Body Image Disturbance, Objectification, and College Women

MISSING IMAGE

Material Information

Title:
Body Image Disturbance, Objectification, and College Women Comparing Collegiate Female Athletes and Non-Athletes
Physical Description:
1 online resource (167 p.)
Language:
english
Creator:
Varnes, Julia Rae
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Health and Human Performance, Health Education and Behavior
Committee Chair:
Stellefson, Michael L
Committee Members:
Dodd, Virginia Jones
Pigg, Robert M, Jr
Janelle, Christopher M
Miller, David

Subjects

Subjects / Keywords:
athlete -- body -- college -- female -- image -- objectification
Health Education and Behavior -- Dissertations, Academic -- UF
Genre:
Health and Human Performance thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Body image disturbance (BID) is highly prevalent among college-aged women, so much that researchers have described it as normative. Prior to 2001, research suggested that athletic participation served to decrease BID among females. Since then, female athletes have become increasingly sexually objectified in the media, and the sociocultural beauty ideal has shifted to emphasize appearing both athletic and thin. The present work used objectification theory to progress BID research in the college female population by investigating the current status of BID, and its measurement, among college-aged females, with an emphasis on identifying how participation in collegiate athletics might promote or protect from BID and self-objectification. These investigations extend the literature and attempt to address limitations of previous research. Chapters 1 and 2 provide a foundation for the research presented in Chapters 3 and 4. First, the potential impact of increased sexualization of the female athlete on the athlete’s BID and self-objectification is explored. Second, the research identifies salient and temporally relevant measures of BID for research in college female populations and college female athlete populations. Overall findings support that female participation in collegiate athletics continues to provide protection from BID. Additionally, athletic participation appears to plays a role in decreasing self-objectification tendencies; however, this may depend on the sport. These differences indicate that competitive athletes participating in sports that are more objectified by society could be at greater risk for certain psychological outcomes and health behaviors than athletes who are less objectified by society. However, more research needs to be done to understand the true mechanism of this difference. Through the work presented here, it appears that our understanding of the relationship between the objectification theory constructs is deficient when it comes to the female athlete population, and specifically athletes participating in more objectified sports. This line of inquiry provides a foundation for objectification research in female athlete populations, and provides recommendations for next steps.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Julia Rae Varnes.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: Stellefson, Michael L.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2014-08-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045691:00001


This item is only available as the following downloads:


Full Text

PAGE 1

BODY IMAGE DISTURBANCE OBJECTIFICATION, AND COLLEGE WOMEN: COMPARING COLLEGIATE FE MALE ATHLETES AND NONATHLETES By JULIA RAE VARNES 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 2013 1

PAGE 2

2013 Julia Rae Varnes 2

PAGE 3

To my wonderful parents Paul and Jill Varnes Thank you for your support, advice, and confidence. I love you. In memoriam of Kathryn Carson, 19792012. 3

PAGE 4

ACKNOWLEDGMENTS I could not have done this without such a strong network of support. There are many people who have been important in this process so many that it would take another dissertation to thank them all! At the top of the list ar e my parents who provided me with unwavering support, encouragement, wisdom, and listening ears; as well as coming to my rescue on uncountable occasions to help with childcare! I am grateful to my children, Austin and Grayson, who have endured this process with me and whose hugs and kisses made everything better I thank my committee members for their continued guidance and support I am grateful to my committee chair, Dr. Michael Stellefson, for his time and energy; and for pushing me and challenging me to move beyond my own perceived limits. I am appreciative of Dr. Virginia Dodd for her support, humor, wisdom, and negotiation skills. I am thankful for Dr. Chris Janelle for our helping me to understand a different side of the content ; and for encouraging me to search deeper into the literature. This helped me to learn and understand more than I ever expected. I am indebted to Dr. David Miller who patiently answer ed my questions and put up with my long emails, especially th e ones I followed up with, Never mind. I think I got it! I am especially thankful to Dr. Morgan Pigg who stepped up when I needed him. I am so happy to have him finish this process with me as he has been with me throughout my college education (and befor e!), help ing me grow and always giving me guidance when I asked and encouragement when I needed it most. I thank Dr. Steve Dorman, who took me in w hen I returned to the department to pursue my doctorate I t was Dr. Dormans initial guidance, support, and mentorship that made me feel like I was coming home. I also thank Lynda Tealer, whose help was integral for my final study. I also wish to thank those who helped me throughout my academic career. There are several, but I especially want to mention Dr. Bo b Gold for his early mentorship and 4

PAGE 5

encouragement; and Dr. Bill Chen for giving me my first opportunity to teach as a Masters student. It was an experience that helped me realize my calling. I thank my friends and peers for helping me to maintain my menta l health! Especially Mandi, Bethany, and Holly for being there when I needed it for long lunches, for girls night outs and for long conversations about life and statistics (Mandi!). Finally, I thank my wonderful Edward Arp for being my cheerleader and my motivator; for unwavering support and help (with the kids, the yard, the dishes, the laundry, the dog, and a lot more!) ; for listening to me read my papers, vent my frustrations, and think out loud. I am grateful for every word, every moment, every look, everything. 5

PAGE 6

TABLE OF CONTENTS ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF TABLES ...........................................................................................................................9 LIST OF FIGURES .......................................................................................................................10 LIST OF ABBREVIATIONS ........................................................................................................11 ABSTRACT ...................................................................................................................................12 CHAPTER 1 INTRODUCTION ..................................................................................................................14 General Introduction ...............................................................................................................14 The Sociocultural Model ........................................................................................................16 Feminist Theories ............................................................................................................17 Social Comparison Theory ..............................................................................................21 Prominent Measures of BID ...................................................................................................23 Objectified Body Consciousness Scale ...........................................................................23 Sociocultural Attitudes Towards Appearance Scale .......................................................25 Eating Disorder I nventory ...............................................................................................26 Summary of BID Instruments .........................................................................................27 Prevalence of Body Image Disturbance .................................................................................28 BID, Objectification, and the Female Athlete ........................................................................29 Rationale and Organization of the Present Studies .................................................................33 2 A SYSTE MATIC REVIEW OF STUDIES COMPARING BODY IMAGE CONCERNS AMONG FEMALE COLLEGE ATHLETES AND NONATHLETES, 19972012 ..............................................................................................................................40 Part I: Sociocultural Changes and the Female Athlete ...........................................................44 Sexual Objectification of Female Athletes ......................................................................44 The New Ideal .................................................................................................................48 Part II: Systematic Re view of Comparative Studies ...............................................................49 Method ....................................................................................................................................49 Search Procedures ...........................................................................................................50 Re sults .....................................................................................................................................52 Description of Studies Included in the Review ...............................................................52 Differences in BIC between Athletes and Non Athletes .................................................55 Relationship between BIC and Sport Type .....................................................................56 Relationship between BIC and Competition Level .........................................................57 Discussion ...............................................................................................................................58 The Athletic Physique .....................................................................................................60 Competition Level Differences .......................................................................................61 Sport Differences .............................................................................................................62 6

PAGE 7

Limitations .......................................................................................................................64 Conclusions and Future Research ...................................................................................65 3 VALIDITY OF THE EATING DISORDER INVENTORY FOR NONCLINICAL RESEARCH IN POPULATIONS OF WESTERNIZED WOMEN: A CRITICAL REVIEW OF FACTOR ANALYTIC STUDIES ...................................................................71 Methods ..................................................................................................................................74 Search Process (Phase 1) .................................................................................................74 Evaluation of Study Methods (Phase 2) ..........................................................................75 CFA Evaluation ...............................................................................................................79 Results .....................................................................................................................................81 Eligible Studies (Phase 1) ................................................................................................81 Methodological Quality and Results of Reviewed Studies (Phase 2) .............................81 Synthesis (Phase 3) ..........................................................................................................87 Discussion ...............................................................................................................................88 Low Self Esteem .............................................................................................................88 Body Dissatisfaction ........................................................................................................90 Interpersonal Distrust ......................................................................................................91 Perfectionism ...................................................................................................................92 Eating Disorder Risk .......................................................................................................93 Removed Items ................................................................................................................94 Limitations .......................................................................................................................96 Conclusion ..............................................................................................................................97 4 BODY ESTEEM AND SELF OBJECTIFICATION: USING OBJECTIFI CATION THEORY TO COMPARE COLLEGIATE FEMALE ATHLETES AND NONATHLETES ..........................................................................................................................106 Overview of Objectification Theory .............................................................................108 Sexual Objectifi cation of the Female Athlete Varies by Sport .....................................110 The Present Study ..........................................................................................................111 Methods ................................................................................................................................113 Recruitment and Procedure ...........................................................................................113 Participants ....................................................................................................................117 Measures ........................................................................................................................118 Data Analysis .................................................................................................................121 Results ...................................................................................................................................122 Preliminary Analyses .....................................................................................................122 Testing Hypothesis 1 (Multivariate Analysis) ...............................................................123 Testing Hypotheses 2, 3, & 4 (Pairwise Comparisons) .................................................125 Discussion .............................................................................................................................126 Strengths and Limitations ..............................................................................................130 Conclusion and Next Steps ............................................................................................131 5 SUMM ARY, CONCLUSIONS, AND RECOMME N DATIONS .......................................137 7

PAGE 8

Summary ...............................................................................................................................137 Conclusion ............................................................................................................................141 Recommendations for Researchers .......................................................................................143 APPENDIX A DEMOGRAPHIC AND BACKGROUND QUESTIONS ...................................................145 B OBJECTIFIED BODY CONSCIOUSNES SCAL E ............................................................147 C BODY ESTEEM SCALE .....................................................................................................149 D INTERNALIZATION GENERAL SUBSCALE OF THE SOCIOCULTURAL ATTITUDES TOWARDS APPEARANCE QUESTIONNAIRE 3 ....................................151 LIST OF REFERENCES .............................................................................................................152 BIOGRAPHICAL SKETCH .......................................................................................................167 8

PAGE 9

LIST OF TABLES Table page 11 Statistically significant correlations between the subscales of the Objectified Body Consciousness Scale and referent measures ......................................................................36 12 Statistically significant correlations between the Body Esteem Scale and referent measures .............................................................................................................................37 21 Characteristics of reviewed articles ...................................................................................68 22 Included sports in the reviewed studies alongside Parsons and Betz (2001) objectification scores .........................................................................................................70 31 EDI, EDI 2, and EDI 3 subscale correlations and item assignmen t (Garner 2004) ..........99 32 Studies retained for EDI review .......................................................................................100 33 Methodological evaluation of retained EDI factor analytic stu dies .................................101 34 Factor structure results .....................................................................................................102 35 Proposed EDI structure for nonclinical populations .......................................................104 41 Demographic counts and BMI by group ..........................................................................134 42 Means and standard deviations by group for body esteem ..............................................134 43 F to remove results for body esteem ...............................................................................135 44 Functions at group centroids for the three groups on body esteem data ..........................135 45 Structure rs for body esteem ...........................................................................................135 46 Mean and standard deviation for each group as a function of outcome; multiple comparisons, and eta squared values ...............................................................................135 9

PAGE 10

LIST OF FIGURES Figure page 11 Model of key objectification theory tenets (Szymanski et al., 2011) ................................38 12 Amended Objectification Theory Model. Note. For parsimony, where indirect chains of relations are depicted between two constructs, possible additional direct paths are not included. (Moradi, 2010). ............................................................................................39 31 Flow chart of search and screen process ..........................................................................105 41 Linear discriminant functions for the body esteem constructs of physical condition, sexual attractiveness, and weight concern .......................................................................136 10

PAGE 11

LIST OF ABBREVIATION S BES Body Esteem Scale BID Body Image Disturbance BIC Body Image Concerns EDI Eating Disorder Inventory EDI BD Eating Disorder Inventory Body dissatisfaction subscale EDI DT Eati ng Disorder Inventory Drive for thinness subscale EDR Eating Disorder Risk EWR Eating and Weight Related LO S Less Objectified Sports MO S More Objectified Sports OBC Objectified Body Consciousness OBCS Objectified Body Consciousness Scale OBCT Objectified Body Consciousness Theory SATAQ Sociocultural Attitudes Towards Appearance Questionnaire SPA Social Physique Anxiety 11

PAGE 12

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requ irements for the Degree of Doctor of Philosophy BODY IMAGE DISTURBANCE, OBJECTIFICATION, AND COLLEGE WOMEN: COMPARING COLLEGIATE FEMALE ATHLETES AND NONATHLETES By Julia Rae Varnes August 2013 Chair: Michael L. Stellefson Major: Health and Human Performance Body image disturbance ( BID ) is highly prevalent among college aged women, so much that researchers ha ve described it as normative Prior to 2001, research suggested that athletic participation served to decrease BID among females. Since then, female athletes have become increasingly sexually objectified in the media, and the sociocultural beauty ideal has shifted to emphasize appearing both athletic and thin The present work used objectification theory to progress BID research in the college femal e population by investigating the current status of BID, and its measurement, among college aged females with an emphasis on identifying how participation in collegiate athletics might promote or protect from BID and self objectification The se investigat ions extend the literature and attempt to address limitations of previous research. Chapters 1 and 2 provide a foundation for the research presented in Chapter s 3 and 4. First, the potential impact of increased sexualization of the female athlete on the at hletes BID and self objectification is explored. Second, the research identif ies salient and temporally relevant measures of BID for research in college female populations and college female athlete populations Overall f indings support that female parti cipation in collegiate athletics continues to provide protection from BID Additionally, athletic participation appears to plays a role in 12

PAGE 13

decreasing self objectification tendencies; however, this may depend on the sport. These differences indicate that co mpetitive athletes participating in sports that are more objectified by society could be at greater risk for certain psychological outcomes and health behaviors than athletes who are less objectified by society. However, more research needs to be done to understand the true mechanism of this difference. Through the work presented here, it appears that our understanding of the relationship between the objectification theory constructs is deficient when it comes to the female athlete population, and specifically athletes participating in more objectified sports This line of inquiry provides a foundation for objectification research in female athlete populations, and provides recommendations for next steps 13

PAGE 14

CHAPTER 1 INTRODUCTION Image is everything. A ndre Agassi t ennis player General Introduction The concept of body image has been a topic of exploration and research in the academic literature for decades; though the psychological, behavioral, and emotional effects of body image have been noted as far back a s the early 1800s (Brumberg, 1997). Despite the endurance of the body image concept it remains a complex issue that has been difficult to define and understand. While what is known about the psychological and emotional repercussions of poor body image hav e varied very little over time the behaviors associated with attempting to change the body to meet a self perceived ideal appear to change frequently according to cultural ideals Researchers have adopted Cash and Szymanskis (1995) definition of body i mage as a multidimensional construct encompassing how one feels, thinks, and acts regarding their body. There are two distinct modalities of body image ( Cash & Szymanski, 1995) : 1) perceptual : having to do with body size or body part size estimation; and 2) attitudinal : including body image evaluation (i.e., emotions and feelings regarding one s appearance) and investment (i.e., degree to which an individual deems physical attributes as important ) Due to the dynamic, broad nature of the body image conce pt, t here are many constructs that can be included under this umbrella term ( Thompson, Heinberg, Altabe, & Tantleff Dunn, 1999). These include, but are not limited to: weight (dis)satisfaction, body (dis)satisfaction, appearance (dis)satisfaction, appearan ce evaluation, body esteem, body dysmorphia, and body schema. The constructs of body image are many; thus, measurement becomes difficult and is often done in a fragmentary way examining one or several constructs. E vidence of body size 14

PAGE 15

misperception and/or negative attitudes or feelings towards the body and its appearance or physical attributes is considered an indication of body image disturbance (BID). BID has been linked to several negative health behaviors and psychological concerns, including, but not limited to: disordered eating (Calogero, Davis, & Thompson, 2005; Grabe, Hyde, & Ward, 2008; Schwitzer, Ro driguez, Thomas, & Salimi, 2001), cigarette smoking (Turnbull, 1999), alcohol use ( Carr & Szymanski, 2010), depression (Ilaria, 2006; Schwitzer et al., 2001; Szymanski & Henning, 2007), overweight and obesity (Annis, Cash, & Hrabosky, 2004), decreased self esteem (Clay, Vignoles, & Dittmar, 2005; Fingeret & Gleaves, 2004; Tylka & Sabik, 2010), anxiety (Ilaria, 2006), stress (Ruggiero et al., 2008), decr eased flow in athletes (Dorland, 2006; Szymanski & Henning, 2007) and consideration of cosmetic surgery (Swami, 2009; Thompson et al., 1999) T he majority of body image literature focus es on college aged white females or patients with an eating disorder. This is justified, as BID is highly prevalent in these populations; and research indicates that general dissatisfaction with the bodys shape or size has even become normative in college populations (e.g., Espelage, Mazzeo, Aggen, Quittner, Sherman, & Thom pson, 2003; Grabe et al., 2008; Fitzsimmons Craft, 2011). The present work seeks to further the research in this area by investigating BID and its measurement in college aged females, with an emphasis on collegiate female athletes. Chapter 1 presents an overview on prominent theories that attempt to explain the etiology of BID among females These theories, which are included under the Sociocultural Model include Social Comparison Theory and Feminist Theories of Objectification Theory and Objectified Bod y Consciousness Theory. Chapter 1 also introduces the reader to the demographic prevalence of BID, factors related to BID, and information on how society and culture might affect BID among female athletes. 15

PAGE 16

The introduction presented in Chapter 1 informed the subsequent aim of Chapter 2, which i s to present a systematic review of literature from 1997 to 2012 that compar e s BID among collegiate female athletes and non athletes. Chapter 3 provides an in depth look at the psychometric properties of one of the m ost prominently used instruments in BID research, the E ating Disorder Inventory (EDI) Chapters 2 and 3 informed the objectives and methods of the research study repor ted in Chapter 4, which examines multidimensional body image, and indicators of self objectification in NCAA Division I female athletes and non athletes. Chapter 5 summarizes and synthesizes the work presented in Chapters 2 through 4, and provides recommendations for next steps The research studies reported in Chapters 2 through 4 were deve loped into manuscripts with the purpose of submi ssion for publication in scholarly journals. The Sociocultural Model Several theories have been proposed to account for BID. The bulk of these theories are based on the premise that society defines and proje cts a cultural ideal of feminine beauty (known as the sociocultural ideal), with an inherent expectation that this ideal is a goal to be attained by all women (Heinberg, 1996; Thompson et al., 1999). At the core of BID is the degree to which a woman internalizes (i.e., takes within and believes to be self defined) th is definition of beauty, currently described as the thin ideal (Allen, 2012; Cafri, Yamamiya, Brannick, & Thompson, 2005; Clay et al., 2005; Fingeret & Gleaves, 2004; Fitzsimmons Craft et al., 2 012; Fredrickson & Roberts, 1997; Glauert, Rhodes, Byrne, Fink, & Grammer, 2009; Grabe et al., 2008; Low et al., 2003; McKinley & H yde, 1996; Tsiantas & King, 2001). Internalization of sociocultural beauty ideals means that the individual believes their de sire to attain th e ideal body is a personal choice rather t han a result of social pressure (McKinley & Hyde, 1996). Internalization of the thin ideal is considered to be the most consistent and significant predictor of BID in women (Cafri et al., 2005; Gra be et al., 2008) and stands as the founda tion of the Sociocultural Model 16

PAGE 17

According to the sociocultural model internalization of the thin ideal means the individual not only believes that thin is indeed ideal, but also that thinness is associated with su ccess (Fitzsimmons Craft, 2011). The sociocultural model arose from early research indicating that eating pathology increased in prevalence over the same decades that Miss Americas and Playboy centerfolds became thinner (Altabe, 1996). For this reason, int ernalization of the thin ideal may be a causal factor in the etiology of eating disorders and BID (Altabe, 1996; Fitzsimmons Craft, 2011; Fitzsimmons Craft et al., 2012). Projections of the thin ideal are seen across media channels, including television, movies, magazines, and the Internet (Cusumano & Thompson, 1997; Grabe et al, 2008; Heinberg, 1996). There is strong evidence that societys perception of the thin ideal is partially defined, and significantly perpetuated, by the media (Altabe, 1996; Fitsimmons Craft, 2011; Heinberg, 1996), and that increased media exposure strengthens ones level of internalization (Clay et al., 2005; Grabe et al., 2008). I t is apparent, however, that individuals are not passive recipients of this pressure; but rather, exposure to, and thus internalization of, these pressures is self perpetuated (Cusumano & Thompson, 1997; Fitzsimmons Craft, 2011; Janelle, Hausenblas, Ellis, Coombes, & Duley, 2009; Mercurio & Rima, 2011). Self perpetuation occurs via a litany of social cognitive mechanisms that influence the various ways individuals perceive and interpret their social contexts (Fitzsimmons Craft, 2011). Social cognitive mechanisms relevant to college women can be further explained by feminist theories (e.g., Objectification Theor y and Objectified Body Consciousness Theory) and Social Comparison Theory. Feminist Theories At the most basic level, feminist theories assert that Westernized women are socialized to equate physical attractiveness with self worth (Heinberg, 1996). This socialization begins with 17

PAGE 18

the first indication of maleness or femaleness, and the exclamation of, Its a boy! or Its a girl! Feminists assert that, with that first gendered comment, inherent gender role expectations are projected onto the fetus, baby, or child. Undoubtedly, sex is a biological construct determined chromosomally; however, gender is a socially defined concept. Thus, the male and female bodies exist within social and cultural contexts, and are constructed by society as such (Fredricks on & Roberts, 1997). Feminist theories purport that society has historically oppressed women through its tendency to decontextualize women's experiences by focusing on the female appearance and diminishing nonfeminine successes (Busanich & McGannon, 2010) As a result of this gendered socialization, women over identify with their bodies, making their self value contingent on conforming to societal norms of beauty (Thompson et al., 1999). The idea of this gender role socialization and orientation is of sign ificance in the development of BID, particularly for women. An example of this phenomenon is portrayed in a study conducted by Mahalik and colleagues (2005). Through the development of the Conformity to Feminine Norms Inventory (CFNI) instrument, Mahalik e t al. (2005) identified eight qualities that women define as female gender roles. These gender roles include: developing friendly, supportive relationships (relational); pursuit of the thin ideal (thinness); not calling attention to ones talents and abili ties (modesty); maintaining the home (domesticity); take care of and be with children (care for children); investment of the self in a romantic relationship (romantic relationship); be sexually monogamous (sexual fidelity); and commitment to maintain and i mprove physical appearance (invest in appearance). Most notably, the thinness, relational, romantic relationship, and invest in appearance attributes are significantly related to BID constructs such as decreased body esteem (Steinfeldt, Zakrajsek, Carter, & Steinfeldt, 2011) drive for thinness, bulimia, body dissatisfaction, and lack of internal awareness (Mahalik et al., 18

PAGE 19

2005). Objectification Theory (Fredrickson & Roberts, 19997) and Objectified Body Consciousness Theory (McKinley & Hyde, 1996) further e xplain how gender role expectations and sociocultural norms and pressures influence BID among women. Objectification t heory Objectification theory is a feminist theory that posits that the societal pressures expressed through feminine gender norms and the sexual objectification of females contributes significantly to mental health problems (i.e., eating disorders, depression, sexual dysfunction, and substance abuse) among diverse groups of women (Fredrickson & Roberts, 1997; Szymanski, Moffit, & Carr, 2011) Sexual objectification occurs when a woman is recognized predominantly for her physical attributes (her body or body parts), and is viewed as a physical object of sexual desire (Szymanski et al. 2011). The experience of sexual objectification can be per sonally realized or observed as occurring to other women, as it is a sociocultural phenomenon (Fredrickson & Roberts, 1997) ; thus, the effect of sexual objectification on mental health risks can be direct and overt or indirect and subtle ( Szymanski et al., 2011). The direct path portends from personal sexual objectification experiences that range from being gazed at sexually to apparent sexual violence (Fredrickson & Roberts, p.177178, 1997). The indirect path involves observed or environmental sexual ob jectification experiences that result in self objectification (Kozee, Tylka, Augustus Horvath, & Denchik, 2007; Moradi, Dirks, & Matteson, 2005; Szymanski et al., 2011 ) This self objectification is manifested as persistent body surveillance (monitoring th e appearance of the body) and internalization of the thin ideal. Among women, higher levels of self objectification have been linked to disordered eating (Moradi et al. 2005; Noll & Fredrickson, 1998; Tiggeman & Kuring, 2004 ; Tylka & Hill, 2004), depressi on ( Carr & Szymanski, 2010; Ilaria, 2006; Impett, Henson, Breines, Schooler, & Tolman, 2011; Szymanski & Henning, 2007; Tiggemann & Kuring, 2004), poor self esteem 19

PAGE 20

(Impett et al., 2011; Tylka & Sabik, 2010), sexual dissatisfaction and dysfunction (Fredrickson & Roberts, 1997; Szymanski et al., 2011), and substance use ( Carr & Szymanski, 2010). The link between self objectification and these mental health risks are typically mediated by psychological consequences, s uch as appearance anxiety and body shame, r educed flow, diminished internal awareness, and anxiety about physical safety ( Calogero, 2004; Fredrickson & Roberts, 1997; McKinley & Hyde, 1996; Moradi, 2010; Szymanski et al., 2011; Tylka & Hill, 2004) Figure 1 1 provides the mediation model for object ification theory principles as designed by Szymanski and colleagues (2010) Figure 1 2 is a more detailed model developed by Moradi (2010). Objectified b ody c onsciousness t heory While objectification theory explains the foundation of objectification as a cultural phenomenon experienced on a public and private level (i.e., feminine gender stereotype and norms and s exual objectification ), objectified body consciousness theory ( OBC T) attempts to describe the interrelationships between three specific attribu tes that represent the construct of objectified body consciousness (OBC) The three components of OBC as defined by McKinley & Hyde (1996) include body surveillance body shame and appearance control beliefs Theoretically, higher levels of OBC are indi cative of greater risk for mental health problems OBCT purports that, to meet the thin ideal, it becomes necessary for women to self survey themselves from the perspective of others ( body surveillance ) This is done to ensure compliance with cultural body standards and thereby reduce the possibility of being negatively judged. While attempting to attain the sociocultural ideal, women compare themselves to an internalized ideal and strive to change their own body to meet this frame of reference. Inevitable failure can lead to feelings of guilt and lowered self esteem, and eventually body shame 20

PAGE 21

(McKinley & Hyde, 1996). The positive correlation between body shame and body surveillance has been supported by research ( r = .53 .66; McKinley & Hyde, 1996; Tylka & Hill, 2004). Also connected to body shame is the idea that a woman is responsible for how her body looks and that she possesses the ability to conform to cultural standards with sufficient effort. Since this idea of attainment becomes a choice, the thou ght that one can control their own appearance becomes instilled in their beliefs ( appearance control beliefs ). Theoretically, women who have higher appearance control beliefs feel more ashamed of their body, especially if they feel they have not attained t heir own personalized ideal. This positive relationship is also somewhat supported in past research findings ( r = .23, p < .05; McKinley & Hyde ; 1996). Social C omparison T heory Social comparison theory asserts that human beings are driven to assess their progress and status in life ; and, i n the absence of an objective measure to compare oneself individuals will evaluate their own self worth and abilities by comparing them to others (Festinger, 1954) Qualities lacking objective measures, such as physical attractiveness, are susceptible to being defined by the norms of others (e.g ., society or peers). A norm is generally described as what is typical, standard, or accepted, as it is a derivation of the word normal. There is evidence that an individuals perception of normal is defined and modified by what they are expose d to. For example, Glauert and colleagues (2009) found that mere exposure to thin images alters ones perception of a normal body size. In their study, participants were placed in one of t wo groups, and then were exposed to either fatter (Group 1) or thinner (Group 2) pictorial images. The results indicated that, regardless of group assignment, participants changed their personal definition of normal body size when exposed to the images Those exposed to fatter images changed their definition of normal body size to be larger than originally stated, while those 21

PAGE 22

exposed to thinner images changed their definition of normal body size to be smaller than originally stated. These findings are concerning considering that the projected media images of women have continued to shrink over the past forty plus y ears (Cafri et al., 2005) When media figures (e.g., models, actresses, etc) do not meet the pre specified ideal of beauty technological modification via picture distortion (or photoshopping) is a common practice that occurs not only in advertising media, but also in movies and television (Kilbourne & Jhally 2010). Th us, many of the image s we are exposed to do not accurately represent a ppearance. Further, a dvertisers often piece tog ether parts of various models to c reate what is perceived to be an id eal, beautiful woman (Kilbourne & Jhally 2010). This perpetuates a societal ideal that is unattainable. In a study by Krcmar, Giles, and Helme (2008), it was demonstrated that social comparison mediated the relationship between media consumption (specifically, magazines) and body esteem. T he researchers identified an inverse relationship between exposure to a variety of magazines (i.e., cel ebrity, fashion, and news) and body esteem but the relationship was mediated by social comparison. This suggests that comparison is actually the mechanism by which esteem is lowered, and not necessarily exposure to the magazine. This finding supports the contention that BID is somewhat self perpetuated; and in this instance, comparison acts as the social cognitive mechanism enabling that self perpetuation. Other research also supports the relationship between self comparison to BID and body shame (Fitzsimmons Craft, 2011; Heinberg, 1996; Thompson et al., 1999; Tylka & Sabik, 2010). C omparison to a self defined ideal is not the only form of comparison that has been investigated. Individuals may use others opinions as referent points to their own progress. F or 22

PAGE 23

example, peers, parents or other important others (e.g., coaches, teammates) provide appearance feedback in the form of criticism or teasing (negative), or compliments (positive ; Cordero & Israel, 2009; Muscat & Long, 2008; Tylka & Sabik, 2010). This a ppearance feedback informs the female as to how she compares to external appearance expectations, and can lead her to compare herself to the ideals projected by others resulting in BID and/or disordered eating ( Cordero & Israel, 2009; Muscat & Long, 2008; Thompson et al., 1999; Tylka & Sabik, 2010). Research indicates, however, that the relationship between appearance feedback and BID/disordered eating is mediated by thin ideal internalization (Cordero & Israel, 2009; Tylka & Sabik, 2010) Prominent Measur es of BID Researchers have numerous instruments to choose from when measuring BID constructs. As with any research study, psychometric properties are an important consideration when attempting to appropriately measure a construct. When considering that BI D is such a broad concept, proper selection of an instrument for BID research is a lso partly contingent on which BID related construct is of interest The most frequently used instruments are survey based, relying on self response. In this section, four of the most prominently used instruments in BID research are described. Objectified Body Consciousness Scale The Objectified Body Consciousne ss Scale (OBCS) is based on OBCT, which asserts that the feminine body is constructed by society as an object to be viewed. Because of this, women tend to view their own bodies as outside observers (McKinley & Hyde, 1996). The instrument was developed and deemed valid for use in populations of undergraduate college women (McKinley & Hyde, 1996) as well as middle aged wo men (McKinley & Hyde, 1996). 23

PAGE 24

The OBCS contains 24items across three subscales, which measure body surveillance body shame and control beliefs (OBC constructs defined previously) Reliability in the form of internal consistencies of the OBCS subscales w ere have been shown to be retest reliability (two weeks) of the instrument (McKinley & Hyde, 1996), the subscales remained stable and significant at both time points ( r = .73 to .79) indicating good test retest reliabi lity. Further convergent validation research reveals that the OBC scales are significantly related to public body consciousness ( awareness of observable aspects of the body ), while none are related to private body consciousness ( awareness of internal sensations ). These findings are consistent with the theoretical foundation of the OBCS in that body surveillance is based on perceived appearance to others, and is distinct from internal awareness See Table 11 for additional convergent and divergent research results. Body Esteem Scale Body esteem is a component of body image that is highly related to self esteem (Franzoi & Herzog, 1986; Franzoi & Shields, 1984). It is apparent that body esteem involves personal perceptions and feelings concerning the bodys ap pearance and function. The three components of body esteem for women are: 1) sexual attractiveness ; 2) weight control ; and 3) physical condition. Both sexual attractiveness and weight concern are concerned ones thoughts and feelings regarding their physic al appearance. Sexual attractiveness refers to an individuals perception or feeling regarding aspects or functions of the body associated with physical attractiveness (Franzoi & Shields, p177, 1984). The sexual attractiveness component concerns the indi viduals feelings about parts of the body that typically c annot be changed with exercise and diet (e.g., lips and breasts) W eight concern is also related to physical appearance, but 24

