expertise 33.99 33.94
SD 5.0 5.1

Character 28.25 28.61
SD 6.3 6.4

Dynamism 19.72a 18.53a
SD 3.8 3.3

Total Credibility 80.68 79.91
SD 13.4 13.6



Note: Means with common subscripts are significantly
different at the R < .001 level, one-tailed
probability.

Hypothesis Three

Hypothesis Three was not supported. T-tests

conducted on the differences between male and female

source scores based on their dress condition revealed that

differences between scores for males were not

significantly smaller than those for females. In fact,

differences in the character dimension between

conservatively and casually dressed men were larger than

those for women. It seemed that character scores for












casually dressed males were higher than those for

conservatively dressed males. In other words, for the

character dimension, the difference between dress

conditions for males (M = -.95) was significantly larger

than the difference between dress conditions for females

(M = .23), t(142) = 2.11, E = .036 (see Table 4-2).

When the dimension scores for male targets were

examined, it was found that when male targets were dressed

casually, they received higher scores in the character

Table 4-2
Mean Difference Scores for Conservative and Casual Dress
Conditions of Male and Female Target Sources



Mean Difference Score

Dimension Females Males t df prob.


Expertise .14 -.42 .85 148 .395
SD 6.0 6.4

Character .23 -.95 2.11 142 .036
SD 4.4 5.1

Dynamism -.12 -.02 -.20 144 .839
SD 4.5 4.2

Total
credibility .63 -1.69 1.56 149 .122
SD 12.6 14.5



Note: Mean difference scores for each dimension were
calculated by the following equation: Conservative
score casual score = difference score











dimension (M = 29.18) than they did when dressed in

conservative clothing (M = 28.18), t(147) = 2.38, p < .02

(see Table 4-3).

Closer examination of the difference in character

scores between conservatively and casually dressed male

targets showed that older subjects gave higher character

ratings to casually dressed males. Older males rated

casually dressed male targets (M = 30.89) significantly

higher in character than they did conservatively dressed

males (M = 28.39), t(35) = 2.35, E = .025. Older females

also rated casually dressed male targets (M = 27.76)

significantly higher in character than they did

conservatively dressed male targets (M = 26.13),

t(37) = 2.09, E = .04. No significant differences between

the two dress conditions were found among younger subjects

of either sex.

In sum, the findings for Hypothesis Three showed that

dress did not make more of a significant difference in the

credibility ratings of female targets than it did for male

targets. Regarding character, the opposite was true:

Among older subjects, dress did make a significant

difference in how they rated male targets. Older men and

women regarded male targets dressed casually to be higher

in character than male targets dressed conservatively.











Table 4-3
Mean Credibility Dimension Scores for Male and Female
Targets by Dress Condition



Male Targets Female Targets


Dimension Conservative Casual Conservative Casual



Expertise 33.80 34.18 34.14 33.84
SD 5.8 6.0 6.0 5.9

Character 28.18a 29.18a 29.46 29.21
SD 5.9 6.2 5.5 5.6

Dynamism 19.82 19.79 18.57 18.81
SD 4.0 4.3 4.0 3.8

Total
credibility 79.91 81.46 80.38 79.64
SD 14.9 15.5 15.0 15.1



Note: Means with common subscripts significantly differ
at the E < .05 level, two-tailed probability.

Summary of Hypothesis Findings

Findings concerning the three hypotheses show that

subjects rated male and female targets equally in terms of

expertise, character, and total credibility. The only

dimension in which male sources had higher scores was

dynamism. These differences were mainly due to lower

ratings given to one of the female targets.

Clothing of female targets had no effect on how

subjects judged their credibility. Male targets were











rated as higher in character by older subjects when

dressed in casual rather than conservative attire, but

their expertise, dynamism, and total credibility were not

affected by what they wore.

Results for Research Questions

The two research questions were posed to examine

differences between older and younger subjects and male

and female subjects. To accommodate unequal cell sizes

based on age and sex, multiple analyses of variance

(MANOVA) were conducted for each of the dependent

variables based on target sex and dress condition, with

subjects' age and sex as factors.

If interaction effects of subject age and sex were

present, t-tests were used to probe significant

differences between subject cell means (Weaver, 1981). If

no interaction effects were revealed but differences were

suspected, t-tests were used to uncover significant

findings. All t-test procedures were based on two-tailed

probability.

Male Targets. Regardless of Dress

Expertise. No significant main effects for subject

age or sex were found for ratings of male targets'

expertise. However, a subsequent t-test conducted between

ratings given by younger females (M = 35.5) and older

females (M = 33.56) revealed a significant difference at











the E < .05 level, t(89) = 2.01. A comparison of means

for male target expertise by subject age and sex is found

in Table 4-4. Men, regardless of age, seemed to rate male

targets similarly on expertise.

Character. A two-way analysis of variance revealed a

significant interaction between subject age and sex on the

character dimension for male targets, F(1, 146) = 5.23,

E = .02. Mean scores for this dimension are found in

Table 4-4. T-tests used to probe differences between

subgroups found that younger females (M = 30.03) tended to

give higher ratings in the character dimension to male

targets than did younger males (M = 28.17), t(60.0) =

-1.98, p = .05, and older females (M = 26.10), t(63) =

2.78, E < .01. Both older and younger men seemed to rate

male targets similarly.

Dynamism. No significant main effects for age or sex

were found for the dynamism of male targets. However, t-

tests again revealed a significant difference between

younger and older females, with younger females (M =

20.57) giving higher ratings of dynamism to male targets

than did older females (M =18.47), t(63.2) = 2.19, p = .03

(see Table 4-4).

Total credibility. A two-way ANOVA for overall

credibility scores for male targets revealed a significant

main effect for subject age, F(l, 149) = 7.37, R = .007.












Younger females tended to give male targets the highest

total credibility scores (see Table 4-4). Subsequent t-

tests found that younger females' ratings (M = 86.1) of

male targets were significantly higher than those given by

younger males (M = 81.5), t(69) = -1.96, E = .05, by older

males (M = 79.63), t(59.0) = -2.28, E = .03, and by older

females (M = 76.19), t(68.1) = 3.72, p < .001. In sum,

younger female subjects gave significantly higher total

credibility ratings to male target sources than did the

other three subgroups.

Female Targets. Regardless of Dress

Expertise. A significant two-way interaction between

subject age and sex was found for expertise of female

targets F(1, 149) = 8.65, p = .004. Younger females gave

significantly higher expertise ratings (M = 36.10) to

female targets than did younger males (M = 32.89), t(69)

-3.06, E = .003, older males (M = 34), t(68) = -2.17, =

.03, and older females (M = 32.41), t(70.8) = 3.63, E =

.001 (see Table 4-4).

Character. Interaction effects of subject age and

sex were also significant for character ratings of female

targets, F(1, 147) = 8.02, E = .005. T-tests conducted

between the four subgroups uncovered a plethora of

significant differences (see Table 4-4). Younger females

gave significantly higher character ratings to female











targets (M = 31.20) than did younger males (M = 28.57),

t(69) = -2.31, p = .02, older males (M = 28.72), t(83) =

-2.02, E < .05, and older females (M = 25.30), t(66.1) =

4.01, R < .001. In addition, older female subjects (M =

25.30) gave significantly lower character scores to female

targets than did both younger males (M = 28.57), t(60.5) =

2.23, E = .03, and older males (M = 28.72), t(78) = 2.01,

E < .05. To sum, younger female subjects gave the highest

character scores to female targets, while older females

gave the lowest scores. Older and younger men seemed to

rate female targets similarly on character.

Dynamism. A significant two-way interaction between

subject age and sex was found for the dynamism dimension

of female targets, F(1, 146) = 7.12, E < .01. Younger

female subjects rated female targets significantly higher

in dynamism (M = 19.85) than did younger males (M =

18.17), t(69) = -3.04, p = .003, older males (M = 18.55),

t(60.3) = -2.03, p < .05, and older females (M = 17.30),

t(59.2) = 3.37, E = .001 (see Table 4-4).

Total credibility. A significant two-way interaction

between subject age and sex was found for total

credibility scores of female targets, F(1, 149) = 8.78,

R = .004. Younger males (M = 79.63) and older females

(M = 73.57) rated the total credibility of female targets

significantly differently, t(63.9) = 2.08, E = .04.











Younger female subjects (M = 87.16) rated women targets

higher than did younger males (M = 79.63), t(69) = -3.47,

E = .001, older males (M = 78.76), t(58.1) = -3.06,

E = .003, and older females (M = 73.57), t(64.5) = 5.0,

E < .001. In sum, younger female subjects gave

significantly higher ratings of total credibility to

female target sources than did the other three subgroups

(see Table 4-4).

Conservatively Dressed Female Targets

Expertise. A significant two-way interaction between

subject age and sex was found for expertise of

conservatively dressed female targets, F(1,147) = 5.24,

E = .02. Younger females (M = 36.02) rated conservatively

dressed female targets significantly higher in expertise

than did younger males (M = 32.74), t(69) = -2.59,

E = .01, and older females (M = 32.88, t(70.2) = 2.45,

p < .02) (see Table 4-5). Older and younger male subjects

tended to rate conservatively dressed female targets

similarly on expertise.