PAGE 25

focuses on body parts that c an be altered t hrough exercise or food intake (e.g., appetite, waist, weight) The third component physical condition, involves an individual s feelings regarding endurance, strength, and health ( Franzoi & Shields, 1984). P hysical condition differs from the other two body esteem constructs in that it is not an aspect typically evaluated by others ; except in situations where physical qualities can be demonstrated and assessed, as is the case in athletics (Franzoi & Shields, 1984). This is the opposite of self objectifying, considering the body as an object to be viewed; thus, a better feeling about physical condition theoretically indicates lower levels of OBC A significant and inverse relationship has been identified between overall body esteem, as measured by the BES, and both body surveillance ( r = .39, p < .001) and body shame ( r = .51, p < .001) subscales of the OBCS (McKinley & Hyde, 1996) See Table 12 for further convergent and divergent validity results Sociocultural Attitudes Towards Appearance Scale The most prominent measure purportin g to assess sociocultural attitudes regarding appearance is the Sociocultural Attitudes Towards Appearance Scale ( SATAQ; Heinberg, Thompson, & Stormer, 1995). The SATAQ and its updated version, SATAQR (Cusumano & Thompson, 1997), measure an individuals awareness of societal pressure regarding beauty ideals and level of internalization regarding this ideal. The SATAQ was revised a third time (SATAQ3) to account for changes in exposure to mass media channels, as well as an evolving definition of ideal beauty (Thompson, van den Berg, Roehrig, Guarda, & Heinberg, 2003). The SATAQ3 is comprised of four subscales: Pressures, Information, InternalizationGeneral, and InternalizationAthlete The Information subscale measures ones perception that the media is a source of information about appearance standards; while the Pressures subscale assesses ones sense of pressure from the media to modify their personal appearance. Internalization general measures the level at which the individual has internalized the thi n ideal, 25

PAGE 26

as it is projected by the media Internalization athlete specifically assesses the level at which the individual is influenced by the medias projection of athletic female images. At least one independent factor analysis study has confirmed the fa ctor structure of the SATAQ (Markland & Oliver, 2008) Additionally, items within each subscale have been deemed internally consistent ( Cronbachs ; Calogero Davis, & Thompson, 2004; Markland & Oliver, 2008; Thompson et al., 2003). Additionally, conv ergent and divergent validity studies indicate the construct validity of the SATAQ 3 subscales (Calogero et al., 2004; Thompson et al., 2003). Eating Disorder Inventory The Eating Disorders Inventory was developed in the early 1980s as a tool to assist in diagnosis of anorectics both calorie restrictors and those with the complication of bulimia. It was validated for use within populati ons with clinical eating disorders (Garner, Olmstead, & Polivy, 1983, p20). The original EDI is comprised of eight s ubscales. The five subscales that measure psychological trait s include: perfectionism ( overemphasis on goal attainment) i neffectiveness ( feelings of insecurity or inadequacy) i nteroceptive awareness (inability to recognize, or feeling uncomfortable with, strong emotions) m aturity f ears (desire to retreat to childhood) and i nterpersonal distrust (difficulty expressing thought or feelings to others ) The three su bscales measure that measure behavioral traits are body dissatisfaction, bulimia, and drive for thinness ) a ssociated with eating disorders Body dissatisfaction is indicated by discontent with the shape and size of body regions that are typically of concern to individuals with eating disorders (i.e., stomach, hips, thighs, buttocks). The bulimia subscale assesses the devotion of thought and behavior regarding uncontrollable eating (i.e., binge eating). Drive for thinness represents an (a) extreme desire to be thinner, (b) concern about dieting, (c) weight preoccupation, and (d) an intense fear of gaining weight. 26

PAGE 27

Since its inception, the EDI has been modified twice. The EDI 2 expanded from 64 to 91items and added three new personali ty (psychological trait) subscales. The three added psychological trait subscales were asceticism (virtue seeking through self restraint and denial) impulse regulation (mood instability, recklessness, and self destructiveness) and social insecurity (dis comfort or reticence in social situations ) The EDI 3 maintain ed the 91item structure, but rearranged the items into 12 subscales instead of 11. Some EDI 3 subscales are unchanged from the EDI 2, yet others (and their labels) have been modified. The thr ee EWR subscales of the EDI have remained virtually unchanged across the three EDI iterations. Validity and reliability research on the EDI and its subscales within clinical populations is strong and long standing (Garner et al., 1983; Garner 1991; Garner 2004) ; however, researchers question the validity of the EDI when it is administered in nonclinical populations, which occurs frequently (Campbell & Hausenblas, 2009; Grabe & Hyde, 2006; Hausenblas & Downs, 1983; Roberts Cash, Feingold, & Johnson, 2006). Specifically, the item level properties of the EDI appear to vary between clinical and nonclinical populations (Clausen Rosenvinge, Friborg, & Rokkedal 2011; Espelage et al., 2003). It has been noted that individuals in these two populations interpret E DI items differently (Kashubeck West, Mintz, & Saunders, 2001), likely because the instrument emphasizes the conceptual relationship of constructs to clinical level eating disorders (Garner, 2004) Furthermore, there has been no confirmation regarding the validity of using the EDI in nonclinical BID research. Summary of BID Instruments The four instruments tha t are briefly reviewed here appear to be the most relevant in BID research. Three of the instruments ( BES, OBCS, SATAQ 3) appear to be psychometrical ly strong when administered among non clinical female populations ; however, appropriate 27

PAGE 28

administration of the EDI in nonclinical populations ha s not yet been demonstrated and warrants further investigation. Prevalence of Body Image Disturbance While BID is prevalent in most populations, it is most pre valent in Westernized countries (Altabe, 1996; Holmqvist & Frisen, 2010; Thompson et al., 1999). Westernized countries are those that promote an individualistic lifestyle that is high in economic consumption. These countries include the United States (U.S.), Canada, Australia, and some countries in Western Europe. Acculturation to Westernized customs has been linked to BID in college aged women (Turnbull, 1999). As such, body image research is most prevalent in W esternized countries, and is most notably focused on females. The focus on females has resonated from implications that: (a) BID is gendered in nature ( prevalence and indicators var y between males and females ; Smolak & Murnen, 2008 ); (b) body image appears to be a much more complex issue for females; and (c) females seem to have a higher personal stake in the issue of body image (Garner, 1997). T he majority of body image literature focus es on college aged white women or patients with an eating disorder. BI D is highly prevalent in these populations, and research indicates that general dissatisfaction with the bodys shape or size has actually even become normative in college populations (e.g., Espelage et al., 2003; Grabe et al., 2008; Fitzsimmons Craft, 2011; Klemchuk Hutchinson, & Frank, 1990). Because of this, research is increasing among female subgroups to examine differences in age, race/ethnicity, and sexual orientation This l imited research indicates that some minorities have lower levels of BID th an Whites (Altabe, 1996), and Blacks may have l ower BID overall ( Crissey & Honea, 2006; Grabe & Hyde, 2006) It also appears that lesbians have lower levels of BID than heterosexual women and gay men ( Engeln Maddox, Miller, & Doyle, 2011) Additionally, it appears that different constructs may predict BID and self objectification among lesbian women (Kozee & Tylka, 2006) 28

PAGE 29

B ID Objectification, and the Female Athlete Research on female athletes is slowly increasing. Investigations in this area of inquiry ar e largely focused on eating pathology in specific appearance focused sports, such as gymnastics (e.g., Anderson, Petrie, & Neumann, 2011; Berry & Howe, 2000; Doughty & Hausenblas, 2007). The sparse literature that does examine BID in collegiate female athl etes is inconsistent especially with respect to specific differences that may exist between college athletes and the general college female population. A prior meta analysis of studies from 1975 to 2000 indicated that athletes have less body image distur bance (BID) than nonathletes, with no differences among sport types (Hausenblas & Downs, 2001). Since the time of the review two noteworthy sociocultural shifts have occurred: (1) female athletes have become increasingly sexually objectified in the media (American Psychological Association [APA], 2010; Kim, Segas, & Walker, 2010); and (2) the definition of ideal beauty now reflects being athletic and thin (Daniels, 2009; Thompson et al., 2003). While media images of male athletes focus on athletic prowes s, female athletic images are highly sexualized and objectified in the media (Kim et al., 2010). It has been proposed that the sexualization of female athletes is an extension of the cultural tendency to sexualize females in general, and represents an attempt to devalue female athletics and maintain patriarchal power over women (Daniels, 2009). In support of this feminist notion, Harrison & Secarea (2010) found that, compared to female athletes presented in a nonsexualized manner, a highly sexualized fem ale athlete is more likely to be perceived as : a) having a more female oriented gender role b) possessing more feminine stereotypic characteristics, and c) be ing heterosexual Further, Jones & Greer (2011) demonstrated that while women were more intereste d in reading an article about a female athlete who appeared more masculine, men preferred to read an 29

PAGE 30

article about a female athlete when the athlete pictured was more feminine and the sport discussed was also more feminine. There has been some research attempting to investigate the effect that sexualization of female athletes has on various populations For example, Daniels (2009) investigated the impact of the sexualization of female athletes on other nonathlete females level of self objectification. Adolescent girls and college women were assigned to one of four different groups. Participants in each of these groups were exposed to pictures of unknown women in one of the following four categories: (a) sexualized athletes, (b) performance athletes, (c) sexualized models, and (d) nonsexualized models. The female in the picture was provided a false name and sport or occupation (model). After viewing the images, participants were asked to make statements about themselves. These statements were categori zed into three general types (physicality, beauty, and nonbody statements), and two emotional tone types (positive and negative). The results demonstrated that females who viewed sexualized images were more likely to make negative comments about their own body, and these comments were more likely to be beauty type statements (e.g., I am ugly.). The results indicated that viewing sexualized images could be problematic for female viewers, and sexualized images of female athletes perpetuate self objectification in girls and women (Daniels, 2009). Research on how increased sexualization a ffects self objectification among female athletes appears to be almost nonexistent. Given the recent sociocultural changes, it could be important to investigate whether or not self objectification is higher in female athletes than non athletes when considering the mental health risks that result from self objectification Dorland (2006) examined the effect of female athletes self objectification on flow (i.e. an athletes focus and motivation while performing in her sport) and found that flow is related to body surveillance. 30

PAGE 31

Dorland (2006) reported that athletes w ho monitored their body and appearance more also experienced decreased flow while the same was not true for ath letes who engaged in less body surveillance (Dorland, 2006). The opposite of self objectification would be viewing/valuing the body for its abilities which is a direct measure o f the physical condition subscale of the BES (Franzoi & Shields, 1984) I t wou ld stand to reason that higher levels of body esteem should indicate lower levels of OBC. However, the research regarding the nature of this relationship (direct vs. indirect), and differences in athletes and non athletes, is sparse. In fact, only two studies have compared athletes and nonathletes with the BES and these studies are more than a decade old (D iNucci, Finkenberg, McCune, McCune, & Mayo, 1994; Wiggins & Moode, 2000) DiNucci et al. (1994) compared four undergraduate college female groups: nona thletes, basketball players, volleyball players, and softball players. While athletes tended to feel more positive ly about their physical condition when compared to the nonathletes, the only statistically significant difference was between basketball athl etes and nonathlete s Additionally, non athletes felt more negatively about their weightrelated body parts as compared to all three athlete groups, indicating that the athletes felt more positively about their body weight and body parts affected by weigh t. A major limitation of this study wa s its very small sample size, with each sport group size ranging from nine to twelve The second study (Wiggins & Moode, 2000) compared a group of female athletes to a group of nonathlete females. The athlete group in cluded participants in such sports as basketball, softball, track, cross country, and volleyball. Comparison across the three BES subscales ( physical condition, sexual attractiveness, and weight concern) indicated that the athletes felt more positively abo ut all body esteem components but these differences were only 31

PAGE 32

statistically significant for physical condition, with female athletes reporting feeling better about the condition of their body (Wiggins & Moode, 2000). P hysical condition is a measure of f eelings regarding ones own endurance, strength, and agility, which is the opposite of self objectifying (valuing the body as an object to be viewed). Th us, understanding that athletes feel better about their physical condition, points to the possibility that female athletes could be protected from higher levels of self objectification. Researchers should further examine the relationship between self objectification and body esteem for both athletes and nonathletes. First, further research needs to confirm the previous studies findings regarding body esteem, as both studies were under represented in terms of the number of sport teams and both were conducted over a decade ago. For athletes participating in higher aesthetic sports (e.g., swimming, tennis, an d gymnastics), where appearance is a focus, self objectification could be of greater concern, as these women may experience higher levels of sexualization and objectification (Kim et al., 2010; Parson & Betz, 2001). Conversely, it is also possible that par ticipation in athletics provides a protective mechanism against negative body image which means athletes are less likely to experience BID. To test these hypotheses, a comparison between representative sport and nonsport samples is necessary (Smolak, Mur nen, & Ruble, 2000). It is unsurprising that athletes may feel better about the physical condition of their body. Athletes presumably train hard to maintain a physique that is optimal for athletic performance. Additionally, other researchers have noted that athletes have a more functional orientation towards their bodies ( de Bruin, Oudejans, Bakker, & Woertman, p. 212, 2011), and thus interpret the bod y as tool for successful performance. This means that s election of a ppropriate body image measures could prove to be difficult when conducting research in the female athlete 32

PAGE 33

population because of indications that athletes experience body image differently than nonathletes ( de Bruin et al., 2011,). In other words, there are different contributors to body ima ge for athletes as compared to non athletes, causing participant responses to different body image measures to vary depending on athletic status. This phenomenon has been demonstrated through both correlational (DArcy, 2007; Hoag, 2012; Krane, Stiles Shipley, Waldron, & Michalenok, 2002) and meta analytic (Smolak et al. 2000) research. For this reason, it is necessary for researchers to use a multi dimensional body image measure that is salient for both athletes and nonathletes. Further work is necessar y to better understand the differences in BID between athletes and nonathletes. Gaining a deeper understanding of female athletes body image issues can be of great benefit, particularly for taking steps towards understanding whether athletic involvement serves as protection from BID. While the proposed research would not provide a causal link between athletic involvement and protection from BID, it will provide an important step in determining the presence (or absence) of a link between the two. Further, it will serve to guide future research efforts looking into how sexual objectification of female athletes in the media affects college level female athletes. Rationale and Organization of the Present Studies Researchers debate the relationship between ath letic performance and BID among females. Some research shows female youth participation in sports is linked with enhanced self esteem and body image ( Miller, Sabo, Melnick, Farrell, & Barnes, 2000; Womens Sports Foundation [WSF], 2009 ) as well as reducti ons in depression (WSF, 2009). Yet other researchers consider sport related weight pressures a contributor to BID, particularly in college female athletes (Reel, SooHoo, Petrie, Greenleaf, & Carter, 2010). 33

PAGE 34

Further confounding the BID issue is its complex multidimensional nature, which leads to measurement difficulties. The present work was developed in an effort to progress BID research in the college female population by investigating BID, and its measurement, among collegeaged females, with an emphasis on identifying how participation in collegiate athletics might promote or protect from BID. To do this, we present the following aims: 1) describe the impact of increased sexualization of the female athlete on the comparative BID of collegiate female athl etes and nonathletes; 2) identify salient and temporally relevant measures of BID for research in college female populations and college female athlete populations ; and 3) to examine how current BID and self objectification levels vary between highly competitive collegiate female athletes and female non athletes. The investigation in Chapter 2 presents a systematic literature review of 10 recent (1998 to 2012) studies comparing BID of collegiate female athletes to nonathletes, in an attempt to identify th e current status of BID in female athletes. The Chapter strives to update the work of Hausenblas and Downs (2001) in the context of collegiate female athletics, given the noted sociocultural changes since their article was published. Findings reported on i n Chapter 2 confirmed that t he EDI is the most commonly used instrument in BID research in college female populations. Given this finding it was determined that the EDI should be more closely explored for potential use in the final study of this work. Pri or to administrating the EDI, it was necessary to more closely examine its appropriateness for use in nonclinical female populations. Thus, t he purpose of Chapter 3 wa s to critically review factor analytic studies that examined the internal structure of th e EDI in nonclinical samples of women from Westernized countries. Chapter 4 presents a research study in which primary data were collected, with the purpose of examining multidimensional body image and self objectification among NCAA Division I female ath letes 34

PAGE 35

and nonathletes. Finally, Chapter 5 presents a synthesized discussion of the major findings across Chapter 2, 3, and 4, and provides practical i mplications and recommendations for future research. The research studies described in Chapters 2 through 4 w ere written in manuscript form for the purposes of submission for publication in scholarly journals. 35

PAGE 36

Table 1 1. Statistically significant correlations between the subscales of the Objectified Body Consciousness Scale and referent measures Measure Body Shame Body Surveillance Control Beliefs Fear of Fat X c X c Endorsement of Cultural Standards X c X c Restrictive Eating X c Dieting X c Exercise to lose weight X c Wearing make up X c Wearing certain clothes to look thinner X c Weig ht/Shape efficacy X c Public Self Consciousness X c Appearance Orientation X c Ideal / Actual Weight Discrepancy X b, X b X b, Body Competence X X c Body Esteem X a,b,c X a,b,c X a,b,c Weight Discrepancy X c X c Self Acceptance X b X b X b View ing media as an information source for beauty standards X d X d Perceive pressure from media to conform to beauty standards X d X d Thin Ideal Internalization X d X d Athletic Ideal Internalization X d aMcKinley & Hyde, 1996 bMcKinley 2004 cMcKinley, 2006 d Forbes, Jove, & Revak, 2006 36

PAGE 37

Table 1 2. Statistically significant correlations between the Body Esteem Scale and referent measures Measure Sexual Attractiveness Weight Concern Physical Condition Comparative attractiveness X b X b Body Competence X b X b X b Private Body Consciousness X b X b Amount of Aerobic Exercise X b Amount of Anaerobic Exercise X b Views food as Weight X b Views food as Energy X b X b Current weight X c X c Weight satisfaction X c Self Rated P hys attractiveness X c X c SCS Public X c SCS Social Anxiety X c X c EDI Drive for Thinness X c EDI Bulimia X c X c EDI Body Dissatisfaction X c X c FRS current vs. ideal X c X c FRS current X c X c VCA Fr ontal Size Dissatisfaction X c X c VCA Profile Size Dissatisfaction X c X c Self Esteem X a,b X a, b X b SCS=Self Consciousness Scale; EDI=Eating Disorder Inventory; FRS=Figure Rating Scale; VCA=Video Camera Assessment a Frazoi & Herzog, 1984 b Fr anzoi & Shields, 1986 c Thomas & Freeman, 1990 37

PAGE 38

Figure 1 1. Model of key objectification theory tenets ( Szymanski et al., 2011) 38

PAGE 39

Figure 1 2. Amended Objectification Theory Model. Note. For parsimony, where indirect chains of relations are depicted bet ween two constructs, possible additional direct paths are not included. (Moradi, 2010). 39

PAGE 40

CHAPTER 2 A SYSTEMATIC REVIEW OF STUDIES COMPARING BODY IMAGE CONCER N S AMONG FEMALE COLLEGE ATHLETES AND NONATHLETES 19972012 1 Female participation in sports has been linked to improved academic performance and enhanced self esteem, as well as reductions in depression and risk for osteoporosis and breast cancer (WSF, 2009). In addition, sports participation provides females with opportunities to learn valuable life and work skills, such as teamwork, goal setting, and perseverance (WSF, 2009). Some research indicates athletic participation could enhance body image ( Miller et al. 2000); yet, other researchers consider sport related weight pressures to be a contr ibutor to body image concerns (BIC), particularly in college female athletes (Reel et al., 2010). Body image is a broad and dynamic concept that extends across perceptual and attitudinal modalities. BIC are indicated by body size misperception and/or nega tive attitudes or feelings toward the human body and its appearance (Cash & Szymanski, 1995) The term body image is an umbrella term encompassing many constructs. These include, but are not limited to: weight (dis)satisfaction, body (dis)satisfaction, bod y shame, appearance (dis)satisfaction, appearance evaluation, body esteem, body appreciation, body dysmorphia, and body schema (Avalos, Tylka, & WoodBarcalow, 2005; McKinley & Hyde, 1996; Thompson et al. 1999). Among college aged females, BI C are associated with several heath issues, such as disordered eating ( Grabe et al. 2008) substance abuse ( Carr & Szymanski, 2010), depression (Carr & Szymanski, 2010), decreased self esteem ( Tylka & Sabik, 2010), and consideration of cosmetic surgery (Swami, 2009) To explore the relationship between athletic participation and body image, Hausenblas and Downs (2001) conducted a meta analysis of studies published between 1975 and 2000. They 1 Body Image : An International Journal of Research (accepted for publication). Reprint in authors dis sertation is within authors rights as outlined by E lsevier Published by Elsevier, 1600 John F Kennedy Boulevard Suite 1800 Philadelphia PA 191032879 USA. Copyright 2013. 40

PAGE 41

compared body image of nonathletes and athletes across gender, age, and competition level (i.e., adolescent/high school, club/recreational, elite, and college). The overall findings suggested athletes possessed a more positive body image than nonathletes, with an equivalent disparity between athletes and non athletes of each gender No comparison was made on body image among males versus females; yet, other research indicates that body image is gendered in nature (Smolak & Murnen, 2008, p. 258). In other words, the level at which BI C occur varies between males and females, and is indicated by different motivators (e.g., the pursuit of thinness versus pursuit of muscularity). Although Hausenblas and Downs (2001) results did not address gender differences, their research suggested athletic participation was associated with more pos itive body image. One limitation of Hausenblas and Downs (2001) study was their inability to analyze data from all identified studies, due to more than 25 different types of body image measures being administered across the reviewed studies. As a result, most of the aggregated data came from studies using the Eating Disorder Inventorys Body Dissatisfaction subscale (EDI BD; Garner et al. 1983). Hausenblas and Downs (2001) and others (e.g., Bailey, Goldberg, Swap, Chomitz, & Houser, 1990) have noted that the EDI BD is not an adequate measure of overall body image, as the items focus on the lower body (hips, thighs, etc.) to the exclusion of other body parts. Also, the EDI BD appears specific to body shape dissatisfaction (Bailey et al., 1990). Despite the se notable limitations, Hausenblas and Downs (2001) provided a comprehensive synthesis comparing body image among athletes and nonathletes at the time. However, since the turn of the 21st Century, two notable sociocultural shifts have taken place: (a) th e sexual objectification of female athletes in the media has increased (APA, 2010; Kim et al., 2010); and (b) the definition of ideal female beauty now includes appearing athletic 41

PAGE 42

in addition to being thin (Daniels, 2009; Thompson et al. 2003) Given the sociocultural changes that have occurred since Hausenblas and Downs (2001) meta analysis, a timely, relevant update is necessary to understand the current status of body image among female athletes. The purpose of this paper is two fold. In Part I, we provide a brief overview of body image literature as it relates to the noted sociocultural changes and the female athlete. In Part II, we conduct a comprehensive narrative synthesis (Center for Reviews and Dissemination [CRD], 2009) of recent studies compar ing BIC in collegiate female athletes and non athletes. We elected to conduct a narrative synthesis as part of a systematic review because the varying nature of body image measurement in studies of female athletes and non athletes makes pooling of all data within a meta analysis impossible (Hausenblas & Downs, 2001). A meta analysis of aggregated data from studies using such heterogeneous outcome measures could likely produce misleading results (CRD, 2009) while a narrative synthesis within a systematic re view will facilitate a better comprehe nsive understanding of current body image literature among female athletes. Also, several new instruments (e.g., Objectified Body Consciousness Scale, Self Objectification Questionnaire, Drive for Leanness Scale, Inter nalization Athlete subscale of the Sociocultural Attitudes Towards Appearance Scale 3) have been developed since the work of Hausenblas and Downs (2001). Many of these instruments are not specific to body image but measure constructs highly related to bod y image and are thus used when examining BI C We hope to determine if there are differences in BIC between athletes and non athletes that are specific to the different body image constructs and measures, which would be difficult to determine through a meta analysis. We focused exclusively on females. Due to the gendered nature of body image (Smolak & Murnen, 2008), a single sex focus will allow for a more in depth review of the literature and 42

PAGE 43

will allow for more informed comparisons by sport. Our focus on college women is warranted since this population is considered to be at highrisk for BI C (Fitzsimmons Craft et al., 2012) Also, research indicates that there are discrepancies in BI C among collegiate female athletes depending on university competition l evel (i.e., Division I, II, and III; Kato, Jevas, & Culpepper, 2011); thus, comparisons of BI C between female athletes across competition levels and nonathletes are warranted. University competition level is best explained by the divisional structure of the National Collegiate Athletic Association (NCAA). The NCAA oversees competitive athletics throughout the United States and Canada and its membership creates rules in Division I, Division II, and Division III (NCAA, 2012) Division I is the most competitive, as it includes the largest athletic programs that provide the most athletically related financial aid. Division II universities have somewhat smaller athletic programs (less sports) and offer more limited financial aid. Division III universities do not offer any athletically related financial aid. Division I and II also have special academic eligibility requirements, because of the financial aid offered, with Division I rules being more stringent. Because there is no potential financial aid associated with participation at the Division III levels, eligibility requirements are set by the individual college or university. Research on how BIC differs among female athletes and non athletes can help identify potential mediating and moderating effects of the relationship between athletic status and specific body image constructs. If differences in BIC among female athletes and non athletes are found to vary depending on the construct, then researchers will need to determine how psychological, environmental, or sociocultural factors are related to specific constructs. 43

PAGE 44

Part I: Sociocultural Changes and the Female Athlete Sexual Objectification of Female Athletes Objectification theory (Fredrickson & Roberts, 1997; Szymanski et al., 201 1) suggests that the huma n body is not only constructed biologically; it is also developed through sociocultural contexts such as gender roles and sexual objectification. Female gender roles, as perceived by women, include exhibiting traits that emphasize submissiveness, modesty, thinness, and appearance investment (Mahalik et al., 2005) Sexual objectification reinforces these gender roles by focusing on the female appearance and diminishing nonfeminine achievements. Sexual objectification occurs when the body is separated from t he person and recognized predominantly for its physical attributes (body or body parts), with its worth being equated to its representation of an object of sexual desire (Fredrickson & Roberts, 1997). It has been proposed that the sexual objectification of the female athletes body extends from the cultural tendency to sexualize females in general, and is an effort to maintain patriarchal power over women via masculine hegemony (Daniels, 2009). Further, sexual objectification of female athletes devalues female athletics (Hardin, Lynn, & Walsdorf, 2005). Increased sexual objectification of female athletes received much attention following the release of the 1997 Sports Illustrated Swimsuit Edition the first issue in which the magazine began using female athletes as swimsuit models (Kim et al., 2010). Since 1997, there has been other research demonstrating that female athletes experience greater sexual objectification than male athletes, usually via sexually suggestive images and articles in print media ( Da niels & Wartena, 2011; Fink & Kensicki, 2002; Hardin et al., 2005) and Internet sports blogs (Clavio & Eagleman, 2011). Even the highest level female athletes have been overtly sexualized. For example, during the 2012 Olympic Games, internet blogs pondered if Olympian javelin thrower Leryn Franco had 44

PAGE 45

breast implants, and noted that, despite not making the finals in her event she still held the top spot for hottest female Olympian ( Gratereaux 2012, para. 1). Sexual objectification contributes to the internalization of sociocultural beauty ideals (Fredrickson & Roberts, 1997; Moradi et al. 2005 ). When something is internalized it is taken within and believed to be self defined. Internalization of sociocultural beauty ideals m ay lead individuals to believe their desire to attain th e ideal is a personal choice rather t han a result of social pressure (McKinley & Hyde, 1996). Further, sexual objectification and internalization are theorized to lead females to self objectify or define their own body as an obj ect to be viewed by others (Fredrickson & Roberts, 1997; McKinley & Hyde, 1996; Szymanski et al., 2011). Objectification theory posits that pronounced self objectification leads to adverse psychological consequences, including but not limited to social phy sique anxiety (SPA; Calogero, 2004) and body shame (Calogero, 2004; Tylka & Hill, 2004; Tylka & Sabik, 2010). SPA reflects concern by individuals that their body or physical appearance is being negatively evaluated by others. Body shame occurs when individuals experience strong feelings of guilt or embarrassment from not meeting an internalized ideal beauty. These psychological consequences are in turn associated with disordered eating (e.g., Moradi et al., 2005; Tylka & Sabik, 2010), depression (Carr & Szy manski, 2010; Szymanski & Henning, 2007), and sexual dysfunction ( Szymanski et al., 2011). Researchers posit that the enhanced sexual objectification of athletes could most negatively affect the body image of female athletes playing sports that focus on ap pearance, leanness, or both (Karr, Davidson, Bryant, Balague, & Bohnert, 2013; Reel et al., 2010). It seems that appearancefocused (i.e., aesthetic) sports are adjudicated sports in which appearance and physique are emphasized for success in competition, and include cheerleading, gymnastics, 45

PAGE 46

figure skating, and diving (Harrison & Fredrickson, 2003; Hausenblas & Carron, 2002; Petrie & Greenleaf, 2012). Cheerleading, gymnastics, and figure skating all emphasize several aspects of appearance, including the aesthetics of technical form (i.e., gracefulness), physical build, make up and hair, and costume (i.e., uniform). The emphasis of appearance in diving is exclusive to technical form and physical build. Thus, the extent to which appearance is emphasized in each of these sports differs. The definition of a lean focused sport is even less clear. Some researchers define a lean sport as an endurance type sport, such as swimming and longdistance running, where being lighter in weight and leaner in fat contrib utes to improved performance (Engel et al., 2003). Other researchers use the term lean to describe sports that emphasize weight as part of appearance, and include sports such as gymnastics, cheerleading and figure skating (Harrison & Fredrickson, 2003; Hom an, McHugh, Wells, Watson, & King, 2012). Still others have indicated that this term could include sports that emphasize leanness for either competition or appearance, which would encompass all of the aforementioned sports (Reel et al., 2010; Reinking & Al exander, 2005). It is apparent that the definition of the terms lean sport and aesthetic sport emphasize the competition aspect. In other words, athletes are grouped into these categories based on expectations specific to competition within the sport. However, athletes may experience additional sociocultural expectations that are related to appearance, but unrelated to athletic performance (Hardin & Greer, 2009; Parsons & Betz, 2001; Petrie & Greenleaf, 2012 ). Through a survey of college students, Par sons and Betz (2001) attempted to determine societys comparative objectification of athletes participating in a variety of sports. Results indicated that female athletes participating in tennis, swimming, gymnastics, and cheerleading experienced 46

PAGE 47

greater o bjectification by peers. As noted previously, swimming, gymnastics, and cheerleading have been defined as lean and/or appearance focused sports in the literature. Tennis, however, has consistently been grouped separately from such sports (Harrison & Fredr ickson, 2003; Hausenblas & Carron, 2002). Parsons and Betzs (2001) findings indicate that there may be external sociocultural expectations of attractiveness and femininity projected onto female athletes that are unrelated to competition. In fact, several researchers have indicated that the same athletes identified as highly objectified by peers (Parsons & Betz, 2001) are also considered more feminine (Clavio & Eagleman, 2011; Hardin & Greer, 2009; Kane, 1988; Kim et al., 2010). Additionally, athletes from tennis and swimming appear to be some of the most sexually objectified in the media (Fink & Kensicki, 2002; Kim et al., 2010). Researchers have also noted volleyball (Kane, 1988; Kim et al., 2010) and golf (Clavio & Eagleman, 2011; Fink & Kensicki, 2002; K ane, 1988; Kim et al., 2010) as two feminine sports in which athletes are more objectified in the media. Early researchers defined feminine sports as being less physical and aggressive, lower risk, and nonstrength based (Kane, 1988). More recent research indicates that societys perception of an athletes femininity is affected by the medias portrayal of the athlete. For example, Harrison and Secarea (2010) found that a sexualized high status female basketball player was more likely to be perceived as st ereotypically feminine and heterosexual than a nonsexualized female basketball player. Basketball is typically perceived as a more masculine sport (Hardin & Greer, 2009; Kane, 1988; Kim et al., 2010; Parsons & Betz, 2001; Petrie & Greenleaf 2012); thus, these findings indicate that the sexual objectification of an athlete, regardless of sport type, can lead viewers to label the athlete as more feminine. Their findings, however, provide no indication as to societys general categorizing of sports as femini ne or masculine, just 47