Character. A significant two-way interaction between

subject age and sex was found for the character dimension

of conservatively dressed female targets, F(1, 143) =

7.42, p = .007. Significant differences were once again

found between younger females (M = 31.08) and older

females (M = 26.15), t(68.6) = 3.39, E = .001. Younger












Table 4-4
Mean Credibility Dimension Scores for Male and Female
Targets Regardless of Dress Condition by Subject Age and
Sex



Male Subjects Female Subjects


Dimension Older Younger Older Younger


Male Targets

Expertise 33.54 33.39 33.56a 35.50a
SD 5.4 4.8 5.0 4.2

Character 28.97 28.17a 26.10b 30.03ab
SD 6.5 3.2 8.1 4.6

Dynamism 19.92 19.94 18.47a 20.57a
SD 3.1 2.9 5.6 3.0

Total
credibility 79.63a 81.50b 76.19c 86.10abc
SD 15.6 9.3 15.1 9.3



Female Targets

Expertise 34.0 a 32.89b 32.41c 36.10abc
SD 5.0 5.0 5.7 3.7

Character 28.72ad 28.57be 25.30cde 31.20abc
SD 6.4 3.4 8.4 4.9

Dynamism 18.55a 18.17b 17.30c 19.85abc
SD 3.3 2.0 4.4 2.3

Total
credibility 78.76a 79.63bd 73.57cd 87.16abc
SD 15.2 8.0 15.7 8.8


Note: Means along dimensions with
significantly differ at E <
probability.


common subscripts
.05 level, two-tailed











females (M = 31.08) also gave significantly higher

character ratings to conservatively dressed female sources

than did younger males (M = 28.39, t(69) = -2.1, E = .04.

Younger and older male subjects again tended to behave

similarly (see Table 4-5).

Dynamism. A significant two-way interaction was

found between subject age and sex for dynamism scores

given to conservatively dressed female targets, F(1, 144)

= 5.36, E = .02. Younger females gave higher dynamism

scores to conservatively dressed female targets (M =

19.44) than did older females (M = 17.24), t(69.8) = 2.36,

E = .02; older and younger men seemed to rate these

targets similarly on dynamism (see Table 4-5).

Total credibility. As for total credibility scores

of conservatively dressed female targets, a two-way

interaction between subject age and sex was significant at

the E = .01 level, F(1, 147) = 6.86. Younger females

(M = 86.54) gave significantly higher overall credibility

scores to conservatively dressed female targets than did

younger males (M = 79.0), t(69) = -2.79, R = .007, older

males (M = 79.97), t(59.4) = -2.0, E = .05, and older

females (M = 74.62, t(69.5) = 3.89, E < .001 (see Table 4-

5). In sum, younger females rated conservatively dressed

female targets higher in overall credibility than did the

three other subgroups.












Casually Dressed Female Targets

Expertise. Significant interaction effects between

subject age and sex were found among expertise scores of

casually dressed female targets, F(1, 149) = 7.59,

E = .007. Younger females (M = 36.19) gave significantly

higher expertise ratings to casually dressed female

targets than did younger males (M = 33.04), t(69) = -2.73,

E < .01, and older females (M = 31.98), t(71.7) = 3.97,

E < .001 (see Table 4-5). Differences between mean

expertise scores for casually dressed female targets given

by younger females (M = 36.19) and older males approached

significance, (M = 33.97), t(85) = -1.93, p = .057. No

differences were found between older and younger men in

their ratings of casually dressed females' expertise.

Character. A significant interaction between subject

age and sex was found for the character dimension of

casually dressed female targets, F(1, 142) = 6.10,

R < .02. Younger females gave significantly higher

character scores to casually dressed female targets

(M = 31.31) than did younger males (M = 28.74), t(69) =

-2.02, R < .05, older males (M = 28.78), t(82) = -2.14,

R < .04, and older females (M = 26.41), t(67.3) = 3.5,

E = .001 (see Table 4-5).

Dynamism. A significant main effect was found for

subject age on the dynamism dimension of casually dressed










67

female targets, F(1, 144) = 4.78, E = .03. Subsequent t-

tests revealed that younger female subjects rated casually

dressed female targets significantly higher in dynamism

(M = 20.27) than did younger males (M = 18.48),

t(69) = -2.5, E < .02, older males (M = 18.22), t(63.1) =

-2.67, E = .01, and older females (M = 17.8), t(63.0) =

2.96, E = .004 (see Table 4-5).

Total credibility. A significant interaction between

subject age and sex for total credibility of casually

dressed female targets was found, F(1, 149) = 7.54,

E = .007. Younger female subjects found casually dressed

female targets to be higher in overall credibility

(M = 87.77) than did younger males (M = 80.26), t(69) =

-3.11, E = .003, older males (M = 77.82), t(58.3) = -3.48,

E = .001, and older females (M =72.49), t(61.8) = 5.1,

E < .001. Significant differences were also found between

older females (M = 72.49) and younger males (M = 80.26,

t(63.6) = 2.27, E < .03 (see Table 4-5). Older male and

female subjects seemed to rate casually dressed female

targets similarly on total credibility, as did older and

younger males. Older females rated casually dressed

female targets significantly lower in total credibility

than did younger males and females. It should be noted

that as with conservatively dressed female targets,

younger female subjects rated casually dressed females











significantly higher in total credibility than did the

three other subgroups.

Conservatively Dressed Male Targets

Expertise. No significant interaction effects were

found among expertise scores for conservatively dressed

male targets. T-tests conducted between subgroups did

show significant differences: Younger females rated

conservatively dressed male targets significantly higher

in expertise (M = 35.73) than did older males (M = 32.51),

t(67.2) = -2.44, p < .02, and older females (M = 33.23),

t(89) = 2.23, p < .03 (Table 4-6).

Character. A significant two-way interaction between

age and sex was found on the character dimension of

conservatively dressed male targets, F(1, 144) = 8.53,

p = .004. Younger females (M = 30.27) rated

conservatively dressed males higher in character than did

younger males (M = 27.57), t(63.3) = -2.74, p < .01.

Older males (M = 28.39) gave significantly higher

character ratings than did older females (M = 25.07, t(75)

= 2.03, p < .05. Younger females also gave conservatively

dressed males significantly higher character scores (M =

30.27) than did older females (5 = 25.07), t(65.0) = 3.58,

E = .001 (see Table 4-6).












Table 4-5
Mean Credibility Dimension Scores for Female Targets
According to Dress Condition by Subject Age and Sex


Dimension




Expertise
SD

Character
SD

Dynamism
SD

Total
credibil:
SD


Male Subjects Female


Older Younger Older


Conservative Dressed Female Targets

34.16 32.74a 32.88b
6.0 5.6 7.0

29.39a 28.39b 26.15ac
7.0 4.0 7.7

18.97 17.87 17.24a
4.7 2.7 5.0


ity 79.97a 79.0 b 74.62c
17.7 9.5 16.9


Casually Dressed Female

Expertise 33.97* 33.04a
SD 6.7 5.7

Character 28.78a 28.74b
SD 5.4 4.4

Dynamism 18.22a 18.48b
SD 4.0 2.8

Total
credibility 77.82a 80.26bd
SD 15.8 10.1


Targets

31.98b
5.9

26.41c
7.3

17.80c
4.6


72.49cd
17.6


Subjects


Younger


36.02ab
4.7

31.08bc
5.5

19.44a
3.5


86.54abc
11.2


36.19ab*
3.9

31.31abc
5.3

20.27abc
2.9


87.77abc
9.2


Note: Means along dimensions with common subscripts
significantly differ at E < .05 level, two-tailed
probability.
*Means are different at the E = .057 level.











Dynamism. No significant interaction effects were

found among dynamism scores for conservatively dressed

male targets. T-tests run between subgroups found that

younger females gave significantly higher dynamism scores

(M = 20.54) to conservatively dressed male targets than

did older females (M = 18.22), t(72.3) = 2.33, R = .02.

Older males also gave significantly higher dynamism

ratings (M = 20.33) than did older females (M = 18.22),

t(69.9) = 2.09, p = .04 (see Table 4-6).

Total credibility. A significant main effect for age

was found among total credibility scores of conservatively

dressed male targets, F(1, 149) = 10.73, R = .001.

Subsequent t-tests revealed that younger females rated

these targets significantly higher in credibility

(M = 86.54) than did younger males (M = 81.22), t(69) =

-1.97, E = .05, older males (M = 77.49, t(62.1) = -2.81,

E < .01, and older females (M = 74.51), t(73.4) = 4.07,

R < .001. Also, younger males (M = 81.22) gave

significantly higher ratings than did older females (M =

74.51), t(63.4) = 2.11, R < .04 (see Table 4-6). In sum,

younger female subjects rated conservatively dressed male

targets significantly higher in total credibility than did

the other three subgroups.












Casually Dressed Male Targets

No significant interaction effects were found for any

of the credibility dimensions for casually dressed male

targets. Subjects seemed to rate male sources in casual

dress similarly on expertise, character, and dynamism.

However, a t-test run between older females and younger

females revealed that younger females gave significantly

higher total credibility scores (M = 85.67) to casually

dressed males than did older females (M = 77.86), t(67.4)

= 2.45, E < .02. Casually dressed male targets seemed to

be equally credible to both younger and older men. A

summary comparison of mean scores for all three dimensions

and total credibility of these targets is found in Table

4-6.

Summary of Research Question Findings

In testing for differences between age groups and sex

of subjects, several trends seemed be evident in this

experiment. Significant interactions between age and sex

were found for most of the analyses done on credibility of

female target sources. In none of the analyses of both

male and female targets' credibility scores did older and

younger men differ from each other in their ratings of

targets.

Younger females gave significantly higher ratings

than did the other groups on all dimensions in separate












analyses of casually and conservatively dressed females,

on character and total credibility of conservatively

dressed male targets, on all dimensions of female targets

in general, and on character and total credibility of male

targets in general. Older males and females, those in the

35 and over category, generally did not differ

significantly in their opinions.