PAGE 48

the athlete in question. Although there are indications that the media more frequently objectify athletes from feminine sports (Fink & Kensicki, 2002; Hardin & Greer, 2009; Kim et al., 2010), we do not know if these sports are conside red feminine because of this objectification, or if they are objectified because they are considered more feminine. In summary, the traditional focus of body image research among female athletes has been on weight and appearancerelated pressures specifi c to competition; however, recent research points to the objectification of female athletes as potentially imposing additional pressures on other female athletes to exemplify stereotypic female characteristics, including ideal appearance. Furthermore, it a ppears that these sociocultural expectations are not exclusive to athletes participating in sports traditionally considered appearance or weight focused. The New Ideal Research demonstrates t hat societys perception of ideal beauty is partially defined and significantly perpetuated by the media ( Grabe et al., 2008) The new ideal of athletic and thin was partly evident with the evolution of mainstream health and fitness magazines, such as Womens Health and Fitness (Thompson et al., 2003). The images, a rticles, and ideals of these magazines, combined with the general medias portrayal of the new beauty ideal has put more pressure on girls and women to attain a glorified, unreachable physical ideal (Kilbourne & Jhally, 2010). The detrimental effect of thi s pressure has been noted among females in general (e.g., Daniels, 2009; Harrison & Fredrickson, 2003; Swami, 2009), and early research indicates female athletes have acknowledged and heeded societys message that appearance is at least as important as per formance (Gibson, 2007; Larabee, 2011). I n recognition of the evolving beauty ideal, prominent body image researchers have revised existing body image survey instruments and created new ones in an effort to include measures specific to th e new athletic id eal For example, in its third revision, the Sociocultural 48

PAGE 49

Attitudes Towards Appearance Questionnaire 3 (SATAQ3; Thompson et al., 2003) added a new subscale called InternalizationAthlete. This subscale assesses the level at which the athletic ideal has b een internalized by asking participants how much they try, or want, to look like sport and athletic figures An example of a nother instrument is the Athletic Image Scale (AIS; Lenart, Goldberg, Bailey, Dallal, & Koff, 1995), a three dimensional silhouette instrument designed to assess a females actual and ideal physiques These instruments can assist researchers in identifying how the desire to attain an athletic physique is related to body image in both female athletes and nonathletes. Part II: Systemati c Review of Comparative Studies Research prior to 2001 indicates that athletes experienced a more positive body image than nonathletes. Since this time, the sexualization of female athletes in the media has increased, and the sociocultural beauty ideal h as come to include appearing athletic in addition to being thin. These sociocultural changes further pressure women and girls to attain a glorified, unreachable ideal, while also pressuring female athletes to exemplify these projected beauty ideals. For the second part of this paper, we conducted a systematic and comprehensive literature review to synthesize recent body image research that compares collegiate female athletes and nonathletes to identify the current status of BI C among female athletes. As pr evious research indicates that these group differences may vary depending on which sport type (e.g., lean, feminine, aesthetic) or university competition level (i.e., Division I, II, and III) is considered, we also will explore differences between athletes and nonathletes in the context of sport type and competition level. Method In this systematic review, competitive athletics is defined as a college or university level sport or emerging sport team governed by the NCAA, or a similar governing body (e .g., 49

PAGE 50

Canadas Collegiate Intrauniversity Sport [CIS]). The term collegiate female athlete is defined as a female athlete participating in a college or university sport team governed by the NCAA, or a similar governing body. T his distinction is important, because past research indicates that varsity athletes experience less BI C than club/recreational athletes attending college (Hausenblas & Downs, 2001). Since varsity athletes and recreational/club athletes experience BI C at significantly different levels, the two groups of athletes should not be collapsed together as one homogenous group. Also, club sports are not nationally governed; they are student run and university regulated with eligibility requirements that differ by university and sport. For exampl e, some universities only allow students to participate in club sports, while other universities allow staff, faculty, and/or alumni to participate. Furthermore, some club sports require individuals to try out for team membership, while other club sports a re open to anyone regardless of skill level. Due to these differences, we felt that participation in club sports could not be considered inherently equal to varsity sport participation. Search Procedures Hausenblas and Downs (2001) used the key words athle te sport body image, body image disturbance, body dissatisfaction, and body esteem, in a computer search of the following databases: Dissertation Abstracts Online, Educational Resources in Completion, MedLine, PsychINFO, and SPORTDiscus. We used their se arch terms to query the same databases, with the addition of the key words internalization objectification thin ideal, and athletic ideal. We also searched the following additional databases: Academic Search Premier, Communication and Mass Media Complete, Masters Abstracts International, Psychological and Behavioral Sciences Collection, PubMed, and Womens Studies International. We also searched the reference lists of all retained articles. Studies were required to meet the following inclusion criteria: ( a) peer reviewed manuscript, article, dissertation, or thesis available in the English language; (b) 50

PAGE 51

published or documented between 1997 and 2013 (J anuary ); (c) measured a component of body image; and (d) included a sample of collegiate female athletes wi th a non athletic comparison group. The beginning cutoff of 1997 was chosen due to our interest in examining how BIC differs between collegiate female athletes and non athletes after the noted sociocultural shifts in beauty ideals and sexual objectificatio n of female athletes. While we concede that this shift did not occur at a single moment in time, the noted change in focus of the 1997 Sports Illustrated Swimsuit Edition to include athlete models provided what we felt to be an appropriate starting point f or our literature search. We intended to include any research studies that may have been inadvertently excluded by Hausenblaus and Downs (2001) from 1997 to 2000. The database search returned 333 studies spanning from 1998 to 2012. One hundred thirty eigh t ( n = 138) studies were eliminated prior to abstract review because they focused on males and/or adolescents ( n = 100) or non college athletes (e.g., ballet dancers, auxiliary unit members; n = 38). After this primary exclusion, abstracts for 195 studies were reviewed, which led to the elimination of 83 additional studies that did not include an athlete group. Full articles were obtained for the remaining 112 studies. Of these 112 studies, four studied an adolescent sample; 66 did not measure body image; 13 did not have a nonathlete comparison group; and 19 did not include sports that met our definition of competitive athletics. Ten studies ultimately satisfied all selection criteria and were included in this review. The following information was extracted from each article: (a) study purpose; (b) participant makeup (sample size, sport type, and grouping); (c) school size, type, and/or region; (d) measurement tools for assessing body image constructs; (e) research design; (f) data analysis procedures; and ( g) results. Comparisons were then made on various body image constructs by sport type and competition level. 51

PAGE 52

Results Description of Studies Included in the Review Ten studies spanning from 1998 to 2012 met the inclusion criteria for this review. Six studi es were reported in peer reviewed manuscripts (DiBartolo & Shaffer, 2002; Finkenberg, DiNucci, McCune, Chenette, & McCoy, 1998; Krane et al. 2002; Robinson & Ferraro, 2002; Steinfeldt et al. 2011; Wiggins & Moode, 2000), three as Masters theses (DArcy, 2007; Fellows, 1999; Rose, 2008), and one was part of a doctoral dissertation (Hoag, 2012). Two studies (Fellows, 1999; Finkenberg et al., 1998) occurred in the two year overlap period between the beginning of the search time window and Hausenblas and Dow ns (2001) meta analysis. One study (Finkenberg et al., 1998) was listed by Hausenblas and Downs (2001) as being reviewed; however, the study was not included in the data analysis section of their meta analysis, nor referenced within their manuscript. Tabl e 2 1 outlines the author(s), year, sample size/type, school type, body image measure(s) administered, and associated study results. Research d esign All 10 studies were cross sectional, quantitative, and collected self report data. Seven used paper/penci l format to collect data (DArcy, 2007; DiBartolo & Shaffer, 2002; Krane et al., 2002; Robinson & Ferraro, 2004; Rose, 2008; Steinfelt et al., 2011; Wiggins & Moode, 2000), while one used an Internet format (Hoag, 2012). Survey format was not identified in two studies (Fellows, 1999; Finkenberg et al., 1998). Sampling methodology was not reported for either the athlete or nonathlete group in one study (Finkenberg et al., 1998). All nine remaining studies recruited athletes through team coaches, although one obtained initial permissions at a higher administrative level (i.e., NCAA conference representatives; Hoag, 2012). For the nonathlete group, eight studies used non52

PAGE 53

random sampling and one combined snowball sampling (requested nonathlete names from ath letes) with non random sampling (Hoag, 2012). Three studies (DiBartolo & Shaffer, 2002; Finkenberg et al., 1998; Krane et al., 2002) did not operationally define the term female athlete One of these studies ( Finkenberg et al., 1998) did not identify the competition level (Division) or sports in which the surveyed athletes participated. The two other studies identified as Division I (Krane et al., 2002) or Division III (DiBartolo & Shaffer, 2002), but reported including sports never governed, nor listed as emerging, by the NCAA (precision figure skating and squash respectively; NCAA, 2012). Thus, while these studies appear to include some sports considered eligible for NCAA membership (NCAA, 2012), this apparent eligibility does not confirm they are sanctioned under the auspices of the NCAA. For example, softball, which is an NCAA eligible sport, is also offered as an intramural or club sport; colleges and universities can have both clubsoftball and an NCAA governed team. Th us the extent to which nonNCAA athletes comprised the samples in these studies (DiBartolo & Shaffer, 2002; Finkenberg et al., 1998; Krane et al., 2002) is unclear. One of these studies (Krane et al., 2002) used a non athlete comparison group comprised solely of aerobic exercisers, while the other nine studies used a nonathlete comparison group comprised of female non athlete undergraduate students. This difference in research design, combined with the unknown extent to which the athlete group is comprised of club athletes, led our resea rch team to determine that the Krane et al. (2002) study did not fit the general characterization of athletes versus nonathletes as defined in other reviewed investigations. Due to the studys noted incompatibility with the other nine reviewed studies, we chose to exclude the results from the Krane et al. (2002) study. 53

PAGE 54

Sport groupings Most studies predominantly included athletes from sports considered less objectified (e.g., basketball, soccer, and track). Table 2 2 provides data from Parsons and Betz (2001) that report how athletes in each sport are objectified according to their peers. These data provide an accessible comparison of mean objectification ratings for each rated sport that is included in each reviewed study. Also, athletes were grouped diffe rently in the reviewed studies. Five studies grouped all athletes from several sports into one large athlete group (DArcy, 2007; DiBartolo & Shaffer, 2002; Finkenberg et al., 1998; Steinfeldt et al., Wiggins & Moode, 2000). Two studies divided the athlete group to facilitate comparisons based on body related pressures ( Fellows, 1999) or determination of success (Robinson & Ferraro, 2004) Two other studies used athlete group(s) representative of only one sport each (Hoag, 2012; Rose, 2008). These types of comparisons are warranted, as researchers have posited that BIC could vary according to sport type (e.g., Parsons & Betz, 2001; Reel et al., 2010). Measurement of body image The most commonly used measure to assess body image was the EDI BD (body shape dis satisfaction), which was administered in five studies. Three studies used traditional (thin) ideal/actual figure measures (e.g., Figure Ratings Scale [FRS]; Stunkard, Sorenson & Schulsinger, 1983), and two studies used athletic ideal/actual figure measures (i.e., AIS) to compare athletes and non athletes on body size dissatisfaction. Three studies measured internalization of the thin ideal: two with the SATAQ 3 internalization general subscale, and one with the body shame subscale of the Objectified Body Co nsciousness Scale (OBCS; McKinley & Hyde, 1996). The two studies administering the thininternalization subscale of the SATAQ 3 also administered the other three SATAQ 3 subscales: pressures, information, and internalization athlete. Three studies measured weight preoccupation with the EDI Drive for 54

PAGE 55

Thinness subscale ( EDI DT ), and three others measured SPA with the Social Physique Anxiety Scale ( SPAS; Hart, Leary, & Rejeski 1989) Of the three studies that administered the Body Esteem Scale (BES; Franzoi & Shields, 1984), one reported on overall body esteem, another reported on the three constructs of body esteem (i.e., physical condition, sexual attractiveness, and weight concern), and the third reported on both overall body esteem and its three components The Self Objectification Questionnaire (SOQ; Noll & Fredrickson, 1998) and Body Appreciation Scale ( BAS; Avalos et al. 2005) were used in one study each. Differences in BI C between A thletes and Non A thletes In nine reviewed studies (excluding Krane et al., 2002 as previously noted), athletes and nonathletes differed s ignificant ly on at least one body image construct. Of these nine studies athletes overall scored more positively in all but one study (i.e., Hoag, 2012). Specifically, the three studies me asuring ideal/actual body size discrepancy with a traditional rating scale found nonathletes to be more dissatisfied with their body size (DArcy, 2007; DiBartolo & Shaffer, 2002; Robinson & Ferraro, 2004). Compared to nonathletes, athletes reported more positive overall body esteem (Fellows, 1999; Steinfeldt et al., 2011); as well as more positive feelings about their bodys physical condition, a component of body esteem (Fellows, 1999; Wiggins & Moode, 2001). Two other body esteem components, weight con cern (body parts modifiable with diet and exercise) and sexual attractiveness (body parts not modifiable with diet and exercise), trend ed higher (more positive) in athletes but were not statistically significant in one of the two studies (Wiggins & Moode, 2000) Fellows (1999) however, noted statistically significant differences between athletes and non athletes on all three BES subscales, reporting moderate to larg e effect sizes for each outcome (physical condition d = 1.11, p < .001; weight concern d = 0.80, p < .001; and sexual attractiveness d = 0.39, p = .001) 55

PAGE 56

Although athletes and nonathletes did not differ in their athletic ideal/actual figure discrepancies, athletes did identify a larger and more muscular current (actual) figure and ideal figure than did nonathletes (DArcy, 2007; Hoag, 2012). Additionally, the two studies measuring internalization of the athletic ideal found that athletes internalized the athletic ideal more so than nonathletes (DArcy, 2007; Hoag, 2012). With regards to internalization of the thin ideal, the three studies measuring this construct reported no statistical differences between athletes and nonathletes (DArcy, 2007; Hoag, 2012; Rose, 2008). Also, according to two studies, SPA appeared unrelated to athletic status ( Finkenbe rg et al., 1998; Hoag, 2012). These results, and others, are further described in the context of sport type and competition level. Relationship between BI C and Sport Type Three studies found that athletes competing in the endurance sports of swi mming, track, and water polo experienced less BIC than non athletes. Specifically, endurance athletes reported less self objectification and weight preoccupation (Robinson & Ferraro, 2004; Rose, 2008); and better general body esteem, including more positive feelings about their physical condition, sexual attractiveness, and weight affected body parts (Fellows, 1999). On the other hand, three of four studies that compared nonathletes to athletes participating in the more feminine sports of gymnastics, tenni s, and volleyball found no differences in BIC. This was noted for constructs specific to appearance of the body and body parts (Fellows, 1999), body appreciation (Hoag, 2012), and self objectification (Rose, 2008). The fourth study (Robinson & Ferraro, 2004) reported that golf and volleyball athletes experienced less weight preoccupation than nonathletes. Between sport type comparisons revealed that athletes defined as participating in appearancefocused sports (gymnastics and tennis) experienced lower positive body esteem and greater weight concern than athletes in sports labeled as non appearance/non lean focused 56

PAGE 57

( volleyball, soccer softball, track and field, and basketball ; Fellows, 1999). In another study, combined body shame and self objectification scores were able to predict group membership to a nonappearance focused endurance sport (water polo) or highly appearance focused sport (gymnastics; Rose, 2008). Relationship between BI C and Competition Level Three studies reported that athletes at lo wer competition levels (e.g., Division II or III) were more satisfied with their body shape (via the EDI BD) than non athletes (DArcy, 2007; DiBartolo & Shaffer; Robinson & Ferraro, 2004), while a fourth study reported no significant difference in body sh ape dissatisfaction between non athletes and volleyball players at either the Division II or III level (Hoag, 2012). On the other hand, Division I volleyball players were significantly more dissatisfied with their body shape than nonathletes (Hoag, 2012). The single study comparing non athletes to volleyball players across all three competition levels noted a linear trend between body shape dissatisfaction and competition level, with Division I athletes reporting the highest body shape dissatisfaction (Hoa g, 2012). These findings could indicate that competition level and body shape dissatisfaction as experienced by athletes are related. Or, these results might not be specific to competition level, but to volleyball, or sports that could be grouped as simila r to volleyball. A negative linear trend was also noted for competition level when looking at body appreciation in volleyball players. That is, volleyball players body appreciation increased as competition level decreased, although these differences were not significant (Hoag, 2012). Two reviewed studies found that athletes at various levels idealized an athletic physique more so than nonathletes (DArcy, 2007; Hoag, 2012), with one of these studies (Hoag, 2012) reporting that the difference was more pro nounced as competition level increased. Comparison between the groups on the SATAQ 3 measures of pressures and information were mixed. While 57

PAGE 58

one study (Hoag, 2012) reported no difference between nonathletes and athletes across all three competition levels on these measures another study (DArcy, 2007) found that less competitive athletes perceived less pressure from the media and reported using the media less as a source of information about beauty than nonathletes Again, the differences in the two studies could be attributable to the type of sports included in the athlete groups, and not just competition level. Discussion F emale athletes have been marginalized in the media for decades ; however, in the last 15 years, the sexual objectification of the f emale athlete seems to have risen in concurrence with a modified beauty ideal of athletic and thin (APA, 2010; Kim et al., 2010) We aimed to identify changes in BI C among female athletes since this shift. A thorough search process revealed 10 studies span ning from 1998 to 2012 that compared various body image constructs among collegiate female athletes and non athletes When considering that BIC are associated with increased risk of adverse psychological (e.g., depression) and behavioral (e.g., disordered eating) consequences ( Fitzsimmons Craft et al., 2012; Grabe et al., 2008; Szymanski et al., 2011), the deficiency of body image research in the athlete population is a bit startling. Moreover, during our search process, we found that most research conducte d between 1997 and 2012 comparing athletes to nonathletes was focused on eating path ology and not differences in body image This is unfortunate, given that BI C are more common than eating disorders and often a strong predictor of eating pathology as well as other disorders (e.g., depression, sexual dysfunction; Carr & Szymanski, 2010; Szymanski et al., 2011 ). Thus, assessing and targeting body image may help identify and help those at risk before disorder develops ( Farrell Shafran, & Lee 2006; Thompson et al., 1999). Five of the reviewed studies assessed body image with the EDI BD, making the EDI BD the most commonly used instrument administered across the 10 reviewed studies. This was 58

PAGE 59

surprising to us, given that researchers have questioned whether the EDI BD can adequately assess overall body image (Bailey et al., 1990; Hausenblas & Downs, 2001). Three studies also reported assessing body size dissatisfaction using a type of figure rating or silhouette scale. Researchers in these three studies asked par ticipants to se lect one of seven (Robinson & Ferraro, 2004) to nine (DArcy, 2007; DiBartolo & Shaffer, 2002) figure s that most closely represent ed their actual body size and select the figure that most closely represent ed the ir ideal size The discrepancy between perceived and ideal size wa s used to determine body size dissatisfaction The psychometric validity of such scales, however, has been questioned. Some scholars suggest these scales do not assess the distress participants feel when they have not met the indicated ideal, but rather an awareness of the attractiveness stereotyp e ( Hildebrandt & Walker, 2006). Others note that t he number of response options is too limited and that successive images do not occur on a true interval scale, potentially causi ng inflate d test retest reliability scores and scale coarseness (over or under estimation of moderated regression effect sizes; Gardner Friedman, & Jackson, 1998). There are other indications that these scales do not represent ethnic and racial diversity of body size preferences for specific body parts, such as breasts or buttocks (Grabe & Hyde, 2006; Swami, Jones, Einon, & Furnham, 2009). Other researchers findings indicate that the scales results can be misleading when considered alone (Forbes & Jung, 2008; Grabe & Hyde, 2006). A dvancements in body image research within the last 15 years have sparked the development of several new instruments for measuring body image; as such, we expected that some of these newer instruments would have been used in pla ce of the EDI BD. We were also surprised to find that only one study compared female athletes and non athletes using self objectification measures. Considering the increase in sexual objectification of female athletes, it 59

PAGE 60

would seem that more researchers w ould be exploring female athletes self objectification. Despite these concerns, our systematic and narrative analytic approach led us to identify two critical findings: (a ) female athletes desire a slightly heavier, more athletic physique and (b) BIC of female athletes may differ across competition level and sport type on various body image constructs The Athletic Physique Athletes, compared to nonathletes, desire a more muscular body (DArcy, 2007; Hoag, 2012) that more closely resembles their current f igure (DArcy, 2007; DiBartolo & Shaffer, 2002; Robinson & Ferraro, 2004). A s indicated by Hausenblas and Downs (2001) in their meta analysis, the results of this review also suggest that athletes are more satisfied with their body shape in general Despit e the increased sexualization and objectification of female athletes in the past decade, i t appears female ath letes continue to be more satisfied with the overall shape of their body than non athletes. As such, it comes as no surprise that collegiate femal e athletes also feel more positively about the functions and abilities of their body (e.g., strength, stamina, and coordination; Fellows, 1999; Wiggins & Moode, 2000). As would be expected, collegiate female athletes internalize the athletic ideal more so than nonathletes (DArcy, 2007; Hoag, 2012). However, studies at the highest competition level (Division I) noted that athletes and nonathletes have similar sociocultural attitudes related to thin ideal internalization, perceived media pressure, and usi ng the media as a source for beauty information (Hoag, 2012; Rose, 2008). Research in the general female population has demonstrated a positive directional relationship between sociocultural attitudes and body size dissatisfaction (Forbes et al. 2006). Ye t, for female athletes, the desire to look athletic might protect them from body size dissatisfaction, despite their similar attitudes regarding sociocultural variables. Whether or not this is accurate is difficult to discern, as previous literature has not 60

PAGE 61

determined whether athletic internalization or desiring an athletic figure contributes positively or negatively to body image. Competition Level Differences University competition ( Division ) level differences were not examined by Hausenblas and Downs ( 2001) Our findings indicate there may be competition level differences within the college athlete group. This finding is in line with other research that indicates Division I and Division III college athletes experience BIC differently, depending on the c onstruct of measure (Kato et al. 2011) Specifically, prior research indicates that athletes at the Division I level experience more body dissatisfaction than athletes at the Division III level (Kato et al., 2011) Our results supported this, as Division I athletes were the only athletes to report being more dissatisfied with their body shape (Hoag, 2012), while athletes at other competition levels reported feeling more satisfied with their body shape (DArcy, 2007; DiBartolo & Shaffer, 2002; Robinson & Fe rraro, 2004). The issue of weight concern may also differ across competition levels. However, it was the Division I and II athletes that felt more positively regarding weight concern and preoccupation (Fellows, 1999; Robinson & Ferraro, 2004), while other athletes did not (DArcy, 2007; Wiggins & Moode, 2000). This finding supports other research that indicates Division III athletes experience greater weight preoccupation than Division I athletes (Kato et al., 2011). Notably, the results of our review indi cate that the difference in weight concern may disappear among Division I athletes and nonathletes, when the athlete group is comprised of athletes in more feminine sports such as gymnastics and tennis (Fellows, 1999). Among females in general, the relat ionship between greater weight preoccupation and body dissatisfaction to thin ideal internalization has been documented (Grabe et al., 2008; McKinley & Hyde, 1996). For that reason, our finding that competitive athletes are less 61

PAGE 62

concerned with their weight yet more dissatisfied with their body shape is co unterintuitive, considering these athletes experience levels of thin internalization similar to non athletes (Hoag, 2012; Rose, 2008). This apparent contradiction could be partly explained by highly compet itive female athletes struggling to reconcile two separate identities: being an athlete and female (Krane, Choi, Baird, Aimar, & Kauer, 2004). Research has demonstrated the positive linear relationship between a females athletic identity and BIC (Krane et al., 2004; Larabee, 2011). The more pronounced the athletic identity and the more internalized the thin ideal, the more likely the athlete will self objectify and will experience body shame (Larabee, 2011). Thus, the highly competitive female athletes in ternalization of the thin ideal, which is similar to the non athletes, is conflicted with her identity of being an athlete, thereby linked to greater BIC. It is also possible that these competition level differences are confounded by the different types o f sports included in each study. In other words, competition level and sport type may have been confounded, or may have had an interactive affect. Thus, w hile the results of this review indicate that Division I athletes appear to be at greater risk for bod y shape dissatisfaction than their college aged peers, this same difference might be explained by the fact that these Division I athletes participated in more feminine sports. Sport Differences Despite conjecture that lean focused athletes and appearance focused athletes are at greate r risk for BIC the results of Hausenblas and Downs (2001) meta analysis indicated no difference between aesthetic (appearance) endurance (lean) and ball game sports. O ur findings differed from Hausenblas and Downs (2001) as we found that BIC varied by sport grouping with a thletes in more feminine sports (e.g., gymnastics, volleyball, and tennis) found to be at greater risk for BI C and athletes in endurance sports (e.g., swimming, longdistance running, and water polo) at l ower risk for BIC. Specifically, endurance athletes reported higher body esteem, 62

PAGE 63

less self objectification, and greater overall body satisfaction than nonathletes (Fellows, 1999; Rose, 2008; Robinson & Ferraro, 2004), while feminine athletes did not (Fell ows, 1999; Hoag, 2012; Rose, 2008). Further, athletes in more feminine sports felt less positively about their weight affected body parts when compared to athletes in less feminine sports (Fellows, 1999). These results indicate that athletes in more femini ne sports may feel more negatively about parts of the body that are significantly changed by weight fluctuations (e.g., weight, figure, and waist) when compared to other athletes. As noted previously, the grouping of sports as appearance or lean focused is inconsistent in the literature, as different researchers use these terms to refer to different sports. This confusion was inherent in at least one reviewed study. Fellows (1999) compared athletes by appearancefocus, lean focus, and nonlean/nonappearance focus. Fellows (1999) inclusion of tennis as an appearance focused sport was not clear, since the author defined appearance focused as sports (gymnastics, diving, ballet) where performance and appearance are judged, and the degree to which the athlet e meets the prevailing ideal for body shape and size may greatly influence her score (p. 11). Additionally, no other literature has identified tennis as an appearance sport. However, several researchers do indicate that both gymnastics and tennis are cons idered more feminine sports (Clavio & Eagleman, 2011; Kim et al., 2010; Parsons & Betz, 2001). In light of this research, we refer to this grouping as more feminine sports. The term lean focused is also inconsistently used in the literature. Thus, we use the term endurance to include sports that emphasize leanness for performance, or require training that is mostly aerobic (e.g., long distance running and swimming). B ody image research among athletes has traditionally focused on the performance asp ect of pressure related to appearance, yet has failed to examine the sociocultural aspect. Our results 63

PAGE 64

indicate that the pressures associated with being an athlete extend beyond performance expectations to sociocultural expectations as projected and perpet uated by the medias sexualization and objectification of athletes performing in sports considered more feminine. It appears that it is now necessary to revise groupings of sport types, as sports that have not traditionally been considered appearance focus ed (e.g., tennis volleyball ) are now being overtly sexualized in the media (Clavio & Eagleman, 2011; Fink & Kensicki, 2002; Kim et al., 2010), objectified by peers (Parsons & Betz, 2001), and seem to be at increased risk for BI C Limitations One consiste nt limitation of this review was that all studies used non random sampling techniques to recruit nonathlete comparison groups. As such, the generalizability of studies analyzed in this review may be limited. Also, almost all studies collected survey respo nses from athletes during or immediately following sport practices and/or meetings. Previous literature indicates response bias may be of concern in these types of environments (Garner, 2004), because athletes may not feel assured of response anonymity or confidentiality in these settings. One of the studies (Robinson & Ferraro, 2002) reported that coaches were not present during survey administration, and athletes were assured of confidentiality. T hree reviewed studies (DiBartolo & Shaffer, 2002; Finkenbe rg et al., 1998; Krane et al., 2002) did not operationalize the term collegiate female athlete, diminishing the ability of our review to make unequivocal conclusions about the research. Without an oper ational definition of the term athlete, we are unab le to confirm if all athletes included in these studies were varsity level, NCAA sport athletes. In many cases, athletes were grouped differently across studies, which made specific between sport comparisons difficult. Additionally, the variation of includ ed sports at different competition levels means that we might have attributed some differences to 64

PAGE 65

competition level, when the difference was actually due to sport type. In other words, competition level and sport type may have been confounded, or may inter act. Conclusions and Future Research Evidence indicates participation in athletics is related to enhanced body esteem and valuing the human body for its physical abilities. The latter is not surprising given that successful athletic performance requires a certain level of bodily skill, endurance, stamina, and strength. Perceiving the body as a useful instrument is the antithesis of self objectifying, so it seems plausible that athletic involvement could provide some protection against self objectification This is unknown, however, as only one reviewed study compared athletes and nonathletes on self objectification. In fact, the most commonly used scale in the reviewed studies was the EDI BD, despite indications that the subscale is not a good single meas ure of body image or even general body dissatisfaction. Given the sexual objectification of individuals from this population, researchers should consider instruments that assess outcomes that are more salient and temporally relevant to female athlete populations. Selection of such measures can also assist researchers in examining the relationship between BI C and the medias increased sexualization of female athletes. Examples of constructs that need to be measured include self objectification and body compa rison, as well as body esteem. The latter focuses on the traditional body image measures of perceived self attractiveness as well as demonstrable physical abilities (e.g., strength, endurance). In a similar vein, researchers should consider measuring other positive body image constructs such as body appreciation and body functionality Additionally, re sults from this review suggest that enhanced athletic internalization may not indicate decreased risk for BIC; but rather there may be other, unexplored vari ables serving to moderate this relationship. Possible examples include sport objectification level (Parsons & Betz, 2001) and gender norm conformity (Steinfeldt et al., 2011). Looking more closely at 65

PAGE 66

female athletes conformity to gender norms (both male and female norms) could be particularly informative considering that female athletes may find it difficult to reconcile their female identity from their athletic identity (Krane et al., 2004; Larabee, 2011); and that female athletes are engaging in some per formance enhancing and appearancemodifying behaviors that are similar to their male athlete counterparts (Muller, Gorrow, & Schneider, 2009). Considering societys ideali zation of the athletic physique, continued research in this area is critical as is t he use of instruments that attempt to measure constructs salient to the athletic ideal. In addition to the AIS and internalization athlete subscale, researchers should consider administering Smolak and Murnens (2008) Drive for Leanness Scale. Results of this review indicate that there may be body image differences across the competition (NCAA Division) levels; although these differences might be confounded by the types of sports included in each Division level. Thus, more research should be conducted comp aring Division I female athletes to female athletes in other Divisions and to non athletes on relevant body image constructs with consideration to the level at which the included sports are perceived as feminine and/or the level at which athletes from the sport are objectified by society and the media. This is important because the results of this review indicate that athletes in more feminine (e.g., gymnastics) and objectified (e.g., tennis) sports experience greater BI C than other athletes. It is apparen t that societys view of what defines an athlete as feminine has changed, and sports that were once considered to be more feminine (e.g., golf; Kane, 1988) may no longer be viewed as such (Parsons & Betz, 2001). Further, recent research indicates that soci etys perception of an athletes femininity is affected by the medias portrayal of the athlete (Daniels & Wartena, 2011; Hardin & Greer, 2009; Harrison & Secarea, 2010). Thus, researchers should 66

PAGE 67

examine how current definitions of athletic femininity are d eveloped. Specifically, researchers should study the relationship between what society defines as a feminine sport and the medias objectification of athletes from these sports. Additionally, researchers should seek to determine how the medias sexual obje ctification of high profile, sport specific female athletes affect the BI C of female athletes participating in that sport. Continued research examining body image and female athletes can help researchers and clinicians better understand the ways in which athletic involvement can protect against BI C. This enhanced understanding can also contribute to improved programming for nonathlete females. For example, understanding that athletes value the abilities of their body more than nonathletes suggests that i nterventionists should identify ways to enhance non athletes feelings regarding their bodys abilities. Participation in college level varsity sports is not possible for every female, and regular participation in general exercise and physical activity may not be sufficient to enhance nonathletes feelings toward the utility of their body. Researchers and interventionists should explore how dedication to perfecting a physical skill (e.g., roller blading or rock climbing), even outside of the sports realm, m ay enhance body esteem. Continued empirical and conceptual efforts will develop new knowledge that can assist researchers and practitioners in the selection of effective methods for addressing BIC in female athletes, especially with respect to internalized sociocultural beauty norms and objectification. 67

PAGE 68

Table 2 1. Characteristics of r eviewed a rticles Study Sample Type & Size School, Type or Region (Divisi on) Measures a Results DArcy (2007) b Athletes, n = 88 Nonathletes, n = 105 York Univers ity, Canada (N/A) BMI FRS AIS EDI BD EDI DT SATAQ 3 Athletes had a higher BMI; larger current figure (AIS) and ideal figure (AIS); and internalized the athletic ideal more. Non athletes had significantly greater body dissatisfaction (FRS & EDI BD); used the media for information about appearance more; and perceived greater pressure from the media to conform to sociocultural beauty ideals. DiBartolo & Shaffer (2002) Athletes, n = 115 Non athletes, n = 94 Liberal arts coll ege, northeastern (Division III) EDI BD BIS Non athletes experienced greater BI C as evidenced by greater body dissatisfaction (EDI BD; BIS). Fellows (1999) b All athletes, n = 111 Lean sports, n = 21 Appearance sports, n = 21 Other sport, n = 69 Non athlete, n = 164 University, western U.S. (Division I) BES WC SA PC All athletes had more positive body esteem and PC than non athletes. Lean athletes also had more positive SA and WC than non athletes. Other athletes had more positive overall body esteem and WC than appearancefocused athletes. No differences between other athletes and lean athletes. Finkenberg et al. (1998) Athletes, n = 108 Kinesiology majors, n = 87 Control, n = 63 University, sou thwestern (not noted) SPAS SPA was lower for athletes; however the groups were not discriminated by SPA. Hoag (2012) b Volleyball players, n = 209 Division I, n = 33 Division II, n = 83 Division III, n = 92 Non Athlete, n = 117 33 U niversities, across the U.S. (All Divisions) BMI AIS EDI BD SPAS BAS SATAQ 3 Athletic status was related to body image after controlling for BMI, as was competition level. Compared to non athletes, athletes had a larger, more athletic actual and ideal figure (AIS), as well as greater internalization of the athletic ideal. Division I athletes had greater body shape dissatisfaction (EDIBD) than nonathletes. Krane et al. (2002) All athletes, n = 204 Revealing uniform, n = 85 Baggy uniform, n = 65 Mixed uniform, n = 54 Aerobic exercisers, n = 198 4 Universities, Midwest (Division I) EDI BD EDI DT SPAS Body satisfaction did not differ between groups, nor did physique about anxiety. The two body satisfaction var iables were the strongest predictors of social physique anxiety in both exercisers and athletes. 68

PAGE 69

Table 2 1. Continued Study Sample Type & Size School, Type or Region (Divisi on) Measures a Results Robinson & Ferraro (2004) All at hletes, n = 53 Speed focused, n = 37 Technique focused, n = 16 Non athlete, n = 55 University of North Dakota (Division II) EDI BD EDI DT Current/ideal weight Current/ideal body type Non athletes were more weight preoccupied (EDI DT), and dissatisfied with body weight and shape than both athlete groups. Non athletes had higher BMI than speed focused athletes. No differences between the two athletes groups. Rose (2008) b Gymnastics, n = 29 Water polo, n = 41 Non athlete, n = 22 Two universities, West coast (Division I) SOQ OBCS BS Non athletes self objectified more than water polo athletes, but similarly to gymnasts. Combined self objectification and body shame predicted group membership with 46% accuracy. Ste infeldt et al. (2011) Athletes, n = 78 Non athletes, n = 65 Private college (Division III) BES Body esteem significantly correlated to athletic status ( r = .28, p < .001). Wiggins & Moode (2000) Athletes, n = 57 Non athletes, n = 67 Two universities, southern medium sized (Not noted) WC SA PC Athletes felt more positively about their physical condition than non athletes. No difference between groups on the body esteem constructs of WC or SA. BMI = Body Mass Index; FRS = Fig ure Ratings Scale; AIS = Athletic Image Scale; EDI BD = Eating Disorders Inventory Body Dissatisfaction Scale; EDI DT = Eating Disorder Inventory Drive for Thinness; SATAQ 3 = Sociocultural Attitudes Toward Appearance Questionairre3; BIS = Body Image Survey; BIC; Body Image Concerns; BES = Body Esteem Scale; WC = Weight Concern; SA = Sexual Attractiveness; PC = Physical Condition; SPAS = Social Physique Anxiety Scale; BAS = Body Appreciation Scale; SOQ = Self Objectification Questionnaire; OBCS BS = Object ified Body Consciousness Scale Body Shame. aOnly measures used to compare BI C in athlete and non athlete groups were included. bUnpublished Doctoral dissertation or Masters thesis. 69

PAGE 70

Table 2 2. Included sports in the reviewed studies alongside Parsons and Betz (2001) objectification scores Sport Objectification Score Reviewed Study DArcy a,b,c DiBartolo & Shaffer d Fellows Hoag Krane et al. e Robinson & Ferraro Rose b Steinfeldt et al. c Wiggins & Moode Golf 2.09 X X Lacrosse 2.12 X Crew f 2.17 X Field Hockey 2.25 X X X Basketball 2.27 X X X X X X Soccer 2.33 X X X X X Cross Country 2.51 X X X X X X Softball 2.58 X X X Track 2.60 X X X X X X Volleyball 2.93 X X X X X X X X Diving 2.99 Swimming 3.17 X X X Tennis 3.29 X X X Gymnastics 4.02 X X X Overall Mean 2.67 Notes : Objectification scores averaged from appearance and femininity scores for each sport. Rating scales were based on 1 to 5 ra tings (1=lit tle focus on appearance or least feminine, 5=strong focus on appearance or most feminine; n = 195 first year undergraduate students at a large public University). Fourteen of the 17 sports from Parsons and Betzs (2001) study are included in Table 22. Dan ce Team, Synchronized Swimming, and Cheerleading were removed (non NCAA sports). Mean score = 2.67 for these 14 sports. Finkenberg et al., 1998 is not included, due to no sport list provided by the authors. a Included, but not listed in Parsons and Betz (2001): Ice Hockey b Included, but not listed in Parsons and Betz (2001): Water Polo c Included, but not listed in Parsons and Betz (2001): Rugby d Included, but not listed in Parsons and Betz (2001): Skiing; Squash e Included, but not listed in Parsons and Betz (2001): Precision Figure Skating f Designated as Rowing by NCAA 70

PAGE 71

CHAPTER 3 VALIDITY OF THE EATI NG DISORDER INVENTORY FOR NONCLINICAL RESEARCH IN POPULATIONS OF WESTERNIZED WOMEN : A CRITICAL REVIEW OF FACTOR ANALYTIC STUDIES Body image disturbance (BID) is a multidimensional disorder characterized by an individuals negative attitude toward, or evaluation of, weight, shape, or size of their body (Cash & Szymanski, 1995; Garner, 2004) BID is a diagnostic criterion for eating disorders such as anore xia nervosa (AN), bulimia nervosa (BN), and eating disorders not otherwise specified (EDNOS ; American Psychiatry Association [APA], 1994) Research ers ha ve widely reported the positive relationship between disordered eating (subclinical level eating pathol ogy) and BID in nonclinical populations of women (Calogero et al. 2005; Fitzsimmons Craft, 2011; Grabe et al., 2008). Both BID and disordered eating have been linked to several negative health behaviors and psychological concerns among women, including a lcohol use (Carr & Szymanski, 2010) depression (Carr & Szymanski, 2010; Ilaria, 2006; Szymanski & Henning, 2007; Tiggeman & Kuring, 2004) decreased self esteem (Tylka & Sabik, 2010) anxiety (Ilaria, 2006) and stress (Ruggiero et al. 2008). Due to the relationship between BID and disordered eating, the two constructs are often assessed concurrently using numerous types of survey instruments. Most BID instruments were developed and validated for use in nonclinical populations; while most eating disorder instruments were developed and validated in clinical settings for females suffering from clinicallevel eating disorders (Tury, Gulec, & Kohls, 2010) As such, most eating disorder survey instruments are administered to assess symptoms, plan for treatment, or evaluate treatment effectiveness among individuals with and without eating disorders (Espelage et al., 2003; Tury et al., 2010) Because eating disorder instruments can generally discriminate between those with and without eating disorders, nonclinical researchers also use these screening assessments. Over 71

PAGE 72

the last several decades, one of the most popular eating disorder instruments used to identify groups at greater risk for eating pathology and/or BID has been the Eating Disorder Inventory (EDI ; Campb ell & Hausenblas, 2009; Garner et al. 1983; Hausenblas & Downs, 2001; Roberts et al. 2006). The original EDI is a 64item s urvey organized into eight subscales (Garner et al., 1983) Five subscales measure psychological traits ( i.e., perfectionism, ine ffectiveness, interoceptive awareness, maturity fears and interpersonal distrust) and three subscales measure behavioral traits ( i.e., body dissatisfaction, bulimia, and drive for thinness ) or eating and weight related (EWR) subscales associated with eating disorders (Garner et al., 1983; Garner, 2004) The psychological traits subscales have been criticized for their inability to differentiate eating disorder patients from patients with other psychological disorders, calling into question their use fulness for identifying subclinical eating pathology or BID (Cooper, Cooper, & Fairburn, 1985; Hurley, Palmer, & Stretch, 1990; S ch oemaker, Verbraak, Breteler & Van der Staak 1997) On the other hand, the frequent administration of the EWR subscales have shown these scales to be highly relevant to eating pathology and BID in both clinical and nonclinical populations (Campbell & Hausenblas, 2009; Grabe et al., 2008; Hausenblas & Downs, 2001; Roberts et al., 2006; Smolak et al. 2000). Each of the three EW R subscales are often administered in dividually in nonclinical research to determine BID risk ; however, all three subscales can be administered together in nonclinical research to obtain an overall disordered eating risk score. Since its inception, the ED I has been modified twice. In its first revision, the EDI 2 added 27 new items ( new total = 91 items ) ascribed to three additional psychological trait subscales ( asceticism impulse regulation, and social insecurity ; Garner, 1991). The EDI 3 maintained the 72

PAGE 73

same 91 items as the EDI 2, but rearranged the items into 12 subscales instead of 11. Some EDI 3 subscales are unchanged from the EDI 2, yet others (and their labels) have been modified. Correlations between subscales included in each EDI iteration are hi gh (see Table 31), suggesting that researchers and clinicians can assume equivalence (Garner, 2004, p. 140) in the three iterations, allowing for score comparisons across EDI versions. It should be noted that nonclinical researchers continue to frequent ly use subscales from the original EDI more often than updated versions due to its availability length, and relative comparability to the EDI 2 and EDI 3 (De Bara rdis, Serroni, & Campanella, 2009; Gordon, Castro, Sitnikov, & Holm Denoma, 2010) Despite t he longstanding popularity of the EDI among nonclinical researchers, there is debate as to whether this clinical survey instrument is appropriate for measuring BID and eating pathology in nonclinical populations. Specifically, researchers have noted that EDI items may be interpreted differently by individuals without eating disorders ( Clausen et al. 2011; Kashubeck West et al. 2001) Moreover, data acquired from patients with eating disorder s versus nonclinical comparison groups m ay lead to different EDI structures (Garner, 2004; Garner et al., 1983) One possible reason for these differences is because the EDI item to subscale assignment was primarily based on clinical experience and conceptual fit (Garner, 2004) For example, despite statistical indications (itemto factor correlations) that item 47 belongs with other drive for thinness items, it was assigned to the body dissatisfaction subscale of the EDI 3 due to content related considerations exclusive to individuals with clinically diagnosed eating disorders. The aforementioned issues have led researchers to suggest the need for item level investigations of the EDI subscales when administered in nonclinical populations (Clausen et al., 2011; Espelage et al., 2003). 73

PAGE 74

To date, no studies have critically reviewed factor analytic studies due to content related considerations exclusive to individuals with clinically diagnosed eating disorder s T he aforementioned issues have led researchers to suggest the need for item level investigations assessing the admi nistration of the EDI in nonclinical populations A critical psychometric analysis of studies using the EDI will assist researchers in identifying which EDI items and subscales may be inappropriate for measuring eating pathology and BID in nonclinical population s of women from Westernized countries. Limiting this cr itical review to studies of women from Westernized countries is appropriate because development of the EDI and its subsequent revised iterations were tested in females from Westernized countries (Garner, 1991; Garner, 2004; Garner et al., 1983) In addition, f actors affecting the development of BID, and subsequently eating disorders, are different for females in Westernized versus nonWesternized and Asian countries (Forbes & Jung, 2008; Holmqvist & Frisen, 2009). Methods Due to its aggregated nature, a meta analysis is not appropriate for examining survey instruments at the itemlevel; thus, a critical review of factor analytic studies is the optimal method for exploring the internal structure of the EDI The critical review described here was conducted in three phases. In Phase 1, we systematically identified research studies In Phase 2, we systematically evaluated the methodological quality of all retained studies. In Phase 3 we synthesized the analytic results of the reviewed studies to propose a theoretical structure for using the EDI when conducting assessments in nonclinical populations of women. Search Process (Phase 1) Phase 1 consisted of a twostep search protocol. First, a cross search of Academic Search Premier, Communication and Mass Media Complete, Psychology & Behavioral Sciences Collection, PsychINFO, PubMed, Sociological Collection, and Womens Studies International 74

PAGE 75

was conducted by inputting the following search terms: eating disorder inventory, EDI, validity, factor, and psychometric. Second, the reference sections of all studies identified in the first step were searched to locate additional studies. I nclusion criteria required that each study: (a) aimed to explore the iteml evel factor structure of the 64 or 91item EDI in a nonclinical population of women; and (b) was a vailable in the English language. S tudies were excluded if they: (a) were conducted in nonWe sternized or Asian countries; (b) sampled females under 18 year s of age; (c) investigated only clinical populations; (d) did not analyze all EDI items (either 64 or 91 items); or (e) did not provide EDI itemlevel information (e.g., item to factor loadings). Figure 31 describes the search and screen process. Ultimately, six articles ( n = 6) met the inclusion and exclusion criteria and were retained for evaluation and review. Evaluation of Study Methods (Phase 2) Phase 2 of our critical review involved evaluating the methodological quality of studies retained after P hase 1 t o determine the relative strength of each study when evaluated against a set of factor analytic (Birrell, Meares, Wilkinson, & Freestno, 2011; Sveinbjornsdottir & Thorsteinsson, 2008) and methodological (Mokkink et al., 2009) criteria S everal stat istical resources were also consulted to identify specific best practice methods for factor analysis to use in our evaluation (Costello & Osborne, 2005; Crocker & Algina, 2008; Gorsuch, 1983; Hu & Bentler, 1999; Kachigan, 1986; Nunnally, 1978; Pett, Lackey & Sullivan, 2003; Stellefson, Hanik, Chaney & Chaney, 2009; Wheaton, 1987) Through our review of these resources, we identified three broad evaluation categories to consider: (1) samp ling (2) data analysis and (3) result reporting. This evaluation ena bled the research team to determine which studies were less rigorous and therefore given less consideration when synthesizing results from the reviewed studies in Phase 3. 75

PAGE 76

Types of factor analysis To develop appropriate evaluation criteria, it was necess ary to identify the types of factor analyses used in the studies retained from Phase 1. The retained studies used either Principal Components Analysis (PCA) or Confirmatory Factor Analysis (CFA). Both PCA and CFA studies were considered relevant to this cr itical review, because item level psychometric information can be obtained using either method. However, the methods for conducting a CFA versus a PCA differ; thus, the two types of factor analytic studies were evaluated separately. PCA is considered a mo re exploratory method of factor analysis, primarily because a prespecified model is not typically expected. In CFA, items are expected, a priori to load onto certain factors; and emphasis is on confirming (or refuting) a theoretically existing factor str ucture, rather than exploring the data to identify a best fitting model (Gorsuch, 1983) The criteria described within each of the following categories are specific to PCA studies. The CFA criteria will be described separately. For all studies, criteria were scored as being met ( X ), not met (0), not performed (NP), information insufficient to evaluate ( ) or not applicable (NA). Sampling methods Evaluation of a studys sampling method is necessary to determine the validity of the study sample (i.e., if the sample is appropriate based on study aims). First, we evaluated whether or not the sampling method for each study was conducted in a way that reduced sampling error. Second, we examined the demographic data of each studys sample to evaluate whether or not the sample was representative of the population of interest Third, we evaluated sample size. T he best approach for researchers planning a factor analytic study is to assess the suitability of the collected data for factor analysis with a statistical te st (e.g., Kaiser Meyer Olkin ; Gorsuch, 1983; Pett et al., 2003). Studies indicating application and interpretation of such a n a priori statistical test met the parameters of the sample size criterion ( X ). Unfortunately, these tests are 76

PAGE 77

underutilized or underreported ; thus it could be necessary to assess sample size with a participant to item ratio (Gorsuch, 1983; Pett et al., 2003) Thus, s tudies that did not test for sample adequacy were scored as not performed (NP), and a sample size criterion of ten participants per item was applied. This participant to item ratio (10:1) is more likely to result in a correct factor structure, with fewer misclassified items and factor loading errors than a smaller ratio (i.e., 2:1 or 5:1; Costello & Osborne, 2005; Nunna lly, 1978). Studies that met the ten to one minimum ratio met this criterion ( X ) Data analytic methods In factor analytic studies, researcher s must make several assumptions about data. These include expected correlations of factors, normality of the data, and if a pre specified factor structure is anticipated. D escription of these assumptions assists the reader in interpreting study findings. Studies explicitly describing these assumptions met the assumption criterion ( X ) Next, we considered whether the reviewed studies identified a minimum itemto factor correlation. I tem to factor correlation indicates the relationship of the item to the factor, with high correlations (.70 .80) useful for defining the resulting factor. A minimum itemto factor corre lation of .30 is commonly applied in factor analysis (Crocker & Algina, 2008; Kachigan, 1986) A higher cut off of .40 is typicall y observed in PCA due to the tendency for PCA to result in inflated itemto factor correlations (Costello & Osborne, 2005; Gor such, 1983) In this review, studies applying the .40 minimum met the itemto factor correlation criterion (X). Studies indicating a lower minimum were scored as criterion not met (0). Studies failing to indicate a cut off were noted as providing insuffici ent information ( ). Another consideration of item retention is whether or not the item is complex. Complex items are items loading above the itemto factor minimum on two or more factors. When deciding to retain or drop a complex item, some researchers su ggest considering the items 77

PAGE 78

content relevance and its contribution to the internal consistency (or reliability) of the factor (Kachigan, 1986; Pett et al., 2003) D evelopers of the EDI retained several complex items, assigning them to the most conceptually relevant subscale (Garner, 2004) For this reason, we scored the evaluated studies on this criterion based on whether the complexity issue was a ddressed Evaluated studies met this criterion if complex items were noted ( X ). Next, we examined each stud ys method of component extraction. A clear explanation of these methods allows the reader to judge the validity of the proposed model structure (Costello & Osborne, 2005; Gorsuch, 1983; Pett et al., 2003; Stellefson et al., 2009) There are many tests available to assist researchers in determining the number of components to retain. The the least accurate (Costello & Osborne, 2005; Gorsuch, 1983; Stellefson et al., 2009) Another commonly used method, generally considered superior to the eigenvalue method is Cattells (1966) scree test (Costello & Osborne, 2005) Other methods include: the standard error scree test (Zoski & Jurs, 1996) Velicers Map criteria (1976) a nd parallel analysis (Horn, 1965) Due to varying results observed when using these different extraction methods, researchers should consider several methods in determining the number of factors to retain (Stellefson et al., 2009) Thus, evaluated studies had to apply at least two factor extraction methods to meet this criterion ( X ) Another important consideration is internal consistency, which is generally reported (Gorsuch, 1983; Nunnally, 1978) This statistic i s necessary to evaluat e the reliability and validity of each retained factor (Gorsuch, 1983) As suggested for instruments administered in the social sciences, we considered an appropriate (Nunnally 1978) Studies using the .70 78

PAGE 79

cutoff met this criterion ( X ). Studies indicating a higher or lower cut off, or failing to provide internal consistency scores, did not meet this criterion (0). Several reviewed resources suggest the number of items loading onto each factor should be considered when deciding to retain or exclude a factor (Costello & Osborne, 2005; Gorsuch, 1983; Pett et al., 2005) Factors with fewer than three items are generally weak and unstable, while factors with at least five items ar e desirable and indicat e a solid factor (Costello & Osborne, p. 5, 2005) We designated a minimum of five items per extracted component for this criterion to be met (X) Result r eporting In o ur final evaluation catego ry, we examined the results and result interpretations reported in each study. First, we evaluated whether a visual interpretation of results was provided. Visual interpretation is an integral part of valid interpretation and can be achieved with table displays of data, graphs, plots, etc (Croc ker & Algina, 2008; Gorsuch, 1983; Kachigan, 1986; Pett et al., 2003) Within each reviewed study, we looked for graphic displays of item to factor correlations; as well as inter factor correlations. For the second criterion in this category, we assessed whether the final model constructs were operationally defined in a meaningful way. A meaningful definition is one that is made explicitly through a description of the constructs relationship to other variables (Crocker & Algina, 2008; Gorsuch, 1983) Revi ewed studies that recommended a model structure differing from the clinically derived EDI met this criterion ( X ) if all new constructs were defined in a way considered meaningful to eating pathology and BID research. CFA Evaluation Some of the evaluative criteria for PCA studies are also applicable to CFA C rit eria for PCA studies that are not appropriate for evaluating CFA studies include: factor extraction 79

PAGE 80

method; itemto factor ratio; and meaningful operationalization of new constructs. One PCA criterio n requiring modification for evaluatin g CFA studies is the itemto factor correlation cut off. While this cut off for the PCA studies was designated at .40, it is acceptable for studies using common factor analysis to set this cut off at .30 (Crocker & Alg ina, 2008; G orsuch, 1983; Pett et al., 2003) Thus, CFA studies applying an item to factor correlation minimum of 0.30 met this criterion ( X ) A statistical criterion specific to CFA, but not PCA, is goodness of fit. Goodness of fit tests can indicate a su pported (or unsupported) factor structure. The most basic measure for goodness of fit is chi square; however, this method is highly susceptible to error with larger sample size s (Gorsuch, 1983; Hu & Bentler, 1999; Wheaton, 1987) T o offset this effect of s ample size, relative chi square (chi 2 / df) is frequently calculated and reported; however, there are indications that relative chi square can also be a ffected by sample size (Wheaton, 1987) As such, there are multipl e other recommended methods for assessing a models goodness of fit (Hu & Bentler, 1999; Wheaton, 1987) For Maximum Likelihood (ML) CFA studies Hu and Bentler (1999) recommend assessing model fit using a ML based Standardized Root Mean Square Residual (S RMS) in addition to at least one of the following indices: Comparative Fit Index (CFI), Tucker Lewis Index (TLI), Bollens Fit Index (BL89), Relative Noncentrality Index (RNI), Gamma Hat, McDonalds Centrality Index (Mc), or Root Mean Square Error of Approximation (RMSEA). CFA studies following Hu and Bentlers (1999) recommendation of assessing goodness of fit with SRMS and a second (aforementioned) index met the goodness of fit criterion ( X ). 80

PAGE 81

Results Eligible Studies (Phase 1) S ix ( n = 6) eligible studies published between 1987 to 2003, were identified in the Phase I search (see Table 3 2). The six studies included a total of nine nonclinical samples of women, with the median age for each sample rang ing from 18 (Klemchuk et al. 1990; Raciti & Norcross, 1987) to 26 (Welch, Hall, & Walkey, 1988) F ive studies (Klemchuk et al., 1990; Raciti & Norcross; Welch et al., 1988; Bennett & Stevens, 1997; Limbert, 2004) analyz ed data from the EDI with PCA while one (van Strien & Ouwens, 2010) analyzed data with ML CFA The five PCA studies focused on the original 64item EDI T he CFA study explored both the 64item EDI and the 91item EDI 2. Methodological Quality and Results of Reviewed Studies (Phase 2) Table 33 summarizes the evaluation results from Phase 2. For each study, we summed the total number of met criteria ( X ) for a criteria met score. This was repeated for the unmet criteria (0) to obtain a criteria not met score. Also of interest were the number of criterion marked as insufficient to evaluate ( ). Application of the evaluation criteria revealed that the methodological rigor of the studies varied. For the single CFA study in this review (van Strien & Ouwens, 2010) six of the ten applicable criteria were met ( X ), while two were not met (0), and one had insufficient information to evaluate ( ). Out of the twelve criteria applied to the PCA studies, the criteria met ( X ) score ranged from four (Bennett & Stevens, 1997) to seven (Raciti & Norcross; Welch et al., 1988) ; criteria not met (0) ranged f rom two (Klemchuk et al., 1990; Raciti & Norcross) to five (Limbert, 2004) ; and t he number of criteria scored as insufficient information to evaluate ( ) ranged from one (Limbert, 2004; Raciti & Norcross) to four (Klemchuk et al., 1990) Based on these r esults, the three studies noted as being the strongest methodologically were Raciti and Norcross (1987; X = 7/12; 0 = 3/12; = 1/12), van Strien and 81

PAGE 82

Ouwens ( 2010; X = 6/1 0; 0 = 3/10; = 1/10), and Welch et al. ( 1988; X = 7/12; 0 = 3/12; = 2/10). Across the studies, sampling methods appeared to be the weakest dimension. Despite four studies (Klemchuk et al., 1990; Limbert, 2004; Raciti & Norcross; van Strien & Ouwens, 2010) appropriately describing sampling strategy, Klemchuk et al (1990) was the only s tudy that provided a sample demographic detailed enough to determine if the sample was representative of the study aims. Another common weakness related to sampling methods was sample size. None of the evaluated studies reported applying a statistical test to assess the adequacy of the sample for factor analysis; thus, all studies were evaluated on sample size using the ten to one (10:1) participant to item ratio. Limberts (2004) study was the only one to meet this minimum. Klemchuk et al (1990) came very close to meeting this criterion with the participantto item ratio just under tento one for each of the two samples in their study (N1 = 9.7:1; N2 = 9.9:1). The only criterion noted as being met by all studies was explicit assumptions about the data. T he studies varied in their scores for other criteria in the data analytic methods category, as well as the result reporting category. Bennett and Stevens (1997) Based on our quantitative analysis, Bennett and Stevens study was less rigorous ( X = 4/12; 0 = 4/12; = 3/12). While most reviewed studies reported running analyses several times with varying constraints and factor rotation methods, Bennett and Stevens description of their analysis was much less specific. The authors reported generally that PCA with varimax rotation was used, applying an item to factor correlation minimum of .4 and an eigenvalue cut off of 1.5. The authors further report, the method of extraction produced the most parsimonious solution with seven factors extracted (para. 15) provided. Overall, the description of the ir data analytic method is insufficient to allow 82

PAGE 83

researchers to attempt replication of the study. Further, this missing information makes valid interpretation o f the results difficult for the reader. On the other hand, one strength of Bennett and Stevens study, which is true for only one other study in this review (van Strien & Ouwens, 2010) was their detailed reporting of itemto factor correlations for complex items. O ther studies reported items removed due to complexity, but did not note which factors these items correlated with. The final model structure reported by Bennett and Stevens differed from the clinically derived eight subscale EDI, and was notably difficult to interpret; thus, Bennett and Stevens concluded the EDI to be inappropriate for use in nonclinical populations. Klemchuck et al (1990) Of the six reviewed studies, Klemchuk and colleagues had the most evaluation criteria marked as insufficient information to evaluate ( X = 5 /12; 0 = 2/12; = 4/12). T he ir factor extraction method criterion was particularly puzzling While sole use of the eigenvalue method was reported, the resulting 16factor solution was discarded due to ten of the factors being uninterpretable and accounting for little variance. The authors reran the data three separate times constraining items to load on five six and eight factors, subsequently rotating each solution both orthogonally and obliquely. T he six factor model provided the most meaningful and easily interpretable solution (p. 299). Costello and Osborne (2005) recommend an extraction approach similar to that reported by Klemchuk and colleagues, with a few differences. First, the application of a scree test (or mor e rigorous extraction method) is recommended over the eigenvalue method to inform factor extraction. Second, based on Costello and Osbornes (2005) recommendations, it seems a seven factor solution should have also been examined. Since Klemchuk et als met hodological process for factor extraction was difficult for our research team to interpret this criterion was marked as having insufficient information to evaluate ( ). 83

PAGE 84

Two other criteria marked insufficient were the itemto factor ratio and visual interp retation of itemto factor correlations. T he authors reported the items that loaded onto only the first two factors, and itemto factor correlations for only the second factor. The authors elected to focus on these results due to the second factors signi ficant departure from the clinically derived ei ght scale structure of the EDI (p. 299) which is noted by its inclusion of items from three EDI subscales ( bulimia drive for thinness and interoceptive awareness ). T he first factor was noted as being a repl ication of the nine item body dissatisfaction subscale. T he final four factors were indicated as highly similar to the ineffectiveness, interpersonal distrust, maturity fears, and perfectionism scalesof the EDI (p. 299). This high similarity is supporte d by high correlations between the studys four derived factors and their corresponding EDI subscales. These correlations were noted in a table that displayed the inter correlations of the studys six factors to the eight subscales of the EDI. The provision of this table resulted in the criterion of visual interpretation of factor to factor correlations being marked as met ( X ). Also missing from Klemchuk et als report was notation of complex items and al limitations, this study had five criteria that were fully met, with a sixth almost met (sample size). Overall, the study appears methodologically strong; however, our ability to interpret the results of Klemchuk et al at the itemlevel is limited becau se of the noted missing information. Limbert (2004) Limbert had several criteria marked as not met ( X = 5/12; 0 = 5/12; = 1/12). Notably, some of the se criteria being unmet (e.g., visual interpretation, notation of complex items, item to factor ratio) co uld have resulted from the aims of the study, which was to determine the replicability of the clinically derived EDI in a nonclinical population. Additionally, Limbert used the psychometric standards indicated for clinical EDI use to evaluat e the instruments 84

PAGE 85

replicability. was ascribed for internal consistency. S uch a rigorous cut off is only necessary in applied settings (i.e., to determine treatment, enhance learning, or evaluate treatment ; Nunnally, 1978). A lower cut off of .70 is acceptable for research in nonclinical populations. V isual interpretation of itemto factor correlations was not met (0); however, a table was provided indicating which items loaded onto which factors. This, combined with a known itemto factor correl ation minimum of .40, allows for a basic item level interpretation. Limbert concluded that the eight subscale EDI was not successful ly replicated despite several subscales being replicated ( body dissatisfaction and perfectionism ) or nearly replicated ( bul imia, drive for thinness and interpersonal distrust Raciti and Norcross (1987) Raciti and Norcross also attempted to replicate the eightsubscale structure of the clinically derived EDI; however, their provis ion of detailed information (e.g., factor to item correlations and complex items) resulted in the study being noted as methodologically strong ( X = 7/12; 0 = 3/12; = 1/12). One strength was their application of two factor extraction methods (scree test a supported the original eight factor structure. Welch et al (1988) = 2/12) also implemented a factor extraction method, FACTOREP, that increased the strength and rigor of their investigation. This method differs from those indicated as acceptable by our designated evaluation criteria. However, it was determined that the authors application of the FACTOREP procedure should be included as meeting the factor extraction method criterion. FACTOREP is a program that conducts a series of factor analyses, varying the number of factors used in each rotation until a structure meeting 85

PAGE 86

the resear chers specifications is found that replicates across several i ndependent groups (in this case, J = 3; Walkey & McCormick, 1985). The final model proposed by Welch and colleagues differed from the original EDI It included only 44 items comprising three components ( perfectionism; selfesteem; and weight, shape, and eating concern) A ll items from two subscales ( maturity fears and interoceptive awareness ) were removed from the final model due to an unclear pattern of loadings (p. 53) as were two EWR items for complexity ( bulimia #53) and low item to factor correlat ions ( drive for thinness #1). The authors included a table noting the final models itemto factor correlations for each of the three samples, resulting in the visual interpretation criterion being marked as met ( X ). However, the table was uninterpretable due to item number and factor loading inconsistencies. The nature of the errors indicate s that it is likely they occurred during the printing process, and were not a result of author error. Unfortunately, this error coupled with the authors use of the .30 itemto factor correlation minimum makes our interpretation of itemlevel specifics difficult. van Strien and Ouwens (2010) Based on the evaluation criteria, van Strien and Ouwens study appears methodologically strong (X = 6/10; 0 = 3/10; = 1/10). A str ength of the ir study was the provision of a detailed table denoting itemto factor correlations above .40 for all items, including complex items. van Strien and Ouwens comparison of the 64item EDI structure to the 91 item EDI 2 structure led them to co nclude that items from the three additional EDI 2 psychological trait subscales did not add anything unique for nonclinical populations beyond the original eight subscales. It was also reported that the clinically derived eight factor structure of the EDI was not a good fit for data from nonclinical populations, despite most items loading principally with their assigned subscales (p. 79). This poor fit was indicated by the relative chi square being just over a designated 2.00 maximum ( 2/df = 2.04). A more rigorous goodness of fit method was 86