The most recurrent and outstanding pattern of

significant differences involving the age and sex

variables was found for older and younger females. In

almost every comparison of mean dimension scores, younger

females gave significantly higher ratings than did older

females. This was most notably evident in ratings of

women sources dressed casually. Younger females seemed to

prefer casually dressed female targets more than the other

groups, in that they consistently gave these targets

significantly higher ratings of expertise, character, and

dynamism. Older females, on the other hand, seemed to

like casually dressed female sources less than did younger

females.

What may be most interesting is that younger females

gave significantly higher ratings than the other three

sample subgroups for casually dressed females, female

targets in general (regardless of dress), and total

credibility of all targets except for casually dressed











Table 4-6
Mean Credibility Dimension Scores for Male Targets
According to Dress Condition by Subject Age and Sex



Male Subjects Female Subjects


Dimension Older Younger Older Younger


Conservatively Dressed Male Targets

Expertise 32.51a 33.48 33.23b 35.73ab
SD 6.9 5.0 5.7 5.0

Character 28.39a 27.57b 25.07ac 30.27bc
SD 6.0 3.2 8.0 5.0

Dynamism 20.33a 20.17 18.22ab 20.54b
SD 3.5 3.1 5.3 3.9

Total
credibility 77.49a 81.22bd 74.51cd 86.54abc
SD 17.4 9.5 16.3 11.2



Casually Dressed Male Targets

Expertise 34.56 33.3 33.88 35.27
SD 6.0 5.9 6.0 5.3

Character 30.19 28.78 28.0 29.79
SD 7.5 4.0 7.8 5.0

Dynamism 19.57 19.70 19.02 20.60
SD 3.4 3.6 6.2 3.6

Total
credibility 81.77 81.78 77.86a 85.67a
SD 17.3 11.5 18.1 10.9


Note: Means along dimensions with
significantly differ at p <
probability.


common subscripts
.05 level, two-tailed











males. It could be concluded that among the subjects,

those who were college-aged women tended to give higher

ratings to all targets, especially those female targets in

casual dress, though these ratings were not significantly

higher in all instances.

Do males and females and older and younger people

differ in the way they rate credibility of sources? Based

on the results presented above, the answer would be a

resounding "yes." While men tended to behave similarly in

their rating of target sources, the most prominent

findings show that older and younger females differed in

the way they rated credibility of differently dressed

targets, regardless of target sex or dress condition.

Credibility of Dummy Sources

The final experiment's focus was on male and female

target sources who appeared in different modes of dress.

Dummy sources, those who appeared in neutral dress, were

placed before each target to serve as distractions. In

order to provide a thorough examination of the effects

that a source's sex and appearance may have on how others

rate his or her credibility, several analyses were

conducted concerning the dummy sources appearing in the

videotape stimulus.

First, Hypothesis One was again tested to see if male

and female dummy sources were rated differently in terms












of credibility. The effects of subject age and sex on

ratings of these sources were also tested through two-way

analyses of variance using the multiple analysis of

variance (MANOVA) procedure conducted for male and female

dummy sources' expertise, character, dynamism, and total

credibility. T-tests again were used to probe differences

between cell means.

Male vs. Female Dummy Sources

Composite female dummy scores by sex were obtained by

combining both female dummies' credibility scores, and the

scores of male dummies. T-tests conducted between male

and female dummy means on the three credibility dimensions

and total credibility scores uncovered no significant

differences. All subjects rated male and female dummy

sources similarly on expertise, character, dynamism, and

total credibility (see Table 4-7).

Female Dummy Sources

Expertise. No significant interaction or main

effects were found for subject age or sex for the

expertise dimension of female dummy sources. Subsequent

t-tests did reveal that scores given by young females to

female dummy sources (M = 34.56) were significantly higher

than those given by younger males (M = 32.43), t(69) =

-2.01, R < .05, and older females (M = 32.02), t(89) =

2.73, E < .01 (see Table 4-8).












Character. A significant two-way interaction was

found between subject age and sex for the character

dimension of female dummy sources at the E < .05 level,

F(l, 144) = 4.09. Younger females (M = 30.63) gave

significantly higher character scores to female dummy

sources than did younger males (M = 28.17), t(69) = -2.26,

p = .03), and older females (M = 26.66), t(58.5) = 2.81,

E < .01 (see Table 4-8).

Dynamism. No significant interaction or main effects

were found among dynamism scores of female dummy sources.

Subsequent t-test probing failed to reveal any differences

between subgroups (see Table 4-8).

Total credibility. A significant main effect for

subject age was found among total credibility scores of

female dummy sources, F(1, 149) = 6.35, E = .01. Younger

females (M = 82.44) gave significantly higher total

credibility scores to female dummies than did younger

males (M = 77.63), t(69) = -2.02, E < .05, older males (M

= 75.32), t(62.4) = -2.46, p < .02, and older females (M =

73.19), t(66.5) = 3.09, E = .003 (see Table 4-8).

Male Dummy Sources

Expertise. A significant main effect for age was

found on the expertise dimension of male dummy sources,

F(l, 149) = 3.77, E = .05. The only two subgroups to

differ in their ratings were older and younger females.