PAGE 87

indicated by our evaluation criteria; thus, van Strien and Ouwens, did not meet this criterion. As state d previously, the relative chi square test can lead to inaccurate results (Wheaton, 1987) Fur ther, the dependence of relative chi square on sample size makes general threshold guidelines impossible to suggest, which means satisfactory fit cannot be clearly defined with a cut off score (e.g., 2.00; Wheaton, p. 128, 1987). Synthesis (Phase 3) In P hase 3 of our critical review, we synthesized the internal structure results presented in each of the six studies to identify a potentially stable structure for the EDI in nonclinical populations of Westernized women. Table 3 4 displays the internal struct ure results from each study [ i.e., which items loaded together onto which component ( if provided) and items that were either noted as complex or otherwise removed from the final model of each study ] Bennett and Stevens (1997) and van Strien and Ouwens (2010) were the only two reviewed studies to report which factors the complex items loaded onto. Thus, complex items listed in these studies are inc luded in the I tems column with the corresponding factor, as well as the Remo ved & Complex Items column. To identify a potentially stable model structure, we examined the itemmapping patterns across the six reviewed studies, as well as reliability and validity evidence (e.g., itemto factor correlations, item loading, and internal consistency coefficients) presented in each study. Based on our methodological evaluation, we gave the least consideration to the results of Bennett and Stevens (1997) study. We further considered that Klemchuk and colleagues (1990) pr ovided itemlevel specifics for only two of the final six factors. Thus, the greatest consideration was given to the results presented by the four remaining studies (Limbert, 2004; Raciti & Norcross; van Strien & Ouwens, 2010; Welch et al., 1988). Synthesis of the reviewed studies results revealed f ive components represented by 51 items (see Table 3 5) The first component is 87

PAGE 88

evident in the first or second factor in five studies (refer to Table 3 4; Bennett & Stevens, 1997; Limbert, 2004; Raciti & Norcross; van Strien & Ouwens, 2010; Welch et al., 1988). The second, third, and fourth components reflect existing EDI subscales ( body dissatisfaction, perfectionism and interpersonal distrust ), which were replicated in each of the reviewed studies. The fifth and final c omponent was less clear through item mapping comparisons; however, examination of other validity and reliability evidence suggested the combination of several EWR items. Discussion The validity of administering the clinically derived EDI in nonclinical populations has been questioned and evaluated by BID and eating pathology researchers for decades. We aimed to critically review factor analytic studies that examin ed the internal structure of the EDI in nonclinical populations of Westernized women, in an effort to i dentify a stable EDI structure for this population. A systematic search process revealed six studies eligible for review. Our systematic evaluation of the methodological quality of the studies indicated they varied in terms of methodological rigor. We note however, that the exclusion or under reporting of some criteria by each study could be attributable to publication restraints and the relative importance of reporting other aspects of the study. In our final step, we synthesized the internal structure results presented by the reviewed studies. Through this synthesis, we were able to identify a 51 item EDI structure comprising five factors: low selfesteem, body dissatisfaction, perfectionism, interpersonal distrust, and eating disorder risk Evidence for each component is provided below with consideration given to the reviewed studies results and the broader body of BID and eating pathology literature. Low Self Esteem The first component of our proposed model low selfesteem is comprised of ten ineffect iveness items, four interoceptive awareness items and one maturity fears item. All 15 of 88

PAGE 89

these items loaded together on the first or second component in three studies ( Limbert, 2004; Raciti & Norcross; van Strien & Ouwens, 2010) with high internal consist Bennett and Stevens (1997) also present a large first component with 12 of these items. Ten of the 15 items load ed and colleagues (1988) Notably, Welch et al. reported removing all maturity fears and interoceptive awareness items from their final model due to an unclear pattern of loadings across three samples (p. 53) The combination of items from the ineffectiveness and interoceptive awareness subscales is not surprising as the two constructs measured by these subscales overlap significantly (Bruch, 1973; Garner et al., 1983 ) Despite the strong relationship between the two constructs, EDI developers purported to measure them separately because the subscal e scores were considered valuable in discriminating between different types of eating disorders (Garner et al., 1983) For the nonclinical researcher who is interested in group level comparisons of general BID or eating disorder risk, separation of the tw o constructs seems superfluous. This is apparent in the results of the reviewed factor analytic studies, which demonstrate that these items consistently load together with one additional item (#35) This low selfesteem component appears to evaluate the le vel of insecurity about ones self worth (e.g., I feel inadequate.) and emotions (e.g., I get confused about what emotion I am feeling.). Construct validity for this subscale is seen in research that confirms the relationship of the component s items to other validated self esteem measures (i.e., Rosenberg Self Esteem Scale [RSES] ; Garner, 2004). Inclusion of this component is relevant for nonclinical BID and disordered eating researchers due to the significant amount of nonclinical research linking sel f esteem to BID (Fingeret & Gleaves, 2004; Franzoi & Herzog, 1986; Shea & 89

PAGE 90

Pritchard, 2007; Tylka & Sabik, 2010) and disordered eating ( Berland, Thompson, & Linton, 1986; Tissot & Crowther, 2008; Tylka & Sabik, 2010) Body Dissatisfaction A body dissatisfac tion component was consistently replicated as a first or second factor ( Bennett & Stevens, 1997; Klemchuk et al., 1990; Limbert, 2004; Raciti & Norcross ) In a fifth study, these nine items loaded together on a large first component with other EWR items (Welch et al., 1988) Notably, the nine item EDI body dissatisfaction subscale appears to be the most frequently used measure in BID research (Campbell & Hausenblas, 2009; Hausenblas & Downs, 2001; Roberts et al ., 2006) with its relationship to eating pathology62,63 and BID ( Berland et al., 1986; Cafri et al. 2005; Kelly et al., 2011; Thompson et al. 2003) often noted. T his subscale is often administered as a sole measure of BID (Bailey et al. 1990; Hausenbla s & Downs, 2001) ; yet researchers should consider its singular use carefully, as its ability to fully encapsulate the concept of BID has been questioned ( Bailey et al., 1990; Hausenblas & Downs, 2001; Tissot & Crowther, 2008). Additionally, t he title of th e subscale could be a misnomer, as it doesnt appear to measure general body dissatisfaction (Bailey et al., 1990) It also does not appear to measure body weight dissatisfaction, as research indicates that body dissatisfaction does not decrease with weigh t loss (Bailey et al., 1990; Rathner & Rumphold, 1994). In fact, it appears that the subscales true measure is body shape dissatisfaction, distinct from body size dissatisfaction (Bailey et al., 1990; Rathner & Rumphold, 1994) Our findings support that t he body dissatisfaction subscale does not appear to measure size or weight dissatisfaction, because, across most reviewed studies, items loaded separately from other diet or weight specific items. The focus of the items on body shape and lower body parts ( i.e., hips, thighs, stomach, and 90

PAGE 91

buttocks) further indicate that the subscale is a narrow measure of BID specific to lower body shape dissatisfaction, and distinct from preoccupation or problems with food or weight. EDI 3 update The EDI 3 adds a tenth ite m (#47, I feel bloated after eating a normal meal) to the body dissatisfaction subscale, despite consistent replication of the subscale across both nonclinical and clinical studies (Eberenz & Gleaves, 1994; Espelage et al., 2003) The change, as indicated by EDI developers (Garner, 2004) was made primarily because of the item s conceptual fit and clinical relevance for eating disorder patients. This was despite statistical evidence (e.g., item to factor correlations) that the item fit with a drive for th inness factor for clinical groups. Three of our reviewed nonclinical studies reported this item was highly correlated with drive for thinness and/or bulimia items (Klemchuk et al., 1990; Limbert, 2004; van Strien & Ouwens, 2010) None of the six reviewed s tudies report ed this item as loading with body dissatisfaction items. Our evidence indicates that this item fits better on the proposed eating disorder risk component for nonclinical populations Interpersonal Distrust Six items from the interpersonal dist rust subscale of the EDI comprise the third component extrapolated from synthesis of the reviewed studies. Five studies reported that four (Bennett & Stevens, 1997; Raciti & Norcross, 1987) five (van Strien & Ouwens, 2010) or all six (Limbert, 2004; Welc h et al., 1988) of these items loaded together onto one factor. The study noting the six item component reported a moderate internal consistency ( ; Limbert, 2004). Interpersonal distrust represents an avoidance of, and inability to form, close relat ionships, due to the individual s discomfort in expressing emotions toward others (Garner et al., 1983) Interpersonal difficulties a re known to characterize those with AN (Garner et al., 1983) There is also evidence that interpersonal difficulties shoul d be considered in disordered 91

PAGE 92

eating research (Broberg, Hjalmers, & nevonen, 2001; P i ran & Cormier, 2005) The modest amount of construct validity research in nonclinical populations suggests that the interpersonal distrust items are related to BID risk fa ctors of exercise dependence (Grandi, Clementi, Guidi, Benassi, & Tossani, 2011) and pursuit of thinness (Mahalik et al., 2005) Convergent validity has also been demonstrated between the interpersonal distrust subscale and feelings of inadequacy ( r = .32) and sexual anxiety ( r = .39) as measured by the Anorexia Nervosa Inventory Scale (ANIS )68; as well as anxiety ( r = .45), depression ( r = .44), and insufficiency ( r = .43) as measured by t he Symptom Checklist90 (SCL 90; Garner, 2004). These correlations and the consistent replication of the interpersonal distrust subscale across the reviewed studies provide support for its use in BID and disordered eating research in nonclinical populations. Perfectionism The six item perfectionism subscale of the EDI was replicated or nearly replicated in five studies (Klemchuk et al., 1990; Limbert, 2004; Raciti & Norcross, 1987; van Strien & Ouwens, 2010) with var .79). Inclusion of a perfectionism component is supported by the recognized relationship of perfectionism to both BID (Boone, Soenens, & Braet, 2011; Grandi et al., 2011; Hewitt, Flett, & Ediger, 1995; Rathner & Rumphol d, 1994) and disordered eating (Berland et al., 1986; Boone et al., 2011; Hewitt et al., 1995; Schwarz, Aruguete, & Gold, 2005; Sherry, Hewitt, Besser, McGee, & Flett, 2004) in nonclinical populations. In support of the subscales construct validity, noncl inical researchers have noted its moderate correlation to feelings of inadequacy ( r = .43) and obsessive compulsive traits ( r = .41) as measured by the ANIS (Rathner & Rumphold, 1994) as well as insufficiency ( r = .35, SCL 90; Garner, 2004) Further, nonclinical research demonstrates that the six items of the EDI perfectionism subscale are a multidimensional measure of perfectionism (Boone et al., 2011; 92

PAGE 93

Sherry et al., 2004) Thus, the perfectionism subscale appears to be an adequate measure of perfecti onism in nonclinical populations, and relevant for BID and eating pathology research. Eating Disorder Risk A consistent mapping pattern was difficult to discern for the items comprising the bulimia and drive for thinness subscales of the EDI. R emoval of it em one (1; drive for thinness ; I eat sweets and carbohydrates without feeling nervous) however, was indicat ed based on the items low factor loadings in three studies (Bennett & Stevens, 1997; van Strien & Ouwens, 2010; Welch et al., 1988) and inconsist ent item mapping in the remaining studies. E xamination of the remaining thirteen bulimia and drive for thinness items revealed frequent cross loadings among these items and two interoceptive awareness items (47 and 64; Bennett & Stevens, 1997; Klemchuk et al., 1990; Limbert, 2004; Raciti & Norcross, 1987; van Strien & Ouwens, 2010). Moreover several of these 15 items were reported as complex, loading onto separate components representative of either drive for thinness or bulimia ( Bennett & Stevens, 1997; Ra citi & Norcross, 1987; van Strien & Ouwens, 2010. Considering the confirmed relationship between the three subscales represented by these items, item cross loading and/or complexity can be expected (Garner, 2004) However, the frequency with which this see ms to occur point s to potential problems with each subscales psychometric properties, calling into question the valid use of these subscales to individually assess their purported constructs in nonclinical populations. This is further concerning when cons idering that BID researchers frequently use only one or two of the three subscales in their research. Combining the bulimia and drive for thinness items (sans item #1) with items 47 and 64 in to a single component potentially alleviates the cross loading an d complexity issues seen with these items. This combined component is further justified when considering that three studies ( Bennett & Stevens, 1997; Klemchuk et al., 1990; Welch et al., 1988) reported most, if not all, of these items loaded together as a larger 93

PAGE 94

component; and two studies reported moderately high inter factor correlations on an eight factor structure ( R = .49 .59; Limbert, 2004; Raciti & Norcross, 1987). Also of note is that both item 47 and 64 were moved to the EWR subscales i n the EDI 3. As part of the clinically derived EDI, the drive for thinness and bulimia subscales assist in diagnosis of AN and BN respectively. A broader measure of eating disorder risk is more relevant for nonclinical populations of women, because researchers have n oted that a combination of disordered eating behaviors (e.g., periodic caloric restriction and episodic binging and / or purging) are practiced by women wit h subclinical eating disorders (Schwitzer et al., 2008) Combining the 13 items from these two subsc ales with items 47 and 64 to form the eating disorder risk component provides a more relevant measure for nonclinical populations where specific eating disorder diagnoses (e.g., AN and BN) are not indicated. This eating disorder risk component measure s pre occupation with food or weight (e.g., I am terrified of gaining weight.). Klemchuk et al. (1990) reported a similar component noting that it appeared to constitute a multidimensional syndrome of maladjustment in young women (p. 303). Removed Items Item s removed from our proposed model include seven from maturity fears items, four from interoceptive awareness one from interpersonal distrust and one from drive for thinness Generally, the maturity fears subscale was unstable and ill defined for nonclini cal populations of Westernized women. This was demonstrated by three studies reporting a trivial maturity fears component (< 5 items ; Bennett & Stevens, 1997; Limbert, 2004; Raciti & Norcross, 1987). Though on e study report ed an adequately replicated matur ity fears component the components internal consistency was ; Limbert, 2004). A nothe r study removed all maturity fears items because of inconsistent itemmapping across the three samp les included in their analysis (Welch et al., 1988) Other nonclinical research supports remov ing the subscale, as the maturity 94

PAGE 95

fears construct is considered less relevant to BID and disordered eating in nonclinical populations (Broberg et al., 2001; Crowther, Lilly, Crawford, & Shepherd, 1992; Grandi et al., 2011; Rathner & Rumphold, 1994) One maturity fears item (#35) was retained as part of the low self esteem component. The remaining seven items (3, 6, 14, 22, 35, 39, 48, and 58) were removed. Several studies reported difficult y interpreting the mapping of ite ms represented by the interoceptive awareness subscale ( Bennett & Stevens, 1997; Limbert, 2004; van Strien & Ouwens, 2010; Welch et al., 1988) In fact, only one study (Raciti & Norcross, 1987) identified an interoceptive awareness component; however, this component represented four EDI subscales Other nonclinical research studies analyzing the EDI at the subscale level have also reported difficulties with the interoceptive awareness measure (Crowther et al., 1992; Podar & Allik, 2009) These findings sugg est that the clinically derived interoceptive awareness subscale is unstable for nonclinical populations. While the construct does not appear to be a relevant measure, six interoceptive awareness items were valuable as part of two other apparent factors ( l ow self esteem and eating disorder risk ). The inconsistent mapping of the remaining four items (21, 26, 33, and 40) led to their removal from our final proposed model. S ix interpersonal distrust items were retained in our proposed model T he seventh item ( #23; I can communicate easily with others) however, was removed due to unpredictable item mapping (Bennett & Stevens, 1997; Limbert, 2004) and low factor loadings (Raciti & Norcross, 1987; van Strien & Ouwens, 2010) As noted previously, item one (1) from the drive for thinness subscale was also removed. Although the item loaded appropriately with other EWR items in one study (Raciti & Norcross, 1987) it failed to correlate adequately with any factor in three other reviewed studies ( Bennett & Stevens, 1997; van Strien & Ouwens, 2010; Welch et 95

PAGE 96

al., 1988) Notably, this same drive for thinness item is the only EDI item expressly retained by EDI developers despite low itemtotal correlations (<.40; Garner et al., 1983) In sum, our results indicate that these 13 items should be removed because they do not fit with the components being measured by the other 51 items. Limitations Considering the popular administration of the EDI in nonclinical research, we were surprised to find only six studies examining the item level structure of the EDI in nonclinical populations of Westernized women. Of the se six studies, almost all used PCA. The limitations of PCA result in error variance being included as explained variance, which can further result in inflated itemto factor correlations and misspecified items (Costello & Osborne, 2005; Gorsuch, 1983) We attempted to account for this limitation by requiring a higher itemto factor correlation minimum in our review. Some argue that use of PCA is not a severe limitation because strong factor structures should be identifiable by several methods of analysis, including PCA (Gorsuch, 1983; Kachigan, 1986) An additional limitation is the variation in methodological rigor across the six reviewed studies. It is important to n ote that this could be a limitation of our review, and not necessarily a limitation of the reviewed studies. Though we attempted to evaluate each study equally by designating a prespecified set of criteria, other important criteria might have inadvertentl y been excluded. Also, the relative strength of the reviewed studies and their failure to meet specific criteria could be attributable to publication restraints and the relative importance of reporting other aspects of the study. Thus, the results of our m ethodological evaluation should be considered only in the context of this review. Finally, due to the nature of the studies identified, our critical review focused on the original 64item EDI. There were no identified research studies exploring the internal structure 96

PAGE 97

of the EDI 3 in our population of interest and only one study examined the internal structure of the EDI 2. Notably, the study indicated that the additional 27 EDI 2 items appear unnecessary for BID and disordered eating measurement in nonclinical populations (van Strien & Ouwens, 2010) Despite our focus on the original 64 item EDI, the results of the proposed model are relevant to current BID and disordered eating researchers. This is true for several reasons. First, nonclinical researchers c ontinue to administer items from the original EDI due to its availability and comparability to the EDI 2 and EDI 3. Second, nonclinical BID and disordered eating researchers administration of the EDI continues to favor the EWR subscales, which sustained very minor revisions in the EDI 3. Third, in nonclinical research, time for survey completion is an important consideration in participant response rates and survey fatigue; thus, a parsimonious model and shorter survey, such as the one we propose, is favor ed over a more complicated model and extensive survey. Finally, while structural modifications to the EDI 3 potentially enhance its relevance, reliability, and validity within clinical populations, these changes may not be relevant to general nonclinical p opulations. EDI 3 developers noted that item to subscale assignment was dictated by more than just statistical indications (Garner, 2004) Also considered was the conceptual fit of the item to the subscale for clinical eating disorder measurement. As indicated by variation in results for the EDI structure between clinical and nonclinical groups, this conceptual fit appears less relevant to nonclinical populations. Conclusion Although administration of the EDI in clinical populations is strongly supported, researchers have suggested that the item level properties and internal structure of the EDI for use in nonclinical populations should be closely examined. Overall, t he findings presented in this paper indicate significant modifications are needed to enhance the validity and reliability for administration of the EDI in nonclinical populations. O ur in depth critical review of studies that 97

PAGE 98

analyzed the internal structure of the EDI after administration to nonclinical populations of Westernized women allowed u s to identify itemmapping patterns, problematic items, and constructs poorly defined for this population. This resulted in the identification and proposal of a five component, 51item model structure. Three of these proposed constructs ( body dissatisfaction, perfectionism, and interpersonal distrust,) are relatively unchanged from their original form, and there is current evidence of their construct validity. There is also strong support for the low selfesteem component, and provisional support for the ea ting disorder risk component; however, their convergent and divergent validities should be studied further. Future research should also attempt to confirm the five factor proposed model structure with CFA. As it stands, the results of this review suggest t hat the five components of this proposed model appear stable and useful for nonclinical BID and disordered eating research. 98

PAGE 99

Table 3 1. EDI, EDI 2, and EDI 3 subscale correlations and item assignment (Garner 2004) EDI Subscale Items EDI 3 Subscale Items R Body Dissatisfaction 2, 9, 12, 19, 31, 45, 55, 59, 62 Body Dissatisfaction 2, 9, 12, 19, 31, 45, 47, 55, 59, 62 .96 Bulimia 4, 5, 28, 38, 46, 53, 61 Bulimia 4, 5, 28, 38, 46, 53, 61, 64 .73 a Drive for Thinness 1, 7, 11, 16, 25, 32, 49 Drive for Thinness b 1, 7, 11, 16, 25, 32, 49 .97 Maturity Fears 3, 6, 14, 22, 35, 39, 48, 58 Maturity Fears b 3, 6, 14, 22, 35, 39, 48, 58 .97 Perfectionism 13, 29, 36, 43, 52, 63 Perfectionism b 13, 29, 36, 43, 52, 63 .98 Ineffectiveness 10, 18, 20, 24, 27, 37, 41, 42, 50, 56 Low Self Esteem 10, 27, 37, 41, 42, 50 .95 Interoceptive Awareness 8, 21, 26, 33, 40, 44, 47, 51, 60, 64 Interoceptive Deficits 8, 21, 26, 33, 40, 44, 51, 60, 77 .95 Interpersonal Distrust 15, 17, 23, 30, 34, 54, 57 Interpersonal Insecurity 15, 23, 34, 57, 69, 73, 87 .85 EDI 2 Subscale c EDI 2 Items Asceticism 66, 68, 71, 75, 78, 82, 86, 88 Asceticis m d 66, 68, 75, 78, 82, 86, 88 .96 Impulse Regulation 65, 67, 70, 72, 74, 76, 77, 79, 81, 83, 85, 90 Emotional Dysregulation 67, 70, 72, 79, 81, 83, 85, 90 .93 Social Insecurity 69, 73, 80, 84, 87, 89, 91 Personal Alienation 18, 20, 24, 56, 84, 91 .83 Interpersonal Alienation 17, 30, 54, 65, 74, 76, 89 e Note. Correlational data reflects a U.S. adult clinical sample. a Lower correlation is attributed to groups with differing types of eating disorders scoring on greater ends of the subscale continuum (i.e., anorexia nervosa restricted type versus bulimia nervosa). b Subscale is unchanged across all three iterations. c EDI 2 consists of the 8 original EDI subscales and their items, plus three addit ional subscales with new items. d Item 71 from the EDI 2 is included in the EDI 3 item booklet, but is not scored on the EDI 3. e Subscale is highly related to interpersonal distrust ( r = .77) and social insecurity ( r = .72) subscales of the EDI 2. 99

PAGE 100

Tabl e 3 2. Studies retained for EDI review Study Title Country N Factor Analysis Method Bennett & Stevens (1997) The internal structure of the EDI. Unknown N = 310 PCA with varimax r otation. Klemchuk et al. a (1990) Body dissatisfaction and eating related problems on the college campus: Usefulness of the EDI with a nonclinical population. United States N1 = 621 N2 = 636 PCA with oblique and orthogonal rotation. No constraint, then c onstraining to load onto 5-, 6-, and 8-factors. Limbert (2004) The EDI: A test of the factor structure and internal c onsistency in a nonclinical sample. United Kingdom N = 647 PCA with varimax rotation constraining 8 factors. Raciti & Norcross (1987) The EAT and EDI: Screening, interrelationships, and psychometrics. United States N = 238 PCA with varimax rotation. No constraint. Scree and eigenvalue 1. van Strien & Ouwens a (2003) Validation of the Dutch EDI -2 in one clinical and two nonclinical populations. Netherlands N = 492 First and second order ML FA with varimax rotation and Kaiser normalization. Welch et al. (1988) The factor structure of the EDI. New Zealand N1 = 192 N2 = 253 N3 = 142 PCA with varimax rotation for an 8-, 7-, 5-, 4-, 3and 2-factor s olution, with FACTOREP. PCA=Principal Component Anal ysis; EDI=Eating Disorder Inventory; EAT=Eating Attitudes Test; ML=Maximum Likelihood; FA=Factor Analysis. a The study sampled numerous groups. Sample size(s), as reported here is only for nonclinical samples of women whose data was analyzed with factor analysis. 100

PAGE 101

Table 3 3. Methodological evaluation of retained EDI factor analytic studies Criteria Bennett & Stevens Klemchuk et al. Limbert Raciti & Norcross van Strien & Ouwens Welch et al. 1 Sampling Methodology 1.1 Method of sampling minimized err or X X X X 1.2 Representative sample (i.e., age, race, ethnicity) X 1.3 Tested sample for adequacy NP NP NP NP NP NP 1.4 Ratio participants to item Summed Scores Criteria met score 4 4 5 7 6 6 Criteria not met score 4 2 5 3 3 3 Information insufficient to evaluate score 3 5 1 1 1 2 Note : NP = not performed; X = Criterion met; 0 = criterion not met; = information insufficient or information totally missing; NA = not applicable. a Ratio approached 10:1 (N1 = 9.7:1; N2 = 9.9:1); b Criterion not noted in text. Tabled results indicated a probable cut off of 0.40; c Item to factor loads reported in a table only for the largest (18item) extracted factor; d Provided correlation table of the proposed fact ors to original EDI subscales. e An acceptable alternative procedure (FACTOREP) used for factor extraction; f The table provided was uninterpretable, likely due to printer errors. 101

PAGE 102

Table 3 4. Factor structure results Study Factor Items

PAGE 103

Table 3 4. Continued Study Factor Items interoceptive awareness and maturity fears ) across three samples.103

PAGE 104

Table 3 5. Proposed EDI structure for nonclinical populations Subscale Definition Items Low Self Esteem Poor self concept. Inability to recognize feelings. 8, 10, 18, 20, 24, 27, 35, 37, 41, 42, 44, 50, 51, 56, 60. Body Shape Dissatisfactiona D issatisfaction with the shape of body parts affected by weight fluctuations (e.g. hips, thighs, buttocks). 2, 9, 12, 19, 31, 45, 55, 59, 62. Perfectionism a Excessive, likely unrealistic, expectations for ones achievement 13, 29, 36, 43, 52, 63. Interpersonal Distrusta Avoidance of and inability to form close relationships, due to discomfort expressing emotions toward others. 15, 17, 30, 34, 54, 57 Eating Disorder Risk Provides indication of pathological eating behaviors. General feelings of guilt about eating, and an unhealthy relationship with food. 4, 5, 7, 11, 16, 25, 28, 32, 38, 46, 47, 49, 53, 61, 64. aDefinition is unmodified from the original EDI subscale definition. 104

PAGE 105

Figure 3 1. Flow chart of search and screen process 177 abstracts reviewed 86 full articles retrieved for review 8 additional a rticles identified through review of reference lists. 169 studies identified through database search. 91 studies failed to meet inclusion / exclusion criteria 14 Did not utilize the Eating Disorder Inventory 45 Not adult female sample 23 Clinical sample 3 Non Westernized / Asian sample 4 Did not utilize all EDI items 80 studies failed to meet inclusion / exclusion criteria 76 Did not analyze the factor str ucture 3 Did not provide item level psychometrics 1 Included males in the sample 6 studies met inclusion / exclusion criteria and are included in this review 105

PAGE 106

CHAPTER 4 BODY ESTEEM AND SELF OBJECTIFICATION: USING OBJECTIFICATIO N THEORY TO COMPARE C OLLEGIATE FEMALE ATH LETES AND NON ATHLETES After the enactment of Title IX, federal legislation prohibiting sex discrimination in athletics, images of female athletes became more prominent in the media (Kane, 1988); though they continue to be far less publicized than male athletes (Clavio & Eagleman, 2011; Fink & Kensicki, 2002; Kane, 1988). Additionally, recent res earch indicates that media images of male athletes emphasize power and athletic prowess, while media images of female athletes are sexualized and objectified (Clavio & Eagleman, 2011; Fink & Kensicki, 2002; Hardin et al. 2005; Kim et al. 2010). According to objectification theory, societal pressures expressed through feminine gender norms and sexual objectification contribute significantly to mental health problems (e.g., eating disorders, depression, sexual dysfunction, and substance abuse) among diverse groups of women (Fredrickson & Roberts, 1997; Szymanski et al. 2011). Sexual objectification occurs when the body is recognized predominantly for its physical attributes (body or body parts), and is viewed as an object of sexual desire (Fredrickson & Roberts, 1997). It is widely understood that the medias portrayal of feminine beauty contributes significantly to psychological (e.g., body shame) and mental health (e.g., eating disorders; Fitsimmons Craft, 2011; Grabe et al. 2008) problems; however, we do not understand how the female athletes body image and/or self objectification is affected by society s sexual objectification of female athletes participating in different sports (Varnes et al. 2013). Traditional ly, body image research among female ath letes has focused on the relationship between body image and sport related pressures, such as weight or appearancepressures in sport competition (Engel et al., 2003; Harrison & Fredrickson, 2003; Hausenblas & Carron, 2002; Reel et al. 2010; Reinking & A lexander, 2005). Results from these studies have been 106

PAGE 107

equivocal, which has largely been attributed to: (a) inconsistent definitions of various sport classifications (Hausenblas & Carron, 2002; Varnes et al., 2013); (b) failure to consider external societal expectations of female athletes (Hardin & Greer, 2009; Varnes et al., 2013); and (c) less relevant body image measures being administered to the athlete group (Varnes et al., 2013). One example of a sport classification that is commonly used in research is lean focused sport. Some researchers use this term to describe sports that emphasize weight as part of appearance, and include adjudicated sports such as diving, gymnastics, cheerleading and figure skating (e.g., Harrison & Fredrickson, 2003). Yet other researchers define lean focused sports as endurance type sports where being low er in weight and body fat contribute to improved athletic performance, such as swimming and long distance running ( Hausenblas & Carron, 2002; Engel et al., 2003). Still other s have indicated that lean focused sports could include sports that emphasize leanness for either competition or appearance, which would encompass all of the aforementioned sports (Reel et al., 2010; Reinking & Alexander, 2005). In a ny of these instances, the sport definition is specific to a sport related pressure, leaving external societal pressures, such as societys increased sexual objectification of female athletes, outside the scope of these investigations. The last noted limitation of previous femal e athlete research emphasizes the importance of selecting body image measures that are appropriate for the population. This proves difficult when conducting research in the female athlete population, because athletes experience body image differently than nonathletes ( Petrie & Greenleaf, 2012 ). In other words, there are different contributors to body image for athletes versus non athletes, causing participant responses to some measures of body image to vary depending on athletic status. This phenomenon has been demonstrated through both correlational (DArcy, 2007; Hoag, 2012; Krane et al. 2002) and 107

PAGE 108

meta analytic (Smolak et al., 2000) research. For this reason, it is necessary for researchers to use a multidimensional body image measure that is salient fo r both athletes and nonathletes when examining body image differences between the two groups In summary, research is needed that focuses on both self objectification and multidimensional body image among female athletes, with particular consideration g iven to external societal pressures, including sexual objectification. The purpose of this study was to use select constructs from objectification theory (Fredrickson & Roberts, 1997; Szymanski et al., 2011) to investigate how societys sexual objectificat ion of female athletes is related to differences in multidimensional body image and self objectification among collegiate female athletes and non athletes. First, we provide an overview of objectification theory and discuss societys sexual objectificatio n of the female athlete. Overview of Objectification Theory The effect of sexual objectification on mental health risks can be direct and overt or indirect and subtle ( Szymanski et al., 2011). The direct path portends from personal sexual objectification e xperiences that range from being gazed at sexually to apparent sexual violence (Fredrickson & Roberts, p.177178, 1997). The indirect path involves observed or personal sexual objectification experiences that result in self objectification (Moradi et al. 2005; Szymanski et al., 2011 ). Self objectification occurs when an individual takes an observers perspective of their own body, thereby defining their body as an object to be viewed (Moradi et al., 2005; Szymanski et al., 2011). Moradi (2010) noted that self objectification should be considered a process that manifests as body surveillance as well as internalization (i.e., taken within and believed to be self defined) of sociocultural beauty standards (p. 146). Internalization of sociocultural beauty i deals may lead the individual to believe their desire to attain the culturally defined beauty ideal is a personal choice rather than a result of social pressure 108

PAGE 109

(McKinley & Hyde, 1996). There is strong evidence that the sociocultural definition of beauty i s partially defined, and significantly perpetuated, by the media (Fitsimmons Craft, 2011; Grabe et al, 2008). The extant manifestation of internalization and body surveillance resulting from sexually objectifying experiences are in turn related to increased body shame and anxiety, and reduced flow and internal awareness (Fredrickson & Roberts, 1997; McKinley & Hyde, 1996; Moradi, 2010; Szymanski et al., 2011; Tylka & Hill, 2004). Thus, self objectification, via the mechanistic nature of internalization and body surveillance, potentially promotes negative body image or body image disturbance (BID; Cafri et al. 2005; Fredrickson & Roberts, 1997; McKinley & Hyde, 1996; Moradi, 2010; Moradi & Huang, 2008; Szymanski et al., 2011 ). BID is a multidimensional disorder characterized by an individuals negative attitude toward, or evaluat ion of, their body weight, shape, size, or appearance (Cash & Szymanski, 1995). Both BID and self objectification are associated with health risks such as disordered eating (Grabe et al., 2008; Moradi et al., 2005; Tylka & Sabik, 2010), depression (Car r & Szymanski, 2010; Szymanski & Henning, 2007), decreased self esteem (Tylka & Sabik, 2010), anxiety (Muscat & Long, 2008), substance abuse (Carr & Szymanski, 2010), and sexual dysfunction ( Szymanski et al., 2011). It is important to note that, although s elf objectification can contribute to BID, and although both BID and self objectification can result in similar outcomes, high levels of one does not necessarily predict the other. In other words, the psychological and behavioral consequences of self objec tification occur as a result of being preoccupied with ones appearance, regardless of appearance satisfaction (Noll & Fredrickson, 1998). 109

PAGE 110

Sexual Objectification of t he Female Athlete Varies by Sport Sexual objectification reinforces feminine gender roles by emphasizing the importance of female appearance and diminishing nonfeminine achievements. Thus, sexual objectification of female athletes devalues female athletics (Hardin et al., 2005). Notably, the level at which female athletes are sexually objectif ied varies by sport. Female athletes who are depicted in, and objectified by, the media are more likely to be from sports that are considered sex appropriate or feminine (e.g., gymnastics, swimming, tennis, and volleyball; Clavio & Eagleman, 2011; Hardin & Greer, 2009; Kim et al., 2010). Research has indicated that female athletes participating in more feminine sports are also more objectified by their peers than female athletes participating in less feminine sports (e.g., basketball; Hardin & Greer, 2009; Kim et al., 2010; Parsons & Betz, 2001). The term sport objectification refers to the different levels of sexual objectification that occur for female athletes from different sports (Moradi & Huang, 2008; Parsons & Betz, 2001). There has been a deficit in research examining how societys select objectification of female athletes potentially imposes additional pressures for female athletes to exemplify stereotypic female characteristics, such as ideal appearance and femininity. In addition, researchers have identified a need to examine populations in which sexual objectification experiences intersect with a womans conflicting identities i.e., being female and being in another marginalized position (Krane et al. 2004; Moradi & Huang, 2008; Szymanski et al. 2011). Female athletes represent one such conflicted population, as previous research indicates female athletes have difficulty reconciling their dual feminine and athletic identity (Krane et al., 2004; Larabee, 2011). Thus, it is necessary to address t he self and sexual objectification research gap in female athlete research. It is possible that elements of societys (e.g., media and peers) increased sexual objectification of female athletes serves to conjoin the feminine and athletic identities for those 110

PAGE 111

participating in more objectified sports. In other words, sexual objectification could act to reconcile these conflicting identities by putting the female athlete more at ease with her athletic body and athletic abilities, while also assuring her th at she is still feminine. Or, the medias sexual objectification of female athletes could act to promote self objectification and/or BID among athletes, especially those participating in more objectified sports. This phenomenon could activate the potential for adverse psychological consequences and increased mental health risks. Also of interest is how the medias sexual objectification of female athletes affects those participating in less objectified sports. Athletes participating in less objectified spor ts may experience greater levels of self objectification and/or BID because of the medias implication that these athletes do not meet the social standard of attractiveness and femininity compared to athletes from more objectified sports. Or, less objectif ied female athletes may experience less pressure to conform to feminine standards, and care more about what the body can do rather than its appearance, thereby self objectifying less. The Present Study The present study was designed to explore and compare body esteem (physical condition, sexual attractiveness, and weight concern) and the three self objectification process constructs of thin ideal internalization, body surveillance, and body shame among three groups: (1) female nonathletes; (2) NCAA Divisio n I female athletes participating in sports that are considered more objectified; and (3) NCAA Division I female athletes participating in sports that are considered less objectified. Each of the three variables associated with body esteem (physical condi tion, sexual attractiveness, and weight concern) represent three different dimensions of body esteem; however, they are also moderately correlated, with positive feelings on one dimension typically indicating positive feelings on the others ( r = .33 to .40; Franzoi & Shields, 1984). Therefore, we 111

PAGE 112

developed an overall body esteem hypothesis (H ypothesis 1) that included all three body esteem constructs, and subhypotheses related to the specific constructs: Hypothesis 1. Group classification (nonathletes, more objectifie d athlete, less objectified athlete) will have a statistically significant impact on the three measures of body esteem: physical condition, sexual attractiveness, and weight concern. 1a. Because athletes generally feel better about the form and physical function of their body (Fellows, 1999; Petrie & Greenleaf, 2012 ), we posit that physical condition will be the most influential body esteem construct characterizing a statistically significant difference between athletes and non athletes, with weight concern b eing less influential (Fellows, 1999), and sexual attractiveness being least influential. 1b. Differences in weight concern will be more pronounced between nonathletes and athletes participating in less objectified sports, versus comparisons made between nonathletes and athletes participating in more objectified sports (Fellows, 1999). Objectification theory research indicates that the process of self objectification manifests separately as thin ideal internalization and/or body surveillance, which mediate the relationship between sexual objectification experiences and body shame (Moradi, 2010). Thus, thin ideal internalization, body surveillance, and body shame are all considered part of the self objectification process and are related; however, it is impor tant to measure each construct separately, because they represent distinct and important constructs in objectification theory (Moradi, 2010). Additionally, very little objectification theory research has been conducted in the female athlete population. The refore, we proposed three sets of pairwise comparison hypotheses to test differences among our three groups (nonathletes, more objectified athlete s less objectified athlete s ) on the constructs of thin ideal internalization (H ypothesis 2), body surveillan ce ( H ypothesis 3), and body shame ( H ypothesis 4): Hypothesis 2. We hypothesize that: (a) nonathletes will internalize the thin ideal more than athletes participating in less objectified sports; (b) non athletes and athletes participating in more objectified sports will internalize the thin ideal 112

PAGE 113

at similar levels; and (c) athletes participating in more objectified sports will internalize the thin ideal more than athletes from less objectified sports. Hypothesis 3. Because of the additional external pressure to conform to societal a ppearance standards that female athletes in more objectified sports potentially experience, we hypothesize that (a) athletes participating in more objectified sports will monitor their body and appearance more than nonathletes; (b) athletes participating in more objectified sports will monitor their body and appearance more than athletes participating in less objectified sports; and (c) athletes participating in less objectified sports will monitor their body less than non athletes. Hypothesis 4. Although internalizat ion and body surveillance have both been shown to be predict body shame in females (McKinley & Hyde, 1996; Moradi, 2010; Moradi et al., 2005; Tylka & Hill, 2004; Tylka & Sabik, 2010), we hypothesize that athletes participating in more objectified sports will experiences less body shame than nonathletes (due to feeling better about the physical function of their body); (b) there will be no difference in body shame between the two athlete groups; and (c) athletes from less objectified sports will report less body shame than nonathletes. Methods Recruitment and Procedure In this study competitive athletics is defined as a university level sport or emerging sport governed by the NCAA (2012). The term female athlete is defined as a female athlete playing for a university level sport or emerging sport team governed by the NCAA (2012). All potential participants in this study were required to be female, full time undergraduate students (enrolled in 12 or more credit hours) between the ages of 18 and 25, a ttending a large Division I southeastern university in either the Fall 2012 or Spring 2013 semester. All students were given the option to complete the survey online (via Qualtrics) or using pen and paper. Separate web survey links were administered for at hlete and non athletes, to double check student athlete classification accuracy and avoid duplicate survey completion. The study protocol and materials were approved by the Institutional Review Board at the participating university prior to 113

PAGE 114

recruiting part icipants. We used a tailored design method for nonathletes and athletes (Dillman, Smyth, & Christian, 2009). Non athlete group A list of 1,400 randomly selected full time undergraduate female students was requested from the Office of the University Regist rar (OUR), including student email and local mailing addresses. It was requested that this list exclude athletes at the university (athletic status is indicated in the OURegistrar database). After obtaining the sample, a postcard was mailed to the local ad dress of each potential nonathlete participant to provide notification that they would be receiving an email requesting their participation in a survey about body attitudes of female undergraduate students. The email invitation to participate, with the survey link embedded, was sent on a Sunday evening in October, 2012, five days following postcard distribution. Participants were told to contact the principal investigator if they wished to be mailed a paper copy of the survey in lieu of the online version. Two follow up emails were sent later in the week on a weekday afternoon (before 5pm), and a weekday morning (before 11am) to facilitate an optimal response rate (Molasso, 2005). These multiple recruitment communications resulted in a total of 422 surveys received from the nonathlete group (response rate = 30.1%). This was judged to be an excellent response rate considering that the survey was webbased (Sax, Gillmartin, & Bryant, 2003), and response rates are typically low for this age group (Dillman et a l., 2009). Of the received survey responses, 53 records were removed due to respondents reporting characteristics that made them ineligible for this study (i.e., n = 40 graduate students; n = 2 part time students; n = 11 did not meet age range specificatio ns). One additional survey was removed because the participant indicated current participation on an NCAA varsity team at the university level, which was surprising given that the link was only sent to students who were not 114

PAGE 115

classified as athletes in the OU R. Of the remaining 368 eligible records, survey data was complete for 322 nonathletes. Athlete group To recruit athletes from the university, assistance was requested from a senior administrator at the Universitys Athletic Association (UAA). A current list of all 231 collegiate female athletes (including names and emails) who participated in at least one of the 11 NCAA governed sport teams during Fall 2012 was acquired from the UAA. Recruitment of athletes into the study was conducted in two waves. The f irst wave occurred in October 2012 with the UAA administrator making initial contact via email with all 231 female athletes to encourage (but not require) participation in the research. It was noted in this email that, although the UAA supported the resear ch, no one affiliated with the UAA would have access to individual or aggregated sport team data. A subsequent recruitment email was sent three days later from the principal investigator to the student athletes. This email reiterated the measures taken to protect the anonymity of their responses. This recruitment email was worded similarly to the email sent to the nonathlete group; however, a different survey link was embedded in the email to athletes. As was the case for the non athlete group, a paper sur vey option was also provided. The first wave of recruitment resulted in 65 responses (response rate = 28.1%) from athletes participating in 10 of the 11 NCAA governed sports at this university. The second wave of athlete recruitment occurred in January 2013 with the UAA administrator first contacting, via email, the coaches of seven sports (basketball, gymnastics, soccer, softball, swimming and diving, tennis, and volleyball) that were under represented in the first wave. The UAA notified the coaches of these teams that they would be receiving a request for assistance from study administrators to help facilitate athlete participation. Subsequently, the principal investigator sent each of the seven head coaches an email requesting assistance with 115

PAGE 116

distributin g paper surveys at an upcoming team meeting or practice. It was emphasized to the coaches that student athletes would be expected to complete and return the surveys to the study investigators on their own time to preclude coaches from being present during survey completion and submission. Six of the seven coaches agreed to let a program administrator or team manager distribute the paper surveys. These surveys were individually distributed to 97 athletes in a manila envelope with a university mailing address on the front label. In an effort to maintain response anonymity, all athletes were asked to accept the survey regardless of whether or not they had completed it in the first wave. Athletes who had completed the survey in the first wave were asked to recycle the paper survey that they received. Additionally, a question was added to the second wave survey asking if the student athlete had completed a survey on body attitudes (title of the survey) sometime in the previous semester, as extra assurance that res ponses from each participant were included only once. The cover letter of the survey indicated three options for completing and returning the survey: (1) complete the paper copy of the survey and return it in the envelope through campus mail (no postage re quired); (2) complete the paper copy of the survey and hand deliver it to the campus address listed on the envelope; or (3) complete the survey online via a survey link that would be emailed securely to all second wave participants within 48 hours. This se cond wave of athlete data collection resulted in 40 returned paper surveys, and no new online surveys (response rate = 41.2%). Three ( n = 3) of the returned second wave surveys were marked to indicate that the survey had been completed in the first wave; t hus, 37 new responses were received and considered for this study. Of the 102 total athlete responses that were collected following waves one and two of the athlete data collection (final athlete response rate = 44.1%), eight records ( n = 8) were discarded ( n = 5 under age 18; n 116

PAGE 117

= 3 were not registered as full time students). Of the 94 remaining responses, data was complete for 85 records. Participants Based on inclusion and exclusion criteria for this study, the final sample included 322 female non athlete s and 85 female athletes (total n = 407). Every NCAA sport at the university was represented in our athlete sample: basketball ( n = 4), cross country ( n = 2), golf ( n = 5), gymnastics ( n = 13), lacrosse ( n = 6), volleyball ( n = 3), soccer ( n = 20), softbal l ( n = 7), swimming and diving ( n = 6), tennis ( n = 5) and track and field ( n = 14). The athlete group was sub divided into two groups: athletes participating in less objectified sports (LOS athletes) and athletes participating in more objectified sports (MOS athletes). Parsons and Betz (2001) sport objectification findings were used to classify the studys 11 sports into MOS athletes and LOS athletes. Parsons and Bet z (2001) sport objectification scores indicate the level at which college students perceiv e a specific sport is feminine, and how the sport emphasizes the female athletes appearance and body P er Parsons and Betz (2001), the 11 sports included in our study have a mean sport objectification score of 2.74 (range 2.09 4.02) on a Likert type five point scale (1=least objectified to 5=most objectified). Athletes from sports that fell below Parsons and Betzs (2001) 2.74 mean were classified into the LOS athlete group ; and athletes from sports above the 2.74 mean were classified into the MOS athlete group. This resulted in the LOS athlete group being comprised of athletes from basketball, cross country, golf, lacrosse, soccer, softball, and track and field (sport objectification score range 2.09 2.60; n = 58); and the MOS athlete group being compris ed of athletes from gymnastics, swimming and diving, tennis, and volleyball (sport objectification score range 2.93 4.02; n = 27). We cross checked these classifications with other researchers findings of which sports are more or less 117

PAGE 118

objectified by the media (Kim et al., 2010), and which sports are deemed feminine versus nonfeminine (Crissey & Honea, 2006; Hardin & Greer, 2009). Measures All participants responded to demographic questions about weight, height, year in school (first year, second year, third year, fourth year, or other undergraduate student), and race and ethnicity (American Indian or Alaska Native, Asian, Black or African American, Hispanic, Mixed, Native Hawaiian or Other Pacific Islander, or Non Hispanic white). Participants marking t wo or more responses under race were coded as multi racial. Self reported weight and height were used to calculate BMI. Body esteem Franzoi and Shields (1984) Body Esteem Scale (BES) was selected to measure body image because of its multidimensional nature and conceptual fit with objectification theory (McKinley & Hyde, 1996; Noll & Fredrickson, 1998; Varnes et al. 2013). The concept of body esteem includes measures of perceived self attractiveness (sexual attractiveness and weight concern), as well as f eelings regarding demonstrable physical abilities, such as endurance and strength (physical condition). To assess body esteem, the three subscales of the female version of the BES were administered: sexual attractiveness (13 items), weight concern (10 items), and physical condition (9 items). All BES items are measured on a fivepoint Likert type scale (1=strong negative feelings to 5=strong positive feelings). The weight concern subscale is specific to parts or functions of the body that are modifiable thr ough physical activity and food intake (e.g., weight, appetite, figure) Th e sexual attractiveness subscale focuses on parts and functions of the body that are not modifiable through exercise or diet (e.g., body hair, lips, sex drive). The physical conditi on subscale assesses how one feels about their bodys physical abilities (e.g., strength, 118

PAGE 119

agility, energy). The physical condition component differs from the other two, as it represents qualities that are not typically evaluated by others, except in situat ions whe re they can be demonstrated and assessed (as in athletics). Principal components analysis of the BES supports the three aforementioned factors (Franzoi & Shields, 1984). In our study, obtained internal consistency scores were moderately high: sexua from t he BES have also revealed high threemonth testretest reliability ( r = .75 to .87; Franzoi, 1994). Convergent validity with other instruments has been demonstrat ed in the female undergraduate population for each subscale (Franzoi & Herzog, 1986; Thomas & Freeman, 1990). I t should be noted that all three subscales are moderately correlated with self esteem ( r = .22 to .39; Franzoi & Shields, 1984; Franzoi & Herzog, 1986). Thin ideal internalization We administered the Internalization General subscale of the Sociocultural Attitudes Towards Appearance Questionnaire3 (SATAQ3; Thompson et al. 2003) to measure thin ideal internalization. The I nternalization General s ubscale assesses the degree to which the sociocultural ideal of thin beauty is internalized Participants indicate on a five point Likert type scale (1=definitely disagree to 5=definitely agree) how much they want or try to look like people who are on TV, or in movies and magazines (e.g., I would like my body to look like the people who are in the movies). In our study, internal reliability comparable to previous studies (e.g., Cafri et al., 2005; Thompson et al., 2003). The Internalization General subscale has demonstrated satisfactory convergent validity in previous studies (e.g., Thompson et al., 2003). 119

PAGE 120

Body surveillance Body surveillance was measured using the Body Surveillance subscale of the Objectified Body Consciousness Scale (OBCS; McKinley & Hyde, 1996). Body surveillance is the extent to which women monitor their body in terms of how it appea rs to others. The body surveillance subscale contains eight items (e.g., I often worry about whether clothes I am wearing make me look good) measured on a seven point Likert type response scale (1=strongly disagree to 7=strongly agree) with a not applic able (NA) option. Six items are reverse scored, and NA responses are treated as missing values. If more than two responses for the subscale are missing, the entire subscale is counted as missing. Researchers have indicated that this subscale yields valid and reliable scores. For example, womens responses to the subscale are consistent over a two week time period ( r = .79) ; and Body Surveillance has been shown to be highly related to a public self consciousness measure ( r = .73) and unrelated to private se lf consciousness (McKinley & Hyde, 1996). In our sample internal consistency was demonstrated as sufficient (Cronbachs = .84), and consistent with reliability scores reported in previous studies (McKinley & Hyde, 1996; Tylka & Hill, 2004). Body shame B ody shame refers to a womans belief that she is not a good person if she does not attain cultural expectations related to her body (e.g., I feel ashamed when I havent made the effort to look my best). The Body Shame subscale of the OBCS (McKinley & Hyde, 1996) was used to measure body shame. This subscale contains eight items (two are reverse scored) measured on a seven point Likert type response scale (1=strongly disagree to 7=strongly agree) with an NA option. NA responses are treated as missing value s. If more than two responses in the subscale are missing the entire subscale is excluded Researchers have indicated that data collected using the Body Shame subscale are valid and reliable. For example, womens responses to the Body 120

PAGE 121

Shame subscale are consistent over a two week period ( r = .79); and convergent validity has been demonstrated when tested against measures of body esteem ( r = .46 to .51; McKinley & Hyde, 1996). In our sample, the internal reliability of data collected using the Body Shame subscale was satisfactory (Cronbachs .84) and consistent with values reported in previous studies (McKinley & Hyde, 1996; Tylka & Hill, 2004). Data Analysis Statistical tests for analyzing data were performed using SAS Version 9.3; except for descrip tive discriminant analysis (DDA), which was done using SPSS Version 21. First, correlations between the measures were analyzed. Our first H ypothesis was testing with multivariate analysis, using the process outlined by Huberty and Olejnik (2006). Multivariate analysis was used for body esteem because physical condition, sexual attractiveness, and weight concern are considered to be three important dimensions of the multidimensional construct of body esteem (Franzoi & Shields, 1984). Additionally, strong moderate correlations between the three subscales were revealed in the data for our study ( r = .56 to .65, p < .01; Huberty & Olejnik, 2006). We first performed a on e way multiple analysis of variance (MANOVA) to assess the three dependent variables of body esteem (physical condition, weight concern, and sexual attractiveness) using group membership (nonathlete, MOS athletes, LOS athletes) as the independent variable. Prior to assessing MANOVA results, Box M was used to test the assumption of covariance matrix homogeneity across the three groups (Huberty & Olejnik, 2006). Assuming similar covariance matrices and a significant MANOVA, group differences and follow up hy potheses were analyzed with DDA. Thin ideal internalization (Hypotheses 2), body surveillance (Hypothesis 3), and body shame (Hypothesis 4) were evaluated separately, using the following pairwise comparisons for 121

PAGE 122

each outcome: (a) non athletes to MOS athlet es, (b) nonathlete to LOS athletes, and (c) MOS athletes to LOS athletes. The separate analysis of each dependent variable was warranted in this research case because we proposed a separate hypothesis for each of the three constructs; and, despite each co nstruct being related to some degree, researchers propose each dependent variable to be a separate construct in the overall self objectification process with each providing unique information in objectification theorys meditational model (Moradi, 2010). W hen pre planned contrasts are of primary interest, as they were when testing hypotheses 2, 3, and 4 in this study, conducting an unnecessary omnibus F test can result in a loss of power (Huberty & Olejnik, 2006; Myers, Well, & Lorch, 2010). Thus, an omnibus test was not conducted prior to analyzing the pairwise comparisons. Levenes test was used to test t he equ al variances assumption prior to analyzing pairwise contrasts. If the equal variances assumption was upheld, the Bonferroni method was used to adjus t the Type I error rate that is associated with multiple pairwise comparisons (Huberty & Olej nik, 2006; Myers et al., 2010). If Levenes revealed that the equal variances assumption was violated, Games Howell was used in lieu of Bonferroni as t he Games Ho well method accounts for unequal variances and unequal group sizes (Myers et al., 2010). Results Preliminary Analyses Demographic information for the three groups is presented in Table 41. The majority of participants in this study were Non Hispanic Whit e ( n = 259; 63.6%), while 14.7% ( n = 60) were Hispanic, 9.6% (n = 3 9) were Black or African American 7.9% (n = 3 2) were Asian, 3.7% ( n = 15) reported being multi racial, and 0.5% ( n = 2) did not report race or ethnicity. The mean age of participants was 1 9.77 years (SD = 1.44 years). Academic classification was well distributed with 2 8.7% of the sample being first year students, 21.6% second year students, 21.4% third year students, 25.8% fourth year students and 2.5% indicating other undergraduate stat us Average 122

PAGE 123

BMI was 23.0 kg/m2 (SD=3.97 kg/m2), and there was no difference in average BMI across the three groups ( F (2,401) = 1.25, p = .288). Testing Hypothesis 1 (Multivariate Analysis) Table 42 provides descriptive statistics for the three body este em outcomes, including g roup means on each of the three outcomes, and error correlations between the three outcomes. Results of Boxs M test for covariance homogeneity indicated the covariance matrices for the three groups were somewhat different [ F (12, 25418) = 2.04, p = .018]. However, the Box test is overly sensitive to even small departures from covariance equality or multivariate normality (Huberty & Olejnik, 2006), so we examined the (natural) logarithms of the determinants for each covariance matrix The generalized variance of t he LOS athlete group (ln =10.14) was slightly less than either the nonathlete (ln = 11.70) or MOS athlete (ln = 11.38) groups; however, these group variations were not different enough to invalidate our results (Huberty & Ol ejnik, 2006). Considering the slight variation between the covariance matrices and the unequal sizes of the three groups, we proceeded with the one way MANOVA opting to use Pillais trace statistic ( U ) to evaluate multivariate effects. Pillais trace is ro bust to unequal sample sizes, as well as small violations of normality and unequal variance covariance matrices (Haase & Ellis, 1987). The results of the MANOVA provided evidence that the observed differences among the three groups were generalizable to th e populations they represented with respect to the three body esteem variables [ U = .155, F (6, 806) = 11.26, p 2 adj = .08]; thus, H ypothesis 1 (body esteem differences between the groups) was supported. We followed up the statistically significant MANOVA with DDA. The three F to remove values, as reported in Table 43, indicated that physical condition contributed the most, by far, to overall group differences ( F to remove = 24.10) followed by sexual attractiveness ( F to remove = 2.01). Weight concern appeared to contribute the least ( F to remove = .30). We 123

PAGE 124

further explored differences between the three g roups by obtaining two linear discriminant functions (LDFs) for the multivariate body esteem outcome. The first LDF explained 97.1% of the variance, canonical R2 = 14.9; whereas the second explained only 2.9% of the variance, canonical R2 = 0.53. The first LDF was significant in distinguishing between the groups (LDF1: p < .001), while the second LDF was not (LDF2: = 1.077 p = .345). Examination of group centroid plots (Figure 41) confirmed a clear separatio n of the groups on the first function, with the nonathlete group centroid mean (LDF1 mean = 2.14) being distinctly less than the centroid means of both the MOS athlete group (LDF1 mean = .729) and LOS athlete group (LDF1 mean = .849; Table 44). We inter preted the variables underlying these group differences by evaluating the correlations of each of the three outcome variable scores to their respective LDF score (structure r s). Based on the structure rs for our data (presented in Table 45), the first c onstruct as represented by LDF1 was shown to be defined primarily by physical condition ( r = .98), with weight concern also contributing moderately ( r = .52). Sexual attractiveness contributed the least ( r = .37), although not negligibly (Huberty & Olejnik, 2006). Thus, the differences between the nonathlete group and two athlete groups appear to be attributable to the athletes having better body esteem, and feeling much more positively about their physical condition and somewhat more positively about thei r physical appearance. We defined this first construct as physical condition plus appearance These findings provide support for Hypothesis 1a, which posited that the differences between the groups would be mostly attributed to physical condition, and mode rately attributed to weight concern. Notably, the F to remove statistics (refer to Table 43) indicated that sexual attractiveness contributed to group differences more than weight concern; however, weight concern was a 124

PAGE 125

larger contributor than sexual attr activeness in separating the groups on the first statistically significant dimension. Further examination of the structure r s revealed that the sexual attractiveness component is more highly correlated with LDF2 ( r = .84; Table 45), which was not statistically significant, than with LDF1. Figure 41 shows that the second function does appear to indicate a non statistically significant difference between the MOS athletes and LOS athletes, and a more subtle difference between MOS athletes and nonathletes. Thus, it is possible that sexual attractiveness contributes to overall group differences more than weight concern; but weight concern is more highly related to the first, statistically significant, construct, while the sexual attractiveness differences occurred as part of a separate construct. For discussion purposes, we labeled the second construct, as defined by LDF2, sexual attractiveness Our second s ubhypothesis for body esteem (H ypothesis 1b) posited that the differences in feelings about weight conc ern would be more pronounced between non athletes and LOS athletes and less pronounced between nonathletes and MOS athletes This subhypothesis was not supported by the results of this study, as weight concern contributed to separating athletes from nonathletes, but not in separating the two athlete groups. Examination of the group means for the weight concern construct confirms this, as the almost identical group means for the two athlete groups (LOS athletes, M = 34.97; MOS athletes, M = 34.44), were notably higher than the nonathlete group mean (M = 29.79). Testing Hypotheses 2, 3, & 4 (Pairwise Comparisons) Descriptive statistics, including means and standard deviations for the outcomes of thin ideal internalization, body surveillance, and body sham e are reported in Table 46. Levenes test for equal variances was upheld for each of the three outcomes : thin ideal internalization [ F (2, 404) = 3.00, p =.051]; body surveillance [ F (2, 404) = 0.71, p =.491]; and body shame [ F (2, 404) 125

PAGE 126

= 1.92, p =.148] ; thu s, a Bonferroni correction was used in controlling the Type I error rate for t he multiple comparisons of H ypothesis 2, Hypothesis 3, and Hypothesis 4. Results of the pre planned pairwise comparisons for each of the outcomes revealed mixed results for H ypo theses 2 through 4. A s predicted in H ypothesis 2, LOS athletes internalized the thin ideal significantly less than non athletes ( t = 3.94, p < .001); and internalization did not differ between the MOS athletes and nonathletes ( t = 1.10, p = .273). Howev er, contrary to our hypothesis, we found no difference in level of thin ideal internalization between the LOS athletes and MOS athletes ( t = 1.47, p =.143). For H ypothesis our prediction that nonathletes would report higher levels of body surveillance than LOS athletes was supported ( t = 4.96, p < .001). Contrary to our prediction, however, MOS athletes engaged in body surveillance less than nonathletes ( t = 2.51, p = .012), and there was no statistically significant difference between the LOS athlet es and MOS athletes on the body surveillance outcome ( t = 0.88, p = .381). Our fourth H ypothesis predicted that both MOS athletes and LOS athletes would have significantly less body shame than nonathletes. This was upheld for LOS athletes ( t = 2.97, p < .0032), but not for MOS athletes ( t = 0.21, p = .832). These results unexpectedly indicate that MOS athletes and nonathletes experience similar levels of body shame. As expected, there was no difference between the LOS athletes and MOS athletes ( t = 1.64, p =.103) on body shame. Discussion Researchers have traditionally studied BID among athletes in the context of sport related pressures without considering whether societys increased sexual objectification of female athletes may exert additional pressur e on the female athlete to meet societys feminine and sexualized athletic ideal (Hardin & Greer, 2009; Parsons & Betz, 2011; Varnes et al., 2013). We designed the present study to examine differences between nonathletes, collegiate female 126

PAGE 127

athletes partic ipating in sports that are less objectified, and collegiate female athletes participating in sports that more objectified on the outcomes of thin ideal internalization, body surveillance, and body shame, as well as feelings about physical condition, sexual attractiveness, and weight concern. Our results revealed that athletes at this Division I university have higher body esteem than nonathletes, with the physical condition outcome contributing the most to this difference. This was expected, as other research indicates that athletes feel better about the functional abilities of their body (e.g., strength, agility, energy) when compared to nonathletes (Fellows, 1999; Petrie & Greenleaf, 2012; Varnes et al., 2013). Our results also suggest that weight concern and sexual attractiveness both contributed to athletes enhanced body esteem over non athletes. Although we expected physical condition to be the greatest contributor to body esteem differences between athletes and non athletes, the magnitude of this di fference was somewhat surprising. The physical condition outcome was almost perfectly correlated ( r = .98) with the construct defined as physical condition plus appearance while the weight concern outcome was more moderately correlated with this construct ( r = .52), and sexual attractiveness was somewhat correlated with this construct ( r = .37). Collegiate female athletes typically have set strength and conditioning programs related to their sport participation, which likely contributes to their strong pos itive feelings regarding physical abilities and functionality. While physical condition and weight concern were unquestionably most related to the first function ( r > .50), sexual attractiveness appeared to be more highly related to the second function ( r = .84) than to the first ( r = .37); and mean score differences (refer to Table 42) on sexual attractiveness indicated that, of the three groups, MOS athletes felt the most sexually attractive. In this study, however, these differences were not statistical ly significant. This non127

PAGE 128

statistically significant result could be partly attributable to low power due to small sample size in the MOS athlete group. Thus, sexual attractiveness differences may need to be explored separately from physical condition and we ight concern when considering sport objectification. Considering weight concern alone, it is apparent that all athletes in our study have more positive feelings about their weight related body parts when compared to nonathletes. This is somewhat contrary to our prediction that, out of the three groups, LOS athletes would feel the most positive about weight concern, as it appears that both MOS athletes and LOS athletes felt similarly about their weightrelated body parts. This is particularly noteworthy, be cause most (67.9%) of the athletes included in our MOS athlete group ( n = 19 of 27) were from sports that are also considered more appearancefocused (e.g., gymnastics and swimming and diving). Previous research has suggested that these athletes are at hi gher risk for BID and its associated mental health risks ( e.g., Engel et al., 2003; Fellows, 1999; Varnes et al., 2013). Contrary to this, our findings indicate that college athletes in swimming, diving, and gymnastics do not experience greater BID than no nathletes, and are not more pre occupied with their weight affected body parts. It seems plausible that higher body esteem, and specifically better feelings about weightrelated body parts, should also translate to lower body shame. The body parts include d on the weight concern subscale of the BES (e.g., waist, thighs, body build, buttocks, hips, legs, figure, appearance of stomach) are those that both men and women perceive as important in judging a womans attractiveness (Franzoi & Herzog, 1987), indicat ing that these are the parts that are potentially objectified and self objectified. Thus, it may seem somewhat surprising that MOS athletes felt better about weight concern than nonathletes, but experienced similar feelings of guilt regarding their potential inability to attain the cultural beauty expectations. Yet these results 128

PAGE 129

are not entirely surprising, as they are not necessarily conflicting. In fact, these results provide support for other researchers indications that the self objectification proces s is defined as being preoccupied with appearance, and not necessarily satisfied or dissatisfied with appearance (Noll & Fredrickson, 1998). In other words, although MOS athletes are more satisfied with their appearance than nonathletes, the MOS athletes and the nonathletes are similarly preoccupied with the appearance of their body. This preoccupation results in both MOS athletes and nonathletes experiencing similar levels of shame or guilt when feeling they have failed to live up to sociocultural expectations related to their appearance. Although body shame differences were statistically significant in comparing LOS athletes (M = 2.87, SD = 1.04) and nonathletes (M = 3.35, SD = 1.18), the body shame difference between LOS athletes and MOS athletes (M = 3.30, SD = 0.94) was not statistically significant. Considering body shame group means, it is probabl e that the difference between LOS athletes and MOS athletes was not statistically significant because of small sample sizes resulting in power too low to detect a difference between the two groups. It was surprising to find that MOS athletes monitor their appearance less than non athletes, despite similar feelings of body shame and thin ideal internaliz ation Considering that both MOS athletes and LOS ath letes engaged in less body surveillance than nonathletes, it appears that female participation in competitive athletics generally contributes to decreased levels of body surveillance, possibly due to increased positive feelings about physical condition. T hus, it is plausible that participation in athletics, in general, decreases self objectification tendencies. However, it appears that the higher levels of sport objectification experienced by MOS athletes may negate the positive effects of high body esteem and low body surveillance when it comes to body shame and thin ideal internalization. These findings indicate that societys 129

PAGE 130

sexual objectification of MOS athletes is a concerning factor in the female athletes self objectification process. Strengths and Limitations Recruitment strategies and efforts were one strength of this study, particularly the use of random sampling for recruitment of the nonathlete group, and the all inclusive sampling of the finite female athlete population at this large southeastern university. However, because nonathletes and athletes were sampled at only one university, the sample can only be assumed to be representative of the female student population at the university where the study took place, and may or may not be general izable to other individual universities. Despite a high response rate and our ability to recruit a diverse group of athletes into our study, the athlete group sample size and its make up could be considered somewhat limiting. While the nonathlete group w as fairly large ( n = 322), the MOS athlete group was much smaller ( n = 2 7); and, of the four sports included in the MOS athlete group, gymnasts represented almost half of the total sample ( n = 13). The LOS athlete group was proportionately larger ( n = 58), but was mostly comprised of soccer athletes ( n = 20) and track and field athletes ( n = 14). Based on the demographic makeup of the sampled university, we believe that our nonathlete sample was an adequate representation of the overall female student p opulation. Additionally, the racial/ethnic distribution of our sample appeared to be more diverse than those reported in other similar studies (Lenart et al. 1995; Steinfeldt et al. 2011). Despite this, there were some apparent ethnic/race differences ac ross the three groups. The nonathlete group included 32 participants ( 7.9%) who identified only as Asian, while no one in the athlete group responded being of Asian descent. Also of note is that the MOS athlete group did not include any Hispanic participa nts, while the nonathlete ( n = 52) and LOS athlete ( n = 8) groups did (16.2% and 13.8% respectively). We were unable to explore the effect of race/ethnicity in our 130

PAGE 131

outcomes because there was not enough power to do so; h owever, researchers have indicated t hat race and ethnic differences may play a role in body image and self objectification (Harrison & Fredrickson, 2003; Roberts et al. 2006). Additionally, at least one study has found body image differences between adolescent girls of different races parti cipating in the same sport (Crissey & Honea, 2006). Further, the NCAA (2012) indicates that the breakdown of race and ethnicity varies by sport. As such, researchers should consider examining race and college athletic status in an interactive model predict ing BID or self objectification. Lastly, we went to extensive lengths to assure anonymity of participant responses. This was done not only for ethical reasons, but also because prior re search indicates response bias should be considered, as female athletes may be concerned that their responses will jeopardize their eligibility (Garner, 2004). Despite our attempts to assure anonymity, response bias is somewhat of a concern because the second wave of recruitment was conducted with assistance from individuals who were directly involved with the sport teams. We did not conduct comparisons between responses from the first and second waves of data collection, because the second wave data collection was sport focused and attempted to gain responses from athletic teams that were under represented in the first wave. Conclusion and Next Steps In general, it appears that female participation in competitive athletics is related to increased body esteem and decreased body surveillance. Additionally, per our findi ngs, at hletes participating in less objectified sports self objectify less than nonathletes. This was indicated by LOS athletes internalizing the thin ideal less, engaging in less body surveillance, and feeling less body shame than nonathletes. Because LOS athl etes self objectify less and experience greater body esteem, we can conclude that they also appear to be at decreased risk for the psychological and behavioral risks associated with self objectification and BID. 131

PAGE 132

It is apparent that athletes participating in more objectified sports also experience better body esteem than nonathletes. Considering the make up of our MOS athlete group, our findings appear to refute previous researchers indications that appearance focused athletes in gymnastics and swimming a nd diving are at higher risk for BID than nonathletes. Despite these findings, it does not appear that this high body esteem translates to decreased mental health risks for athletes participating in more objectified sports as their levels of body shame w ere similar to non athletes. In other words, according to the mediation model of objectification theory ( Szymanski et al., 2011; Moradi, 2010), MOS athletes and non athletes appear to be at similar risk for the mental health concerns that result from body shame. Overall, our results indicate that the increased sexual objectification of certain sports serves to promote thin ideal internalization among athletes participating in those more objectified sports thereby enhancing body shame without increasing body surveillance or BID. These findings suggest that researchers and interventionists should put more focus on sexual and self objectification of collegiate female athletes participating in these sports, as their psychological and mental health risks appea r to be associated with objectification and not necessarily BID. Researchers should continue this line of work by examining how other factors might mediate or moderate these relationships. For example, body comparison has been indicated as a mediator between sexual objectification and body shame (Tyka & Sabik, 2010). Thus, researchers should examine this relationship for female athletes, striving to identify who more and less objectified athletes compare themselves to, and how that mediates the relationship between objectification theory constructs. Researchers should also consider the female athletes perception of how she is viewed or judged by others. If an athlete perceives that others are critical of her appearance or weight, she may experience increased body shame (Muscat & Long, 2008), despite her own 132

PAGE 133

positive feelings about her body condition and appearance. In a similar vein, research exploring female athletes awareness and perception of how they are depicted in the media will allow for a better understanding of the self objectification process of female athletes. 133

PAGE 134

Table 4 1. Demographic c ounts and BMI by group Non Athlete ( N = 322) MOA ( N = 27) LOA ( N = 58) N % N % N % Race Asian 32 9.9% 0 0 Black / African American 30 9. 3% 2 7.4% 7 12.1% Non Hispanic White 197 61.2% 21 77.8% 41 70.7% M ulti Rac ial 9 2.8% 4 14.8% 2 3.5% Hispanic 52 16.2% 0 8 13.8% Missing 2 0.6% 0 0 Undergraduate Status 1 st Year 100 31.1% 3 11.1% 14 24.1% 2 nd Year 63 19.6% 6 22.2% 19 32.8% 3 rd Year 69 21.4% 7 12.3% 11 19.0% 4 th Year 82 25.5% 11 40.7% 12 20.7% Other Undergraduate 8 2.5% 0 2 Body Mass Index Mean (SD) 23.12 (4.26) 22.28 (2.29) 22.39 (2.67) Note: No participants indicated their race as American Indian or Alaska Native or Native Hawaiian or Other Pacific Islander, so they are not included here. Table 4 2. Means and s tandard deviations by g roup for body e steem Group Means/SDs Error Correlations Varia ble NA Athletes MOS Athletes LOS Weight Concern Physical Condition Sexual Attractiveness 46.5 1 (7. 77 ) 5 0 78 (8. 63 ) 48. 91 (6. 90 ) .55 .56 Weight Concern 29.79 (9.4 6 ) 34. 44 (9. 91 ) 3 4 97 ( 8 31 ) .63 Physical Condition 29. 68 (7.46) 3 6 5 2 (6.66) 36. 95 (5. 0 8) NA=Non athlete; MOS=More Objectified Sports; LOS = Athlete Less Objectified Sports. a df1 = 2, df2 = 406 b p < .01 c p < .001 134

PAGE 135

Table 4 3. F to r emove r esults for body e steem Variable F to Remove Physical Condition 24.103 Sexual Attractiveness 2.005 Weight Concern 0.295 Table 4 4. Functions at g roup c entroids for the t hree groups on body e steem data Group LDF 1 2 Non Athlete .214 .002 MOS A .729 .241 LOS A .849 .099 Table 4 5. Structure rs for body esteem Vari able LDF 1 a LDF 2 Sexual attractiveness .37 .84 b Weight Concern .52 b .02 Physical Condition .98 b .17 a Significant at p < .001 b Largest absolute correlation between each variable and any discriminant function. Table 4 6. Mean and s tandard deviati on for each g roup as a f unction of outcome ; m ultiple c omparisons, and e ta s quared values Group Means/SDs Diff a p < .05 using Bonferroni adj ustment 135

PAGE 136

Figure 4 1. Linear discriminant functions for the body esteem constructs of physical condition, sexual attractiveness, and weight concern 136

PAGE 137

CHAPTER 5 SUMMARY, CONCLUSIONS AND RECOMME N DATIONS Summary Body image disturbance ( BID ) is highly prevalent among collegeaged women, so much that researchers ha ve described it as normative (e.g., Espelage et al., 2003; Grabe et al., 2008; Fitzsimmons Craft, 2011). The present work was developed in an effort to progress BID research in this population b y investigating BID, and its measurement, among college aged females, with an emphasis on identifying how participation in collegiate athletics might promote or protect from BID. The following steps were undertaken to achieve this goal: (1) explore recent BID literature comparing collegiate female athletes and non athletes with consideration to the changes in sociocultural beauty ideals in the last 15 years; (2) identify salient and temporally relevant measures of BID for research in college female populati ons; (3) critically evaluate the most commonly used BID measure for appropriateness of use in research among college females; and (4) using temporally relevant and appropriate measures, compare current BID levels among highly competitive collegiate female athletes and female non athletes. Three investigations were developed and presented that expand on the current literature, address weaknesses of previous research, and provide direction for future research. Chapter 2 presented a systematic literature review of studies from 19972012 that explored BID among collegiate female athletes given the two recent sociocultural changes: (a) the increase in sexual objectification of female athletes in the media; and (b) the evolution of ideal female beauty to include appearing athletic in addition to being thin The review aimed to provide an in depth and focused update of research previously presented by Hausenblas and Downs (2001) that indicated athletes (across age, gender, and competition level) experienced lower l evels of BID than nonathletes. The update, as written in Chapter 2, focused specifically 137

PAGE 138

on comparing BID among collegiate female athletes and college female non athletes. Findings indicated that female involvement in collegiate athletics still provides s ome protection from BID; however, this protection appeared to attenuated for athletes in more feminine sports (e.g., gymnastics) and at higher competition levels (e.g., Division I). The work presented in Chapter 2 also identified 13 different instruments used to evaluate BID across the 10 studies. Of these, the Eating Disorder Inventory (EDI) was used by half of the studies, supporting previous indications that the EDI is the most commonly used survey instrument in BID research. There were indications, howe ver, that the EDI may not be appropriate for BID measurement in general female populations due to the instruments clinical nature. Also, given the increased sexual objectification of female athletes and the scarcity of research regarding this it was dete rmined that there may be more optimal survey instruments to evaluate the current status of BID. Specifically, it was noted that researchers should consider measures that focus on self objectification, positive body image, and thin and athletic ideal inter nalization The manuscript titled A Systematic Review of Studies Comparing Body Image Concerns Among Female College Athletes and Non Athletes, 19972012 was provisionally accepted for publication in Body Image: An International Journal of Research ( May 2 9, 2013) As indicated by the findings of Chapter 2, the EDI has been the most frequently and popularly used measure of BID for research in female populations; thus, it warranted consideration for administration in the final study of this work. However, be cause of indications that the instrument may not be appropriate for use in nonclinical populations, further evaluation of the EDI was needed, prior to its administration. The aim of Chapter 3 was to examine the relevance and appropriateness of the EDI in n onclinical BID research. Chapter 3 presented a c ritical review of factor analytic studies that examined the internal structure of the EDI for use in 138

PAGE 139

nonclinical populations of women from Westernized countries, in an effort to synthesize the results of the reviewed studies to identify a stable EDI structure for this population. Thorough and systematic methodological evaluation of the six studies identified for review revealed that methodological rigor varied by study. Final results of the reviewed studies al so varied somewhat. Examination of i tem mapping patterns and validity and reliability results across the six reviewed studies revealed that significant modifications were indicated for using the EDI in nonclinical research. Thus it was decided that the ED I would not be administered as part of the final study (Chapter 4). The manuscript titled Validity of the Eating Disorder Inventory for Nonclinical Research in Populations of Westernized Women: A C ritical R eview of F actor A nalytic S tudies was submitted f or publication in International Journal of Eating Disorders (revision for resubmission in progress). Chapter 4 described an original research study conducted with 322 randomly selected female undergraduate non athletes, and 85 NCAA female athletes from a large southeastern Division I university. The study used objectification theory as a guide in comparing non athletes and two groups of athletes (athletes participating in sports that are more objectified by society and athletes participating in sports that are less objectified by society) on three constructs of body esteem (sexual attractiveness, weight concern, and physical condition), and three constructs identified as part of the self objectification process: thin ideal internalization, body surveillance and body shame. The study was the first to consider external sociocultural pressures (i.e., sexual objectification of female athletes) when comparing BID and self objectification among athletes and nonathletes. This was also the only known study to exam ine the construct of body surveillance among female athletes. 139

PAGE 140

The first H ypothesis tested the differences between three groups (non athletes, athletes participating in more objectified sports (MOS), and athletes participating in less objectified sports (L OS)) on the constructs measuring body esteem. Multivariate analysis revealed significant differences across the three groups on body esteem. Descriptive discriminant analysis (DDA) results indicated that the nonathletes experienced less positive body este em than both athlete groups with physical condition being the primary contributor to this difference. Weight concern (feelings about parts of the body modifiable through diet and exercise) also contributed moderately to this difference, and sexual attracti veness (feelings about parts of the body not modifiable through diet and exercise) contributed minimally. Results also suggested that athletes participating in MOS feel more sexually attractive than either of the two groups; although this dimension of diff erence was not statistically significant. The second, third, and fourth H ypothesis compared groups on the self objectification outcomes of thin ideal internalization, body surveillance, and body shame respectively. It was found that athletes participating in LOS self objectify less than non athletes. This was indicated by LOS athletes reporting less body surveillance, less body shame, and less internalization of the thin ideal. There were no statistically significant differences between LOS athlete and MOS athletes. Comparisons between nonathletes and MOS athletes were somewhat mixed. Athletes participating in MOS also reported less body surveillance than the non athletes; however, their levels of body shame and thin ideal internalization were similar to th e nonathletes. This is particularly interesting because thin ideal internalization has been shown to be a strong predictor of body surveillance (Fitzsimmons Craft et al., 2012; Moradi et al., 2005), and body surveillance has been shown to be a somewhat st ronger predictor of body shame than has thin ideal internalization (Moradi et al., 2005; Tylka & Hill, 2004). While past research indicates that body 140

PAGE 141

surveillance is slightly more positively correlated with body shame than is internalization (Moradi et al. 2005), our results may indicate that, for the athlete group, thin ideal internalization is a better predictor of body shame than is body surveillance. These results could indicate that the increased sexual objectification of certain sports serves to prom ote thin ideal internalization among athletes participating in MOS, thereby enhancing body shame without increasing body surveillance. Overall, the findings of the study in Chapter 4 indicate that athletes have better body esteem than non athletes and that athletes participating in LOS self objectify less than nonathletes. The findings also indicate that researchers using objectification theory should consider that body surveillance may not be a strong indicator of self objectification among female athlete s participating in MOS. The manuscript titled Body E steem and S elf O bjectification: Comparing C ollegiate F emale A thletes and N onA thletes in the C ontext of Objectification Theory is being prepared for submission to a scholarly journal Conclusion The ove rall findings from this line of research indicate that female participation in collegiate athletics continues to provide protection from BID, despite societys increased sexual objectification of female athletes. Findings of this work also support other re searchers implications that female athletes generally experience better body esteem (Fellows, 1999; Steinfeldt et al., 2011), are more satisfied with their weight and weight affected body parts (Fellows, 1999; Robinson & Ferraro, 2004), and desire a more athletic, physically fit figure than nonathletes (DArcy, 2007; Hoag, 2012). Also, as indicated in Chapter 2 of this work and by other researchers (DArcy, 2007; Hoag, 2012; Homan, 2010), it appears that the athletes enhanced athletic internalization may not be related to decreased risk for BID. 141

PAGE 142

This work also supports that athletes feel better about their bodys physical form and function (Fellows, 1999; Petrie & Greenleaf, 2012 ; Wiggins & Moode, 2000). There is consensus among researchers that valuing the body for its abilities is the antithesis of self objectifying (taking an external observer perspective of the body); however, it is apparent through this work that better feelings about physical condition do not necessarily translate into decreased sel f objectification for all athletes. A lthough our overall findings indicated that all athletes felt better about the bodys abilities when compared to nonathletes, we also found that not all athletes experienced less overall self objectification. While LOS athletes experienced less self objectification, findings for MO S athletes were mixed, and indicat ed that sport objectification experiences could be contributing to thin ideal internalization and body shame. Since Fredrickson and Roberts (1997) first intr oduced objectification theory, researchers, psychologists, and clinicians have tested its tenets and principles providing strong support for a path model that shows sexually objectifying experiences lead to psychological disorders and mental health risks through the self objectification process. The research supporting this theory has largely been conducted in general college female populations though research across races, cultures, and sexual orientation ha ve found support for the objectification theory model in minority populations as well (e.g., Moradi et al., 2005; Moradi & Huang, 2008; Kozee et al., 2007; Szymanski et al., 2011). Through the work presented here, it appears our understanding of how objectification theory constructs are related is defi cient when it comes to the female athlete population, and specifically MOS athletes. While our results support the relationship between thin ideal internalization and body shame, some correlates appear to be refuted (e.g., negative relationship of physical condition to body surveillance and thin ideal internalization ) ; and others difficult to interpret because they oppose findings in this work (e.g., body surveillance and body 142

PAGE 143

shame of MOS athletes ). There are many possible reasons construct relationships d iffer for athletes and nonathletes, including athletic status acting as a moderator that diminishes body surveillance, while sport objectification acts as a moderator in the self objectification process that affects body shame and thin ideal internalization. In conclusion, it is apparent that female participation in competitive collegiate athletics is related to lower levels of BID. It is also apparent that role athletic participation plays in self objectification varies by sport. These differences indicat e that competitive athletes participating in sports that are more objectified by society could be at greater risk for certain psychological outcomes and health behaviors than athletes who are less objectified by society despite them feeling better about t heir body abilities and appearance. T he mechanism of this difference, however, is not yet understood and more research needs to be done. Recommendations for Researchers We continue to lack understanding about how the female athletes body image is developed, enhanced, or worsened. Thus, we need more foundational information to better guide research studies and intervention possibilities. More research is necessary that explores how societal definitions of athletic femininity are developed. For example, ide ntifying the relationship (c orrelational and causal ) between what society defines as a feminine sport and the medias objectification of athletes from these sports. Additionally, it is necessary to further determine how current societal definitions of athl etic femininity affect the female athletes overall identity and how this relates to self objectification. In a similar vein, researchers should examine sport specific female athletes awareness and perceptions of the medias depiction of female athletes, and if female athletes perception contributes to their personal feelings of being sexually objectified. One athlete group, specifically, that may be at higher risk than other athletes are volleyball players (Hoag, 2012 and Chapter 2 of this work). Future r esearch should examine this 143

PAGE 144

more closely to identify why. Last, in terms of foundation research, we are still lacking in understanding how nonathletes and athletes interpret and respond to different BID assessment instruments Enhancing this understanding can serve to inform researchers on which instruments will provide the best interpretable information on outcomes of interest Despite deficiencies in the broader body of literature, this work has helped to identify next steps for practical/intervention re search that could be explored among nonathletes. U nderstanding that athletes value the abilities of their body more than nonathletes suggests that interventionists should identify ways to enhance nonathletes feelings regarding their bodys abilities. S pecifically, intervention researchers should explore how dedication to perfecting a physical skill (e.g., roller blading or rock climbing), even outside of the sports realm, may enhance body esteem among nonathletes There are also recommendations for next steps in moving forward from the findings of this work. These include: (1) c onduct a confirmatory factor analysis on the EDI s tructure presented in Chapter 3; (2) determine if objectification theory constructs are upheld in a path model for collegiate f emale athletes; (3) examine the interactive effects of competition level, sport, and race on body image and self objectification outcomes; (4) e xamine how the medias sexual objectification of high profile, sport specific female athletes affects the BID of impressionable female athlet es participating in that sport; (5) e x plore the role of body comparison in predicting athletes self objectification; and (6) i dentify who athletes are comparing themselves to, and what how that affects other self objectificati on constructs. 144

PAGE 145

APPENDIX A DEMOGRAPHIC AND BACKGROUND QUESTIONS The following items were included for Chapter 4 only. 1. What is your sex? Female Please proceed to Question 3 (this section) Male Thank you for your participation, but this survey is for fe males only. You may stop the survey at this time, and you do not need to return the survey. 2. Are you registered for at least 12 credits at the University of Florida for the current semester? Yes Please proceed to Question 4 (this section) No Thank you for your participation, but this survey is for full time students only. You may stop the survey at this time, and you do not need to return the survey. 3. What is your classification (check all that apply)? 1st year undergraduate 2nd year undergraduate 3rd year undergraduate 4th year undergraduate International Student Transfer student Other undergraduate student I am not an undergraduate student Thank you for your participation, but we are surveying undergraduate students only. You may stop the survey at this time, and you do not need to return the survey. 4. How old are you today? Years 5. Are you a student athlete who plays an NCAA varsity level sport for the University of Florida? Yes, I am a student athlete --Continue to Question #2 (this sec tion) No, I am not a student athlete --Skip Question #2 and go to Question #3 (this section) 6. Which University of Florida NCAA varsity athletic team are you a member of? Basketball Cross Country Diving Golf Gymnastics 145

PAGE 146

Lacrosse Soccer Softball Swimming T ennis Track Volleyball 7. Are you Hispanic or Latina? Yes No Regardless of your response to the previous question, please continue to answer the following, if applicable, by marking one or more boxes to indicate what you consider your race to be. 8. How would you describe yourself? (Choose one or more from the following racial groups) American Indian or Alaska Native Asian Black or African American Native Hawaiian or Other Pacific Islander White 9. How tall are you (please respond in feet and inches. For exa mple 5 feet 2 inches)? 10. What is your approximate weight (please respond in pounds)? 146

PAGE 147

APPENDIX B OBJECTIFIED BODY CONSCIOUSNES SCALE This scale was used for Chapter 4 only. INSTRUCTIONS: On a scale of 1 to 7 (1 = strongly disagree; 4 = neither agree no r disagree; 7 = strongly agree), please clearly circle the response that corresponds to how much you agree or disagree with each statement. IMPORTANT : Circle NA only if the statement does not apply to you. Do not circle NA if you don't agree with a statem ent. For example, if the statement says "When I am happy, I feel like singing" and you don't feel like singing when you are happy, then you would circle one of the disagree choices. You would only circle NA if you were never happy. Strongly Disagree Nei ther Agree nor Disagree Strongly Agree Does not Apply 1. I rarely think about how I look. 1 2 3 4 5 6 7 NA 2. When I cant control my weight, I feel like something must be wrong with me 1 2 3 4 5 6 7 NA 3. I think it is more important that my clothes are comfortable than whether they look good on me 1 2 3 4 5 6 7 NA 4. I think a person is pretty much stuck with the looks they are born with 1 2 3 4 5 6 7 NA 5. I feel ashamed of myself when I havent made the effort to look my best.. 1 2 3 4 5 6 7 NA 6. A large part of being in shape is having that kind of body in the first place.. 1 2 3 4 5 6 7 NA 7. I think more about how my body feels than how my body looks.. 1 2 3 4 5 6 7 NA 8. I feel like I must be a bad person when I dont look as good as I could.. 1 2 3 4 5 6 7 NA 9. I rarely compare how I look with how other people look 1 2 3 4 5 6 7 NA 10. I think a person can look pretty much how they want to if they are willing to work at it... 1 2 3 4 5 6 7 NA 147

PAGE 148

Strongly Disagree Neither Agree nor Disagree Strongly Agree Does not Apply 11. I would be ashamed for people to know what I really weigh 1 2 3 4 5 6 7 NA 12. I really dont think I have much control over how my body looks... 1 2 3 4 5 6 7 NA 13. Even when I cant control my weight, I think Im an okay person... 1 2 3 4 5 6 7 NA 14. During the day, I think about how I look many times 1 2 3 4 5 6 7 NA 15. I never worry that something is wrong with me when I am not exercising as much as I should. 1 2 3 4 5 6 7 NA 16. I often worry about whether the clothes I am wearing make me look good 1 2 3 4 5 6 7 NA 17. When Im not exercising enough, I question whether I am a good enough person. 1 2 3 4 5 6 7 NA 18. I rarely worry about how I look to other people. .. 1 2 3 4 5 6 7 NA 19. I think a persons weight i s mostly determined by the genes they are born with.. 1 2 3 4 5 6 7 NA 20. I am more concerned with what my body can do than how it looks.. 1 2 3 4 5 6 7 NA 21. It doesnt matter how hard I try to change my weight its probably always going to be about the same. 1 2 3 4 5 6 7 NA 22. When Im not the size I think I should be, I feel ashamed 1 2 3 4 5 6 7 NA 23. I can weigh what Im supposed to when I try hard enough 1 2 3 4 5 6 7 NA 24. The shape you are in depends mostly on your genes. 1 2 3 4 5 6 7 NA 148

PAGE 149

APPENDIX C BODY ESTEEM SCALE This scale was used for Chapter 4 only. INSTRUCTIONS: In this section, several body parts and functions are listed. Please read each item, and clearly circle the number corresponding with how you feel about this part or function of your own body using the following scale: 1 = Have strong negative feelings 2 = Have moderate negative feelings 3 = Have no feel ing one way or the other 4 = Have moderate positive feelings 5 = Have strong positive feelings Have strong negative feelings Have moderate negative feelings Have no feelings one way or the other Have moderate positive feelings Have strong positive feeling s 1. Body Scent 1 2 3 4 5 2. Appetite 1 2 3 4 5 3. Nose 1 2 3 4 5 4. Physical Stamina 1 2 3 4 5 5. Reflexes 1 2 3 4 5 6. Lips 1 2 3 4 5 7. Muscular strength 1 2 3 4 5 8. Waist. 1 2 3 4 5 9. Energy Level 1 2 3 4 5 10. Thighs 1 2 3 4 5 11. Ears 1 2 3 4 5 12. Biceps 1 2 3 4 5 13. Chin 1 2 3 4 5 14. Body Build. 1 2 3 4 5 15. Physical Coordination. 1 2 3 4 5 149

PAGE 150

Have strong negative feelings Have moderate negative feelings Have no feelings one way or the other Have moderate positive feelings Have strong positive feelings 16. Buttocks. 1 2 3 4 5 17. Agility. 1 2 3 4 5 18. Breasts.. 1 2 3 4 5 19. Appearance of eyes 1 2 3 4 5 20. Cheeks / cheekbones. 1 2 3 4 5 21. Hips 1 2 3 4 5 22. Legs 1 2 3 4 5 23. Figure. 1 2 3 4 5 24. Sex Drive... 1 2 3 4 5 25. Sex Organs 1 2 3 4 5 26. Appearance of Stomach.. 1 2 3 4 5 27. Health. 1 2 3 4 5 28. Sex Activities. 1 2 3 4 5 29. Body Hair... 1 2 3 4 5 30. Physi cal Condition 1 2 3 4 5 31. Face 1 2 3 4 5 32. Weight 1 2 3 4 5 150

PAGE 151

APPENDIX D INTERNALIZATION GENERAL SUBSCALE OF THE SOCIOCULTURAL AT TITUDES TOWARDS APPEARANCE QUESTIONNAIRE 3 This instrument was used for Chapter 4 only. INSTRUCTIONS: Please read each of the following items carefully and indicate the number that best reflects your agreement with the statement. Definitely Disagree = 1 Mostly Disagree = 2 Neither Agree Nor Disagree = 3 Mostly Agree = 4 Definitely Agree = 5 Definitely Disagr ee Mostly Disagree Neither Agree nor Disagree Mostly Agree Definitely Agree 1. I do not care if my body looks like the body of people who are on TV. 1 2 3 4 5 2. I compare my body to the bodies of people who are on TV... 1 2 3 4 5 3. I would like my body to look like the models who appear in magazines 1 2 3 4 5 4. I compare my appearance to the appearance of TV and movie stars 1 2 3 4 5 5. I would like my body to look like the people who are in movies. 1 2 3 4 5 6. I do not co mpare my body to the bodies of people who appear in magazines 1 2 3 4 5 7. I wish I looked like the models in music videos. 1 2 3 4 5 8. I compare my appearance to the appearance of people in magazines.. 1 2 3 4 5 9. I do not try to look l ike the people on TV.. 1 2 3 4 5 151

PAGE 152

LIST OF REFERENCES Allen, J. L. (2012). Sociocultural influences on body image across the lifespan. Dissertation Abstracts International: Section B: The Sciences and Engineering, 72(7B). Altabe, M. N. (1996). Issues in the assessment and treatment of body image disturbance in culturally diverse populations. In J. K. Thompson (Ed.), Body image, eating disorders, and obesity (pp. 129148) Washington D.C.: American Psychological Association. American Psychiatry Associati on. (1994). The diagnostic and statistical manual of mental disorders Washington, D. C.: Author; 1994. American Psychological Association, Task Force on the Sexualization of Girls. (2010). Report of the APA Task Force on the Sexualization of Girls Retri eved from http://www.apa.org/pi/women/programs/girls/reportfull.pdf. Anderson, C. M., Petrie, T. A., & Neumann, C. S. (2011). Psychosocial correlates of bulimic symptoms among NCAA Division I female collegiate gymnasts and swimmers/divers. Journal of Sport & Exercise Psychology, 33, 483505. Annis, N. M., Cash, T. F., & Hrabosky, J. I. (2004). Body image and psychosocial differences among stable average weight, currently overweight and formerly overweight women: The role of stigmatizing experiences. Body Image, 1, 155167. Avalos, L., Tylka, T. L., & WoodBarcalow, N. (2005). The Body Appreciation Scale: Development and psychometric evaluation. Body Image, 2, 285297. Bailey, S. M., Goldberg, J. P., Swap, W. C., Chomitz, V. R., & Houser, R. F. (1990). Relationships between body dissatisfaction and physical measurements. International Journal of Eating Disorders, 9, 457461. Bennett, K., & Stevens, R. (1997). The internal structur e of the Eating Disorder Inventory. Health Care for Women International, 18, 495500. Berland, N. W., Thompson, J. K., & Linton, P. H. (1986). Correlation between the EAT 26 and the EAT 40, the Eating Disorder Inventory, and the Restrained Eating Inventor y. International Journal of Eating Disorders, 5, 569574. Berry, T. R., & Howe, B. L. (2000). Risk factors for disordered eating in female university athletes. Journal of Sport Behavior, 23, 207218. Birrell, J., Meares, K., Wilkinson, A., & Freeston, M. (2011). Toward a definition of intolerance of uncertainty: A review of factor analytic studies of the Intolerance of Uncertainty Scale. Clinical Psychology Review, 31, 11981208. 152

PAGE 153

Boone, L., Soenens, B., & Braet, C. (2011). Perfectionism, body dissatisfac tion, and bulimic symptoms: The intervening role of perceived pressure to be thin and thin idealization. Journal of Social and Clinical Psychology, 30, 10431068. Broberg, A. G., Hjalmers, I., & Nevonen, L. (2001). Eating disorders, attachment, and interp ersonal difficulties: A comparison between 18to 24year old patients and normal controls. European Eating Disorders Review, 9, 381396. Bruch, H. (1973). Eating disorders: Obesity, anorexia nervosa, and the person within. New York: Basic Books: Brumber g, J. J. (1997). The body project: An intimate history of American girls. New York: Random House, Inc. Busanich, R., & McGannon, K. R. (2010). Deconstructing disordered eating: A feminist psychological approach to the body, food, and exercise relationship in female athletes. Quest, 62, 385405. Cafri, G., Yamamiya, Y., Brannick, M., & Thompson, J. K. (2005). The influence of sociocultural factors on body image: A meta analysis. Clinical Psychology: Science and Practice, 12 421433. Calogero, R. M. (2004). A test of objectification theory: Effect of the male gaze on appearance concerns in college women. Psychology of Women Quarterly, 28, 1621. Calogero, R. M., Davis, W. N., & Thompson, J. K. (2004). The Sociocultural Attitudes Toward Appearance Question naire (SATAQ 3): Reliability and normative comparisons of eating disordered patients Body Image 1, 193198. Calogero, R. M., Davis, W. N., & Thompson, J. K. (2005). The role of self objectification in the experience of women with eating disorders. Sex R oles, 52, 4350. Campbell, A., & Hausenblas, H. A. (2009). Effects of exercise interventions on body image: A meta analysis. Journal of Health Psychology, 14, 780794. Carr, E. R., & Szymanski, D. M. (2010). Sexual objectification and substance abuse in young adult women. The Counseling Psychologist, 39, 3966. Cash, T. F., & Szymanski, M. L. (1995). The development and validation of the Body Image Ideals Questionnaire. Journal of Personality Assessment, 64, 466477. Cattell, R. B. (1966). The scree tes t for the number of factors. Multivariate Behavioral Research, 1, 245276. 153