Table 4-7
Mean Credibility Dimension Scores for Male and Female
Dummy Sources



Dummy Sources


Credibility Dimension Males Females



Expertise 32.887 32.891
SD 5.1 4.9

Character 28.09 28.71
SD 5.2 5.7

Dynamism 17.13 17.44
SD 3.0 3.4

Total Credibility 76.39 77.05
SD 12.5 14.1




Younger females (M = 34.41) rated male dummies

significantly higher on expertise than did older females

(M = 31.4), t(89) = 3.05, p = .003 (see Table 4-8).

Character. A two-way analysis of variance revealed

significant interaction effects for subject age and sex on

the character dimension of male dummy sources,

F(l, 145) = 8.0, p = .005. Subsequent t-tests revealed

differences within age groups (see Table 4-8). Older

males (M = 28.51) gave significantly higher character

ratings to male dummies than did older females (M =

25.46), t(76) = 2.17, p = .03. Younger females (M =











29.69) gave significantly higher character ratings than

did younger males (M = 27.61), t(69) = -1.97, p = .05).

Older and younger females differed in their opinions of

male dummy sources. Younger females (M = 29.69) gave

significantly higher character ratings to male dummies

than did older females (M = 25.46), t(87), = 3.85,

E < .001 (see Table 4-8).

Dynamism. No significant interaction or main effects

were found for dynamism of male dummies. No significant

differences between subgroups were revealed by subsequent

t-tests.

Total credibility. A two-way interaction between

subject age and sex on total credibility of male dummies

was found at the p < .05 level, F(1, 149) = 4.04. Younger

females (M = 81.28) gave significantly higher total

credibility scores to male dummies than did older females

(M = 71.31), t(89) = 4.3, E < .001. Older females (M =

71.31) gave significantly lower ratings to male dummies

than did younger males (M = 77.57), t(64.0) = 2.56,

R = .01 (see Table 4-8). In short, older females rated

male dummies significantly lower in overall credibility

than did younger people of both sexes.

Summary of Dummy Source Findings

Results for these sources generally reflected those

found for target sources. The credibility ratings given









79

by the total sample to the two male dummy sources combined

did not significantly differ from the ratings given to the

two female dummy sources combined. In terms of Hypothesis

One, then, male dummies were not rated significantly

higher in credibility than were female dummies.

Patterns of significant differences between sample

subgroups based on age and sex were similar to those found

among ratings of target sources. Significant differences

between mean credibility scores given by younger and older

females were found for expertise, character, and total

credibility of female dummies.

Younger and older females also differed in their

ratings of male dummies' expertise, character, and total

credibility. Ratings of male and female dummy source

dynamism given by older and younger females were not

significantly different.

Several instances of differences between younger

females and younger males were also found for ratings of

female target sources, and on character scores of male

dummies. When compared to both older females and younger

males, younger females gave significantly higher scores.

Indeed, across all analyses of dummy sources, younger

females tended to give the highest ratings.

It should also be noted that while older and younger

males did not differ in their ratings of male and female











target sources, the same was true for their ratings of

male and female dummy sources. No significant differences

in ratings given by older and younger male subjects were

revealed in any of the analyses involving the sources

appearing in the videotape stimulus.

Additional Analyses

Because the two groups which comprised the final

experimental sample came from different sources, one being

composed mainly of college students, the other of adults

having different education levels, there was a concern

that education level may have been a confounding factor in

the interactions found between subject subgroups. To

investigate the effect of education level of subjects on

their ratings of target sources, one-way analysis of

variance tests were conducted among both male and female

subjects in the civic organization sample (n = 89) for

targets in different dress conditions based on education

levels. This was basically a repetition of the research

question analyses, with education level substituted for

age and sex.

Four levels of education were created from the six

possible responses on the questionnaire: (1) completion

of high school, (2) completion of some college and

vocational or trade school, (3) completion of a bachelor's

degree, and (4) formal education beyond college or












Table 4-8
Mean Credibility Dimension Scores for Male and Female
Dummy Sources by Subject Age and Sex



Male Subjects Female Subjects


Dimension Older Younger Older Younger


Male Dummy Sources

Expertise 32.80 33.04 31.40a 34.41a
SD 5.9 4.4 5.3 4.1

Character 28.51a 27.61b 25.46ac 29.69bc
SD 6.5 3.4 5.9 4.5

Dynamism 17.58 16.91 16.61 17.27
SD 3.0 2.4 3.7 2.6

Total
credibility 75.71 77.57a 71.31ab 81.28b
SD 15.8 6.9 12.9 9.0



Female Dummy Sources

Expertise 32.65 32.44a 32.02b 34.56ab
SD 5.5 4.2 4.7 4.2

Character 28.31 28.17a 26.66b 30.63ab
SD 6.1 4.0 7.9 4.4

Dynamism 17.61 17.02 17.13 17.48
SD 3.3 2.4 4.2 3.5

Total
credibility 75.32a 77.63b 73.19c 82.44abc
SD 15.6 7.6 17.2 10.1



Note: Means along dimensions with common subscripts
significantly differ at R < .05 level, two-tailed
probability.











completion of a graduate degree. One-way analyses of

variance comparing subjects in the four education levels,

regardless of age group or sex revealed no significant

differences in the way they rated target sources. Thus,

it was concluded that education level was probably not the

main source of variance among older and younger subjects

in the combined sample of 165 subjects.

The relationship between younger and older female

subjects borne out by the recurring pattern of significant

differences between these groups posed some concern to the

researcher. Why did younger females consistently give

significantly higher ratings to target sources than did

older females? And why did these two groups differ so

much in their opinions of sources while older and younger

males did not significantly differ at all? To try and

answer these questions, additional analyses concerning the

women in the civic organization sample were conducted. It

was thought that since most of the younger women subjects

were from the college-student sample, perhaps the mere

fact that more younger female subjects were in college

could account for the findings.

To see if younger and older women from the civic

organization sample also differed in their ratings of

target sources, women from the Orlando sample were put

into two categories: younger (18-44) and older (45 and











over). The younger category was extended to included

those women aged 35 to 44 because so few women were

between the ages of 18 and 34 (n = 8). There were only 10

women between 18 and 44, while 35 were 45 years old or

older. T-tests were run between the two age groups to see

if younger women from this sample differed from older

women on the same dependent variables based on target sex

and dress condition used in answering the research

questions.

No significant differences were revealed between

scores given by younger women and those given by older

women, except on the character dimension of conservatively

dressed males. The older women in the Orlando sample gave

significantly lower ratings of character to conservatively

dressed male sources (M = 23.73) than did the younger

women (M = 27.0), t(40.6) = 2.12, E = .04. However,

scores given by older women, those 45 and older, were

lower than those given by women in the 18-44 age group for

23 of the 32 comparisons made. Though these differences

were not significant, they did not contradict the trend

found in the results which showed younger females giving

higher ratings than did older females.















CHAPTER 6
DISCUSSION


This study investigated the relationship between

variations of a communicator's appearance and perceptions

of his or her credibility, and how a message receiver's

age and sex may affect those perceptions. This was done

by creating videotaped interviews in which male and female

sources, posing as college professors, appeared in casual

and conservative modes of dress. After viewing these

sources, subjects were asked to rate their credibility in

terms of expertise, character, and dynamism.

General Implications

Male target sources were predicted to receive

significantly higher ratings of credibility than female

target sources. This prediction was not wholly supported

by the findings: Males were not rated significantly

higher in expertise, character, or total credibility than

were females, but males were rated significantly more

dynamic than females.

One explanation for this finding is that the female

sources were less dynamic in their performances than the

men. Specifically, one of the female targets was rated

significantly lower in dynamism than both male targets,

84











regardless of dress condition. The same target was also

rated significantly lower in dynamism than her female

counterpart when both were dressed conservatively. This

illustrates a potential disadvantage of using

nonprofessional actors in an experimental stimulus.

Overall, however, the female target sources were

considered to be less dynamic than the male target

sources. Dynamism, as measured in this study, included

characteristics such as aggressiveness and boldness--

traits which have traditionally been regarded as more

appropriate for men than for women (Richardson, 1988).

This general societal expectation for men to be more

aggressive and bold may have influenced the way subjects

rated the sexes on this dimension. That is, subjects may

have expected the male sources to be higher in dynamism

than the female sources and rated them as such, thus

creating a self-fulfilling prophecy. Researchers of

future studies could further explore this aspect of sex

differences in source perception.

While the performances by the female targets in this

study were rated less dynamic than those given by the men,

of more importance concerning the societal implications of

the findings for Hypothesis One is that both sexes were

considered by subjects to be equal in expertise and

character. Previous research has demonstrated that









86

competent women are more likely to be devalued in serious,

realistic contexts and when evaluations of them are

salient to raters, such as in hiring situations (Lott,

1985). In this study, subjects watched a television

program and rated the people they saw; there were no

interpersonal contacts with the actors and no real

consequences to the subjects as a result of their ratings.

Hence, it may have been that women sources were not rated

as being different in credibility than were men because

there were no serious consequences for the subjects.

Another possible explanation for the lack of

differences between male and female sources is social

desirability. The people in this study may have wanted to

rate the male and female sources equally because they did

not want to appear sexist. An awareness of "politically

correct" attitudes toward women may have influenced some

people's opinions, especially those in college. In

addition, the fact that the researcher was a woman may

have sensitized the subjects enough for them to rate both

sexes equally.

While these explanations are all plausible, what the

results may truly imply is that the stereotype of women as

being less competent than men is changing. In this study,

equal credentials yielded equal evaluations; the women

professors and men professors were rated the same in terms









87

of their expertise and character. What is most telling is

that women were thought to be just as intelligent, bright,

informed, trained, competent, and expert as men.