PAGE 154

Center for Reviews and Dissemination [CRD]. (2009). Systematic reviews: CRDs guidance for undertaking reviews in healthcare. Retrieved from http://www.york.ac.uk/inst/crd/SysRev/!SSL!/WebHelp/SysRev3.htm Clausen, L., Rosenvinge, J.H., Friborg, O., & Rokkedal, K. (2011). Validating the Eating Disorder Inventory 3 (EDI 3): A comparison between 561 female eating disorders patients and 878 females from the general population. Journal of Psychopathology and Behavioral Assessment, 33, 101110. Clavio, G., & Eagleman, A. N. (2011). Gender and sexually suggestive images in sports blogs. Journal of Sport Managem ent, 7, 295304. Clay, D., Vignoles, V. L., & Dittmar, H. (2005). Body image and self esteem among adolescent girls: Testing the influences of sociocultural factors. Journal of Research on Adolescence, 15, 451477. Cooper, Z., Cooper, P.J., & Fairburn, C .G. ( 1985). The specificity of the Eating Disorder Inventory. British Journal of Clinical Psychology, 24, 129130. Cordero, E. D., & Israel T. (2009). Parents as protective factors in eating problems of college women. Eating Disorders, 17, 146161. Coste llo, A. B., & Osborne, J. W. (2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment Research and Evaluation, 10, 19. Crissey, S. R., & Honea, J. C. (2006). The relationshi p between athletic participation and perceptions of body size and weight control in adolescent girls: The role of sport type. Sociology of Sport Journal, 23, 248272. Crocker, L., & Algina, J. (2008). Introduction to classical and modern test theory. Mas on, OH: Cengage Learning. Crowther, J. H., Lilly, R. S., Crawford, P. A., & Shepherd, K. L. The stability of the Eating Disorder Inventory. International Journal of Eating Disorders, 12, 97101. Cusumano, D. L., & Thompson, J. K. (1997). Body image and body shape ideals in magazines: Exposure, awareness, and internalization. Sex Roles, 37, 701721. Crissey, S. R., & Honea, J. C. (2006). The relationship between athletic participation and perceptions of body size and weight control in adolescent girls: The role of sport type. Sociology of Sport Journal, 23, 248272. DArcy, A. J. (2007). Body image: Athletes and nonathletes (Masters thesis). Retrieved from Library and Archives Canada. (ISBN: 9780494 295571). 154

PAGE 155

Daniels, E. A. (2009). Sex objects, athle tes, and sexy athletes: How media representations of women athletes can impact adolescent girls and college women. Journal of Adolescent Research, 24, 399422. Daniels, E. A., & Wartena, H. (2011). Athlete or sex symbol: What boys think of media represent ations of female athletes. Sex Roles, 65, 566579. De Barardis, D., Serroni, N., Campanella, D., Carano, A., Gambi, A. V., Conti, C., Ferro, F. M. (2009). Alexithymia and its relationships with dissociative experiences, body dissatisfaction and eating d isturbances in a nonclinical female sample. Cognitive Therapy Research, 33, 471479. de Bruin, A. P. K., Oudejans, R. R. D., Bakker, F. C., & Woertman, L. (2011). Contextual body image and athletes disordered eating: The contribution of athletic body im age to disordered eating in high performance women athletes. European Eating Disorders Review, 19 201215. DiBartolo, P. M., & Shaffer, C. (2002). A comparison of female college athletes and nonathletes: Eating disorder symptomatology and psychological w ell being. Journal of Sport & Exercise Psychology, 24 3341. Dillman, D. A., Smyth, J. D., & Christian, L. M. (2009). Internet, mail, and mixedmode surveys: The tailored design method. Hoboken, NJ: Wiley & Sons, Inc. DiNucci, J. M., Finkenberg, M. E., McCune, S. L., McCune, D. E., & Mayo, T. (1994). Analysis of body esteem of female collegiate athletes. Perceptual and Motor Skills, 78, 315319. Dorland, J. M. (2006). Objectification theory: Examining the relation between selfobjectification and flow for college aged women athletes. (Unpublished doctoral dissertation). University of Akron, Akron, OH. Doughty, J. H., & Hausenblas, H. A. (2007). Perceived social influences on disordered eating correlates of collegiate female gymnasts. Sport & Exercise P hysiology In NASPA Free Communications: Verbal and Poster. Eberenz, K. P., & Gleaves, D. H. An examination of the internal consistency and factor structure of the Eating Disorder Inventory 2 in a clinical sample. International Journal of Eating Disorder s, 16, 371379. Engel, S. G., Johnson, C., Powers, P. S., Crosby, R. D., Wonderlich, S. A., Wittrock, D. A., & Mitchell, J. E. (2003). Predictors of disordered eating in a sample of elite Division I college athletes. Eating Behaviors, 4, 333343. Engeln Maddox, R., Miller, S. A., & Doyle, D. M. (2011). Tests of objectification theory in gay, lesbian, and heterosexual community samples: Mixed evidence for proposed pathways. Sex Roles, 65, 518532. 155

PAGE 156

Espelage, D. L., Mazzeo, S. E., Aggen, S. H., Quittner, A. L., Sherman, R., & Thompson, R. (2003). Examining the construct validity of the Eating Disorder Inventory. Psychological Assessment, 15 7180. Farrell, C., Shafran, R., & Lee, M. (2006). Empirically evaluated treatments for body image disturbance: A rev iew. European Eating Disorders Review, 14, 289300. Fellows, C. L. (1999). Comparing the differences of self esteem, body esteem, and disordered eating behaviors between college female athletes and college female non athletes (Masters thesis). Retrieved from UMI Films (UMI Number: 1397170). Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117140. Fingeret, M. C. & Gleaves, D. H. (2004). Sociocultural, feminist, and psychological influences on womens body satisfaction: A structural modeling analysis. Psychology of Women Quarterly, 28, 370180. Fink, J. S., & Kensicki, L. J., (2002). An imperceptible difference: visual and textual constructions of femininity in Sports Illustrated and Sports Illustrated for Women. Mass Communication and Society, 5, 317339. Finkenberg, M. E., DiNucci, J. M., & McCune, S. L. (1998). Commitment to physical activity and anxiety about physique among college women. Perceptual and Motor Skills, 87, 13931394. Fitzsimmons Craft, E. E. (2011). Social psychological theories of disordered eating in college women: Review and integration. Clinical Psychology Review, 31, 12241237. Fitzsimmons Craft, E. E., Harney, M. B., Koehler, L. G., Danzi, L. E., Riddell, M. K., & Bardone Cone, A. M. (2012). E xplaining the relation between thin ideal internalization and body dissatisfaction among college women: The role of social comparison and body surveillance. Body Image, 9, 4349. Forbes, G. B., Jobe, R. L., & Revak, J. A. (2006). Relationships between dis satisfaction with specific body characteristics and the Sociocultural Attitudes Toward Appearance Questionnaire 3 and Objectified Body Consciousness Scale. Body Image, 3, 295300. Forbes, G. B., & Jung, J. (2008), Measures based on sociocultural theory and feminist theory as predictors of multidimensional measures of body dissatisfaction among Korean and U.S. college women. Journal of Social and Clinical Psychology, 27, 70103. Franzoi, S. L. (1994). Further evidence of the reliability and validity of the body esteem scale. Journal of Clinical Psychology, 50, 237239. 156

PAGE 157

Franzoi, S. L., & Herzog, M. E. (1986). The body esteem scale: A convergent and discriminant validity study. Journal of Personality Assessment, 50, 2431. Franzoi, S. L., & Herzog, M. E. (1987). Judging physical attractiveness: What body aspects do we use? Personality and Social Psychology Bulletin, 13, 1933. Franzoi, S. L., & Sheilds, S. A. (1984). The body esteem scale: Multidimensional structure and sex differences in a college populati on. Journal of Personality Assessment, 48, 173178. Fredrickson, B. L., & Roberts, T. A. (1997). Objectification theory: Toward understanding womens lived experiences and mental health risks. Psychology of Women Quarterly, 21, 173206. Gardner, R. M., F riedman, B. N., & Jackson, N. A. (1998). Methodological concerns when using silhouettes to measure body image. Perceptual and Motor Skills, 86, 387395. Garner, D. M. (1991). EDI 2 professional manual Lutz, FL: Psychological Assessment Resources, Inc. Garner, D. M. (1997). Survey says: Body image poll results. Psychology Today, 31. Retrieved from http://www.psychologytoday.com/articles/199702/survey says body imagepoll results Garner, D. M. (2004). EDI 3 professional manual Lutz, FL: Psychological Assessment Resources, Inc. Garner, D. M., Olmsted, M. P., & Polivy, J. (1983). Development and validation of a multidimensional eating disorder inventory for anorexia nervosa and bulimia. International Journal of Eating Disorders, 2, 1534. Gibson, A. K. (2007). The influence of sports magazines and thinideal images on the body image of Division IA female athletes. (Unpublished doctoral dissertation). Univers ity of Southern Mississippi. Hattiesburg, MS. Glauert, R., Rhodes, G., Byrne, S., Fink, B., & Grammer, K. (2009). Body dissatisfaction and the effects of perceptual exposure on body norms and ideals. International Journal on Eating Disorders, 42, 443452. Gordon, K. H, Castro, Y., Sitnikov, L., & Holm Denoma, J. M. (2010) Cultural body shape ideals and eating disorder symptoms among White, Latina, and Black college women. Cultural Diversity and Ethnic Minority Psychology, 16, 125143. Gorsuch, R. L. (1983). Factor analysis. (2nd ed.). Lawrence Erlbaum Associates: Hillsdale, NJ. Grabe, S., & Hyde, J. S. (2006). Ethnicity and body dissatisfaction among women in the United States: A meta analysis. Psychological Bulletin, 132, 622640. 157

PAGE 158

Grabe, S., Ward, L. M ., & Hyde, J. S. (2008). The role of the media in body image concerns among women: A meta analysis of experimental and correlational studies. Psychological Bulletin, 134, 460476. Grandi, S., Clementi, C., Guidi, J., Benassi, M., & Tossani, E. (2011). Per sonality characteristics and psychological distress associated with primary exercise dependence: An exploratory study. Psychology Research, 189, 270275. Gratereaux, A. (2012, August 10). Olympics 2012: Leryn Franco does not make finals, still hottest female Olympian. Fox News. Retrieved from http://latino.foxnews.com/latino/sports/2012/08/10/olympics london2012javelin thrower leryn franco does not makefinals still/ Haase, R. F., & Ellis, M. V. (1987). Multivariate analysis of variance. Journal of Counseling Psychology, 34, 404413. Hardin, M., & Greer, J. D. (2009). The influence of gender role socialization, media use, and sports participation on perceptions of gender appropriate sports. Journal of Sport Behavior, 32, 207226. Hardin, M., Lynn, S., & Walsdorf, K. (2005). Cha llenge and conformity on contested terrain: Images of women in four womens sport/fitness magazines. Sex Roles, 53, 105117. Harrison, K., & Fredrickson, B. L. (2003). Womens sports media, self objectification, and mental health in Black and White adol escent females. Journal of Communication, 53, 216232. Harrison, L. A., & Secarea, A. M. (2010). College students attitudes toward the sexualization of professional women athletes. Journal of Sport Behavior, 33, 403425. Hart, E. A., Leary, M. R., & Rej eski, W. J. (1989). The measurement of social physique anxiety. Journal of Sport & Exercise Psychology, 11, 94104. Hausenblas, H. A., & Carron, A. V. (2002, Winter). Assessing eating disorder symptoms in sport groups: A critique with recommendations for future research. International Sports Journal, 6574. Hausenblas, H. A., & Downs, D. S. (2001). Comparison of body image between athletes and nonathletes: A meta analytic review. Journal of Applied Sport Psychology, 13, 323339. Heinberg, L. J. (1996). T heories of body image disturbance: Perceptual, developmental, and sociocultural factors. In J. K. Thompson (Ed.), Body image, eating disorders, and obesity. Washington D.C.: American Psychological Association. 158

PAGE 159

Heinberg, L. J., Thompson, J. K., & Stormer, S. (1995). Development and validation of the sociocultural attitudes towards appearance questionnaire. International Journal of Eating Disorders, 17, 8189. Hewitt, P. L., Flett, G. L., & Ediger, E. (1995). Perfectionism traits and perfectionistic self pr esentation in eating disorder attitudes, characteristics, and symptoms. International Journal of Eating Disorders, 18, 317326. Hildebrandt, T., & Walker, D. C. (2006). Evidence that ideal and attractive figures represent different constructs: A replication and extension of Fingeret, Gleaves, and Pearson (2004). Body Image, 3, 173182. Hoag, M. M. (2012). Body image in female collegiate volleyball players: Are athletics detrimental or beneficial? (Doctoral dissertation). Retrieved from ProQuest Dissertations and Theses (UMI Number: 3502536). Holmqvist, K., & Frisen, A. (2010). Body dissatisfaction across cultures: Findings and research problems. European Eating Disorders Review, 18, 133146. Homan, K. (2010). Athletic ideal and thinideal internalizatio n as prospective predictors of body dissatisfaction, dieting, and compulsive exercise. Body Image, 7, 240 245. Homan, K., McHugh, E., Wells, D., Watson, C., & King, C. (2012). The effect of viewing ultra fit images on college womens body dissatisfaction. Body Image, 9, 50 56. Horn, J. L. (1965). A rationale and test for the number of factors in factor analysis. Psychometrika, 30, 179 185. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional crit eria versus new alternatives. Structural Equation Modeling, 61, 155. Huberty, C. J., & Olejnik, S. (2006). Applied MANOVA and Discriminant Analysis. (2nd ed.). John Wiley & Sons, Inc: Hoboken, NJ. Hurley, J. B., Palmer, R. L., & Stretch, D. (1990). The specificity of the Eating Disorder Inventory: A reappraisal. International Journal of Eating Disorders, 9, 419424. Ilaria, L. M. (2006). Body image, body objectification, and depression. (Doctoral dissertation). Retrieved from http://commons.pacificu.edu/spp/40. Impett, E. A., Henson, J. M., Breines, J. G., Schooler, D., & Tolman, D. L. (2011). Embodiment feels better: Girls body objectification and wellbeing across adolescence. Psychology of Women Quart erly, 35 46 58. 159

PAGE 160

Janelle, C. M., Hausenblas, H. A., Ellis, R., Coombes, S. A., & Duley, A. R. (2009). The time course of attentional allocation while women high and low in body dissatisfaction view self and model physiques. Psychology & Health, 24, 351366. Jones, A., & Greer, J. (2011). You dont look like an athlete: The effects of feminine appearance on audience perceptions of female athletes and womens sports. Journal of Sport Behavior, 34, 358377. Kachigan, S. K. (1986). Statistical analysis: An i nterdisciplinary introduction to univariate & multivariate methods. New York, NY: Radius Press. Kane, M. J. (1988). Media coverage of the female athlete before, during, and after Title IX: Sports Illustrated revisited. Journal of Sport Management, 2, 8799. Kashubeck West, S., Mintz L. B., & Saunders, K. J. (2001). Assessment of eating disorders in women. The Counseling Psychologist 29, 662694. Kato, K., Jevas, S., & Culpepper D. (2011). Body image disturbances in NCAA Division I and III female athletes. Sport Journal, 14. Retrieved from SportDiscus (Accession number: 71202326). Kelly, N. R., Mitchell, K. S., Gow, R.W., Trace, S. E., Lydecker, J. A., Bair, C. E, & Mazzeo, S. (2011). An evaluation of the reliability and construct validity of eating dis order measures in white and black women. Psychological Assessment Advance online publication. doi: 10.1037/a0026457. Kilbourne, J. (Producer), & Jhally, S. (Director). (2010). Killing us softly 4: Advertising images of women [Motion picture]. (Available from http://www.mediaed.org/cgibin/commerce.cgi?display=user1). Kim, K., Sagas, M., & Walker, N. A. (2010). Replacing athleticism with sexuality: Athlete models in Sports Illustrated swimsuit issues. International Journal of Sport Communication, 3, 148162. Klemchuk, H. P., Hutchinson, C. B., & Frank, R. I. (1990). Body dissatisfaction and eating related problems on the college campus: Usefulness of the eating disorder inventory with a nonclinical population. Journal of Counseling Psychology, 37, 297305. Kozee, H. B., & Tylka, T. L. (2006). A test of objectification theory with lesbian women. Psychology of Women Quarterly, 30, 348357. Kozee, H. B., Tylka, T. L., Augustus Horvath, C. L., & Denchik, A. (2007). Development and psychometric evaluation of the Interpersonal Sexual Objectification Scale. Psychology of Women Quarterly, 31, 176189. 160

PAGE 161

Karr, T. M., Davidson, D., Bryant, F. B., Balague, G., & Bohnert, A. M. (2013). Sport type and interpersonal and intrapersonal predictors of body dissatisfaction i n high school female sport participants. Body Image, 10, 201219. Krane, V., Choi, P. Y. Y., Baird, S. M., Aimar, C. M., & Kauer, K. J. (2004). Living the paradox: Female athletes negotiate femininity and muscularity. Sex Roles, 50 315329. Krane, V., S tiles Shipley, J. A., Waldron, J., & Michalenok, J. (2002). Relationships among body satisfaction, social physique anxiety, and eating behaviors in female athletes and exercisers. Journal of Sport Behavior, 24, 247264. Krcmar, M., Giles, S., & Helme, D. (2008) Understanding the process: How mediated and peer norms affect young womens body esteem. Communication Quarterly, 56, 111130. Larabee, A (2011). Negotiating the female athlete paradox: Examining gender identity and body image (Doctoral dissertat ion). Retrieved from ProQuest Dissertations and Theses (UMI Number: 3453929). Lenart, E. B., Goldberg, J. P., Bailey, S. M., Dallal, G. E., & Koff, E. (1995). Current and ideal physique choices in exercising and nonexercising college women from a pilot a thletic image scale. Perceptual and Motor Skills, 81, 831848. Limbert, C. (2004). The Eating Disorder Inventory: A test of the factor structure and internal consistency in a nonclinical sample. Health Care for Women International, 25, 165178. Low, K. G., Charanasomboon, S., Brown, C., Hiltunen, G., Long, K., & Reinhalter, K. (2003). Internalization of the thin ideal, weight and body image concerns. Social Behavior and Personality, 31, 8190. Mahalik, J. R., Morray, E. B., Coonerty Femiano, A., Ludlow, L. H., Slattery, S. M., & Smiler, A. Development of the Conformity to Feminine Norms Inventory. Sex Role, 52, 417435. Markland, D., & Oliver, E. J. (2008). The sociocultural attitudes towards appearance questionnaire 3: A confirmatory factor analysis. B ody Image, 5, 116 121. McKinley, N. M. (2004). Resisting body dissatisfaction: fat women who endorse fat acceptance. Body Image, 1, 213 219. McKinley, N. M. (2006). The developmental and cultural contexts of objectified body consciousness: A longitudinal analysis of two cohorts of women. Developmental Psychology, 42, 679687. McKinley, N. M., & Hyde, J. S. (1996). The Objectified Body Consciousness Scale: Development and validation. Psychology of Women Quarterly, 20, 181215. 161

PAGE 162

Mercurio, A., & Rima, B. (2011). Watching my weight: Self weighing, body surveillance, and body dissatisfaction. Sex Roles, 65, 4755. Miller, K., Sabo, D. F., Melnick, M. J., Farrell, M. P. & Barnes, G. M. (2000). The Womens Sports Foundation report: Health risks and the teen athlete Womens Sports Foundation. Retrieved from: http://www.womenssportsfoundation.org/home/research/articles and reports/mentaland physical health/health risks and the teen athlete Mokkink, L. B., Terwee, C. B., Stratford, P. W., Alonso, J., Patrick, D. L., Riphagen, I., de Vet, H. C. W. (2009). Evaluation of the methodological quality of systematic reviews of health status measurement instruments. Quality of Life Research, 18, 313333. Molasso, W. R. (2005). Ten tangible and practical tips to improve student participation in web surveys. Student Affairs Online, 6. Retrieved from http://studentaffairs.com/ejournal/Fall_2005/StudentParticipationinWebSurveys.html Moradi, B. (2010). Objectification Theory: Areas of promise and refinement The Counseling Psychologi st, 39, 153163. doi: 10.1177/0011000010384279 Moradi, B., Dirks, D., & Matteson, A. V. (2005). Roles of sexual objectification experiences and internalization of standards of beauty in eating disorder symptomatology: A test and extension of objectification theory. Journal of Counseling Psychology, 52, 420428. Moradi, B., & Huang, Y. (2008). Objectification Theory and psychology of women: A decade of advances and future directions. Psychology of Women Quarterly, 32, 377398. Muller, S. M., Gorrow, T. R., & Schneider, S. R. (2009). Enhancing appearance and sports performance: Are female collegiate athletes behaving more like males? Journal of American College Health, 57, 513520. Muscat, A. C., & Long, B. C. (2008). Critical comments about body shape and weight: Disordered eating of female athletes and sport participants. Journal of Applied Sport Psychology, 20, 124. Myers, J. L., Well, A. D., & Lorch, R. F. (2010). Research design and statistical analysis. (3rd ed.). New York, NY: Routledge. National Collegiate Athletic Association. (2012). Who we are. Retrieved from: http://www.ncaa.org/wps/wcm/connect/public/NCAA/About+the+NCAA/Who+We+Are / Noll, S. M., & Fredri ckson, B. L. (1998). A mediational model linking self objectification, body shame, and disordered eating. Psychology of Women Quarterly, 22, 632636. Nunnally, J. C. (1978). Psychometric Theory. (2nd ed.). New York, NY: McGraw Hill. 162

PAGE 163

Parsons, E. M., & Betz, N. E. (2001). The relationship of participation in sports and physical activity to body objectification, instrumentality, and locus of control among young women. Psychology of Women Quarterly, 25, 209222. Petrie, T. A., & Greenlea f, C. (2012). Body image and sport/athletics. In T. F. Cash (Ed.), Encyclopedia of body image and human appearance (pp. 160165). San Diego, CA : Academic Press. Pett, M. A., Lackey, N. R., & Sullivan, J. J. (2003). Making sense of factor analysis: The use of factor analysis for instrument development in health care research. California: Sage Publications, Inc. Piran, N., & Cormier, H. C. The social construction of women and disordered eating patterns. Journal of Counseling Psychology, 52, 549558. Podar, I., & Allik, J. (2009). A cross cultural comparison of the Eating Disorder Inventory. International Journal of Eating Disorders, 42, 346355. Raciti, M. C., & Norcross, J. C. The EAT and EDI: Screening, interrelationships, and psychometrics. International Journal of Eating Disorders, 6, 579586. Rathner, G., & Rumphold, G. (1994). Convergent validity of the Eating Disorder Inventory and the Anorexia Nervosa Inventory for Self Rating in the Austrian nonclinical population. International Journal of Eating Disorders, 16, 381393. Reel, J. J., Soohoo, S., Petrie, T. A., Greenleaf, C., & Carter, J. E. (2010). Slimming down for sport: Developing a weight pressures in sport measure for female athletes. Journal of Clinical Sport Psychology, 4, 99111. Reinking M. F., & Alexander, L. E. (2005). Prevalence of disorderedeating behaviors in undergraduate female collegiate athletes and nonathletes. Journal of Athletic Training, 40, 4751. Roberts, A., Cash, T. F., Feingold, A., & Johnson, B. T. (2006). Are BlackWhite differences in females body dissatisfaction decreasing? A meta analytic review. Journal of Consulting and Clinical Psychology, 74, 11211131. Robinson, K., & Ferraro, F. R. (2004). The relationship between types of female athletic participation and female body type. Journal of Psychology, 138, 115128. Rose, J. A. (2008). Self objectification and body shame as predictors of sport participation. (Masters thesis). Retrieved from ProQuest Dissertations and Theses (UMI Number: 1459710). 163

PAGE 164

Ruggiero, G. M., Bertelli, S., Boccalari, L., Centorame, F., Ditucci, A., La Mela, CSassaroli, S. (2008). The influence of stress on the relationship between cognitive variables and measures of eating disorders (in healthy female university students): A quasiexperim ental study. Eating and Weight Disorders, 13, 142148. Sax, L. J., Gilmartin, S. K., & Bryant, A. N. (2003). Assessing response rates and nonresponse bias in web and paper surveys. Research in Higher Education, 44, 409432. doi: 03610365/03/08000409/0. Schoemaker, C., Verbraak, M., Breteler, R., & Van der Staak, C. (1997) The discriminant validity of the Eating Disorder Inventory 2. British Journal of Clinical Psychology, 36, 627629. Schwarz, H. C., Aruguete, M. S., & Gold, E. S. (2005). Eating attit udes, body dissatisfaction, and perfectionism in female college athletes. North American Journal of Psychology, 7, 345352. Schwitzer, A., Hatfield, T., Jones, A. R., Duggan, M. H., Jurgens, J., & Winninger, A. (2008). Confirmation among college women: The Eating Disorders Not Otherwise Specified diagnostic profile. Journal of American College Health, 56, 607615. Schwitzer, A. M., Rodriguez, L. E., Thomas, C., & Salimi, L. (2001). The eating disorders NOS diagnostic profile among college women. American College Health, 49, 157. Retrieved online from Academic Search Premier. Shea, M. E, & Pritchard, M. E. (2007). Is self esteem the primary predictor of disordered eating? Personality and Individual Differences, 42, 15271537. Sherry, S. B., Hewitt, P. L., Besser, A., McGee, B. J., & Flett, G. L. (2004). Self oriented and socially prescribed perfectionism in the Eating Disorder Inventory Perfectionism subscale. International Journal of Eating Disorders, 35, 6979. Smolak, L., & Murnen, S. K. (2008). Drive for leanness: Assessment and relationship to gender, gender role and objectification. Body Image, 5, 251260. Smolak, L., Murnen, S. K. & Ruble, A. E. (2000). Female athletes and eating problems: A meta analysis. International Journal on Eating Disorders, 27, 371380. Steinfeldt, J. A., Zakrajsek, R., Carter, H., & Steinfeldt, M. C. (2011). Conformity to gender norms among female student athletes: Implications for body image. Psychology of Men & Masculinity, 4, 401416. Stellefson, M. L., Hanik, B. W., C haney, B. H., & Chaney, J. D. (2009). Factor retention in EFA: Strategies for health behavior researchers. American Journal of Health Behavior, 33, 587599. 164

PAGE 165

Stunkard, A., Sorensen, T., & Schulsinger, F. (1983). Use of the Danish Adoption Register for the study of obesity and thinness. Association for Research in Nervous & Mental Disease 60, 115120. Sveinbjornsdottir, S., & Thorsteinsson, E. B. (2008). Adolescent coping scales: A critical psychometric review. Scandinavian Journal of Psychology, 49, 533548. Swami, V. (2009). Body appreciation, media influence, and weight status predict consideration of cosmetic surgery among female undergraduates. Body Image, 6, 315 317. Swami, V., Jones, J., Einon, D., & Furnham, A. (2009). Mens preferences for womens profile waist to hip ratio, breast size, and ethnic group in Britain and South Africa. British Journal of Psychology, 100, 313325. Szymanski, D. M., & Henning, S. L. (2007). The role of self objectification in womens depression: A test of objectification theory. Sex Roles, 56, 4553. Szymanski, D. M., Moffitt, L. B., & Carr, E. R (2011). Sexual objectification of women: Advances to theory and research. The Counseling Psychologist, 39, 638. Thomas, C. D., & Freeman, R. J. (1990). The body esteem scal e: Construct validity of the female subscales. Journal of Personality Assessment, 54, 204212. Thompson, J. K., Heinberg, L. J., Altabe, M., & Tantleff Dunn, S. (1999). Exacting beauty: Theory, assessment and treatment in body image disturbance. Washington D.C.: American Psychological Association. Thompson, J. K., van den Berg, P., Roehrig, M., Guarda, A. S., & Heinberg, L. J. (2003). The Sociocultural Attitudes Towards Appearance Scale 3 (SATAQ 3): Development and Validation. International Journal on Eat ing Disorders, 35, 293304. Tiggeman, M., & Kuring, J. K. (2004). The role of body objectification in disordered eating and depressed mood. British Journal of Clinical Psychology, 43, 299311. Tissot, A. M., & Crowther, J. H. (2008). Self oriented and socially prescribed perfectionism: Risk factors within an integrative model for bulimic symptomatology. Journal of Social & Clinical Psychology, 27, 734755. Tsiantas, G., & King, R. M. (2001). Similarities in body image in sisters: The role of sociocultura l internalization and social comparison. Eating Disorders, 9, 141158. Turnbull, D. B. (1999). [Abstract] Weight control strategies and eating disorder tendencies among college women in four ethnic groups. (African American, Asian American, European Ameri can, Hispanic American). Dissertation Abstracts International: Section B: The Sciences and Engineering, 59(10B). 165

PAGE 166

Tury, F., Gulec, H., & Kohls, E. (2010). Assessment methods for eating disorders and body image disorders. Journal of Psychosomatic Research, 69, 601611. Tylka, T. L., & Hill, M. S. (2004). Objectification theory as it relates to disordered eating among college women. Sex R oles, 51, 719730. Tylka, T. L., & Sabik, N. J. (2010). Integrating social comparison theory and self esteem within obj ectification theory to predict womens disordered eating. Sex Roles, 63, 1831. Varnes J. R., Stellefson, M. L., Janelle, C. M., Dorman, S. M., Dodd, V., & Miller, M. D. ( in press ). A systematic review of studies comparing body image disturbance among fem ale college athletes and nonn athletes. Body Image Velicer, W. F. (1976). Determining the number of components from the matrix of partial correlations. Psychometrika 41, 321327. Walkey, F. H., & McCormick, I. A. (1985). A logical solution for a number s of factors problem. Muitivariate Behavioral Research, 20, 5767. Welch, G., Hall, A., & Walkey, F. (1988). The factor structure of the Eating Disorders Inventory. Journal of Clinical Psychology. 44, 5156. Wheaton, B. (1987). Assessment of fit in overi dentified models with latent variables. Sociological Methods & Research, 16, 118154. Wiggins, M. S., & Moode, F. M. (2000). Analysis of body esteem in female college athletes and nonathletes. Perceptual and Motor Skills, 90, 851854. Womens Sports Foundation. (2009). Womens sports facts. Retrieved from http://www.womenssportsfoundation.org/home/research/articles and reports/athletes/womens sports facts Zoski, K., & Jurs, S. (1996). An objective counterpart to the visual scree test for factor analysis: The standard error scree. Educational & Psychological Measurement, 56, 443451. 166

PAGE 167

BIOGRAPHICAL SKETCH Julia Rae Varnes was born (1980) and raised in Gainesville, Florida to Dr. Paul Ray Varnes (University of Florida, TRSM retired) and Dr. Jill Wilson Varnes (University of Florida, HEB retired). She attended the University of Florida for her undergraduate studies earning a Bachelor of Sci ence and Masters in Health Science Education in 2002 and 2003, respectively. Soon after graduation, Julia was accepted into the Department of Public and Community Health, School of Public Health at the University of Maryland, College Park. While studying a t UMCP, Julia worked as a graduate teaching assistant and as a research assistant in the Center for Health Behavior Research. Julia earned her Masters of Public Health from UMCP in 2007. After graduating with her MPH, Julia worked at the University of Flor ida Area Health Education Centers (AHEC) Program as their Education and Training Coordinator for Tobacco Programs. In 2009, Julia began part time studies toward a Ph.D. in Health and Human Performance at the University of Florida. In 2010, she became a ful l time student and worked as a graduate assistant within the Department of Health Education and Behavior, where she taught classes and pursued her dissertation research. She developed a line of research investigating body image among college females, and n arrowed that focus to societal objectification and collegiate female athletes. Julias doctoral dissertation includes several scientific papers based on her research, submitted for publication to scholarly journals. Julia will be granted a Doctor of Philos ophy in Health and Human Performance with an emphasis in Health Education and Behavior and minor in Research and Evaluation Methodology in August 2013. 167