The general societal bias against women which was the

basis for the first hypothesis did not appear here.

Neither did the expectation that differences in the

clothing of female sources would lead to differences in

their credibility. It was predicted that when female

sources were dressed in casual clothing (a knit top and

slacks) rather than in conservative clothing (a suit and

blouse), they would be rated significantly lower in

credibility.

This was not supported by the findings: Women were

rated equally credible whether dressed casually or

conservatively. The "wardrobe engineering" as propagated

in books which provide women with fashion advice as to

what they should wear in order to be successful is called

into question by the results of this study. Molloy (1977)

asserts that, among other types of clothes, "sweaters,

slacks, skirt and blouse outfits...all announce that you

have no ambition" (p. 125). The "dress for success" look,

conservative suits and accessories such as collars that

tie, which was prescribed to women who wanted to look

ambitious and competent, did not significantly enhance

subjects' ratings of women in this experiment.











The importance of appearance as a factor of source

credibility was also predicted to be more salient to

female sources than to male sources. It was hypothesized

that differences in ratings of expertise, character, and

dynamism between dress conditions of women would be larger

than those of men. This was not borne out by the results.

In fact, for the character dimension, men were rated

higher in character when they wore casual clothing (a

shirt and jeans) than when they wore conservative clothing

(a suit and tie).

Upon closer examination, this difference in character

ratings was found to occur mainly among the older people

in the study. This anti-suit-and-tie attitude on the part

of older subjects, those 35 and older, may have been a

function of the topic presented--marine biology--and of

the identification of the experts as being professors of

zoology. Perhaps older people had a stereotype of college

professors who studied marine biology. It may be that

they trusted sources who better fit the image of an expert

in marine biology--someone who dressed as if he actually

worked with marine animals or out of doors. A suit and

tie may not have fit that image, and subsequently male

targets dressed that way may not have been rated higher on

character.











Another possible explanation for this is that the

business suit, at least for men, carries with it certain

connotations, such as authority and bureaucracy. The suit

and tie is the acceptable form of dress for professions

involving politics, government, banking, and the like. On

the other hand, casual clothes such as blue jeans do not

imply such meanings. At least for the older people in

this study, men dressed in jeans seemed more honest,

trustworthy, sympathetic, and virtuous--all indications of

higher character--than those dressed in suits. This

implies a negative stereotype of the business suit for

men, at least in terms of the character of the wearer.

Future studies into the meanings associated with such

clothing would provide information concerning reasons for

the distrust of suits implied by these results.

The findings for the hypotheses show that at least

for experts who are college professors of zoology, dress

does not greatly affect the credibility of sources who are

both seen and heard. The context within which the

stimulus interviews were conducted may be related to

practices used in television news programming. The

sources were seen on television and were identified as

being professors of zoology by a graphic. News coverage

by television reporters often includes soundbites, or

interviews, with people who are deemed experts on the











particular topic of a story. Often, these experts are

interviewed at their place of work. College professors,

scientists, and other similar experts may appear in such

interviews in their everyday work clothes--and this often

does not include formal attire. Subjects may have

expected the experts who appeared in the interviews to

wear clothes they would wear to work. With the exception

of the differences in character ratings for differently

dressed men, it may be that the experts in this study were

so fluent in their presentations that their dress was not

distracting. That is, the subjects focused more on what

the experts were saying than on what the experts were

wearing. Therefore, the effects of different dress styles

were not as great as predicted, especially for women

sources.

Several noteworthy trends were found regarding the

influence of subjects' age and sex on their ratings of

sources. Female subjects between the ages of 18 and 34

tended to give high credibility ratings to all targets,

regardless of targets' sex or appearance. Younger females

differed from younger males in many of their ratings of

sources, with younger females giving significantly higher

ratings. This supports previous research which showed

that college-age women tended to give more generous











ratings of expertise, character, and dynamism to others

than did college-age men (Carocci, 1988).

Significant differences between subjects were most

pronounced for total credibility scores of female targets,

and for casually dressed females in particular. Younger

females seemed to like and gave significantly higher

ratings to women dressed casually than did older females

and older and younger males. This may be an indication of

a pro-female bias on the part of younger women, in that

they may be more favorable to professional women and may

even consider them to be role models. Richardson (1988)

asserts that there is a lack of female professors to serve

as role models for women in college. It may be, too, that

college women have higher regard for women professors and

identify more with them than do college men.

In contrast to younger women, older women tended to

give the lowest credibility ratings. This trend was

especially pronounced among ratings given by older females

to female targets. Perhaps the older women--away from a

campus atmosphere--were less concerned about appearing

sexist or "politically correct" than were the younger

women who might have been exposed to such concerns

regularly on campus.

The finding that older females gave low ratings to

casually dressed females compared to younger females




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs