|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
An Experimental Study of Pro-Dieting a nd Anti-Dieting Psychoeducational Messages: Effects on Immediate and Short-Term Psychol ogical Functioning and Weight Control Practices in College Women by Megan Roehrig A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Psychology College of Arts and Sciences University of South Florida Major Professor: J. Kevin Thompson, Ph.D. Michael Brannick, Ph.D. Ellis Gesten, Ph.D. Doug Rohrer, Ph.D. William Sacco, Ph.D. Date of Approval: May 23, 2006 Keywords: eating disorders, obesity prevention, risk factor, bulimia Copyright 2007, Megan Roehrig
Acknowledgements This dissertation is dedicated to my pa rents, Dennis and Nora Roehrig, who have always believed in me and encouraged me to pursue my dreams. Their unconditional love, support, dedication, and generosity ar e inspiring, and this achievement would not have been possible without them. I would al so like to thank my wonderful sister, Erin, and my dear friends, Erin Thompson and Sylvia Herbozo, whose constant encouragement, understanding, and sense of humor kept me grounded and balanced throughout this process. My dissertation would also not ha ve been possible without my major professor, J. Kevin Thompson, Ph.D., and I am grateful for his guidance, patience, and friendship. Thanks also to all of the members of the Body Image Research Group for being great colleagues and friends and a special thanks to Gu y Cafri for his very helpful statistical advice.
i Table of Contents List of Tables ............................................................................................................... ....iii List of Figures .............................................................................................................. ....iv Abstract ...................................................................................................................... .........v Chapter 1. Introduction....................................................................................................... 1 Overview .................................................................................................................1 Risk Factor Research .............................................................................................6 Etiological Models of Eati ng and Weight Disturbances .......................................10 Causal Risk Factor s for Eating Pathology.............................................................13 The Controversy Surrounding Dieting...................................................................16 Dieting Recommendations: To Diet or Not to Diet .............................................21 Pilot Study..............................................................................................................24 Current Study.........................................................................................................31 Chapter 2. Method............................................................................................................3 4 Participants ............................................................................................................34 Measures ............................................................................................................34 Demographic Information..........................................................................34 Body Dissatisfaction..................................................................................35 Thin-Ideal Internalization..........................................................................36 Sociocultural Pressures ........................................................................... 36 Drive for Thinness.....................................................................................37 Dieting .......................................................................................................38 Negative Affect..........................................................................................38 Bulimic Symptoms....................................................................................39 Eating Disorder Screening.........................................................................40 Healthy Eating...........................................................................................40 Exercise......................................................................................................41 Flu Prevention Intentions and Behaviors...................................................41 Message Rating Form................................................................................42 Attention Check.........................................................................................42 Distraction Task.........................................................................................43 Procedure...............................................................................................................43 Design and Analyses..............................................................................................45
ii Chapter 3. Results............................................................................................................ .50 Preliminary Analyses.............................................................................................50 Planned ANOVA and ANCOVA Analyses...........................................................54 Univariate ANCOVAs...............................................................................54 Repeated Measures ANOVAs...................................................................55 Mediation Analyses ............................................................................................58 Exploratory Analyses.............................................................................................63 Exploratory Moderator Analyses...............................................................63 Exploratory Weight Status Analyses.........................................................67 Chapter 4. Discussion.......................................................................................................73 References..................................................................................................................... .....83 Appendices..................................................................................................................... ..100 Appendix A: The Tripartite Model of Influence..................................................101 Appendix B: The Dual-Pathway Model of Bulimic Pathology ..........................102 Appendix C: Obesity Prevention Message.........................................................103 Appendix D: Eating Disorder Prevention Message............................................105 Appendix E: Flu Prevention Message.................................................................106 Appendix F: Visual Analogue Scales.................................................................108 Appendix G: Positive and Ne gative Affect Scale-Revised.................................109 Appendix H: Modified SATAQ-3......................................................................110 Appendix I: Message Rating Form.....................................................................111 Appendix J: Eating Di sorder Inventory-2...........................................................112 Appendix K: SATAQ-3......................................................................................113 Appendix L: Dutch Eating Behavior Questionnaire-Restraint Scale ................114 Appendix M: Eating Disorder Examination-Questionnaire...............................115 Appendix N: Modified Eating Diso rder Examination-Questionnaire................116 Appendix O: Eating Disorder Inventory-3 Referral Form..................................117 Appendix P: Multidimensional He alth Behavior Inventory...............................118 Appendix Q: Pilot Study Attention Check Items................................................119 Appendix R: Main Study Attention Check Items...............................................120 Appendix S: Distraction Task.............................................................................121 About the Author...................................................................................................End Page
iii List of Tables Table 1. Mean Scores for Pilot Study by Condition......................................................28 Table 2. Means and Standard Deviati ons for Message Rating Form Items by Condition.....................................................................................................52 Table 3. Correlations am ong Pre-Test Measures...........................................................53 Table 4. Means, Standard Deviations, F, p and partial 2 values for Planned ANOVAs.....................................................................................................57 Table 5. Correlations among Diet Me ssage Condition, Perceived Pressure, Intentions, and Behaviors............................................................................62 Table 6. Unstandardized Path Coeffici ents, Standard Errors, and Significance Tests for Mediation Analyses......................................................................63 Table 7. Correlations among Trait Intern alization, Diet Message, and Outcome Variables......................................................................................................65 Table 8. Standardized Beta Weights and R2 values for Moderator Analyses ...............66
iv List of Figures Figure 1. General mediation model.................................................................................59 Figure 2. General moderation model...............................................................................64 Figure 3. Covariate adjusted m eans for perceived pressure............................................69 Figure 4. Covariate adjusted mean s for state body dissatisfaction..................................70
v An Experimental Study of Pro-Dieting a nd Anti-Dieting Psychoeducational Messages: Effects on Immediate and Short-Term Psychological Functioning and Weight Control Practic es in College Women Megan Roehrig ABSTRACT While dieting is relatively normative in our soci ety, it is controversial within the fields of eating disorders and obesity. Dieting for we ight loss has been touted by the obesity prevention field as a solution to the growing obesity epidemic yet a body of research in the eating disorders field has also implicated it in the etiology a nd maintenance of eating pathology. Thus, a divergence in approaches toward dieting has emerged, with both prodieting and anti-dieting messages being r ecommended. Little is known, however, about the impact of these two types of message s on immediate and short-term psychological functioning and weight control intentions and behaviors. The current study sought to explore this gap in the extant literatur e by conducting an experimental study that evaluated the two messages. Undergraduate women (N=139) were randomly assigned to either a pro-dieting, anti-dieting, or no-dieting (control) message condition. Psychological functioning and weight control variables were assesse d at baseline, posttest, and a two-week follow-up. Results indica ted that the pro-die ting message resulted in significantly greater post-tes t perceived pressure to lose weight, dieting intentions, and thin-ideal internalization intentions while the anti-dieting message yielded significantly lower post-test bulimic intentions. Healt hy eating behavior signi ficantly increased from
vi baseline to follow-up in the pro-dieting c ondition while there were no changes in the other two conditions. Post-test perceived pressure was found to fully mediate the relationship between diet message and post-test dieting, bulimic, thin-ideal internalization, and healthy eating intentions as well as follow-up healthy eating behavior. Trait thin-ideal in ternalization levels moderated the relationship between diet message and post-test perceived pressure a nd thin-ideal interna lization intentions. Exploratory analyses revealed that overweight participants in the pro-dieting condition increased significantly from pre to post-te st on state body dissatisfaction and had the highest level of post-test perceived pressure compared to all other groups. Nonoverweight participants in the pro-dieting condition also had si gnificantly greater posttest perceived pressure to lose weight than both weight status gr oups in the other two conditions. Findings are discu ssed in the context of the pr evention goals of the obesity and eating disorders fields. Limitations of th e study and directions for future research are offered.
1 Chapter 1 Introduction Overview Disturbances of eating and weight are a considerable problem in American society and can range from symptoms of ex tremely restrictive dieting, exercising, and binging and purging behaviors to excessive overeating and a complete lack of physical activity (Thompson, 2004a). At one extreme are the eating diso rders of anorexia nervosa and bulimia nervosa. Anorexia nervosa is ch aracterized by weight th at is below 85% of what would be expected given height, an inte nse fear of fatness, a distorted body image, and amenorrhea (American Psychiatric Association, 2000). Bulimia nervosa is characterized by recurrent episodes of binge eating, compensatory behaviors such as purging, excessive exercise, or laxative use to prevent weight gain from binges, and a distorted body image (American Ps ychiatric Association, 2000). Eating disorders are a particular prob lem for adolescent girls and young adult women (Streigel-Moore & Sm olak, 2001; Thompson & Smolak, 2001). They are one of the most prevalent psychiatric disorder s experienced by young females with a 0.5-1% prevalence rate for anorexia nervosa and a 1-3% prevalence rate for bulimia nervosa (American Psychiatric Association, 2000; Thom pson, Roehrig, & Kinder, in press). An additional 10-13% of adolescent and college fe males engage in subclinical, disordered eating practices (Irving & Neumark-Sztainer 2002). Eating disord ered symptoms are
2 associated with a number of negative physic al and psychological consequences, including a chronic course (Fairburn, Cooper, Doll, Norman, & OConner, 2000), psychiatric comorbidity (Fichter & Quadflieg, 1999; Sull ivan, Bulik, Carter, & Joyce, 1996), high rates of mortality and morbidity (Reijonen, Pratt, Patel & Greydanus, 2003), and selfinjury (Favaro & Santonastaso, 1996). At another extreme is obesity, which is characterized by excessive weight for age and height and defined as a body mass index (BMI) over 30 (Devlin, Yanovski, & Wilson, 2000; Flegal, Carroll, Kuczma rski, & Johnson, 1998; World Health Organization, 1998). BMI is standardized by ag e and height and is computed as weight (in kilograms) divided by he ight (in meters) squared (Fie ld, Barnoya, & Colditz, 2002). Similarly, overweight is de fined as a BMI between 25 and 29.9 (Devlin et al., 2000; Flegal et al., 1998; World Health Orga nization, 1998). Excess body fat leading to overweight and obesity results from an im balance of caloric intake and physical expenditure with greater calor ies consumed than used. While most overweight or obese individuals do not suffer from a diagnosable eating disorder, a substantial minority of them do meet criteria for bi nge eating disorder with estimates ranging from 10-33% (Grilo, 2002; Grissett & Fitzgibbon, 1996; Yanovski, Nelson Dubbet, & Spitzer, 1993). Binge eating disorder is characterized by th e presence of out of control binge eating without subsequent compensatory behavior s such as exercise, purging, or fasting (American Psychiatric Association, 2000). Additionally, overweight and obese adults and adolescents are more likely to engage in sub-clinical levels of binge eating (Marcus, 1993) and unhealthy weight cont rol practices (i.e., diet p ills, laxatives, dieretics;
3 Neumark-Sztainer, Story, Faulkner, Beuhri ng, & Resnick, 1999) than those who are not overweight. Rates of overweight and obesity are in creasing at alarming rates, and public health officials have noted these increases are at epidemic proportions (Henderson & Brownell, 2004; World Health Organization, 19 98). The rate of obesity has doubled in Americans since the 1980s, and currently one-thi rd of American adults are obese (Flegal et al., 1998). Data from the 1999-2000 Na tional Health and Nutrition Examination Survey (NHANES) conducted by the Centers fo r Disease Control indicate that 64% of Americans over twenty are overweight and th at 33% of adult women and 28% of adult men are obese, with minority women experiencing even higher rates than Caucasian women ( Flegal et al., 1998). Obesity is associ ated with a number of negative health consequences including heart disease, diabetes, stroke, hype rtension, osteoarthritis, sleep apnea and certain types of cancer (Sarwer, Foster, & Wadden, 2004 ) as well as psychological difficulties, including body dissa tisfaction, low self-esteem, and weightrelated stigmatization (Neuma rk-Sztainer & Haines, 2004; Schwartz & Brownell, 2002) While obesity and disordered eating have both significantly increased in prevalence over the last twenty years, rese archers have historically regarded these problems as orthogonal (Brownell & Rodin, 1994; Irving & Neumark-Sztainer, 2002). Little overlap has occurred between the eating disorder and obesity fields, and distinct etiological theories and methods for the tr eatment and prevention of these difficulties have been implemented (Irving & NeumarkSztainer, 2002; Smolak & Striegel-Moore, 2004). Theoretical as well as practical reas ons have led some researchers to call for greater integration between the two fields, pa rticularly in the domains of etiology and
4 prevention (Battle & Brownell, 1996; Irv ing & Neumark-Sztainer, 2002; Smolak & Striegel-Moore, 2004). Irving and Neumark-Szta iner (2002) note that there is substantial overlap in etiological factors related to eati ng disorders and obesity, and they suggest that disordered eating practices and obesity should no t be viewed as conceptually distinct. At this point, however, the mechanisms involve d are unclear, and future research must systematically investigate shared etiologica l factors and prevention strategies (Smolak & Striegel-Moore, 2004). Accordingly, the current study seeks to br idge the fields of eating disorders and obesity by systematically examining the psychoeducational prevention messages espoused by each group. While both of these messages have the goal of increasing health-related behaviors and decreasing dysf unctional eating patterns, they take very different stances on dieting and weight loss (Irving & Neumark-Sztainer, 2002). In fact, the recommendations of these two messages app ear to be in direct conflict with one another. The obesity prevention message espous es a pro-dieting appr oach to weight loss and maintenance while the eating disorder pr evention message advocat es an anti-dieting approach. The eating disorder preventi on approach was developed based on the consistent finding that perceived pressure to be thin is a risk factor for the development of eating pathology. It promotes acceptance of all body sizes and shapes and seeks to reduce sociocultural pressures to be thin (S tice, 2002; Stice & Hoffman, 2004). Genetics are often discussed as a signifi cant factor in body weight and shape, and participants are encouraged to avoid dieting and to eat a nd exercise in moderation (National Eating Disorder Association, 2004; Stice & Shaw, 2004). In cont rast, the obesity prevention message stemmed from a medical model, wh ich views dieting and weight loss as a
5 solution to the serious health consequences associated with overweight and obesity (Brownell & Rodin, 1994 ). It stresses restricting caloric intake and increasing physical activity to control and lose we ight and tends to de-emphasize s the role of genetics in overweight and obesity (Brownell & Rodi n, 1994; Centers for Disease Control and Prevention, 2004; Irving & Neumark-Sztainer, 2002). Previous research has found that exposu re to psychoeducational messages focused on reducing perceived sociocultural pressure s to be thin has produced decreases in established risk factors for eating disorders such as body dissatisfa ction and thin-ideal internalization as well as eating pathology in some at-risk samples (e.g., Stice & Ragan, 2003; Stice & Shaw, 2004). However, no resear ch was located that examined the effects of the pro-dieting, obesity pr evention message on psychologica l functioning or eating and weight control intentions and behaviors. While prior research suggests that extensive obesity education programs targe ting weight loss in self-selec ted individuals may lead to increases in healthy eating habits such as fruit and vegetable consumption and decreased fat intake over several months (Jaso n, Greiner, Naylor, Johnson, & Van Egeren 1991; Jeffery & French, 1999; Miles, Rapoport, Wa rdle, Afuape, and Duman, 2001), no studies were found that examined the acute effects of either the obesity prevention message or the eating disorder prevention message on heal thy eating and weight control intentions and behaviors. Given the recent explosion in media coverage on dieting and weight concerns, it appears timely to directly examine the psychological and behavioral effects of these messages. Therefore, the current study intends to experimentally manipulate the pro-dieting, obesity preventi on and anti-dieting, eating diso rder preventio n messages and examine the immediate and short-term ef fects on psychological functioning (i.e.,
6 perceived pressure to be th in, body satisfaction, negative aff ect, drive for thinness) as well as intentions and behavior s related to healthy and unheal thy weight control practices (i.e., dieting, bulimic symptoms, healthy eating) in undergraduate women, who are targets of both of these psychoeducational messages. The first section of this paper will in troduce the concepts and nomenclature of risk factor research. Etiological theories a nd risk factor research on eating and weight disturbances will then be discussed with an emphasis on the role of the sociocultural environment. The next section will di scuss the controversy surrounding dieting, specifically its relationship to eating pathology and impact on treatment and prevention recommendations. Lastly, results from a pilot st udy that examined immediate perceptions of diet-related psychoeducational messages will be discussed, and goals and hypotheses for the current study will then be offered. Risk Factor Research Discrepancies among the findings of e xperimental, prospective, and crosssectional studies can occur and can have a si gnificant impact on theo ries of etiology as well as recommendations for treatment and prevention. To address this problem, researchers have called for a standardized nomenclature of risk factor terminology and have outlined strategies for risk factor rese arch. The following sect ion will discuss the nomenclature of risk factor research as we ll as research methodologies that have been recommended to standardize risk factor research. Kraemer, Kazdin, Offord, & Kessler (1997) argue that it is es sential that risk factor terminology be standardized in orde r to promote methodologically sound research.
7 Rigorous risk factor research that uses a common language among investigators has several important implications. For example, it allows researchers to be able to differentiate between variables that are true risk factors and those that are not, which is important in the development and refinement of etiological models of a disorder and are also essential to inform the development of effective prevention and treatment programs (Weissberg, Kumpfer, & Seligman, 2003). Accord ingly, Kraemer et al. (1997) define (1) risk as the probability of an outcome occurring, (2) a correlate as a factor associated with the outcome of interest, (3) a risk factor as a measurable characteristic that temporally precedes the outcome of interest, (4) a variable risk factor as a risk factor that can be changed, and (5) a causal risk factor as a variable risk factor that when manipulated produces changes in the outcome of interest Kraemer et al. (1997) argue that to effectively measure risk, the outcome of intere st must be defined cl early and all variables of interest must be measured usi ng psychometrically sound instruments. Research methodology is critical in dis tinguishing among these various types of risk, and Kraemer et al. (1997) have outlin ed the process for establishing risk-factor status, which includes sequential stages beginning with a correlate and ending with a causal risk factor Different research designs are needed during each stage of the riskfactor research process, and each design has it s own role in the process of establishing the risk factor status of a variable. The following section will describe each stage of the risk factor research process and illustra te the importance of each phase. Cross-sectional designs should be utilized in the first stag e of risk factor research to establish correlate status (Kraemer et al., 1997). B ecause temporal precedence is the critical characteristic of a risk factor, cross-sectional designs cannot be used to establish
8 risk factor status but are impor tant as they establish a rela tionship between two variables in a relatively cheap and easy study. After co rrelate status is confirmed, the second phase of risk factor research in volves determining whether a fa ctor precedes the outcome of interest (Kraemer et al., 1997). A prospective design must be utilized during this phase to examine whether the correlate variable is pr esent before the development of the outcome of interest, and only longitudinal designs can definitively establish temporal precedence (Kazdin, 2003). Because prospective studies are costly and time consuming, it is important that this design is not utilized unt il correlate status has been attained through cross-sectional design. If temporal preceden ce is established in the longitudinal study, then the variable can be deemed a risk fact or for the outcome of interest. If temporal precedence is not established after having been studied prospectively, then Kraemer and colleagues (1997) suggest the terms concomitant or consequence be used to describe the relationship of the correlate variable to the outcome of interest. According to Kraemer et al. (1997), an im portant distinction mu st be made in all empirically established risk factors. Th ey propose that risk factors should be characterized as one of two types: variable or fixed marker A variable risk factor is one that can be changed within an individual either spontaneously (i.e., age) or through intervention (i.e., administrati on of a drug). A fixed marker, on the other hand, is a risk factor that cannot change within an individual such as race or gender. This distinction is important for informing future risk factor research as well as for the development of prevention and treatment programs. The last phase of the research process fo r establishing risk factor status involves using an experimental design to manipulate a va riable risk factor. If the experimental
9 manipulation of the variable risk factor results in a change in the outcome of interest, then the variable should be called a causal risk factor rather than using the term the cause (Kraemer et al., 1997). This di stinction is critical, as it allows for the likelihood of multiple pathways to an outcome of interest. Additionally, it is impor tant to note that the identification of causal risk factors does not suggest knowledge of mechanisms by which causal risk factors exert their influence (Kraemer et al., 1997). Future research must be conducted to ascertain these processes. If the experimental manipulation, howeve r, does not result in a change in the outcome of interest, the term causal risk f actor cannot be used. The variable may be considered a variable marker or may have in fact been a proxy risk factor. A proxy risk factor is defined as a variable that is strongly co rrelated with a true risk factor and thus appears to precede the outcome of interest; howev er, if a proxy risk factor is manipulated, it will not result in changes in the outcome of interest whereas ma nipulation of a causal risk factor will lead to subsequent reductions in the outcome of interest (Kraemer et al., 2001). In sum, each type of research design has a role in the process of establishing risk factor status. While only expe rimental designs can determine whether a risk factor is a causal risk factor according to the criter ia proposed by Kraemer et al. (1997), crosssectional, case-control, and longitudinal de signs must all be c onducted earlier in the process of establishing risk fact or status. As Kraemer et al. (1997) note, risk factors must be characterized into different types of risk factors (i.e., va riable, fixed, causal) in order to inform the development of effective prev ention and treatment programs. Kraemer and
10 colleagues (1997) have called for researchers to continue to se arch for causal risk factors to move towards a greater understandi ng of the etiology of a disorder. Etiological Models of Ea ting and Weight Disturbances Researchers postulate that eating and we ight disturbances develop through a complex interaction among gene tic, cultural, social, be havioral, and psychological mechanisms (Brownell & Wadden, 1992; Bulik, 2004; Cope, Fernandez, & Allison, 2004; Stein, OByrne, Suminski, & Haddock, 2000; Thompson et al., in press). Behavioral genetic studies have verified the substantial ro le that genes play in the development of eating and weight disorders, and researchers are beginning to make advances in knowledge of th e interactions that occur among these genes (Bulik, 2004; Cope et al., 2004). Yet despit e this progress, geneticists caution that the expression of genes is highly dependent upon the environmen t (Cope et al., 2004). This fact coupled with the recent explosion of disturbed eating practices and obesity over the last twenty years has led many researchers to focus on the sociocultural environment, and its role in the etiology of eating and weight disturba nces (Anderson-Fye & Becker, 2004; Battle & Brownell, 1996; Irving & Neumark-Sztainer 2002; Stice, 2001; Thompson, Heinberg, Altabe, & Tantleff-Dunn, 1999). Obesity has been conceptualized as a co mplex condition that has a heterogeneous etiology (Brownell & Wadden, 1992; Devlin et al., 2000). It arises when an individual has a positive energy balance, and therefore consumes mo re energy than is expended (Stein et al., 2000). While this simple equa tion inevitably results in weight gain, the factors that lead to this energy imbalance are multifaceted and complex. For example, a
11 combination of behavioral and biologi cal variables including physical inactivity, excessive caloric intake, high fat diets, lo w resting metabolic rate, low rates of fat oxidation, insulin sensitivity, and high fat cell numbers can contribute to the development and maintenance of obesity (Brownell & Wadden, 1992; Stein et al., 2000; Tataranni & Ravussin, 2002). Despite the mounting evidence that up to 70% of the variance in BMI can be accounted for by genetic variations, Tatara nni and Ravussin (2002) acknowledge that research appears to support a paradigm shift for geneticists, suggesting that obesity is a condition that results from normal physiologi cal variability within a pathoenvironment (p.61). In fact, proponents of an environmenta l explanation for the obesity epidemic have coined the term toxic environment to de scribe modern American society, which is characterized by the widespread availability and marketing of cheap, quick, and tasty energy dense foods that are high in fat and s ugar and low in nutriti onal value, supersize portions, and an increasingly se dentary lifestyle coupled with a glorification of thinness and stigmatization of fatness (Battle & Brownell, 1996; Henderson & Brownell, 2004; Irving & Neumark-Sztainer, 2002; Wadden, Brow nell, & Foster, 2002). As Battle and Brownell (1996) note, it is diffi cult to envision an environmen t more effective than ours for producing nearly universal body dissati sfaction, preoccupation with eating and weight, and clinical cases of eating disorders and obesity (p. 761). In addition to the effect the environmen t has on obesity, researchers have also argued that the sociocultural environment pl ays a significant role in the etiology and maintenance of disturbed eating and weight control practices (A nderson-Fye & Becker, 2004; Heinberg, 1996; Thompson et al., 1999). Theorists have posited the mechanisms
12 by which sociocultural forces foster the develo pment of disturbed eating practices such as excessive restricting, binging, and purging. Tw o competing, yet similar, models of eating disorder symptomatology in females have been proposed, the Tripartite Model (Thompson et al., 1999; van den Berg, Thom pson, Obremski-Brandon, & Coovert, 2002; See Appendix A) and the Dual-Pathway mode l (Stice, Nemeroff, & Shaw, 1996; Stice, 2001; See Appendix B). Both models hypothesize that sociocultural pressures to be thin and internalization of the thin-ideal, which is the extent to which one buys into societal standards of appearance and weight both cognitively and behaviorally (Thompson & Stice, 2001 p.181), contribute to eating disturbances by fostering the development of body dissatisfaction. Body dissatisfaction, in turn, is hypothesized to foster dieting, eating disordered symptoms, and negative aff ect as the ideal is almost impossible to attain for the average female (Heinberg, 1996; Stice, 2001; Thompson et al., 1999). Cross-sectional, structur al equation modeling studies on undergraduate females have found broad support for both the Triparti te Model (van den Be rg et al., 2002) and the Dual-Pathway Model (Stice et al., 1996) Stice (2001) also found support for the Dual-Pathway Model in a twenty-month pros pective study of adol escent girls using random regression growth curve models. Sp ecifically, Stice (2001) found evidence that initial levels of per ceived pressure to be thin and thin-ideal internalization predicted increases in body dissatisfaction over time. Thin -ideal internalization and pressure to be thin were also found to prospectively predict growth in dieting even in the absence of body dissatisfaction, leading Stice (2001) to conc lude that sociocultural pressure to be thin as well as thin-ideal internalization appear to have direct and i ndirect influences on promoting dieting. Additionally, results suppo rted the hypothesis that initial levels of
13 body dissatisfaction predicted subs equent growth in dieting a nd negative affect. Initial levels of negative affect and dieting also prospectively predicted growth in bulimic symptoms, and the relationship between body di ssatisfaction and bulimic symptoms was completely mediated by dieting and negative affect. Initial leve l of dieting, however, only led to a marginally signifi cant growth in negative affect Collectively, findings from structural equation modeling st udies support the theoretical as sertions that sociocultural pressure to be thin, thin-id eal internalization, bod y dissatisfaction, dieting, and negative affect promote the onset of bulimic symptomatology. Causal Risk Factors for Eating Pathology Stice (2002) conducted a meta-analysis on risk and maintenanc e factors for eating pathology. In order to ensure that only true risk factor s were included in the metaanalysis, Stice limited the studies reviewed to longitudinal and experimental studies. It is important to note that all of the studies found in his literature review examined bulimic symptoms, binge eating, or eating disorder co mposites; none of them focused solely on anorexic symptoms. Therefore, Stices ( 2002) findings may only be generalizable to bulimic or binge eating symptoms. Several possible risk factors were exam ined in Stices (2002) meta-analysis, including body mass, perceived sociocultural pressure to be thin, modeling of body image or eating disturbances by parents and/or peers, thin -ideal internalization, body dissatisfaction, dieting, negative affect, perfectionism, early menarche, and impulsivity. This meta-analysis supported the conclusion that several of these variables met Kraemer et al.s (1997) criteria as established risk factors for eat ing pathology. Specifically, Stice
14 (2002) found that perceived socioc ultural pressure to be thin, thin-ideal internalization, body dissatisfaction, negative affect, and perfec tionism are all risk factors for eating pathology; however, he concluded that only pe rceived sociocultural pressure to be thin and thin-ideal internalization meet Kraemer et al.s (1997) criteria fo r causal risk factors for eating pathology. Stices (2002) finding that thin -ideal internalization meets criteria for a causal risk factor corroborates the conclusions of Th ompson and Stice (2001). Thompson and Stice (2001) outline the phases of research on thin-i deal internalization a nd report that early cross-sectional research established it as a correlate of eating dist urbances. They then reviewed the longitudinal res earch on thin-ideal interna lization and conc luded that it prospectively predicts eating pathology, which esta blishes it as a risk factor based on Kraemer et al.s (1997) cr iteria. Lastly, Thompson and Stice (2001) reviewed experimental prevention studies that manipulated thin-ideal internalization over the course of three hour-long sessi ons (Stice, Mazotti, Weibel, & Agras, 2000; Stice, Chase, Stormer, and Appel, 2001; Stice, Trost, a nd Chase, 2003). Because these experimental manipulations of thin-ideal internalization have led to decreases in body dissatisfaction and eating pathology, thin-ideal in ternalization meets Kraemer et al.s (1997) criteria for a causal risk factor. Stices (2002) conclusion that perceived sociocultural pressure to be thin meets criteria for a causal risk factor for eating pathology, however, appears to be somewhat premature. While there is ample evidence to support the conclusion that sociocultural pressure to be thin prospectively predic ts body dissatisfaction and eating disturbances (e.g., Cattarin & Thompson, 1994; Stice, 2001; Stice & Agras, 1998), experimental
15 research on sociocultural pressu re to be thin is limited. In fact, Stices (2002) conclusion that it is a causal risk factor for eating pathology is based prim arily on experimental studies of brief exposure to thin-ideal media images, wh ich assessed body dissatisfaction and negative affect preand postexposu re to the images (see Groesz, Levine, & Murnen, 2002 for a meta-analyti c review). While brief expo sure to thin-ideal media likely exerts some degree of sociocultural pr essure to be thin, none of these studies directly assessed the extent to which participants perceive d pressure from the images, limiting the conclusions that can be drawn from these studies about the pressure construct. Additionally, thin-ideal media images do not appear to exert direct pressure to lose weight and/or maintain a thin body but rather portray an indirect, ubiquitous message that thin is beautiful and a necessary co mponent for a happy, exciting, and fulfilling life. Only one experimental study was located th at directly manipulated sociocultural pressure to be thin. Stice, Maxfield, and We lls (2003) examined the effects of fat talk on undergraduate womens body satisfaction and ne gative affect. Par ticipants in this study engaged in a 3-5 minute scripted conversation with on e of the two study confederates, who were young adult women that both objectively met societal standards of thinness and attractiveness and had worked in the fashion industry. Participants were randomly assigned either to a condition in which the confederate discussed her dissatisfaction with her weight and the extreme ex ercise and diet strate gies she used or to a neutral conversation conditi on in which the confederate discussed classes she was currently taking and her plans for the w eekend. Body dissatisfaction was found to significantly increase from pr eto post-test in the expe rimental condition; however, no significant differences in negative affect we re found between condi tions. Measures of
16 dieting and bulimic symptoms were not obtai ned, so although the findings suggest that increased sociocultural pressure to be thin results in increased body dissatisfaction, which is a strong predictor of eati ng pathology (Thompson et al., 1999 ) it is unclear how social pressure from fat talk a ffects eating behaviors. As Stice et al. (2003) is the first stud y to experimentally manipulate social pressure to be thin, much more research is need ed to elucidate the role of pressure to lose weight and/or maintain a thin body in eating pathology and the associated risk factors of body dissatisfaction, thin-ideal in ternalization, negative affect, and dieting. In addition to more research on fat talk and other forms of pressure from peers, further experimental research on sociocultural pressu re to be thin from other influential sources, including the media, parents, significant others, and health professionals, is warranted. The Controversy Surrounding Dieting In a seminal paper, Polivy and Herman (1985) outlined the tenets of Restraint Theory and proposed that dieting causes binge eating. Restraint Th eory postulates that restrained eaters, a term Polivy and Herman (1985) used interchangeably with dieters (Lowe, 1993), rely heavily on cognitive factors ra ther than physiological cues to maintain control over their eating behavior. Laborator y research has consis tently shown that restrained eaters can maintain their strict dietary guidelines and avoid overeating when demands of the study are low and allow them to follow their diet; however, when restrained eaters must consume a high-calorie pre-load (i.e., a milkshake) prior to a laboratory taste test, they ove reat or even binge. Polivy and Herman (1985) call this phenomenon counter-regulation and assert that these episodes of overeating appear to be
17 due to a violation of the strict dietary rule s of the restrained ea ter---the abstinence violation effect (Marlatt & Gordon, 1985). Counter-regula tion has also been found to occur in laboratory-induced negative aff ect and alcohol consumption (Lowe, 1993; Polivy & Herman, 1985). Non-restrained eate rs, on the other hand, show a more normal eating pattern under laboratory conditions. They eat more in the taste test if there is no high-calorie pre-load but less when there is one. Similarly, non-restrained eaters have been shown to eat less in di stressful situations and follo wing alcohol consumption than restrained eaters (P olivy & Herman, 1985). Following from Restraint Theory, the cognitive-behavioral model of bulimia nervosa proposed that strict dieting is a ke y etiological factor in the development and maintenance of bulimic pathology (Fairbur n, Marcus, & Wilson, 1993). According to Fairburn et al. (1993), extreme dieting behavi ors often develop in individuals with low self-esteem who overvalue weight and shape in an attempt to enhance their self-worth. This severe dieting eventually leads to a vi olation of the strict dietary guidelines and results in a binge episode. Extreme weight control methods such as vomiting or laxative use may then be used to compensate for th e excess calories consum ed during the binge. This binge-purge cycle can become self-perpetu ating and spiral out of control into a fullblown eating disorder (Fairburn et al, 1993). Based on these models, theorists have ge nerally agreed that dieting is a key etiological factor in eating pathology (Hsu, 1996). Empiri cal studies have provided some support for restraint theory and the cognitiv e-behavioral model of bulimia. Several retrospective studies of eating disordered pa tients have shown that dieting frequently precedes binge eating and the subsequent deve lopment of the eating disorder (Brewerton,
18 Dansky, Kilpatrick, & ONeil, 2000; Bulik, Sull ivan, Carter, & Joyce, 1997; Mitchell, Hatsukami, Eckert, & Pyle, 1985); however, retr ospective studies also suggests that binge eating precedes significant die ting behaviors in a substant ial minority of individuals (Brewerton et al., 2000; Bulik et al., 1997; Mussell, Mitchell Weller, & Raymond, 1995) Several longitudinal studies us ing self-reported dietary restra int measures have found that dieting prospectively predicts bulimic symp tomatology (Killen et al., 1994, 1996; Stice, 2001; Stice & Agras, 1998). A recent study, ho wever, did not find dietary restraint to prospectively predict growth in bulimic sy mptomatology when simultaneously compared in a logistic regression e quation with body dissatisfacti on (Johnson & Wardle, 2005). Body dissatisfaction did remain a significant prospective predictor of bulimic symptoms when dieting was controlled. Results from experimental studies of behavior al weight loss programs have also conflicted with the assertion that dieting is a key etiological factor in the development of bulimic symptoms. Studies of overweight and obese individuals placed on low-calorie diets in controlled trials have not shown s ubsequent increases in binge eating (Porzelius, Houston, Smith, Arfkin, & Fisher, 1995; Wadde n, Foster, & Letizia, 199 4). Furthermore, studies on obese individuals with binge eat ing disorder found significant decreases in binging over the course of university-based, be havioral weight loss treatments (Marcus, Wing, & Fairburn, 1995; Porzelius et al., 1995). Presnell and Sti ce (2003) replicated these findings in a non-obese sample of young adult women who were randomly assigned to a six-week, low calorie, beha vioral weight loss treatment or a waitlist control group. Stice, Presnell, Groesz, and Shaw (2005) examined the effects of a three-session weight management diet as opposed to a weight loss diet on bulimic pathology in
19 adolescent girls with elevated body image c oncerns. The intervention did not encourage calorie counting or a reduction in caloric intake as tradit ional behavioral weight loss programs do. Rather, the importance of a h ealthy body weight and balanced diet was stressed, and strategies for making these cha nges were discussed. Results confirmed that weight was indeed maintained in the inte rvention group over a one-y ear period while the measurement-only control group ga ined weight. Consistent with Stice et al.s (2005) hypothesis, significant decreases in bulimic symptomatology were observed at the oneyear follow-up in the intervention condition relative to the measurement-only control group. Collectively, findings from randomized contro lled trials of behavioral weight loss and weight maintenance treatments provide ev idence that contradicts the primary tenet of Restraint Theory (Polivy & Herman, 1985) tha t dieting promotes the onset of bulimic symptomatology. Not only was there no growth in bulimic pathology, but it was actually reduced over the course of these diet trials. Because of these findings from experimental research, Stice (2002) concluded in his meta-analysis that dieting is not a risk factor for eating pathology but rather attenuates overeating tendencies (p.836). The literature on dieting is complicat ed by measurement issues, which may contribute to these conflicting findings. Dieting and restra ined eating are often used interchangeably; however, research suggests th at these are distinct constructs (Lowe, 1993). Dieting has been defined as purposeful restriction of cal oric intake that results in a negative energy balance with the intention of weight loss or weight maintenance (Stice et al., 2005; Wadden et al., 2002). Much of the research that has been conducted on dieting has used one of three measures of re strained eating: the Restraint Scale (Polivy,
20 Herman, & Warsh, 1978), Three-Factor Eati ng Questionnaire-Cognitiv e Restraint Scale (Stunkard & Messick, 1985), and the Dutch Eating Behavior Questionnaire-Restrained Eating Scale (Van Strien, Frijter s, Bergers, & Defares, 1986). Yet, these scales have not been found to assess actual dieting behavi or as defined by a negative energy balance (Lowe, 1993; Stice, Fisher, & Lowe, 2004; S tice, Presnell, Lowe, & Burton, 2006). The restraint scales do appear, howev er, to measure an important albeit unclear construct in the development of bulimic pathology as they have consistently predicted growth in bulimic symptoms (Stice et al., 2006). More re search is needed to elucidate the construct being assessed by the restraint scales as well as to develop a valid measure of dieting that reliably assesses a negative energy state. In addition to measurement issues, the mixe d findings in the literature could have occurred because there are different types of dieting with some type s increasing and other types decreasing the risk for bulimic sympto ms (Stice et al., 2006). Real-world dieting likely differs substantially from dieting in randomized, controlled behavioral weight loss and weight maintenance trials. As Stice et al. (2005) note, dieting as usual often involves meal skipping whereas behavioral weight loss a nd weight maintenance diets promote eating at regular intervals. Real -world dieting may also not follow proper nutrition and possibly exclude certain cla sses of food (i.e., carbohydrates) and may involve more intense caloric restriction th an university-based di et programs. The relationship between self-ini tiated, real-world dieting a nd eating pathology remains unclear, and much more experimental resear ch is needed to address this issue.
21 Dieting Recommendations: To Diet or Not to Diet The controversy surrounding dieting ha s implications for the treatment and prevention of eating disorders and obesity. As mentioned previous ly, the obesity field stems from a medical model and has generally promoted dieting and stressed weight loss for most Americans with a pa rticular emphasis on the hea lth risks of excess weight (Brownell & Rodin, 1994; Irving & Neumark-Sz tainer, 2002). Treatment and prevention efforts have primarily recommended caloric re striction and increased physical activity for the purposes of weight loss or weight maintenance (National Task Force on the Prevention and Treatment of Obesity, 2000). A large body of literature on randomized clinical trials of behavioral weight loss programs have cons istently shown modest success (i.e., 8.5-9.0 kg loss on average) over the course of a 20-week program; however, maintenance of these gains after the termin ation of treatment is poor with patients regaining about one-third of their weight in the year post-treatment and almost all of it within five-years (Bacon et al., 2002; Wa dden et al., 2002). Prevention efforts, particularly those geared towards adults, ha ve also largely focused on weight as the outcome variable of interest with dietary a nd exercise changes promoted as a means of weight loss or weight maintenance and disease prevention (Cogan, 1999; Jeffery & French, 1999). Large-scale obesity prevention trials in adults ha ve generally produced disappointing results (Schmitz & Jeffery, 2002). Concerns about the long-term failure of most diets, th e potential negative health consequences of weight cycling, and the role of dieting in the prom otion and maintenance of eating pathology has lead se veral researchers to promote an anti-dieting or un-dieting approach (Foster & McGuckin, 2002; Polivy & Herman, 1992). The eat ing disorder field
22 has largely endorsed an anti-dieting approach in both treatment and prevention contexts (Fairburn et al., 1993; Irving & Neumark-Stzainer, 2002), and preventative interventions were designed using the etio logical models of eating pa thology which aim to reduce sociocultural pressures to be thin, lessen the importance of weight and shape, and teach participants to be critical consumers of the media (Stice & Shaw, 2004). Anti-dieting approaches have also emphasized: (1) the cessa tion of dieting, (2) learning to attend to physiological cues of hunger and satiety, (3 ) promoting body satisfaction and acceptance of current weight, and (4) e nhancing self-esteem (Bacon et al., 2002; Polivy & Herman, 1992; Wadden et al., 2002). Randomized controlled trials of undieting have largely found improvements in self-esteem, mood, and body imag e with little to no cha nges in body weight over the course of the intervention and follow-up (Foster & McGuckin, 2002). Some studies have also found positive changes in physio logical indicators of health (i.e., blood pressure, lipids, cholesterol) in the absen ce of weight loss (Bacon et al., 2002; Mellin, Croughan-Minihane, & Dickey, 1997; Rapoport, Clark, & Wardle, 2000). A metaanalysis of the effectiveness of the anti-di eting approach in eati ng disorder prevention programs concluded that effective eating disorder prevention programs have been developed that have significantly reduced eat ing pathology and the asso ciated risk factors of body dissatisfaction, thin-ideal internalization, and perceived pressure to be thin (Stice & Shaw, 2004). Interactive, psychoeducational interventions appear to be more effective than didactic formats at reducing eati ng pathology. The didactic anti-dieting psychoeducational programs tended to pr oduce changes in knowledge about eating disorders with few changes in eating disorder risk factors (i.e., body dissatisfaction) and
23 no changes in eating pathology (Stice & Sh aw, 2004). Successful interventions were generally multi-faceted and contained not onl y psychoeducation but some combination of group discussions, coping skills, media liter acy, or peer pressure resistance skills components. With so many different components utili zed in these eating disorder prevention interventions, it is unclear which anti-die ting components were re latively successful and which ones were not. Paxton, Wertheim, Pila wski, Durkin, and Holt (2002) addressed this issue by systematically examining seve n distinct anti-dieti ng messages frequently used in prevention programs and assessing th eir persuasiveness and immediate impact on psychological functioning and dieting intentions in adolescent girls. The messages were presented in brief video format. Results sugge st that the messages were rated as a least somewhat relevant and importa nt by most participants. Intentions to diet were significantly reduced in approxima tely a quarter to a third of the girls while the majority of participants reported no ch ange in their intentions to di et. Furthermore, no changes in body satisfaction were observed from preto post-test. The Paxt on et al. (2002) study appears to be the first to systematically ex amine the immediate perception and impact of the anti-dieting approach by a ssessing seven distinct anti-d ieting messages; however, this study did not address the collective imp act of these anti-dieting messages on persuasiveness, psychological f unctioning, and weight control intentions. Furthermore, its generalizability to adults is unknown. No research was located that examined how the pro-dieting message is perceived as we ll as its impact on immediate psychological functioning and weight c ontrol intentions.
24 To sufficiently address the question of whether health care professionals should recommend to diet or not to diet, more research shoul d be conducted that directly compares these two approaches. A handful of randomized, controlled trials have directly compared these two approaches within th e context of long-term outcome on weight, psychological functioning, and physiological measures (Bacon et al., 2002; Foster & McGuckin, 2002; Lowe et al., 2001); however most people do not seek professional advice or treatment for dieting and weight loss and instead try it on their own (Serdula, Collins, Williamson, Pamuk, & Byers, 1993). They are most likely exposed to information on dieting and weight loss in everyd ay situations such as in the newspaper, on television, or from a health care provider. Therefore, it also appears necessary to examine the impact that exposure to these me ssages may have in a format that is more externally valid such as a brief written articl e or video. Because th e pro-dieting and antidieting approaches diverge substantively on their recommendations towards dieting and weight control, it seems likely that they w ould be perceived differently and potentially yield significant differences in immediat e psychological functioning (i.e., mood, body dissatisfaction) and weight control intentions. The following section will describe a pilot study that systematically examined the eff ects of the two dieting messages on perceived pressure to lose weight, body dissatis faction, and dieting intentions. Pilot Study A pilot study was conducted to test the hypothesis that the pr o-dieting and antidieting psychoeducational messages differ s ubstantially in the amount of pressure perceived by participants to lose weight and/or maintain a thin body, body dissatisfaction,
25 and dieting intentions. The experimental s timuli for the pilot study were developed by compiling available information on obesity prevention, eating disorder prevention, as well as a neutral, flu prevention message from reputable online resources. Specifically, the websites for the U.S. Department of Health and Human Services, the Centers for Disease Control, and the National Eating Diso rders Association were consulted, and the experimental stimuli were derived primarily using material from these agencies. These sources were consulted in an effort to not only provide accurate information but also to increase the external validity of the study by approxima ting as closely as possible the health information being disseminated to the public. In an effort to obtain a strong experime ntal manipulation, material that appeared to clearly advocate weight loss and dieti ng versus non-weight lo ss and non-dieting was selected to be included in the obesity prev ention and eating disorder prevention stimuli. Each of the experimental stimuli was presented on one-page in the format of a health information article and divided into the fo llowing subsections: prevalence and costs, definition, causes, consequences, and what th e individual can do to prevent the problem. In addition, the headlines of each article empha sized the central point of the particular health education message and were equated on wording. For example, the pro-dieting, obesity prevention message stimuli stated, L ose Weight and/or Ma intain a Low Body Weight to Prevent Overweight and Obesit y, while the anti-dieting, eating disorder prevention message headline indicated, Stop Dieting to Prevent Disordered Eating. The flu prevention message also had a simila rly structured headline that stated, Get Vaccinated to Prevent the Flu. Please see A ppendices C, D, and E to review each of the health education messages in its entirety.
26 After the experimental stimuli were created, they were presented to an expert panel of researchers that sp ecialize in the study of body imag e and eating disturbances to verify the content and readability of the me ssages. The expert panel consisted of one licensed clinical psychologist, six doctoral students in c linical psychology, and three undergraduate research assistan ts. Feedback from the expert panel suggested that the stimuli were sufficiently equated. Minor changes in wording were made based on feedback from the expert panel to ensure readability of each message. Sixty-five undergraduate women between the ages of 18 and 47 ( M = 23.95, SD = 5.8) were then randomly assigned to read one of the three psychoeducational dietingrelated messages: (1) pro-dieting, obesity prevention, (2) anti-dieting, eating disorder prevention, and (3) no-dieting, flu prevention. The sample was ethnically diverse and composed of 47.7% Caucasian, 21.5% Hispanic, 18.5% African-American, 6.2% Asian/Pacific Islander, and 6.2% who identifi ed themselves as Other. Self-reported weight and height indicates that 12.3% were underweight, 48.4% were average weight, 25% were overweight, and 14.3% were obese. Participants were compensated with one extra credit point in th eir psychology course. In addition to demographic information, we ight/shape dissatis faction and affect were assessed pre-post test with the Visual Analogue Scale (VAS; Heinberg & Thompson, 1995; see Appendix F) and the Posi tive and Negative Affect Scale-Revised (PANAS-X; Watson & Clark, 1992; see Appendix G) After the pre-test measures were obtained, the participants were asked to read the psychoeducational material and answer several questions about their perceptions of it utilizing five true/false attention check items, a modified version of the Soci ocultural Attitudes Towards Appearance
27 Questionnaire (SATAQ)-3 (Thompson, van de n Berg, Roehrig, Guarda, & Heinberg, 2004; see Appendix H) Pressures and Internal ization subscales, and the Message Rating Form (Sperry, Thompson, Roehrig, & Vandello 2004; see Appendix I). Post-test VAS and PANAS-X measures were then completed, and the participants were debriefed and awarded their extra credit point. Analyses were conducted to assess for any preliminary differences among the groups on the demographic and pre-test variab les. No significant differences were found among the groups on race, 2 (8) = 6.81, p >. 05, BMI, F (2, 62) = .84, p > .05, age, F (2, 62) = .51, p > .05, pre-test diss atisfaction with weight, F (2, 61) = 1.0, p > 05, pre-test dissatisfaction with shape, F (2, 61) = 2.2, p > .05, or pr e-test PANAS-X scores, F (2, 62) = .01, p > .05. Collectively, these preliminary analyses suggest that random assignment was successful. No participants met the a priori exclusion criteria for the attent ion check (< 4 of 5 correct; see Appendix Q), suggest ing that all participants sufficiently attended to the experimental stimuli. Therefore, all partic ipant data is included in the subsequent analyses. To examine for any differences among the three messages on non-specific factors, the Message Rating Form was examined with each item analyzed separately. A significant difference was found among the th ree groups in the extent to which the participants rated the messa ges as easy to understand, F (2, 62) = 4.75, p <. 05, with Fishers LSD post-hoc test suggesting that the no-dieting, flu prevention message was significantly more easy to understand ( M = 4.87) than the pro-dieting, obesity prevention message ( M = 4.38) and the anti-dieting, eating disorder prevention message ( M = 4.55). No other significant differences were found on the Message Rating Form items (see
28 Table 1 for mean scores from pilot study), in dicating that the message s were perceived as equally convincing, effective, applicable, and credible. Although the flu prevention condition was endorsed as easier to understand than the other two conditions, this does not appear to be a significant problem as ex amination of mean scores suggests that all three messages were highly understandable to the participants (all means over 4.37). Overall, results from the Message Rating Fo rm indicate that the three messages were successfully equated on non-specific factors. Table 1 Mean Scores for Pilot Study by Condition Measure Pro-Dieting, Obesity Prevention (N=21) Anti-Dieting, Eating Disorder Prevention (N=22) No-Dieting, Neutral Flu Prevention (N=22) Pre-VAS BD 57.2 (31.13) 45.18 (37.99) 58.00 (30.81)1 Post-VAS BD 62.65 (34.59) 42.77 (37.44) 48.55 (32.32)2 Pre-VAS Shape Dissatisfaction 64.9 (27.69) 45.68 (35.77) 51.27 (25.54) Post-VAS Shape Dissatisfaction 62.38 (32.99) 41.0 (37.9) 45.41 (28.52) MRF-Convincing 4.14 (.73) 4.05 (.79) 4.18 (.85) MRF-Effective 3.9 (.94) 3.45 (.86) 4.0 (.82) MRF-Applicable 3.10 (1.5) 2.77 (1.38) 3.45 (1.38) MRF-Easy to Understand 4.38 (.59)a 4.55 (.6)a 4.86 (.35)b MRF-Credible 3.86 (.57) 3.68 (.84) 3.95 (.84) MRF-Influential 3.48 (1.12) 3.0 (.87) 3.5 (1.01) Pre-PANAS-X 32.29 (11.87) 31.82 (11.3) 32.0 (14.58) Post-PANAS-X 31.70 (10.33) 31.09 (13.48) 29.59 (12.86) SATAQ-3 Perceived Pressure 16.43 (4.20)a 9.45 (2.5)b 9.32 (.89)b SATAQ-3 Internalization 21.48 (5.34)a 13.05 (2.61)b 13.82 (2.11)b Weight Loss Intention 3.05 (1.53)a 1.59 (.91)b 1.05 (.21)b Exercise Intention 4.38 (.92)a 3.18 (1.37)b 2.0 (1.16)c Note : Letter subscripts indicate signi ficant differences ac ross conditions; Number subscripts denote significant differences across time ; VAS BD: Visual Analogue ScaleBody Dissatisfaction; MRF: Message Rating Form; PANAS-X: Positive and Negative Affect Scale-Revised; SATAQ-3: Sociocultural Attitudes Towards Appearance Scale-3
29 The modified SATAQ-3 Pressures subscal e was then analyzed for differences among the three conditions on the pressures construct utilizing a one-way ANOVA. As hypothesized, a significant main effect was found for condition, F (2, 62) = 43.60, p < .001, partial 2 = .58. Fishers LSD post-hoc test indi cates that participants in the prodieting, obesity prevention message felt signifi cantly more pressure to lose weight ( M = 16.43) than those in the anti-dieting, eating disorder prevention message ( M = 9.45) and no-dieting, flu prevention message ( M = 9.32). Differences betw een the eating disorder and flu prevention messages were not statistic ally significant. This finding remained significant even after an ANCOVA was run to control for BMI, F (2, 61) = 48.1, p < .001, partial 2 = .61. Group differences among behavioral intenti ons were then examined. A one-way ANOVA was computed on the modified SA TAQ-3 Internaliza tion subscale. A significant main effect for condition was found, F (2, 62) = 35.57, p < .001, partial 2 = .53, with post-hoc tests revealing that the pr o-dieting, obesity prev ention message elicited significantly greater Intern alization intentions ( M = 21.48) than both the anti-dieting, eating disorder prevention message ( M = 13.05) and the no-dieting, flu prevention message did ( M = 13.82). The difference in mean scor es between the eating disorder and flu prevention messages was non-significant. An ANCOVA revealed that this finding remained significant after controlling BMI, F (2, 62) = 35.22, p < .001, partial 2 = .54. Individual items assessing the extent to which the article made the participant want to start a weight loss di et and increase their physical ac tivity were also analyzed in separate one-way ANOVAs. Significant main effects were found for both the weight loss and physical activity intentions, F (2, 62) = 21.76, p < .001, partial 2 = .41, and F (2, 62)
30 = 22.42, p < .001, partial 2 =.42, respectively. Fishers LSD post-hoc tests revealed that the pro-dieting, obesity prevention message elic ited a greater desire to start a weight loss diet ( M = 3.05) than both the anti-dieting, eating disorder ( M = 1.59) and no-dieting, flu ( M = 1.05) prevention messages. The differe nce between the eating disorder and flu prevention messages was non-significant. An ANCOVA confirmed this finding was independent of BMI, F (2, 61) = 29.44, p < .001, partial 2 = .49. Post-hoc tests also found significant differences in intentions to increase physical activity among all three conditions with the pro-dieting, obesity prevention message being greatest ( M = 4.38), followed by the anti-dieting, eating disorder ( M = 3.18) and no-dieting, flu ( M = 2.0) prevention messages, which was independent of BMI, F (2, 61) = 21.91, p < .001, partial 2 = .42. Pre-post test analyses were then conducte d to assess for state changes in weight and shape dissatisfaction and negative affect. Separate 3 (Condition) X 2 (Time) Mixed Design ANOVAs were computed. A significant time by condition interaction was found for weight/size dissatisfaction, F (2, 61) = 6.21, p < .01, partial 2 = .17. Follow-up paired t-tests utilizing Bonferronis correc tion indicate that the nodieting, flu prevention group reported significant reductions in we ight/size dissatisfaction from pre ( M = 58.0) to post ( M = 48.6) test; although non-signi ficant, mean trends sugge st that the pro-dieting, obesity prevention message elicited some increase in weight/size dissatisfaction from pre to post test ( M1 = 57.2, M2 = 62.7). No significant cha nges were found pre-post on the shape dissatisfaction VAS or the PANAS-X total subscale score for any condition. Overall, findings from the pilot st udy supported the hypothesis that the prodieting, anti-dieting, and no-die ting messages differ significantly in the extent to which
31 participants perceived pressure to lose wei ght and/or maintain a thin body from them. Significant differences in behavioral intentions also emerged with the pro-dieting, obesity prevention message eliciting grea ter internalization, dieting, a nd exercise intentions than the other two conditions. Exploratory analys es revealed a non-si gnificant trend across time that the pro-dieting, obesity preventi on message tended to produce increased body dissatisfaction at post-test. Current Study Based upon the results of the pilot study, experimental manipulation of the three prevention messages provides the opportunity to directly examine the effects of differing dieting messages (i.e., pro-dieting, anti-dieti ng, no dieting) on psychological functioning as well as weight control intentions and be haviors in undergraduate women. The current study builds upon the pilot study by increasing th e sample size and adding a two-week follow-up assessment to examine the short-term effects of the experimental manipulation. A study of this nature is needed for theoretical as well as practical reas ons. First, there is virtually no evaluative work on the pro-die ting, obesity prevention message, which has been widely disseminated by public health ag encies and the media. In light of the findings from the pilot study that exposur e to these messages increased perceived pressure to lose weight, a construct whic h research has consistently found to have deleterious effects on women and girls, it is im perative that the effect s of the pro-dieting, obesity prevention messages on psychological functioning and eating and weight control practices be explored further. Second, no research to date has examined the effects of the pro-dieting and anti-dieting messages on heal thy eating and weight control practices.
32 Given that both messages share a goal of increasing healthy eati ng and weight control behaviors, which in turn promotes the reduction of dis ease, addressing this gap in the extant literature is important. Third, no st udy has directly compared the effects of the pro-dieting and anti-dieting messages on immediate and short-term psychological functioning and weight control intentions and be haviors. Lastly, there is still very little experimental work which has evaluated the eff ect of perceived pressu re to lose weight on internalization of the thin-ideal, body di ssatisfaction, affect, a nd eating and weight control intentions and behaviors. An expe rimental study that indu ces change in this construct allows for the examination of causal risk factor status according to the criteria proposed by Kraemer et al (1997). While perc eived pressure to be thin has received support in cross-sectional and l ongitudinal studies as a risk factor for eating disordered symptoms (Stice, 2002), additional experimental research is needed to examine the effect of the perceived pressures construct on ea ting disordered symptoms and other weight control practices. Accordingly, the goals of the current study are: (1) to experimentally manipulate the dieting and weight loss messages to determin e their immediate effects on (a) perceived pressure to lose weight and/or maintain a low body weight, (b) ps ychological functioning, including body dissatisfaction, negative affect and thin-ideal internalization, and (c) eating and weight-control inten tions, including both healthy and unhealthy st rategies of dieting, exercise, healthy eating, and bulimic sy mptoms, (2) to examine the impact of the dieting and weight loss messages over a tw o-week period on the same psychological variables and eating and weight control behaviors, (3) to test whether perceived pressure to lose weight from the experimental messa ge mediates the relationship between dieting
33 message and weight control intentions and two-week follow-up behaviors, and (4) to evaluate the risk factor st atus of the perceived pressure construct for bulimic symptomatology utilizing Kraemer et al.s (1997) criteria. Based on the literature as well as find ings from the pilot study, the following hypotheses are offered: (1) Participants in th e pro-dieting condition w ill perceive greater pressure to lose weight from the psychoe ducational message than those in the antidietingand no-dieting conditions. (2) Participants in the pr o-dieting cond ition will report significantly greater disturbances in psychological functioning a nd intentions to engage in weight loss strategies immediately after expos ure to the psychoeducational message than those participants in the other two conditions Specifically, it is hypothesized that state body dissatisfaction, intentions to diet, exerci se, eat more healthfully, utilize unhealthy weight control practices, and engage in thin-ide al thinking and behavior will be greater in the pro-dieting condition than the anti-dieti ng and no-dieting conditions. No differences in negative affect among the groups are hypothesi zed at post-test based on the findings of the pilot study. (3) Trait levels of body dissa tisfaction, thin-ideal internalizat ion, drive for thinness, perceived pressure to be thi n, negative affect, die ting, bulimic symptoms, healthy eating, and exercise will increase from pre-test to the two-week follow-up in the pro-dieting condition compared to the an ti-dieting and no-dieting conditions (4) Perceived pressure to lose weight from the psychoeducational message will mediate the hypothesized increase in weight control intentions at post-te st and behaviors at the twoweek follow-up. (5) The findings will indicat e that the pressures construct will meet Kraemer et al.s (1997) criter ia for a causal risk factor for bulimic symptomatology in college women.
34 Chapter 2 Method Participants The participants were 139 undergraduat e females who were recruited from the University of South Floridas Department of Psychology participant pool. They ranged in age between 18 and 30 ( M = 20.63, SD = 2.51). The sample was racially diverse with 18% African-American ( N = 25), 6.5% Asian/Pacific Islander ( N = 9), 49.6% Caucasian ( N = 69), 17.3% Hispanic ( N = 24), 0.7% Native American ( N = 1), and 7.9% Other ( N = 11). Self-reported weight and height reveal ed that the average body mass index (BMI) was in the normal range ( M = 24.31, SD = 6.05) with scores ranging from 17 to 62. 6.5% were underweight ( N = 9; BMI = 18.5 or lower), 61.9% were average weight ( N = 86; BMI = 18.51-24.49), 15.1% were overweight ( N = 21; BMI = 25-29.99), and 16.5% were obese ( N = 23; BMI = 30.0 or higher) with no current or past history of an eating disorder diagnosis or current purging behaviors reported. Participan ts were compensated with extra credit points in thei r psychology course(s). Measures Demographic information Participants were asked to provide demographic information including age, race, height, weight, and year in school. Body mass index
35 (BMI) was calculated using self -reported weight and height with the standard formula: [(weight in pounds/(height in inches)2] X 703. Body dissatisfaction : Two measures of body dissatisfaction were utilized: one trait measure and one state measure. The Eating Disorder Inventory Body Dissatisfaction subscale (ED I-BD, see Appendix J) (Gar ner, Olmsted, & Polivy, 1983) was used as the trait measure of body dissati sfaction. The EDI-BD is a 7-item scale that assesses overall satisfaction with various weight related body s ites. It has demonstrated good reliability (alphas above .80) across varied samples in previous studies (Garner, 1991; Thompson, 1992). Cronbachs alpha in this study was .89. The EDI-BD was administered at baseline and follow-up. The Visual Analogue Scales (VAS, see Appe ndix F) was utilized to assess state dissatisfaction with body weight and shap e (Heinberg & Thompson, 1995). On these scales, participants are asked to indicate their level of dissatisf action on a 100 mm line, with the left-most point being "no weight/s ize dissatisfaction" (" no overall appearance dissatisfaction") and the right-most point being that of "ext reme weight/size dissatisfaction" ("extreme overall appearance dissatisfaction"). The distance from the left-most point on the line (0) measured in millimeters indicates the level of distress (Thompson et al., 1999). The VAS has been found to correlate highly with the Eating Disorder Inventory-Body Dissatisfaction subscale (e.g., Heinberg & Thompson, 1995) and has been widely-used because it may reduce the level of pre-test sensitization on post-test responses (Thompson, 2004b). The VAS assessed weight and shape dissatisfaction preand postexposure to the experimental manipulation of the psychoeducational message.
36 Thin-ideal internalization The Sociocultural Attitudes Towards Appearance Scale-3 (SATAQ-3, see Appendix K)-Internaliza tion subscale was used to assess trait levels of thin-ideal internalization (Tho mpson et al., 2004). This measure focuses specifically on internalization of media message s regarding the thin-i deal, and ratings are made on a five-point Likert scale rangi ng from Definitely Agree to Definitely Disagree. The SATAQ-3 has two internalizat ion subscales with excellent reliability: Internalization-General (Cronb achs alpha = .96) and Interna lization-Athlete (Cronbachs alpha = .95) (Thompson et al., 2004 ). In this sample, Cronbachs alpha revealed good reliability for both subscales: Internalization-General=.94 and InternalizationAthlete=.85. Additionally, 5 items from the SATAQ-3 Internalization-General and Athlete subscales were modified and utilized in the pilot and full studies to assess the impact of the psychoeducational messages on future thin-i deal thoughts and behaviors at post-test (see Appendix H). All SATAQ-3 stem phrases were retained, but wording was changed to reflect the impact of the psychoeducatio nal message on thin-ideal intentions. For instance, one of the modified items state d, Reading this article makes me want to compare my body to that of people in good sh ape. General and athlete items were summed to obtain a composite modified Internaliz ation score. Internal consistency of the modified measure was acceptable in the pilo t study (Cronbachs al pha=.76) and full study (Cronbachs alpha=.90). Sociocultural pressure The Sociocultural Attitudes Towards Appearance Scale-3 (SATAQ-3; Thompson et al., 2004) Pre ssures subscale (see Appendix K) was administered during the pretest and follow-up assessments to examine perceived
37 sociocultural pressures to be thin. The Pressures subscale co nsists of six, Likert scale items and has demonstrated excellent reliab ility (Cronbachs alpha=.94) and convergent validity (Thompson et al., 2004). A modified, five-item version of the SATAQ-3 Pressures subscale (Thompson et al., 2004) was developed for the pilot and fu ll studies to assess th e extent to which participants perceived pressure from the e xperimental message to lose weight and/or maintain a low body weight (see Appendix H). Items modified for this study retained the SATAQ-3 stems but changed the cited source of perceived pressure from TV, movies, and magazines to the psychoeducational message For example, an original item on the Pressures subscale was modified from, Ive felt pressure from TV or magazines to lose weight, to Ive felt pressure from this arti cle to lose weight. Items were summed to obtain a composite pressures scor e. Internal consistency of the modified measure was .78 in the pilot and full study. Item -total analyses revealed that the internal consistency of the measure improved to .90 by deleting item 13 (I felt pressure from this article to avoid dieting.); therefore, al l analyses were conducted on the four-item subscale. Drive for thinness The Eating Disorder Inventor y-Drive for Thinness (EDI-DT; See Appendix J, Garner et al., 1983) was used to measure drive for thinness. This scale measures restricting tendencies, desire to lose weight, and fe ar of weight gain. It has been show to have an internal consiste ncy of .83 for a combined sample of eating disordered individuals and .81-.91 for four samples of nonpatient female controls (Garner, 1991). The EDI-DT was administered at pre-test and the two-week follow-up, and the directions were modified to assess us ual and past two-week drive for thinness. Reliability was excellent with an alpha of .91 in this sample.
38 Dieting The Dutch Eating Behavior Ques tionnaire-Restraint Scale (DEBQ-RS; see Appendix L, van Strien, Frijters, Berger s, & Defares, 1986) was used to assess dieting intentions and behavior. This scal e consists of ten-items that measure the frequency of dieting behaviors using a 5-point Likert scale, which ranges from never to always. The DEBQ has been shown to ha ve good internal consistency (Cronbachs alpha=.95) and test-retest re liability (r=.92) (Allison, Ka linsky, & Gorman, 1992). The original DEBQ-RS was administ ered at pre-test and two-w eek follow-up to assess usual and past 2 week behavior, respectively. Direc tions were modified to assess intentions to diet at post-test. Reliability of the DEBQ -RS was excellent (Cr onbachs alpha= .92) in this sample. Negative affect The Positive Affect and Negative Affect Scale-Revised, Negative Affect subscale (PANAS-X; see A ppendix G, Watson & Clark, 1992) was used to assess both state and trait negative affect. State negative affect was assessed preand postmanipulation of the psychoeducationa l message, and trait negative affect was assessed at pre-test and the two-week follow-up. In this scale, participants rate 20 negative emotional states (e.g., sadness, guilt, and fear/anxiety) curren tly or over the past two weeks. A 5-point Likert scale, which ra nges from very slightly or not at all to extremely, is used. This scale has been f ound to have adequate internal consistency, test-retest reliability, converge nt and divergent validity, and predictive validity (Stice & Agras, 1998; Watson & Clark, 1992). Reliabil ity was very high (Cronbachs alpha=.95) in this sample. Visual Analogue Scales relate d to affect were used as filler questions to disguise the main purpose of the VAS scalesto asse ss state body dissatisfaction (see Appendix
39 F). Following the same procedure described above for the measurement of state weight and shape dissatisfaction, participants will be asked to rate the extent of their current affect on several dimensions, including happi ness, anxiety, energy le vel, disappointment in self, anger, calmness, and irritability. Bulimic symptoms The Eating Disorder Examination-Questionnaire (EDE-Q; see Appendix M, Fairburn & Beglin, 1994) Bulimia Subscale was used to measure bulimic symptoms at pre-test and the two-week follow-up. The EDE-Q is derived from the Eating Disorder Examination (EDE; Fairburn & Cooper, 1993), which is a widely used and validated semistructured in terview. The EDE-Q Bulimia Subscale consists of twelve items that assess the frequency of binge eating and purging (i.e, vomiting, laxative and diuretic use, excessive exerci sing). The frequency is measured in terms of the number of days that binging and/or purging occurred as opposed to the number of individual episodes. The internal consistency of the EDE-Q has been found to be adequate (Cronbachs alpha=.84) (F airburn & Beglin, 1994). In addition, the EDE-Q demonstrates acceptable criterion validity and convergent validity (Black & Wilson, 1996). Alpha was .77 in this sample. A six-item modified version of the ED E-Q was also developed for this study to assess unhealthy weight control intentions (see Appendix N). Items 10-12, which assess compensatory behavior frequency, were adap ted to measure intentions to vomit, use laxatives/diuretics, and excessive exercise to control weight on a five-point Likert scale. Additionally, items related to intentions to use diet pills, fasting, smoking, and meal skipping as weight control practic es were added to the scale. Reliability of this modified
40 measure was low (Cronbachs alpha= .59), and item-total analyses did not indicate any improvements if any item was deleted from the scale. Eating disorder screening In an attempt to minimize any risk associated with the study that might potentially affect individuals with a high level of eating disturbance, potential participants were administered sc reening questions via USF Experiment Trak from the Eating Disorder Inventory (EDI)-3 Referral Form (Garner, 2005, see Appendix O) which is designed to identify individuals at risk for an eat ing disorder or with a past history of an eating disorder. Five Likert -scale items ranging from Never to Once a Day or More were utilized to assess curre nt eating disordered symptomatology such as Over the past three months, how often have you used laxatives to control your weight or shape? Additionally, a yes/no question asked potential participants whether they have ever been diagnosed or treated for an eating disorder. Evidence of past history of an eating disorder or active purging episodes excl uded potential participants from the study, and they were blocked from enrolling in the study. Healthy eating The Multidimensional Health Behavior InventoryDiet subscale (MHBI; Kulbok, Carter, Baldwin, Gilmarti n, & Kirkwood, 1999; see Appendix P) was utilized to assess healthy eating intent ions and behaviors. The MHBI is a psychometrically sound instrument that was de veloped for use in adolescent and collegeaged samples. The MHBI-Diet subscale cons ists of 13 items assessing frequency of healthy nutritional behaviors such as eati ng whole grain foods and limiting sugar intake on a 5-point Likert scale ranging from Never to Always. Internal consistency of the Diet subscale has been found to be very good (Cronbachs alpha=.88) (Kulbok et al., 1999). Directions were modified to assess us ual behavior, intentions, and past two week
41 behavior. In addition to the original MHBI items, two questions regarding fruit and vegetable consumption were added using th e MHBI stems. Cronbachs alpha was found to be .79 in the current study; it em-total analyses revealed th at reliability improved to .84 when item 32 (Eat at least one or more of the following items every day: chips, candy bars, cake, doughnuts, pastries muffins, cookies, ice cr eam, pudding, chocolate) was deleted. All analyses were conducted w ith item 32 deleted from the scale. Exercise The Multidimensional Health Behavi or InventoryEx ercise subscale (MHBI; Kulbok et al., 1999; see Appendix P) wa s used to assess exercise intentions and behaviors. The MHBI-Exercise subscale consis ts of four items on the same five-point Likert scale described above for the MHB I-Diet subscale. Items assess frequency of physical activity such as vigor ous exercise for at least 20 minutes a day, three times a week. Kulbok et al. (1999) demonstrated th e scale has acceptable internal consistency (Cronbachs alpha=.80) and cont ent and convergent validity. Test-retest reliability was not assessed. Directions were changed to assess usual, intende d, and past two-week exercise behavior. Cronbachs alpha was .86 in this sample, suggesting good reliability. Flu prevention intentions and behaviors The MHBICheckup and Stress/Rest subscales (Kulbok et al., 1999; see Appendix P) were utilized to assess intentions and behaviors advocated in the flu prevention messa ge for the purposes of face validity. The Stress/Rest subscale consists of six items that measure freque ncy of self-care and stress reduction behaviors such as sl eeping 7-8 hours per night on a fi ve-point Likert scale. Cronbachs alpha was acceptable (.76) for this subscale (Kulbok et al ., 1999). Directions were modified to assess usual, inte nded, and past two-week behaviors.
42 The Checkup subscale of the MHBI is a 9item scale that assesses the frequency of routine health care such as regular physic al checkups and monthly self breast exams on a five-point Likert s cale. Internal cons istency of this subscale is good (Cronbachs alpha=.82) (Kulbok et al., 1999). Some items fr om the original scale were modified to include behavior related to flu prevention su ch as receive a flu shot and wash hands frequently, and directions were modified to assess usual, intended, and past two-week behaviors. Message rating form A modified version of the Message Rating Form (Sperry et al., 2004; see Appendix I) was utilized in the pilot and full studies to assess non-specific factors of the messages at posttest. The extent to which th e messages were perceived as convincing, effective, applicable, easy to unde rstand, credible, and in fluential were rated on a five-point Likert scale ranging from Definitely Disagr ee to Definitely Agree. An alpha of .65 was obtained in the pilot study, and Cronbachs alpha was .82 in the full study. Attention check Five true/false questions were created for each condition to serve as an attention check. Efforts were made to include relevant information from each message in the attention check and to e quate the items for each condition. Questions related to prevalence, prevention, and sy mptom presentation. These items were administered immediately after the particip ant finished reading the psychoeducational information. Because no participants faile d the attention check in the pilot study, the items were re-worked for the full study in an effort to increase item difficulty to ensure that participants were attending to the messages (see Appendix R). Participants who
43 answered fewer than four out of the five items correctly were excluded from further analyses. Distraction task A distraction task was utilized after all trait measures were obtained as a washout period prior to the administration of the pre-test measures, experimental manipulation, and post-test m easures. Nolen-Hoeksema and colleagues have found that brief (5-8 minutes), ex ternally-focused, ac tive tasks return experimentally-induced dysphoric mood states back to baseline levels (Lyubomirsky & Nolen-Hoeksema, 1993, 1995; Morrow & Nolen-Ho eksema, 1990). Therefore, a similar procedure was used in the curr ent study to counter any negati ve affect induced as a result of completing the pre-test trait measures. Participants were asked to spend 5-8 minutes thinking about the countries of the world and then to write a list of their top ten travel destinations as well as their perceptions of how the media portrays these destinations (see Appendix S). Procedure Participants enrolled in the study via USF Experiment Trak. To minimize any potential risks associated with the study, pot ential participants we re prescreened through Experiment Trak using the EDI-3 RF and a que stion about past eating disorder history. Any participant who reported a current or past history of an eating disorder or current purging behavior was excluded from the study and was unable to enroll in it. Participants were randomly assigned to one of the three experimental conditions: (1) pro-dieting message (obesity prevention) (2) anti-dieting messa ge (eating disorder prevention), and (3) no dieting message (flu prevention). The study was conducted in a
44 group setting in classrooms, and pa rticipants were instructed to sit at least one seat apart so that they were unable to r ead one anothers testing material s. They were told that the study examined mood, health, and the media. Pa rticipants provided th e last four digits of their social security number as their st udy identification number in order to easily link participant data from both sessions. Testing packets for each of the three c onditions were stacked consecutively by condition (i.e., 1, 2, 3, 1, 2, 3, etc.) and handed out randomly to particip ants; the measures were in the following order: demographic in formation, trait measures using the SATAQ3, EDI-BD, EDI-DT, PANAS-X, DEBQ-RS, EDE-Q, and MHBI t he distraction task, the pre-test VAS and PANAS-X measures. Imme diately after completi on of the pre-test measures, participants read the experimental stimuli and co mpleted the attention check items, Message Rating Form, modified SATAQ3 Pressures and Inte rnalization scales, and post-test VAS and PANAS-X measur es. The MHBI, DEBQ-RS, and EDE-Q behavioral intention questionna ires were then administered After participants handed in their completed measures, they were asked to schedule their appointment for the twoweek follow-up assessment. Participants email addresses and phone numbers were obtained at this point in or der to provide reminder calls to minimize attrition rates. The two week follow-up assessment was also conducted in a group setting, and participants were given a packet of qu estionnaires to assess past two-week body dissatisfaction, thin-ideal internalization, ne gative affect, dieting, bulimic symptoms, healthy eating, exercise, drive for thinness, a nd perceived pressure to be thin using the same measures from the baseline assessment. Directions were changed on each measure to instruct participants to answer the que stions based on their feelings and behaviors
45 over the past two weeks. After completing the follow-up packet, participants were fully debriefed and awarded th eir extra credit points. Design and Analyses Any participant who failed the attention check (< 4 out of 5 true/false items correct) or did not attend the second se ssion was dropped from the study analyses. Preliminary analyses were conducted to te st for any initial differences among the conditions as well as to determine if there were differences between participants who were dropped from the study because they failed the manipulation check and those retained for the study. Demographic vari ables and baseline trait levels of body dissatisfaction (EDI-BD), thin-ideal internaliz ation (SATAQ-3), perceived pressure to be thin (SATAQ-3), drive for thinness (EDI-D T), negative affect (PANAS-X), dieting (DEBQ-RS), bulimic symptoms (EDE-Q), h ealthy eating (MHBI), and exercise (MHBI) as well as the pre-test state VAS and PA NAS-X measures were computed by condition using one-way ANOVAs for continuous variables and 2 for categorical variables. Differences among the ratings of the psychoeducational message items (MRF) were also analyzed in separate one-way ANOVAs. A series of ANCOVA analys es were computed to test the hypotheses related to group differences. Hypothesis 1 stated th at there would be significant post-test differences among the groups on perceived pressu re to lose weight with the pro-dieting message eliciting greater pressure than the anti-dieting and no-dieting message conditions. To test this hypothesis, a one-way ANCOVA was computed on the modified
46 SATAQ-3 Pressures scale with the baseli ne SATAQ-3 Pressures scale used as a covariate. To test Hypothesis 2, which stated that post-test body dissatisfaction and intentions to diet, exercise, eat more h ealthfully, utilize unhealthy weight control practices, and engage in thin-i deal thinking and behaviors wo uld be significantly greater in the pro-dieting message condition than the anti-dieting and no-dieting conditions immediately after the experimental mani pulation, separate one-way ANCOVAs were computed using baseline scores as the covari ate on the modified version of the SATAQ-3 Internalization subscale, EDE-Q, DEBQ-R S, MHBI Exercise and Healthy Eating subscales, and EDI-DT. One-way ANCOVAs were computed on post-test VAS weight and shape dissatisfaction and PANAS-X scores w ith the baseline and pre-test state scores used as covariates to analyze for changes in state body dissatisfaction and negative affect. Additionally, exploratory analyses were conduc ted to examine the ro le of participant weight status on these outcomes. The same analyses described above were conducted adding weight status (overweight vs. non-ove rweight) as an additional between-subjects factor, resulting in a series of 3 (Condition) X 2 (Weight Status) Between Subjects ANCOVAs. Hypothesis 3, which stated that body dissa tisfaction, thin-ideal internalization, perceived pressure to be th in, negative affect, dieting, bu limic symptoms, healthy eating, and exercise would increase significantly fr om baseline to follow-up in the pro-dieting message condition compared to the anti-diet ing and no-dieting me ssage conditions, was examined using separate 3 (Experimental Condition) X 2 (Time: Baseline, Two Week Follow-Up) repeated measures ANOVAs on the EDI-BD, EDI-DT, SATAQ-3, EDE-Q,
47 MHBI, DEBQ-RS, and PANAS-X. Separate exploratory analyses were conducted on the above measures with weight status (not overweight vs. overweight) as an additional between subjects factor. To test hypothesis 4, which stated that perc eived pressure to lose weight from the psychoeducational message would mediate wei ght control intentions and behaviors, Baron and Kennys (1986) procedure for assessi ng mediation was utilized. According to Baron and Kenny (1986), the following conditions must be met to establish mediation: (1) the independent variable (diet message) must affect the mediator variable (perceived pressure), (2) the independent variable (diet message) must a ffect the dependent variable (weight control intentions/behaviors), (3) th e mediator (perceived pressure) must affect the dependent variable (weight control inten tions/behaviors), and (4) the effect of the independent variable (diet message) on th e dependent variable (weight control intentions/behaviors) should be near zero when controlling for the mediator variable (perceived pressure). The Sobel test was co mputed for each analysis that met the Baron and Kenny (1986) criteria to test the si gnificance of the mediational effect. Exploratory moderational analys es were also conducted us ing trait levels of thinideal internalization (SATAQ-3) as the moderato r variable, diet message as the predictor, and post-test perceived pressu re and weight control inten tions and behaviors as the outcome variables. According to Baron and Kenny (1986), the moderator hypothesis is supported if the interaction be tween the predictor and the mo derator is significant after controlling for the effects of th e predictor and the moderator in the regression analyses. Based on the recommendations of Kraem er, Wilson, Fairburn, and Agras (2002), which state that treatment groups can be directly compared in mediational and
48 moderational analyses in randomized contro l trials, the two dieting messages (prodieting vs. anti-dieting) we re directly compared in the current mediational and moderational analyses. The pro-dieting me ssage was coded as .5 and the anti-dieting message was coded as -.5 in the regression analyses as recommended by Kraemer et al. (2002). If trait levels of a de pendent variable were assessed at pre-test, these scores were used as a covariate in each of the regressi on equations for that out come variable in the mediational and moderationa l analyses (Kenny, 2006). Hypothesis 5, which stated that perceived pressure to lose weight/maintain a thin body would meet Kraemer et al. s (1997) criteria as a causal risk factor for bulimic symptomatology, was evaluated by examini ng the findings from the one-way ANOVAs by condition for post-test perceived pressure and post-test bulimic intentions and the mixed model ANOVA (condition by time) fo r follow-up bulimic behaviors. For Hypothesis 5 to be supported, two conditions had to be met. First, a significant difference among the groups in post-test percei ved pressure to lose weight had to be found, with the pro-dieting message eliciting significantly higher levels than the antidieting and no-dieting message conditions (Hypothesis 1). S econd, the ANOVAs for post-test bulimic intentions and past tw o-week bulimic behaviors had to reveal significantly greater bulimic symptomatology in the pro-dieting condition than the antidieting and no-dieting conditions. Skewness and kurtosis values were examin ed for all outcome variables, and all variables were within the acceptable ranges. Pearson Product Moment and Point-Biserial correlations were computed for all conti nuous and categorical de pendent variables, respectively. The modified Bonferroni procedure was utilized on all follow-up
49 comparisons to control Type I error rate whil e maintaining a higher degree of statistical power than the traditional, more conser vative Bonferroni co rrection (Kromrey & Dickinson, 1995; Simes, 1986). All an alyses were performed with SPSS 14.0.
50 Chapter 3 Results Preliminary Analyses Sixteen participants were excluded from th e study analyses, leaving a sample size of 123. The sample sizes per condition were: pro-dieting ( N = 46), anti-dieting ( N = 37), and no-dieting ( N = 40). There was little overlap be tween the two exclusion criteria: inattention and attrition. Ten participants failed the atten tion check, and 7 did not return for the second session; only one participant failed the attention check and did not return for the second session. Overall, there was a ma rginally significant effect for condition by exclusion status, 2 (2) = 5.48, p > .05, with 2 participants excluded from the pro-dieting, 9 from the anti-dieting, and 5 from the no-di eting conditions. Upon examining exclusion status more closely, a signi ficant difference among conditions was found for those who failed the attention check, 2 (2) = 6.27, p <.05, with more pa rticipants failing the antidieting ( N = 6) and no-dieting ( N = 4) than the pro-dieting ( N = 0) conditions; however, no differences emerged by condition for attrition rates, 2 (2) = 2.11, p > .05, and rates were roughly equal among the pro-dieting ( N = 2), anti-dieting ( N = 4), and no-dieting ( N = 1) groups. No significant differences were found for exclusion status by race (collapsed into Caucasian vs Non-Caucasian because of small Ns in most cells), 2 (1) = 2.45, p > .05, or age, t (17) = -1.5, p > .05, and BMI, t (16) = -.72, p > .05, after adjusting for the significant inequality of variances based on Levenes test.
51 One-way ANOVAs confirmed there were no significant differences among conditions on age, F (2, 122) = 2.09, p > .05, BMI, F (2, 122) =.29, p > .05, or year in school, F (2, 122) = .78, p > .05. Additionally, no significant difference was found among conditions for race, 2 (8) = 9.06, p > .05. Separate one-w ay ANOVAs on each pre-test trait and state variable re vealed no significant differences among the conditions. Collectively, these findings suggest that random assignment was successful. The Message Rating Form (MRF) items we re analyzed separately in one-way ANOVAs to test for differences in non-specifi c perceptions of the three messages. The MRF items assessed the extent to which par ticipants rated the me ssages as convincing, effective, applicable to themselves, easy to understand, credible, and influential. A significant difference was found among the conditions for the applicable item, F (2, 123) = 5.35, p < .05, with the modified Bonferroni pos t-hoc test revealing that the anti-dieting, eating disorder prevention message wa s perceived as less applicable ( M = 2.57) than the pro-dieting, obesity prevention ( M = 3.26) and no-dieting, flu prevention ( M = 3.53) messages. The messages were found to be equivalent on all other MRF items. Examination of mean values (see Table 2) shows that means ranged between 3 and 4, suggesting that the messages were generally perceived positively. Overall, findings indicate that the messages appear to be e quated on the non-specif ic factors with the exception of the lowered applicability of the anti-dieting, eating disorder prevention message.
52 Table 2 Means and Standard Deviations for Message Rating Form Items by Condition Pro-Dieting Anti-Dieting No-Dieting Convincing 4.02 (1.09)a 3.78 (.98)a 4.10 (.81)a Effective 3.80 (1.03)a 3.62 (.9)a 4.05 (1.01)a Applicable 3.26 (1.56)a 2.57 (1.26)b 3.53 (1.06)a Easy to Understand 4.48 (.75)a 4.46 (.99)a 4.65 (.53)a Credible 3.61 (1.09)a 3.54 (.90)a 3.68 (1.0)a Influential 3.13 (1.36)a 3.11 (1.08)a 3.53 (.78)a Note Letter subscripts indicate signifi cant differences across conditions. Correlations among the pre-te st trait and state measures were examined (see Table 3 ) The correlation between the pre-test VAS weight dissatisf action and body shape dissatisfaction items was very high ( r = .91), suggesting the two items were not independent. Therefore, the two items we re collapsed, and a state body dissatisfaction composite score was created for both pre-test and post-test. The VAS composite state body dissatisfaction scores were used in all subsequent analyses. As Table 3 illustrates, many of the pre-test trait vari ables were significantly correla ted (magnitude of rs ranging from .02 to .78), which was expected based on previous research which has suggested these are theoretically related yet distinct constructs (i.e., body dissatisfaction is associated with dieting yet is a distinct be havior). Healthy eati ng and exercise, however, were not correlated with severa l of the pre-test vari ables. Because all trait measures were not correlated, it appeared to be most appr opriate to proceed with separate univariate ANOVAs as planned rather than co nducting multivariate analyses.
53 The correlations between baseline and pos t-test scores (pos t-test perceived pressure, state body dissatisfaction and negative affect, and intentions) were examined for each variable. All of the correlations were st atistically significant w ith an alpha level of less than .01 (rs ranging from .35-.85). B ecause of these high correlations between baseline and post-test scores, the baseline scores were used as covariates in subsequent analyses to reduce within-group error variance and increase the power to detect the effect of the independent variable (Field, 2000). Table 3 Correlations Among Pre-Test Measures Note. BD: Composite Body Dissatisfaction; PANAS: Positive and Nega tive Affect Scale; MHBI: Multidimensional Health Behavior Inventory; EDI-BD: Eating Disorder Inventor y-Body Dissatisfaction subscale; EDI-DT: Ea ting Disorder Inventory-Drive for Thinnes s subscale; SATAQ: Sociocultural Attitudes Towards Appearance Scale; EDEQ: Eating Di sorder Examination Questionnaire-Bulimia subscale; DEBQ: Dutch Eating Behavior Questionnaire-Restraint scale p<.05 ** p<.01 State BD State PANAS Trait PANAS MHBIExercise MHBIHealthy Eating EDI-BD EDI-DT SATAQPressures SATAQGeneral SATAQAthlete EDEQ DEBQ State BD 1 State PANAS .49** 1 Trait PANAS .51** .86** 1 MHBI-Exercise -.02 -.21* -.16 1 MHBI-Health Eating .09 -.11 -.04 .45** 1 EDI-BD .73** .36** .45** .02 .14 1 EDI-DT .64** .30** .40** .25** .30** .66** 1 SATAQ-Pressures .40** .19* .28** .22* .20* .43** .49** 1 SATAQ-General .36** .28** .30** .19* .13 .30** .37** .74** 1 SATAQ-Athlete .30** .14 .18* .38** .17 .22* .29** .48** .57** 1 EDEQ .75** .39** .48** .10 .09 .62** .65** .44** .37** .39** 1 DEBQ .48** .16 .29** .30** .47** .47** .78** .38** .20* .21* .51** 1
54 Planned ANOVA and ANCOVA Analyses Univariate ANCOVAs. Separate one-way ANCOVAs were computed to examine post-test differences in perc eived pressure and weight co ntrol intentions (Hypotheses 1 and 2). As Table 4 illustrates, a significan t main effect was found for perceived pressure, F (2, 123) = 79.99, p < .0001, partial 2 = .57. The modified Bonferroni revealed significant differences among all three conditions with the pro-dieting message (adjusted M = 12.19) yielding greater perc eived pressure than the an ti-dieting message (adjusted M = 6.33), which was greater than the no-dieting message (adjusted M = 4.7). A significant main effect was also found for dieting intentions, F (2, 121) = 13.64, p < .0001, partial 2 = .19. The pro-dieting condition (adjusted M = 31.75) had significantly greater intentions to diet at post-test than th e no-dieting condition (adjusted M = 29.20), which was significantly greater than the anti-dieti ng condition (adjusted M = 25.13). There was also a significant main effect fo r internalization intentions, F (2, 123) = 36.96, p < .0001, partial 2 = .38, with the pro-dieting condition having significantly greater internalization intentions (adjusted M = 11.61) than the anti-dieting (adjusted M = 7.09), which was significantly greater than the no-dieting condition (adjusted M = 5.54). A significant main effect was found for bulimic intentions, F (2, 122) = 3.22, p < .05, partial 2 = .05. The modified Bonferroni procedure revealed a significant difference between the antidieting (adjusted M = 7.56) and the no-dieting (adjusted M = 8.97). There was no difference between the pro-dieting message (adjusted M = 8.77) and the other two messages. The ANCOVA analyses revealed no significant differenc es by condition for healthy eating intentions or exercise intentions.
55 To assess differences by condition for post-test state measures (body dissatisfaction and negative affect), separa te ANCOVAs were computed with baseline and pre-test state measures entered as cova riates (Hypothesis 2). The hypotheses for state differences at post-test were not supported (see Table 4). There was not a significant main effect for post-test state body dissatisfac tion when baseline and pre-test scores were controlled. Repeated measures ANOVAs. To test Hypothesis 3, separate mixed design, repeated measures ANOVAs were computed to evaluate any changes in psychological functioning and weight control behaviors from pre-test to follow-up (see Table 4). For the psychological functioning variables, there was a significant time by condition interaction for the SATAQ-3 Pressures subscale, F (2, 120) = 4.46, p < .01, partial 2 = .07, with significant decreases in percei ved pressure found for the anti-dieting ( M1 = 18.5; M2 = 16.84) and no-dieting ( M1 = 19.13; M2 = 15.3) conditions, but no changes across time found for the pro-dieting condition ( M1 = 20.3; M2 = 19.15). A significant time by condition interaction was also found for the SATAQ-3 Internalizatio n-Athlete subscale, F (2, 120) = 3.91, p < .05, partial 2 = .06; the no-dieting conditi on exhibited significant decreases from pre-test to follow-up ( M1 = 15.25; M2 = 13.65). A marginally significant decrease in SATAQ-3 Internalization-Athlete scores was also seen in the anti-dieting condition ( M1 = 16.32; M2 = 15.38) while no changes were seen in the pro-dieting condition ( M1 = 16.72; M2 = 16.87). A main effect for time was found for drive for thinness, F (1, 118) = 11.44, p < .001, partial 2 = .09, with drive for thinness decreasing from pre-test ( M = 21.43) to follow-up ( M = 20.05). The time by condition interaction for drive for thinness was non-significant. A main effect over time was also found for
56 negative affect, F (1, 118) = 7.92, p < .01, partial 2 = .06, with negative affect scores decreasing from pre-test to follow-up across conditions. The time by condition interaction for negative affect was non-si gnificant. No significant time or time by condition effects were found for body dissatisfacti on or general thin-ide al internalization. For the weight control outcome vari ables, a significant time by condition interaction was found for healthy eating, F (2, 118) = 7.97, p < .001, partial 2 = .12. The modified Bonferroni procedure revealed significant increases in healthy eating in the prodieting condition ( M1 = 38.5; M2 = 42.17), but no changes in either the anti-dieting ( M1 = 40.0; M2 = 39.43) or no-dieting condition ( M1 = 40.48; M2 = 41.23). A significant main effect for dieting across time was found, F (1, 118) = 12.36, p < .001, partial 2 = .10, with dieting behaviors decrea sing from pre-test ( M = 25.85) to follow-up ( M = 23.88). The time by condition interaction for dieting was non-significant. There were no significant effects for bulimic symptoms or exercise behaviors across time or time by condition.
57 Table 4. Means, standard deviations, F, p, and partial 2 values for planned ANOVAs Univariate ANCOVAS Pro-Dieting (Adjusted means & SE) Anti-Dieting (Adjusted means & SE) No-Dieting (Adjusted means & SE) F, p, partial 2 values Perceived Pressure 12.19 (.43)a 6.33 (.48)b 4.7 (.46)c F(2,123)=79.99, p<.0001, partial 2=.57 Dieting Intentions 31.75 (.85)a 25.13 (.95)b 29.20 (.93)c F(2,121)=13.64, p<.0001, partial 2=.19 Internalization Intentions 11.61 (.50)a 7.09 (.56)b 5.54 (.54)c F(2,123)=36.96, p<.0001, partial 2=.38 Bulimic Intentions 8.77 (.40) 7.56 (.44)a 8.97 (.42)b F(2,122)=3.22, p<.05, partial 2=.05 Healthy Eating Intentions 48.00 (.93)a 44.97 (1.05)a 46.3 (1.0)a F(2,123)=2.37, p=.10 Exercise Intentions 15.11 (.47)a 14.9 (.53)a 14.97 (.51)a F(2,123)=.05, p>.05 State Body Dissatisfaction 93.52 (4.6)a 81.64 (5.21)a 81.81 (4.95)a F(2,122)=2.05, p>.05 State Negative Affect 32.59 (.69)a 30.56 (.76)a 31.19 (.73)a F(2,121)=2.12, p>.05 Repeated Measures ANOVAs Baseline (M & SD) Follow-Up (M & SD) Baseline (M & SD) Follow-Up (M & SD) Baseline (M & SD) Follow-Up (M & SD) 20.30 (7.11)1 18.49 (6.96)1 19.13 (6.01)1 Perceived Pressures 19.15 (7.62)1 16.84 (6.27)2 15.30 (5.95)2 T: F(1,120)=31.71, p<.001, partial 2=.21 C: F(2,120)=1.93, p>.05 TxC: F(2,120)=4.46, p<.05, partial 2=.07 16.72 (4.88)1 16.32 (3.99)1 15.25 (4.79)1 Internalization-Athlete 16.87 (5.44)1 15.38 (5.07)1 13.65 (5.15)2 T: F(1,120)=8.97, p<.01, partial 2=.07 C: F(2,120)=2.67, p>.05 TxC: F(2,120)=3.91, p<.05, partial 2=.06 22.49 (8.77) 21.49 (7.49) 20.15 (9.63) Drive for Thinness 21.91 (8.73) 1922 (8.17) 18.69 (8.9) T: F(1,118)=11.44, p<.001, partial 2=.09 C: F(2,118)=1.22, p>.05 TxC: F(2,118)=1.35, p>.05 38.89 (14.36) 40.24 (15.77) 35.03 (16.43) Negative Affect 36.20 (13.62) 39.30 (20.13) 29.95 (10.67) T: F(1,118)=7.92, p<.01, partial 2=.06 C: F(2,118)=2.69, p>.05 TxC: F(2,118)=13, p>.05 32.48 (10.62) 31.31 (10.56) 30.83 (10.94) Body Dissatisfaction 32.46 (11.69) 30.89 (9.21) 30.30 (10.52) T: F(1,119)=.42, p>.05 C: F(2,119)=.39, p>.05 TxC: F(2,119)=10, p>.05 28.93 (10.17) 2692 (8.29) 27.25 (9.13) Internalization-General 30.74 (12.19) 2792 (9.25) 26.6 (10.65) T: F(1,120)=2.16, p>.05 C: F(2,120)=1.11, p>.05 TxC: F(2,120)=2.27, p>.05 38.5 (8.0)1 40.0 (8.3)1 40.48 (7.46)1 Healthy Eating 42.17 (8.83)2 39.43 (9.37)1 41.23 (9.15)1 T: F(1,118)=8.05, p<.01, partial 2=.06 C: F(2,118)=.18, p>.05 TxC: F(2,118)=7.97, p<.001, partial 2=.12 26.67 (8.95) 25.49 (8.85) 25.41 (9.17) Dieting 25.71 (10.28) 22.68 (9.67) 23.26 (10.63) T: F(1,118)=12.36, p<.001, partial 2=.10 C: F(2,118)=.67, p>.05 TxC: F(2,118)=97, p>.05 15.49 (10.79) 1422 (9.15) 11.85 (10.25) Bulimic Symptoms 15.27 (9.93) 13.03 (8.15) 11.03 (9.9) T: F(1,119)=2.05, p>.05 C: F(2,119)=1.88, p>.05 TxC: F(2,119)=0.3, p>.05 12.87 (3.86) 12.65 (3.94) 12.15 (4.59) Exercise 12.41 (4.03) 1259 (3.88) 12.05 (4.72) T: F(1,120)=.89, p>.05 C: F(2,120)=.24, p>.05 TxC: F(2,120)=37, p>.05 Note. T: Time main effect; C: Condition main effect; TxC: Time by Condition interaction; Letter subscripts indicate significant differences across conditions; Number subscripts denote significant differences across time.
58 Mediation Analyses It was predicted that the e ffect of the dieting message on weight control intentions and behaviors would be mediat ed by post-test perceived pres sure to lose weight. The Baron and Kenny (1986) procedure for tes ting mediation was followed, directly comparing the pro-dieting and anti-dieting me ssages in separate regression analyses for each weight control intention and behavior (i .e., healthy eating, dieting, exercise). Figure 1 depicts the general mediation model. In each model, the dire ct path from dieting message to the weight control intention/behavi or outcome variable wa s tested (path c). The indirect paths from dieting message to perceived pressure (path a) and perceived pressure to the outcome variable (path b) were then tested. Baseline scores for each outcome variable were used as covariates in all regression analyses (Kenny, 2006). Paths a, b, and c must be significant to meet Baron and Kennys (1986) preconditions for mediation. If the preconditions were met for a model, then the path coefficient c was examined after the introduction of perceived pressure into the re gression equation. Full mediation occurs when c is no longer sign ificant after th e introduction of the meditating variable. Sobels test was computed for all m odels in which there was a reduction in c to test the significance of the mediation effect.
59 Figure 1. General mediation model. Table 5 contains the correlation matrix for the variables and conditions (prodieting and anti-dieting) exam ined in the mediation analyses. Baron and Kennys (1986) preconditions for mediation (significant a, b, and c paths) were met for four out of the five intention variables: diet ing intentions, bulimic intentions, internalization intentions, and healthy eating intentions. Only exercise intentions did not meet the preconditions for mediation. Additionally, past two-week h ealthy eating behaviors assessed at follow-up met the preconditions for mediation; howev er, no other follow-up behavior met the necessary criteria with path c being non-significant in each ca se. For each of the five models that met the preconditions, path c became non-significant when perceived pressure was controlled. The Sobel test wa s significant for each model, suggesting the presence of full mediation. Table 6 contains the unstandardized path coefficients, standard errors, and Sobel test z-values for each of the five mediated models. b (Sb) a (Sa) c(Sc) c (Sc) z (Sz) Perceived Pressure Dieting Message ( Pro vs. Anti ) Weight Control Intentions/Behaviors Baseline Measure y (Sy)
60 Mediation implies a causal chain with the e ffects of the independent variable and mediator causing the change in the outcome variable; however, for non-manipulated variables, causality cannot always be a ssumed (Baron & Kenny, 1986; Shrout & Bolger, 2002). Because the dieting message was e xperimentally manipulated in this study, causality can be inferred for the dieting message on perceived pressure and weight control intentions and behaviors; however, the mediator and outcome variables were not experimentally manipulated, and the causal re lationship between perceived pressures and the outcome variables cannot necessarily be assumed (Shrout & Bolger, 2002). In such cases, Kenny (2006) recommends examining th e theoretical plausi bility of reverse causality (i.e., the outcome variable cau sing the mediator) as well as any design considerations that may weaken the possibility of it. If it is plausible that the outcome variable may have caused the mediator, it is of ten useful to intercha nge the mediator and outcome variable in the regression equations and compare the paths to the original model (Kenny, 2006). If the b and c paths are similar to those in the original model, the causal hypothesis cannot be supported. In the current study, the temporal distan ce between perceived pressure and past two-week weight control be haviors renders the reverse causality hypothesis impossible for the follow-up healthy eating mediation model. Furthermore, it is unlikely that weight control intentions caused perc eived pressure to lose wei ght from both a theoretical and design standpoint because the former was asse ssed after the la tter; however, because both variables were measured within a short-time span, the mediator and intention variables were interchanged in the regression equations to test the reverse cau sality hypothesis. All c paths remained signifi cant in these analyses, providing support for the original
61 models of mediation and causa lity hypothesis that percei ved pressure mediates the relationship between dieting message and the weight control intentions and behaviors (i.e., dieting intentions, bulimic intentions internalization inte ntions, healthy eating intentions, and healthy eating behaviors) rather than th e weight control variables mediating the relationship between dieti ng message and perceived pressure.
62 Table 5 Correlations among Pre-Test Measures Note. PP: Post-Test Perceived Pressure; DEBQ: Dutc h Eating Behavior Questionnaire-Restraint scale; EDEQ: Eating Disorder Examination Questionnaire-Bulimia subscale; SATAQ: Sociocultural Attitudes Towards Physical Appear ance Questionnaire; MHBI: Multidimensional Health Behavior Inve ntory; FU: Follow-Up; EDI-BD: Eating Disorder Inventory-Body Dissatisfaction subs cale; EDI-DT: Eating Disorder Inventory-Drive for Thinness p<.05, **p<.01 PP DEBQ Intent EDEQ Intent SATAQ Intent MHBI Eating Intent MHBI Exercise Intent FU DEBQ FU EDI-BD FU EDI-DT FU SATAQ Pressure FU SATAQ General FU Athlete FU EDEQ FU Eating FU Exercise Diet Message Condition PP 1 DEBQ Intent .46** 1 EDEQ Intent .40** .60** 1 SATAQ Intent .75** .44** .38** 1 MHBI Eating Intentions .19 .54** .26* .22* 1 MHBI Exer cise Intentions .27** .50** .24* .37** .69** 1 FU DEBQ .33** .85** .62** .33** .56** .44** 1 FU EDI-BD .40** .53** .41** .27* .32** .32** .48** 1 FU EDI-DT .43** .77** .61** .43** .51** .44** .80** .61** 1 FU Pressure .52** .43** .33** .55** .22* .28** .37** .40** .50** 1 FU SATAQ General Internalization .42** .27* .30** .57** .06 .19 .25* .23* .39** .78** 1 FU SATAQ Athlete Internalization .34** .23* .18 .49** .11 .31** .20 .14 .24* .51** .63** 1 FU EDEQ .38** .59** .47** .39** .26* .25* .61** .59** .73** .43** .33** .20 1 FU MHBI Eating .20 .44** .20 .18 .77** .50** .53** .16 .33** .09 -.04 .13 .10 1 FU MHBI Exercise .12 .26* .09 .22* .40** .56** .34** .01 .21* .10 .12 .28** .05 .46** 1 Diet Message Cond .69** .31** .20 .51** .11 .10 .14 .03 .17 .21 .15 .15 .15 .16 .02 1
63 Table 6. Unstandardized Path Coefficients, Standard E rrors, and Significance Tests for Mediation Analyses Path Label Path Coefficent (Standard Errors) Dieting Intentions Bulimic Intentions Internalization Intentions Healthy Eating Intentions Healthy Eating Behaviors at Follow-Up Dieting Message to Perceived Pressure a (Sa) 6.07 (.77)* 6.04 (.73)* 5.87 (.71)* 6.12 (.80)* 6.12 (.80)* Perceived Pressure to Weight Control Intentions/Behavior b (Sb) .66 (.18)* .23 (.09)* .64 (.11)* .54 (.19)* .36 (.16)* Dieting Message to Weight Control Intentions/Behavior c (Sc) 6.67 (131)* 1.26 (.59)* 4.37 (.82)* 3.05 (1.41)* 4.13 (1.15)* Dieting Message to Weight Control Intentions/Behaviors (controlling for path b) c(Sc) 2.64 (1.62) -.12 (.78) .61 (.93) -.29 (1.78) 1.93 (1.48) Baseline variable to Perceived Pressure y (Sy) .09 (.04)* .14 (.04)* .17 (.04)* -.02 (.05) -.02 (.05) Baseline variable to Weight Control Intentions/Behaviors z (Sz) 1.07 (.07)* .12 (.03)* .24 (.04)* .87 (.09)* .92 (.07)* Correlation between Baseline Variable and Diet Message .08 .07 .11 -.08 -.09 Sobel Test (z-value) 3.32* 2.43* 4.75* 2.66* 2.16* Note. Baseline trait variables were controlled in each model. p<.05 Exploratory Analyses Exploratory moderator analyses Trait level thin-ideal internalization was explored as a possible modera tor of outcome based on prev ious research supporting its causal risk factor status for eating pat hology (Stice, 2002; Thompson & Stice, 2001). Following the procedures of Baron and Kenny (1986), baseline levels of the SATAQ-3 General Internalization subscale was tested as a moderator of post-test perceived pressure and weight control intentions and behaviors. Figure 2 illustrates the general moderation model. Accordingly, regression analyses eval uated the diet message as a predictor (path
64 a), pre-test thin-ideal internalization as a moderator (path b), and the interaction of the product of the predictor and the moderator (p ath c). Baseline scores for the outcome variable were entered as c ovariates (paths w, x, y, z). The moderator hypothesis is supported when the interaction term (path c) is significant (Baron & Kenny, 1986). Figure 2. General moderation model. Initial considerations for the moderator analyses include examining the temporal relationship of the variables as well as es tablishing independence of the predictor and mediator variables. The SATAQ-3 Internaliz ation subscale is a trait measure that was measured at baseline prior to the experiment al manipulation, which is ideal in moderator analyses (Kenny, 2006; Kraemer et al., 2002). Because the predictor (diet message) is randomized, there should be no relationship between the predictor and the mediator (internalization). The point-biserial co rrelation between thes e two variables ( rpb = .1, p > .05) confirms that the predic tor and moderator are indeed independent. Correlations between the hypothesized modera tor and the outcome variables should also be ideally Predictor (Diet Message) Moderator (Internalization) Predictor X Moderator Interaction Outcome (Perceived Pressure, Weight Control Intentions & Behaviors) a b c Baseline Measure w x y z
65 uncorrelated to provide a clearly interp retable interaction term (Baron & Kenny, 1986, p.1174); however, as Table 7 illustrates, the SATAQ-3 trait internalization measure was significantly correlated with al l of the outcome variables except follow-up healthy eating behaviors (rs ranging fr om .14 .87). Although it is desirable that the moderator and outcome variables be uncorrelated, Baron and Kenny (1986) do not state that it is prerequisite that they be uncorrelated in order to conduc t the moderator analyses. Therefore, it appeared to be appropria te to proceed with the analyses. Table 7. Correlations Among Trait Internalization, Diet Message, and Outcome Variables Note. DEBQ: Dutch Eating Behavior Questionnaire-Restraint scale; EDEQ: Eating Disorder Examination Ques tionnaire-Bulimia subscale; SATAQ: Sociocultural Attitudes Towards Physic al Appearance Questionnaire; MHBI: Multidimensional Health Behavior Inventory; EDI-BD: Eating Disorder Inventory-Body Dissatisfaction subscale; EDI-DT: Eating Disorder InventoryDrive for Thinness; PANAS: Positive and Negative Affect Scale p<.05 **p<.01 Separate regression equations were computed using the baseline measure as a covariate for each exploratory model. As Table 8 illustrates, the moderator hypothesis SATAQ-3 General Internalization Diet Message (pro vs. anti) .10 Post-Test Perceived Pressure .30** DEBQ Intentions .32** EDEQ Intentions .28** Internalization Intentions .38** MHBI Eating Intentions .19* MHBI Exercise Intentions .20* Follow-Up EDI-BD .36** Follow-Up EDI-DT .39** Follow-Up SATAQ Pressures .70** Follow-Up SATAQ Internalization .87** Follow-Up EDEQ .35** Follow-Up MHBI Eating .14 Follow-Up MHBI Exercise .18* FU PANAS .25**
66 was supported for two of them: post-test perceived pressure and internalization intentions. For the post-test pr essure model, the interaction term (path c) was significant, standardized = .64, p < .01, when controlling base line SATAQ-3 Pressures, diet message, and baseline SATAQ-3 General Interna lization. The R2 significantly increased from .21 with only the covariate entered into the regression to .62 for the entire model. Additionally, the interaction term (path c) fo r the internalization intentions model was significant, standardized = .75, p < .01, when the pred ictor and moderator were controlled. The R2 for the entire model was .51. All re maining models resulted in nonsignificant interaction terms. Therefore, the findings s uggest that trait thin-ideal internalization moderated the relationship be tween diet message and post-test perceived pressure to lose weight as well as diet messa ge and post-test interna lization intentions. Table 8. Standardized Beta Weights and R2 Values for Moderator Analyses Baseline Variable Diet Message Baseline Internalization Interaction R2 Post-Test Pressure .3** .01 .05 .64** .62 Internalization Intentions N/A -.26 .39** .75** .51 Dieting Intentions .8** .31 .22** -.06 .80 Bulimic Intentions .39** .41 .13 -.23 .24 Exercise Intentions .49** -.38 .13 .42 .33 Healthy Eating Intentions .74** -.1 .02 .28 .57 Healthy Eating Behavior .83** .28 -.02 -.05 .70 Exercise Behavior .78** -.09 .02 .04 .62 Dieting Behavior .80** .32 .18* -.26 .68 Bulimic Behavior .80** .23 .06 -.17 .65 Note. p < .05 ** p < .01
67 Exploratory weight status analyses. Exploratory analyses were also conducted to examine the effects of participant weight status by condition on the non-specific perceptions of the psychoeducational messages (MRF items), post-test perceived pressure and behavioral intentions, and pre-test to follow-up change in psychological functioning and weight control behaviors. The particip ant weight status variable was developed using self-reported BMI and collapsing particip ants into either Not Overweight (BMI < 25) or Overweight (BMI 25). 69.1% ( N = 85) of participants were categorized as Not Overweight while 30.9% ( N = 38) fell into the Overweight category. Cell sizes, denoted as Not Overweight ( N1) and Overweight participants ( N2), were as follows: Pro-dieting condition ( N1 = 32, N2 = 14), Anti-dieting condition ( N1 = 27, N2 = 10), and No-dieting condition ( N1 = 25, N2 = 13). All exploratory analyses were identical to those conducted in the planned analyses section with weight status (Not Overweight/Overweight) added as a between subjects factor. On the MRF items, a significant conditi on by weight status interaction was found for the MRF applicable item, F (2, 123) = 3.43, p < .05, partial 2 = .06. Post-hoc tests revealed that overweight participants in the pro-dieting condition found the message significantly more applicable to them ( M = 4.43) than did those who were not overweight ( M = 2.75); no differences in applicability of the message by weight status were found for the anti-dieting or no-dieting conditions. Th ere was a marginally significant condition by weight status interaction for the MRF influential item, F (2, 123) = 2.9, p = .06, partial 2 = .05. The mean trends suggest that overweight participants in bot h the pro-dieting and anti-dieting message conditions found the messages more in fluential than non-overweight participants (Pro-dieting: Mo = 3.93, Mno = 2.78; Anti-dieting: Mo = 2.96, Mno = 3.5). A
68 main effect for weight status was found on the MRF credible item, F (1, 123) = 4.68, p < .05, partial 2 = .04, with overweight participants fi nding the messages more credible ( M = 3.91) than the not overweight participants ( M = 3.48); the main effect for condition and the weight status by condition interaction were not significan t for the credible item. There was a marginally significant main effect for weight status on the convincing item, F (1, 123) = 3.51, p = .06, with non-significant m ean trends suggesting that overweight participants ( M = 4.21) found the messages more convincing than non-overweight participants ( M = 3.86). The main effect for conditi on and the condition by weight status interaction were non-significant for the MRF c onvincing item. There were no significant main effects or interactions for the MR F effective or easy to understand items. For the post-test perceived pressure scale, the ANCOVA revealed a significant condition by weight status interaction, F (2, 123) = 11.14, p < .0001, partial 2 = .16. Figure 3 illustrates the covariate adjusted mean values by condition and weight status. Post-hoc tests indicated that overweight participants in the pro-dieting condition perceived greater pressure to lose weight at post-test (adjusted M = 15.29) than the nonoverweight participants in the same condition (adjusted M = 10.87). Furthermore, nonoverweight participants in the pro-dieting condition perceived significantly greater pressure to lose weight than both overweight and non-overweight participants in the other two conditions.
69 0 5 10 15 20 123 ConditionPerceived Pressure Not Overweight Overweight Figure 3 Covariate adjusted means for perceived pressure. ANCOVA analyses on the state measures revealed a sign ificant condition by weight status interaction for post-test state body dissatisfaction, F (2, 122) = 4.12, p < .05, partial 2 = .07. As Figure 4 illustrates, overwei ght individuals reported higher state body dissatisfaction in the prodieting condition (adjusted M = 109.15) than not overweight participants in that condition (adjusted M = 86.92). There were no significant differences in post-test state body dissatisfacti on between overweight and non-overweight individuals in the anti-dieting and no-d ieting conditions. Additionally, state body dissatisfaction for overweight participants in the pro-dieting condi tion was significantly higher than for the overweight individual s in the no-dieting condition (adjusted M = 71.55); however, there was not a significant di fference in state body dissatisfaction for overweight individuals when co mparing the pro-dieting vers us the anti-dieting or the anti-dieting versus the no-d ieting conditions. There were no differences in state body dissatisfaction for non-overweight participants among the conditions.
70 0 20 40 60 80 100 120 123 ConditionState Body Dissatisfaction Not Overweight Overweight Figure 4. Covariate adjusted means for state body dissatisfaction. The ANCOVA for state negative affect yi elded a marginally significant main effect for condition, F (2, 121) = 3.02, p = .05, partial 2 = .05. Mean trends suggest that the pro-dieting message (adjusted M = 33.25) elicited greater nega tive affect at post-test than the anti-dieting message (adjusted M = 30.71) and the no-dieting message (adjusted M = 31.19). The main effect for weight st atus and the condition by weight status interaction were non-significant. For the post-test weight control intention variables, a significant main effect for weight status was found for healthy eating intentions, F (1, 122) = 6.67, p < .01, partial 2 = .06, with overweight participants reporting greater healthy eating intentions (adjusted M = 48.71) than non-overweight individuals (adjusted M = 45.51). Consistent with the main analyses, the main effect for conditi on as well as the condition by weight status interaction was non-significant for healthy eati ng intentions. A significant main effect for weight status was also found for exercise intentions, F (1, 123) = 12.71, p < .001, partial 2 = .10. Overweight individuals (adjusted M = 16.52) reported significantly
71 greater exercise intentions than t hose who were not overweight (adjusted M = 14.34). Consistent with the main analyses, the condi tion main effect remained non-significant for exercise intentions, and the condition by wei ght status interaction was non-significant. For bulimic intentions, the main eff ect for condition remained significant, F (2, 122) = 3.48, p < .05, partial 2 = .06; however, post-hoc tests reve aled a different mean trend when weight status was added as a between subjects factor. The pro-dieting condition (adjusted M = 9.05) yielded greater bulimic inten tions than the anti-dieting condition (adjusted M = 7.48), but no other post-hoc differences were found.1 No new findings were found when adding weight status as a between-subjects factor for internalization intentions or dieting intentions. The main effects for condition remained significant in the same directions reported in the main study analyses. For healthy eating behaviors, the time, F (1, 115) = 11.71, p < .001, partial 2 = .09, and time by condition, F (2, 115) = 8.7, p < .001, partial 2 = .13, effects remained significant in the directions reported in the main study analyses. A significant time by weight status interaction was also found for healthy eating behaviors, F (1, 115) = 4.26, p < .05, partial 2 = .04. Post-hoc tests re vealed significant increases in healthy eating across time in overweight individuals ( M1 = 40.22; M2 = 42.75) with no change in the healthy eating behaviors of non-overweight participants ( M1 = 39.56; M2 = 40.07). The time by condition by weight status interaction for healthy ea ting was non-significant. No new findings were yielded in the repeated measures ANOVAs when adding weight status as a between subjects factor for exercise be haviors, dieting, bulimic symptoms or any of 1 In the main study analyses, the no-dieting condition re ported significantly greater bulimic intentions than the anti-dieting condition; however, the pro-dieting cond ition was not significantly different from either of the other conditions.
72 the psychological functioning variables: negative affect, body dissatisfaction, drive for thinness, perceived pressure, and thin-ideal internalization (general or athlete).
73 Chapter 4 Discussion The purpose of the current study was to examine the immediate and short-term effects of dieting-related psychoeducati onal messages on psychological functioning and weight control intentions and behaviors. It was hypothesized that the pro-dieting message would produce greater pos t-test perceived pressure to lose weight and weight control intentions and behaviors as well as greater state body dissatisfaction than the antidieting and no-dieting message conditions. It was also hypothesized that post-test perceived pressure to be thin would medi ate the relationship between dieting message and weight control intentions and behaviors. The experimental nature of the current study also allowed for the examination of the ri sk factor status of the perceived pressure construct. Specifically, it was hypothesized that perceived pressure would meet Kraemer et al.s (1997) criteria as a causal risk f actor for bulimic symptomatology. Exploratory analyses were also conducted in order to exam ine the role of partic ipant weight status on the outcome variables as well as to assess the extent to which trait thin-ideal internalization modera ted the findings. Several of the hypotheses were fully or partially supported. A large effect size (partial 2 = .57) was found for post-test differences in perceived pressure. Differences were in the predicted direc tion with the pro-dieting messa ge yielding greater post-test perceived pressure than the anti-dieting me ssage, which was greater than the no-dieting
74 message. This finding replicates the pilot st udy, and taken together, the results from the two studies suggest a robust difference among the dieting messages on post-test perceived pressure to lose weight. Furt hermore, the current study found that the prodieting message elicited greater dieting and internalization inte ntions at post-test than the anti-dieting and no-dieting messa ge conditions. Bulimic inte ntions were significantly lower in the anti-dieting, eating disorder prevention message than the other two conditions. The hypotheses were not supported, however, for post-test differences for the variables of healthy eating and exercise intentions or state body dissatisfaction. The primary findings from the repeated measures analyses suggest there were significant increases in healt hy eating behaviors from base line to follow-up in the prodieting condition but no change s in the other two conditi ons. Perceived pressure decreased significantly from baseline to fo llow-up in the anti-dieting and no-dieting conditions with no changes in the pro-dieting condition. Similarly, internalization-athlete scores decreased significantly in the no-dieting condition and decreased marginally in the anti-dieting condition while there were no changes in the pro-dieting condition. There were decreases from baseline to followup in drive for thinness, negative affect, and dieting in all conditions. No cha nges were found for body dissatisfaction, general thin-ideal internalization, bulimic symptoms, or exercise behavior. The two active, dieting-related psyc hoeducational messages were directly compared in mediator and moderator an alyses, and the hypotheses were partially supported. Post-test perceive d pressure was found to fully mediate the relationship between the diet message (pro vs. anti) a nd dieting intentions, bulimic intentions, internalization intentions, healthy eati ng intentions, and fo llow-up healthy eating
75 behaviors. The plausibility of reverse cau sation was examined and does not appear to account for the significant findings. The me diator hypotheses were not supported for exercise intentions, bulimic behaviors, die ting behaviors, or exer cise behaviors. Exploratory moderational analys es were also conducted usin g trait level of thin-ideal internalization as a potential moderator of posttest perceived pressure and weight control intentions and behaviors. Thin-ideal inte rnalization was found to moderate post-test perceived pressure and interna lization intentions. No other weight control intentions or behaviors were found to be moderated by baseline internalization levels. Weight status of the partic ipant was also examined as a between subjects factor in exploratory analyses that are intriguing, yet limited due to small sample sizes within each cell. Overweight participants in the prodieting condition rated the psychoeducational message as more applicable to themselves than non-overweight indi viduals in the same condition. Similarly, a marginally significant interaction trend sugge sts that overweight individuals in the pro-dieti ng and anti-dieting conditions pe rceived the psychoeducational message as more influential to them than t hose who were not overwei ght. Main effects for weight status suggest the messages were perceived as more credible and convincing in overweight compared to non-overweight participants. Exploratory analyses also revealed im portant differences among the conditions by weight status on several post-te st and follow-up variables. A significant weight status by condition interaction revealed overweight pa rticipants in the pro-dieting condition perceived the most pressure to lose weight at post-test. Interestingly, non-overweight participants in the pro-dieti ng condition reported significantly greater perceived pressure than both overweight and non-overweight i ndividuals in the other two conditions.
76 Similarly, those who were overweight in th e pro-dieting condition reported significantly greater state body dissatisfaction at post-test than non-overweight individuals in the same condition; they also were more dissatisfied at post-test than overweight individuals in the no-dieting, control condition. Si gnificant main effects for we ight status revealed that overweight individuals reporte d significantly greater h ealthy eating and exercise intentions than those who were not overwei ght. Additionally, signi ficant increases in healthy eating behavior were found in overweight compared to non-overweight participants. A final goal of the study was to evaluate the risk factor stat us of the perceived pressure construct according to Kraemer et al .s (1997) criteria. Pr evious research has identified perceived pressure as a risk f actor for eating pathology in women; however, experimental research is nece ssary to determine whether the construct meets Kraemer et al.s (1997) definition of a causal risk f actor. The current study found significant differences among the three diet message cond itions on perceived pressure at post-test, allowing for the interpretation of the cons tructs effect on bulim ic symptomatology. Significant differences in bulimic intentions were found at post-test; however, the main study analyses did not find higher bulimic intent ions in the pro-dieti ng (greater perceived pressure) condition as hypothesized. Rath er, the anti-dieting (decreased pressure) condition resulted in lower bulim ic intentions compared to the pro-dieting and no-dieting conditions; however, the exploratory analyses, which included participant weight status as a between-subjects factor, did find greater bulimic intentions in the pro-dieting condition than the anti-dieting as predicted. Th e mediational analyses found that post-test perceived pressure fully mediated the relations hip between the diet message (pro vs. anti)
77 and bulimic intentions. There was not, how ever, a significant increase in bulimic symptomatology across time in the pro-dieti ng condition as hypothesized. Collectively, these findings provide some evidence that perc eived pressure plays a substantive role in bulimic symptomatology, but they do not provide direct support for the causal risk factor hypothesis. Because of these mixed findings more experimental research must be conducted to further evaluate the percei ved pressure construct on eating pathology. It is instructive to interpret the findin gs based on the effectiveness of the overall goals of the two dieting messages. The prodieting message is aimed towards preventing obesity, and the overall goals are to increase caloric restriction, exercise, and healthy eating behaviors for the purposes of weight loss and/or maintaining an average body weight. From this perspective, the findi ngs from the current study are encouraging. Dieting intentions were si gnificantly higher in the pr o-dieting message condition compared to the other two conditions. Hea lthy eating increased fr om baseline to followup in those who were in the pro-dieti ng condition while there were no significant increases in bulimic symptomatology. This finding provides further evidence for a mounting body of research that is calling into questi on the widely held belief that dietary restriction is associated with growth in bulimic symptoma tology (e.g., Presnell & Stice, 2003; Stice et al., 2005). There is additional evidence that the pr o-dieting message had particularly strong effects on overweight and obese individuals which could also be interpreted as encouraging, given that the message is target ed towards this populat ion. The pro-dieting message was perceived as more influential and applicable to themselves by overweight participants. Additionally, overweight par ticipants reported significantly greater
78 perceived pressure and state body dissatisfacti on at post-test. While this finding could be viewed negatively because perceived pressu re and body dissatisfaction have been found to be risk factors for eating pathology (Stice, 2002), there is some evidence that moderate body dissatisfaction may be a motivator for increasing healthy lif estyle behaviors (Heinberg, Thompson, & Matzon, 2001). It is al so possible that perceived pressure could be an impetus for positive behavioral cha nge. Findings from the current study provide mixed results for this hypothesis. On th e one hand, it is promising that perceived pressure was found to fully mediate the relatio nship between diet message (pro vs. anti) and dieting intentions, healthy eating inten tions, and healthy eating behaviors in the current study; however, perceive d pressure also fully mediated the relationship between diet message and bulimic intentions, suggesting th at it is also relate d to unhealthy weight control practices. The findings of the study can also be interpreted from an eating disorder prevention perspective. The anti-dieting, ea ting disorder prevention message aims to reduce dysfunctional eating patterns, includ ing strict dieting, improve body image attitudes, and increase hea lthy weight control practices. The anti-dieting message was successful at producing significantly lower pos t-test bulimic intentions and dieting intentions than the other two conditions. Furthermore, posttest perceived pressure and internalization intentions were significantly lower in the anti-dieting condition compared to the pro-dieting group. At follow-up, thos e in the anti-dieting condition reported reduced levels of perceived pr essure and athlete-internalizat ion; however, this reduction was also found in the no-dieting condition. The anti-dieting message was not successful
79 at producing lower state body dissa tisfaction levels. Furtherm ore, no behavioral changes were found from baseline to follow-up in the anti-dieting message condition. From the eating disorder prevention perspec tive, it is alarming that the pro-dieting message elicited greater levels of established risk factors (perceived pressure, dieting and internalization intentions) fo r eating pathology at post-test. Particularly concerning are the high levels of perceived pr essure to lose weight, even in participants who were not overweight. Given the recent widespread dissem ination of this message in the media, it is unclear what the cumulative effect of this message may be on psychological functioning and weight control practices. Although the results of the study are intriguing, there ar e several limitations that warrant discussion. The sample size for the study was somewhat small, and power to detect interaction effects may have been co mpromised, particularly in the exploratory analyses. Statistical power was also reduced in the mediator a nd moderator analyses because a third of the sample (no-dieting condition) was excluded, and Type II errors may have occurred. The sample used in the study also has some limitations. Only undergraduate females were included in the study, which lim its the generalizability of the findings to other populations. Future rese arch should replicate the study in samples that include males, older individuals, and non-college st udents. Furthermore, while the ethnic composition of the sample was fairly diverse, it is possible that th e diet messages have differential effects by race. The current st udy did not have adequate power to examine these potential differences, and future resear ch should ensure adequate numbers of ethnic minorities to examine this question. Lastly, to minimize any potential risks associated
80 with the study, individuals with a past history of an eati ng disorder or current purging behaviors were excluded from the study, whic h may have restricted the range of the sample at the disordered eati ng end of the spectrum while th e range was not restricted on the overweight/obesity side. It is unknow n how many individuals were excluded based on these criteria because it was done automatically through the online participant pool filtering system; however, it is possible that this restriction of range may have affected the results by potentially re ducing any deleterious effects of the messages on more eating disturbed individuals. Future research shoul d include more disturbed samples on both the eating disorder and obesity ends of the spectrum. The study also relied solely on self-re port measures of behavior. This is problematic because it is unclear how actua l behavior overlaps with self-reported behavior on several of the measures, particular ly with respect to eating behavior. It is widely accepted that people tend to underre port the amount of food they consume because of poor accuracy of food quantity a nd caloric value, cognitive processing errors, and social desirability (Klesges, Ec k, & Ray, 1995; Mulheim, Allison, Heshka, & Heymsfield, 1998; Smith, Jobe, & Mingay, 1991; Zegman, 1984). Although caloric intake was not directly assesse d in this study, participants di d self-report on frequency of healthy eating, dietary restrict ion, exercise, and bulimic symptoms, which are also likely plagued by similar limitations. Future research may consider using diary methods as well as including a social desirability scale to more accurately assess actual behavioral change. The follow-up period for the study was also brief, and it is unclear whether differences observed at the two-week follo w-up would be sustained over a longer time interval. Additionally, the psychoeducational messages were created to be very brief in
81 an effort to maintain external validity by cl osely replicating patient education brochures or newspaper articles; however, it is unclear whether the length of the materials would differentially affect the results. Moreover, participants were only exposed to the experimental message once. Future rese arch should study the dose effects of the messages to assess their cumulative effect s on psychological functioning and weight control practices, which is of particular importance for the pro-dieting message given its widespread dissemination to the public. Future research may also consider examin ing the effects of tailoring the dieting messages based on individual difference vari ables (Kreuter, Oswald, Bull, & Clark, 2000). For example, the current study found that trait levels of thin-i deal internalization moderated the extent to which the particip ant perceived pressure from the dieting message. This finding could suggest that individuals with high levels of trait internalization may benefit mo re from the anti-dieting message while those with low levels may benefit more from the pro-dieti ng message. Demographic variables such as race, sex, weight status, and age as well as dieting history and tr ait body dissatisfaction may be other individual differen ce variables that could be explored in future research on tailored messages. Another potential avenue for future research is to examine the psychological and behavioral e ffects of these messages when they are delivered in-vivo by a physician, dietician, or other health professional. In efforts towards bridging the fields of obesity and eating disorder prevention, future research from both perspectives shoul d assess healthy weight control practices as well as disordered eating risk factors and be haviors to better unders tand the interplay of these constructs. Ultimately, these two fields share similar goals of promoting health and
82 wellness through diet and exercise to re duce the likelihood of physical and mental disorders. Developing a more unified appr oach to psychoeducation and prevention will likely benefit all by resulting in a more cost -effective and straightforward program for the consumer as well as greate r potential for the reduction of eating and weight-related disorders.
83 References American Psychiatric Association (2000). Diagnostic and statistical manual of mental disorders (4th ed., text revision). Washington, DC: Author. Anderson-Fye, E.P. & Becker, A.E. (2004). Soci ocultural aspects of eating disorders. In J.K. Thompson (Ed.), Handbook of Eating Disorders and Obesity (pp.565-589). Hoboken, NJ: Wiley. Allison, D.B., Kalinsky, L.B., & Gorman, B.S. (1992). A comparison of the psychometric properties of three m easures of dietary restraint. Psychological Assessment, 4 391-198. Bacon, L., Keim, N.L., Van Loan, M.D., Derri cote, M., Gale, B., Kazaks, A., & Stern, J.S. (2002). Evaluating a non-diet wellness interven tion for improvement of metabolic fitness, psychological well-bei ng and eating and activity behaviors. International Journal of Obesity, 26, 854-865. Baron, R.M. & Kenny, D.A. (1986). The mode rator-mediator variab le distinction in social psychological research: Conceptual, st rategic, and statistic al considerations. Journal of Personality and Social Psychology, 51 1173-1182. Battle, K.E. & Brownell, K.D. (1996). Confr onting a rising tide of eating disorders and obesity: Treatment vs. prevention and policy. Addictive Behaviors, 21 755-765. Black, C.M.D. & Wilson, G.T. (1996). Assessmen t of eating disorders: Interview versus questionnaire. International Journal of Eating Disorders, 20 (1), 43-50.
84 Brewerton, T.D., Dansky, B.S ., Kilpatrick, D.G., & ONeil, P.M. (2000). Which comes first in the pathogenesis of bulimia nervosa: Dieting or bingeing? International Journal of Eating Disorders, 28 259-264. Brownell, K.D. & Rodin, J. (1994). The diet ing maelstrom: Is it possible and advisable to lose weight? American Psychologist, 49 781-791. Brownell, K.D. & Wadden, T.A. (1992). Etiology and treatment of obesity: Understanding a serious, prevalen t, and refractory disorder. Journal of Clinical and Consulting Psychology, 60, 505-517. Bulik, C.M. (2004). Genetic a nd biological risk factors. In J.K. Thompson (Ed.), Handbook of Eating Disorders and Obesity (pp.3-16). Hoboken, NJ: Wiley. Bulik, C.M., Sullivan, P.F., Carter, F.A., & Joyce, P.R. (1997). Initial manifestation of disordered eating behavior: Dieting versus bingeing. International Journal of Eating Disorders, 22 195-201. Cattarin, J.A. & Thompson, J.K. (1994). A th ree-year longitudinal study of body image, eating disturbance, and general psychologi cal functioning in adolescent females. Eating Disorders: The Journal of Treatment and Prevention, 2 (2), 114-125. Centers for Disease Contro l and Prevention (2004). Overweight and obesity. Retrieved March 20, 2004, from http://www.cdc.gov/nccdphp/dnpa/obesity/ Cogan, J.C. (1999). A new national health agenda: Providing the public with accurate information. Journal of Social Issues, 55, 383-400. Cope, M.B., Fernandez, J.R., & Allison, D.B. (2004) Genetic and biological risk factors. In J.K. Thompson (Ed.), Handbook of Eating Disorders and Obesity (pp.323338). Hoboken, NJ: Wiley.
85 Devlin, M.J., Yanovski, S.Z., & Wilson, G.T. (2000). Obesity: What mental health professionals need to know. American Journal of Psychiatry, 157, 854-866. Fairburn, C.G. & Beglin, S.J. (1994). Assessmen t of eating disorders: Interview or selfreport questionnaire? International Journal of Eating Disorders, 16 (4), 363-370. Fairburn, C.G. & Cooper, Z. (1993). The Eati ng Disorder Examination (12th edition). In C. Fairburn & G. Wilson (Eds.), Binge Eating: Nature, assessment, and treatment (pp. 317-360). New York: Guildford Press. Fairburn, C.G., Cooper, Z., Doll, H.A., Norman P., & OConner, M. (2000). The natural course of bulimia nervosa and binge eating disorder in young women. Archives of General Psychiatry, 57 659-665. Fairburn, C.G., Marcus, M.D., & Wilson, G.T. (1993). Cognitive-behavioral therapy for binge eating and bulimia nervosa: A comprehensive treatment manual. In C. Fairburn & G. Wilson (Eds.), Binge Eating: Nature, assessment, and treatment (pp. 361-404). New York: Guildford Press. Favaro, A. & Santonastaso, P. (1996). Pu rging behaviors, suicide attempts, and psychiatric symptoms in 398 ea ting disordered subjects. International Journal of Eating Disorders, 20 (1), 99-103. Fichter, M.M. & Quadflieg, N. (1999). Si x-year course and outcome of anorexia nervosa. International Journal of Eating Disorders, 26 (4), 359-385. Field, A. (2000). Discovering statistics using SPSS for windows. London: Sage Publications.
86 Field, A. E., Barnoya, J., & Colditz, G.A. (2002). Epidemiology and health and economic consequences of obesity. In T.A. Wadden & A.J. Stunkard (Eds.), Handbook of Obesity Treatment (pp.3-18). New York: Guilford Press. Flegal, K.M., Carroll, M.D., Kuczmarski, R. J., & Johnson, C.L. (1998) Overweight and obesity in the United States: Prevalence and trends, 1960-1994. International Journal of Obesity, 22 39-47. Foster, G.D. & McGurckin, B.G. (2002). N ondieting approaches: Pr inciples, practices, and evidence. In T.A. Wadden & A.J. Stunkard (Eds.), Handbook of Obesity Treatment (pp.494-514). New York: Guilford Press. Garner, D.M. (1991). Eating disorders inventory -2: Professional manual. Odessa, FL: Psychological Assessment Resources. Garner (in press). Eating disorder inventory -3: Professional manual. Odessa, FL: Psychological Assessment Resources. Garner, D.M., Olmstead, M.P., & Polivy, J.P. (1983). Development and validation of a multidimensional eating disorder inventory for anorexia nervosa and bulimia. International Journal of Eating Disorders, 2 (2), 15-34. Grilo, C.M. (2002). Binge eati ng disorder. In C.G. Fairburn & K.D. Brownell (Eds.), Eating Disorders and obesity: A comprehensive handbook (pp.178-182). New York: Guilford Press. Grissett, N.I. & Fitzgibbon, M.L. (1996). The clinical significance of binge eating in an obese population: Support for BED and que stions regarding its criteria. Addictive Behaviors, 21 57-66.
87 Groesz, L.M., Levine, M.P., & Murnen, S.K. (2002). The effect of experimental presentation of thin media images on body satisfaction: A meta-analytic review. International Journal of Eating Disorders, 31, 1-16. Heinberg, L.J. (1996). Theories of body imag e disturbance: Perceptual, developmental, and sociocultural factors. In J.K. Thompson (Ed.), Body Image, Eating Disorders, and Obesity: An Integrat ive Guide for Assessment and Treatment (pp.83-107). Washington, DC: Ameri can Psychological Association. Heinberg, L.J., & Thompson, J.K. (1995). Body image and televised images of attractiveness: A controlled laboratory investigation. Journal of Social and Clinical Psychology, 14, 325-338. Heinberg, L.J., Thompson, J.K., & Matzon, J.L. (2001). Body image dissatisfaction as a motivator for healthy lifestyle change: Is some distress beneficial? In R.H. Striegel-Moore & L. Smolak (Eds.), Eating Disorders: Innov ative directions in research and practice (pp. 215-232). Washington, D. C.: American Psychological Association. Henderson, K.E. & Brownell, K.D. (2004). The toxic environment and obesity: Contribution and cure. In J.K. Thompson (Ed.), Handbook of Eating Disorders and Obesity (pp.349-371). Hoboken, NJ: Wiley. Hintze, J. (2001). NCSS and PASS. Number cruncher statis tical systems. Kaysville, UT. www.ncss.com Hsu, L.K.G. (1996). Epidemiol ogy of the eating disorders. Psychiatry Clinics of North America, 19, 681-700.
88 Irving, L.M. & Neumark-Sztainer, D. (2002) Integrating the prevention of eating disorders and obesity: F easible or futile? Preventive Medicine, 34 299-309. Jason, L.A., Greiner, B.J., Naylor, K, Johns on, S.P. & Van Egeren, L. (1991). A largescale, short-term, media-ba sed weight loss program. American Journal of Health Promotion, 5, 432-437. Jeffery, R.W.& French, S.A. (1999). Preven ting weight gain in adults: The pound of prevention study. American Journal of Public Health, 89, 747-751. Johnson, F. & Wardle, J. (2005). Diet ary restraint, body dissatisfaction, and psychological distress: A prospective analysis. Journal of Abnormal Psychology, 114 119-125. Kazdin, A.E. (2003). Research Design in Clinical Psychology (4th ed.). Boston: Allyn and Bacon. Kenny, D.A. (2006). Mediation Retrieved March 5, 2006 from http://davidakenny.net/cm/mediate.htm Killen, J.D., Taylor, C.B., Hayward, C., Hayde l, K.F., Wilson, D.M., Hammer, L., et al. (1996). Weight concerns influence the de velopment of eating disorders: A 4-year prospective study. Journal of Consulting and Clinical Psychology, 64 936-940. Killen, J.D., Taylor, C.B., Hayward, C., Wils on, D.M., Haydel, K.F., Hammer, L. (1994). Pursuit of thinness and onset of eating di sorder symptoms in a community sample of adolescent girls: A threeyear prospective analysis. International Journal of Eating Disorders, 21 167-174.
89 Klesges, R.C., Eck, L.H., & Ray, J.W. (1995) Who underreports dietary intake in a dietary recall? Evidence from the Second National Health and Nutrition Examination Survey. Journal of Consulting an d Clinical Psychology, 63 438444. Kraemer, H.C., Kazdin, A.E., Offord, D.R., & Kessler, R.C. (1997). Coming to terms with the terms of risk. Archives of General Psychiatry 54(4), 337-343. Kraemer, H.C., Stice, E., Kazdin, A.E., Offo rd, D.R., Kupfer, D. (2001). How do risk factors work together? Mediators, mo derators, and independent, overlapping, proxy risk factors. American Journal of Psychiatry, 158, 848-856. Kraemer, H.C., Wilson, G.T., Fairburn, C.G., & Agras, W.S. (2002). Mediators and moderators of treatment effects in randomized clinical trials. Archives of General Psychiatry, 59 877-884. Kreuter, M.W., Oswald, D.L., Bull, F.C., & Clark, E.M. (2000). Are tailored health education materials always more eff ective than non-tailored materials? Health Education Research, 15 305-315. Kromery, J.D. & Dickinson, W.B. (1995). The us e of an overall f test to control type I error rates in factorial anal yses of variance: limitations and better strategies. Journal of Applied Behavioral Science, 31 51-64. Kulbok, P.A., Carter, K.F., Baldwin, J.H., G ilmartin, M.J., & Kirkwood, B. (1999). The multidimensional health behavior inventory. Journal of Nursing Measurement, 7, 177-195. Lowe. M.R. (1993). The effects of dieting on eating behavior: A three-factor model. Psychological Bulletin, 114 100-121.
90 Lyubomirsky, S. & Nolen-Hoeksema, S. ( 1993). Self-perpetua ting properties of dysphoric rumination. Journal of Personality and Social Psychology, 65, 330349. Lyubomirsky, S. & Nolen-Hoeksema, S. (1995). Effects of self-focused rumination on negative thinking and interp ersonal problem solving. Journal of Personality and Social Psychology, 69, 176-190. Marlatt, G.A. & Gordon, J.R. (1985). Relapse prevention: Maintenance strategies in the treatment of addictive behaviors New York: Guilford Press. Marcus, M. (1993). Binge eati ng in obesity. In C.G. Fair burn & G. T. Wilson (Eds.), Binge eating: Nature, as sessment, and treatment (pp.77-96). New York: Guilford Press. Marcus, M.D., Wing, R.R., & Fairburn, C. G. (1995). Cognitive treatment of binge eating vs. behavioral weight control in the treatment of binge eating disorder. Annals of Behavior Medicine, 25 225-238. Mellin, L., Croughan-Minihane, M., & Dickey, L. (1997). Th e solution method: 2-year trends in weight, blood pressure, exerci se, depression, and functioning in adults trained in development skills. Journal of the American Dietetic Association, 97 1133-1138. Miles, A., Rapoport, L., Wardle, J., Afuape T., & Duman, M. (2001). Using the mass media to target obesity: An analysis of the characteristics and reported behaviour change of participants in the BBCs F ighting Fat, Fighting Fit campaign. Health Education Research, 16, 357-372.
91 Mitchell, J.E., Hatsukami, D., Eckert, E.D., & Pyle, R.L. (1985). Characteristics of 275 patients with bulimia. American Journal of Psychiatry, 142 482-485. Morrow, J. & Nolen-Hoeksema, S. (1990). Effects of response to depression on the remediation of depressive affect. Journal of Personality and Social Psychology, 58 519-527. Mulheim, L.S., Allison, D.B., Heshka, S., & Heymsfield, S.B. (1998). Do unsuccessful dieters intentionally unde rreport food intake? International Journal of Eating Disorders, 24, 259-266. Murphy, K.R. & Myors, B. (1998). Statistical power analysis: A simple and general model for traditional and m odern hypothesis testing. Mahwah, NJ: Erlbaum. Mussell, M.P., Mitchell, J.E., Weller, C.L ., Raymond, N.C., Crow, S.J., & Crosby, R.D. (1995). Onset of binge eating, diet ing, obesity, and mood disorders among subjects seeking treatment for binge eating disorder. International Journal of Eating Disorders, 17 395-401. National Eating Disorders Association (2004). Eating Disorders Information Index. Retrieved on March 22, 2004, from http ://www.nationaleatingdisorders.org/ p.asp?WebPage_ID=294. National Task Force on the Prevention and Treat ment of Obesity (2000). Dieting and the development of eating disorders in overweight and obese adults. Archives of Internal Medicine, 160 2581-2589. Neumark-Sztainer, D. & Haines, J. (2004). Ps ychosocial and behavior al consequences of obesity. In J.K. Thompson (Ed.), Handbook of Eating Disorders and Obesity (pp.349-371). Hoboken, NJ: Wiley.
92 Neumark-Sztainer, D., Story, M., Faulkner, N.H., Beuhring, T., & Resnick, M.D. (1999). Sociodemographic and personal characteristic s of adolescents engaged in weight loss and weight/muscle gain behaviors: Who is doing what? Preventative Medicine: An International Journal Devoted to Practice and Theory, 28 40-50. Paxton, S.J., Wertheim, E.H., Pilawski, A., Durk in, S., & Holt, T. ( 2002). Evaluations of dieting prevention messages by adolescent girls. Preventative Medicine, 35 474491. Polivy, J. & Herman, C.P. (1985). Dieting and binging: A causal analysis. American Psychologist, 40 193-201. Polivy, J. & Herman, C.P. (1992) Undieting: A program to help people stop dieting. International Journal of Eating Disorders, 11 261-268. Polivy, J., Herman, C.P., & Warsh, S. (1978). Internal and external components of emotionality in restrained and unrestrained eaters. Journal of Abnormal Psychology, 87 497-504. Porzelius, L.K., Houston, C. Smith, Arfkin, A., & Fisher, E. (1995). Comparison of a standard behavioral weig ht loss treatment and a binge eating weight loss treatment. Behavior Therapy, 26, 119-134 Presnell, K. & Stice, E. (2003). An experiment al test of the effect of weight loss dieting on bulimic pathology: Tipping the scal es in a different direction. Journal of Abnormal Psychology, 112, 166-170. Rapoport, L., Clark, M., & Wardle, J. (2000) Evaluation of a modified cognitivebehavioural programme for weight management. International Journal of Obesity, 24 1726-1737.
93 Reijonen, J.H., Pratt, H.D., Patel, D.R., & Greydanus, D.E. (2003). Eating disorders in the adolescent population: An overview. Journal of Adolescent Research, 18 (3), 209-222. Sarwer, D.B., Foster, G.D., & Wadden, T.A. (2004). Treatment of obesity I: Adult obesity. In J.K. Thompson (Ed.), Handbook of Eating Disorders and Obesity (pp.421-442). Hoboken, NJ: Wiley. Schmitz, K.H. & Jeffery, R.W. (2002). Prev ention of obesity. In T.A. Wadden & A.J. Stunkard (Eds.), Handbook of Obesity Treatment (pp.556-593). New York: Guilford Press. Schwartz, M.B. & Brownell, K.D. (2002). Obesity and body image. In T.F. Cash & T. Pruzinksky (Eds.), Body image: A handbook of theory, research, and clinical practice (pp.200-209). New York: Guilford Press. Serdula, M.K., Collins, E., Williamson, D.F., Anda, R.F., Pamuk, E., & Byers, T.E. (1993). Weight control practices of U.S. adolescents and adults. Annals of Internal Medicine, 119 667-671. Shrout, P.E. & Bolger, N. (2002). Mediat ion in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7 422445. Simes, R.J. (1986). An improved bonferroni pr ocedure for multiple tests of significance. Biometrika, 73 751-754. Smith, A.F., Jobe, J.B., & Mingay, D.J. ( 1991). Retrieval from memory of dietary information. Applied Cognitive Psychology, 5 269-296.
94 Smolak, L. & Striegel-Moore, R.H. (2004). Future directions in eating disorder and obesity research. J.K. Thompson (Ed.), Handbook of Eating Disorders and Obesity (pp.738-754). Hoboken, NJ: Wiley. Sperry, S., Thompson, J.K., Roehrig, M., & Vandello, J. (2004). The influence of communicator weight on psychoeducationa l message acceptance in females with high vs. low levels of preexisting bo dy image disturbance. Unpublished manuscript. Stein, R.J., OByrne, K.K., Suminski, R. R., & Haddock, C.K. (2000). Etiology and treatment of obesity in adults and children : Implications for th e addiction model. Drugs and Society, 15, 103-121. Stice, E. (2001). A prospective test of the dual-pathway model of bulimic pathology: Mediating effects of die ting and negative affect. Journal of Abnormal Psychology, 110 (1), 124-135. Stice, E. (2002). Risk an d maintenance factor s for eating pathology: A meta-analytic review. Psychological Bulletin 128 (5), 825-848. Stice, E. & Agras, W.S. (1998). Predicti ng onset and cessation bulimic behaviors during adolescence: A longitudinal grouping analysis. Behavior Therapy, 29 (2), 257276. Stice, E., Chase, A., Stormer, S., & Appe l, A. (2001). A ra ndomized trial of a dissonance-based eating disord er prevention program. International Journal of Eating Disorders, 29, 247-262.
95 Stice, E., Fisher, M., & Lowe, M. (2004). Are dietary restraint scales valid measures of acute dietary restriction? Unobtrusi ve observational data suggest not. Psychological Assessment, 16, 51-59. Stice, E. & Hoffman, E. (2004). Eating disord er prevention programs. In J.K. Thompson (Ed.), Handbook of Eating Disorders and Obesity New York: Wiley. Stice, E., Maxfield, J., & Wells, T. (2003). Adve rse effects of social pressure to be thin on young women: An experimental investigat ion of the effects of fat talk. International Journal of Eating Disorders, 34, 108-117. Stice, E., Mazotti, L., Weibel, D., & Ag ras, W.S. (2000). Dissonance prevention program decreases thin-id eal internalizat ion, body dissatisfaction, dieting, negative affect, and bulimic symptoms: A preliminary experiment. International Journal of Eating Disorders, 27 (2), 206-217. Stice, E., Nemeroff, C, & Shaw, H. E. (1996). Test of the dual pathway model of bulimia nervosa: Evidence for dietary restrain t and affect regulation mechanisms. Journal of Social & Clinical Psychology, 15 (3), 340-363. Stice, E., Presnell, K., Groesz, L., & Shaw, H. (2005). Effects of a weight maintenance diet on bulimic symptoms in adolescent girl s: An experimental test of the dietary restraint theory. Health Psychology, 24 402-412. Stice, E., Presnell, K., Lowe, M.R., & Burt on, E. (2006). Validity of Dietary Restraint Scales: Reply to van Strien et al. (2006). Psychological Assessment, 18 95-99. Stice, E. & Ragan, J. (2002). A prelimin ary controlled evaluation of an eating disturbance psychoeducational inte rvention for college students. International Journal of Eating Disorders, 31 (2), 159-171.
96 Stice, E. & Shaw, H. (2004). Eating diso rder prevention programs: A meta-analytic review. Psychological Bulletin, 130 206-227. Stice, E., Trost, A., & Chase, A. (2003). Healthy weight contro l and dissonance-based eating disorder prevention programs: Results from a controlled trial. International Journal of Eating Disorders, 33 10-21. Striegel-Moore, R.H. & Smolak, L. (2001). Eating disorders: Innov ative directions in research and practice (Eds.), Washington, DC: American Psychological Association. Stunkard, A.J. & Messick, S. (1985). The Thr ee Factor Eating Questi onnaire to measure dietary restraint, disi nhibition, and hunger. Journal of Psychosomatic Research, 29 71-83. Sullivan, P.F., Bulik, C.M., Carter, F.A., & Joy ce, P.R. (1996). Correlates of severity in bulimia nervosa. International Journal of Eating Disorders, 20 (3), 239-251. Tanco, S., Linden, W., & Earle, T. (1998). We ll-being and morbid obesity in women: A controlled therapy evaluation. International Journal of Eating Disorders, 23 325-339. Tataranni, P.A. & Ravussin, E. (2002). Ener gy metabolism and obesity. In T.A. Wadden & A.J. Stunkard (Eds.), Handbook of Obesity Treatment (pp.). New York: Guilford Press.
97 Thompson, J. K. (1992). Body image: Extent of disturbance, associated features, theoretical models, assessment methodol ogies, intervention strategies, and a proposal for a new DSM diagnostic cate gory Body Image Disorder. In M. Hersen, R. M. Eisler, & P. M. Miller (Eds.), Progress in behavior modification (Vol. 28) (pp. 3-54). Sycamore, IL: Sycamore Publishing, Inc. Thompson, J.K. (2004a). Handbook of Eating Disorders and Obesity. Hoboken, New Jersey: John Wile y & Sons, Inc. Thompson, J. K. (2004b). The mismeasurement of body image. Ten strategies for improving assessment for clinical and research purposes Body Image: An International Journal of Research, 1 7-14 Thompson, J.K., Heinberg, L.J., Alta be, M., & Tantleff-Dunn, S. (1999). Exacting beauty: Theory, assessment, and trea tment of body image disturbance. Washington, DC : American Psychological Association. Thompson, J. K., Roehrig, M., & Kinder, B. (in press). Eating disorders. In M. Hersen & S. Turner (Eds.), Adult psychopathology and diagnosis (5th ed.). New York: Wiley. Thompson, J.K. & Smolak, L. (2001). Body Image, Eating Disorders, and Obesity in Youth: Assessment, Prevention, and Treatment. Washington, D.C.: American Psychological Association. Thompson, J.K. & Stice, E. (2001). Thin-ide al internalization: Mounting evidence for a new risk factor for body-image disturbance and eating pathology. Current Directions in Psychological Science, 10 (5), 181-183.
98 Thompson, J.K., van den Berg, P, Roehrig, M ., Guarda, A.S., & Heinberg, L.J. (2004). The sociocultural attitudes towards appearance questionnaire-3 (SATAQ-3): Development and validation. International Journal of Eating Disorders, 35, 293304. van den Berg, P., Thompson, J.K., Obrems ki-Brandon, K., & Coovert, M. (2002). The tripartite influence model of body imag e and eating disturbance: A covariance structure modeling investigation testi ng the mediational ro le of appearance comparison. Journal of Psychosomatic Research, 53, 1007-1020. van Strien, T., Frijters, J.E., Bergers, G.P ., & Defares, P.B. (1986). The Dutch Eating Behavior Questionnaire (DEBQ) for asse ssment of restrained, emotional, and external eating behavior. International Journal of Eating Disorders, 5 (2), 295315. Wadden, T.A., Brownell, K.D., Foster, G.D. (2002). Obesity: Responding to the global epidemic. Journal of Consulting and Clinical Psychology, 70, 510-525. Wadden, T.A., Foster, G.D., & Letizia, K.A. (1994). One-year beha vioral treatment of obesity: Comparison of moderate and severe caloric restriction and the effects of weight maintenance therapy. Journal of Consulting and Clinical Psychology, 62, 165-171. Watson, D. & Clark, L.A. (1992). Affects separa ble and inseparable: On the hierarchical arrangement of the negative affects. Journal of Personality and Social Psychology, 62 (3), 489-505. Weissberg, R.P., Kumpfer, K.L., & Seligma n, M.P. (2003). Prevention that works for children and youth. American Psychologist, 425-432.
99 World Health Organization (1998). Obesity: Preventing and managing the global epidemic. Geneva, Switzerland: Author. Yanovski, S.Z., Nelson, J.E., Dubbert, B.K., Spit zer, R.L. (1993). Association of binge eating disorder and psychiatric comorbidity in obese subjects. American Journal of Psychiatry, 150 1472-1479. Zegman, M.A. (1984). Errors in food record ing and calorie estimation: Clinical and theoretical implications for obesity. Addictive Behaviors, 9 347-350.
101 Appendix A: The Tripartite Model of Influence Media Influences Parental Influences Peer Influences Com p arison Internalization Body Dissatisfaction Restriction Bulimic Symptoms Global Psychological Functionin g
102 Appendix B: The Dual-Pathway Model of Bulimic Pathology Pressure to be thin Thin-Ideal Internalization Body Dissatisfaction Dieting Negative Affect Bulimic Pathology
103 Appendix C: Obesity Prevention Message U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES CALLS FOR HEALTH PROMOTION LOSE WEIGHT AND/OR MAINTAIN A LOW BODY WEIGHT TO PREVENT OVERWEIGHT AND OBESITY The rates of overweight and obes ity in the United States have reached epidemic proportions. According to Tommy G. Thompson, Secretary of the U.S. Department of Health and Human Services, Overweight and obesity are among th e most pressing new health challe nges we face today. In 1999, 61% of adults in the United States were overweight or obese. 13% of children aged 6 to 11 and 14% of adolescents aged 12 to 19 were overweight. Obesity among adults has doubled since 1980, while overweight among adolescents has tripled. The U.S. Surgeon General, David Satcher, has stated, Overweight and obesity may soon cause as much prev entable disease and death as cigarette smoking. What is Overweight and Obesity? The National Institutes of Health define obesity and overweight using a Body Mass Index (BMI), which is a measure of weight in relation to height. It can be calculated using the formula of ((weight in pounds/height in inches2) X 703. An overweight adult has a BMI between 25 and 29.9, while an obese adult has a BMI of 30 or above. Causes of Overweight and Obesity Overweight and obesity can be contributed to an imbalance between caloric intake and calories used throughout the day. Two common reasons for being overweight are eating too much and not being physically active enough. When individuals eat more calories than the body burns up, the extra calories are stored as fat. Excess fat results in overweight or obesity. Consequences of Overweight and Obesity Overweight and obesity are serious problems, and there are many health consequences that are attributable to them. Approximately 300,000 U.S. deaths per year curren tly are associated with obesity and overweight; this is compared with ap proximately 400,000 U.S. deaths pe r year that are associated with cigarette smoking. Overweight and obesity have been linked with heart disease, certain types of cancer, type 2 diabetes, stroke, arthritis, breathing problems such as asthma and sleep apnea. Additionally, overweight and obesity are associated with increased risks of gall bladde r disease, incontinence, increased surgical risk, and reproductive problems. Overweight and obesity also have several psychological consequences, including an increased risk for depression. Quality of life is also frequently a ffected by overweight and obesity due to limited mobility and decreased physical endurance. Social, academic, a nd job discrimination has also been associated with obesity. What You Can Do to Preven t Overweight and Obesity If you are overweight or obese, lose weight! Weight loss, as modest as 5-15% of overweight or obese persons body weight, reduces the risk factors for some diseases, particularly heart disease. Weight loss can also result in lowered blood pressure, lowered blood sugar, and improved cholesterol levels. Maintain a low body weight if you are currently not overweight or obese. To maintain your weight, your intake of calories must equal your energy output. A difference of one extra 12-oz soda (150 calories) can add 10 pounds to your weight each year, so it is im portant to maintain a balance between caloric intake and energy output even if you are not currently overweight or obese.
104 Appendix C (Continued) Other good habits for health that will he lp prevent overweight and obesity are: Recognize that although genetics do play a role in body size and sh ape, your habits largely impact your weight and are important in the prevention of overweight and obesity. Consult a body mass index chart to determine your ideal weight. Make physical fitness a priority! It is reco mmended that Americans accumulate at least 30 minutes of moderate physical activity most days of the week. More may be needed to prevent weight gain, to lose weight, or to maintain weight loss. Eat well by choosing lower fat, lower calorie foods to attain or maintain a low body weight. Prepare food by broiling or baking more often than frying. Eat lean meat, fish, and poultry without skin. Eat more fruits, vegetables, and whole grains. Reduce your caloric intake by limiting your portion sizes and avoid going back for seconds.
105 Appendix D: Eating Disorder Prevention Message THE NATIONAL EATING DISORDER ASSOCIATION CALLS FOR HEALTH PROMOTION STOP DIETING AND ACCEPT ALL BODY SI ZES TO PREVENT DISORDERED EATING In the United States, conservative estimates i ndicate that 5-10% of girls and women (that translates into 5-10 million girls and women) and 1 million boys and men are struggling with eating disorders including anorexia, bulimia, binge eating diso rder, or borderline conditions. In addition to fullblown eating disorders, many individuals struggle with body dissatisfaction and sub-clinical disordered eating attitudes and behaviors. Studies suggest that approximately 80% of American women are dissatisfied with their appearance. It is estimated that 40-50% of American women are trying to lose weight at any point in time. In fact, Americans spend more than $40 billion dollars per year on dieting and diet-related products. Thats roughly the equivalent the U.S. government spends on education each year. What is Disordered Eating? The symptoms of disordered eating such as rest ricting food intake, compulsively exercising, overeating, purging, and dissatisfaction with body weight and shape are often considered normal and harmless in our culture. Many people who engage in these behaviors may not feel that they have a problem. However, these habits are problematic and unhealthy. Causes of Disordered Eating Disordered eating arises from a complex combination of behavioral, emotional, psychological, interpersonal, and social factors. Cultural pressures that glorify thin ness and place value on obtaining a low body weight have also been implicated as reasons for the development of disordered eating. Consequences of Disordered Eating Disordered eating is a serious problem, and there ar e many health consequences associated with it. Chronic dieting can deprive you of essential nutrients such as calcium, and repetitive cycles of gaining, losing, and regaining weight has been shown to have negative health effects, including increased risk of heart disease and long-lasting negative impact on metabo lism. Disordered eating has also been associated with osteoporosis, dehydration, fatigue, dry skin and hair, muscle loss, electrolyte imbalances, loss of coordination, tooth decay, peptic ulcers and pa ncreatitis, and even death in severe cases. Disordered eating also has several psychological consequences. Research has shown that disordered eating is associated with feelings of depression, low self-est eem, increased stress, and problems with memory and concentration. What You Can Do to Prevent Disordered Eating Stop dieting! Dieting is rarely effective and can lead to disordered eating. 95% or all dieters regain their lost weight and more within 1 to 5 years. Additionally, many studies and health professionals note that patients with eating disorders were dieting at the time of the onset of their eating disorder. While dieting may not cause an eating disorder, the constant concern about body weight and shape, fat grams, and calories can start a vicious cycle of body dissatisfaction and obsession that can spiral into disordered eating all too quickly. Other good habits that will help prevent disordered eating are: Recognize that every body is different and that genetics strongly influence bone structure, body size, shape, and we ight differently. Understand that there is no ideal body size, shape, or weight that every individual should strive to achieve. Dont rely on char ts, formulas, and tables to dictate whats right for you. Exercise moderately by engaging in physical activity that you enjoy. Enjoy your favorite meal without feelin gs of guilt or anxiety over calories. Fuel your body with a variety of foods. Listen to your body. Eat exactly what appeals to you when you are truly hungry. Stop when you are full.
106 Appendix E: Flu Prevention Message CENTERS FOR DISEASE CONTROL AND PREVENTION CALLS FOR HEALTH PROMOTION GET VACCINATED TO PREVENT THE FLU Infection with influenza viruses can result in illness ranging from mild to severe with lifethreatening complications such as pneumonia. An es timated 10% to 20% of U.S. residents get the flu each year. An average of 114,000 peop le are hospitalized for flu-related complications and 36,000 Americans die each year from complications of flu. What is Influenza? Influenza (commonly called the flu) is a contagious respiratory illness caused by influenza viruses. It attacks the respiratory tract in humans (nose, throat, and lung s) and is different from a cold. Influenza usually comes on suddenly and may include these symptoms: fever, headache, tiredness, dry cough, sore throat, nasal congestion, and body aches Gastro-intestinal symptoms such as nausea, vomiting, and diarrhea are much more common in children than adults. Spread of the Flu The main way that influenza viruses are spread is from person to person in respiratory droplets of coughs and sneezes. This is called droplet spread. Th is can happen when droplets from a cough or sneeze or an infected person are propelled (generally up to 3 feet) through the air and deposited in the mouth or nose of people nearby. Though much less frequent, the viruses can also be spread when a person touches respiratory droplets on another person or object and then touches their own mouth or nose (or someone elses mouth or nose) before washing their hands. Scientific studies show that adults can shed the virus from 1 day before developing symptoms to up to 7 days after getting sick. Young children can shed the virus for longer than seven days. In general, however, more virus is shed earlier in the illness than later. Consequences of Influenza Most people who get the flu will recover in one to two weeks, but some people will develop lifethreatening complications as a result of the flu. Anyone can get the flu, and serious problems from influenza can happen at any age. People age 65 years and older, people of any age with chronic medical conditions, and very young children are more likely to get complications from influenza. Pneumonia, bronchitis, and sinus and ear infections are three examples of complications from influenza. The flu can make chronic health problems worse. For example, people with asthma may experience asthma attacks while they have the flu. What You Can Do to Prevent Influenza The single best way to prevent the flu is to get v accinated each fall. In the absence of a vaccine, however, there are other ways to protect against flu. Three antiviral drugs (ama ntadine, rimantadine, and oseltamivir) are approved and commerci ally available for use in preventing flu. All of these medications are prescription drugs, and a doctor should be consulted before the drugs are used for preventing the flu.
107 Appendix E (Continued) Other good habits for health that may prevent the spread of respiratory illnesses like the flu are: Avoid close contact with people who are sick. When you are sick, keep your distance from others to protect them against getting sick. Stay home when you are sick. Cover your mouth and nose with a tissue when coughing or sneezing. Wash your hands often will help protect you from germs. Avoid touching your eyes, nose, or mouth. Germs are spread when a person touches something that is contaminated with germs and then touches his or her eyes, nose, or mouth. Get enough rest. Inadequate rest and sleep can cause the breakdown of your immune system making you more vulnerable to the flu.
108 Appendix F: Visual Analogue Scales Instructions: Place a mark through the area of the line that matches your feelings right now. 1. Happiness None Extreme 2. Anxiety None Extreme 3. Energetic None Extreme 4. Disappointed in Self None Extreme 5. Anger None Extreme 6. Calmness None Extreme 7. Dissatisfied with Weight/Size None Extreme 8. Healthy None Extreme 9. Irritability None Extreme 10. Dissatisfied with Body Shape None Extreme
109 Appendix G: Positive and Ne gative Affect Scale-Revised Please circle the response that indicates ho w you feel currently/generally/over past two weeks. not at all a little moderately a lot extremely 1. Disgusted with self . . 1 2 3 4 5 2. Sad. . . . . . . . . 1 2 3 4 5 3. Afraid . . . . . . . 1 2 3 4 5 4. Shaky. . . . . . . . 1 2 3 4 5 5. Alone. . . . . . . . 1 2 3 4 5 6. Blue. . . . . . . . 1 2 3 4 5 7. Guilty . . . . . . 1 2 3 4 5 8. Nervous. . . . . . 1 2 3 4 5 9. Lonely. . . . . . . 1 2 3 4 5 10. Jittery. . . . . . . 1 2 3 4 5 11. Ashamed . . . . . 1 2 3 4 5 12. Scared . . . . . . 1 2 3 4 5 13. Angry at self . . . . 1 2 3 4 5 14. Downhearted. . . . 1 2 3 4 5 15. Blameworthy. . . . 1 2 3 4 5 16. Frightened . . . . . 1 2 3 4 5 17. Dissatisfied with self. 1 2 3 4 5 18. Anxious. . . . . . 1 2 3 4 5 19. Depressed . . . . . 1 2 3 4 5 20. Worried . . . . . . 1 2 3 4 5
110 Appendix H: Modified SATAQ-3 You will be asked to rate your agreement with ma ny statements about the article you just read. Some of the questions will seem very relevant to what you just read and some will not. Please read each of the following items carefully and indicate the number that best reflects your agreement with the statement to the best of your ability. Definitely Disagree Mostly Disagree Neither Agree Nor Disagree Mostly Agree Definitely Agree 1 2 3 4 5 Pressures subscale 1. Ive felt pressure from this article to lose weight. 2. Ive felt pressure from this article to be thin 3. Ive felt pressure from this article to avoid dieting 4. Ive felt pressure from this article to exercise 5. Ive felt pressure from this article to change my appearance. Internalization subscale 6. Reading this article makes me believe that all body sizes are acceptable. 7. Reading this article makes me want my bod y to look like the models who appear in magazines. 8. Reading this article makes me want my body to look like the people who are in the movies 9. Reading this article makes me want to compar e my appearance to the appearance of TV and movie stars. 10. Reading this article makes me want to compar e my body to that of people in good shape. 11. Reading this article makes me want to accept all body sizes, shapes, and weights. 12. Reading this article makes me want to compare my body to that of people who are athletic. Flu Prevention items 13. This article encourages me to avoid spreading germs. 14. Ive felt pressure from this article to get vaccinated for the flu. 15. Reading this article makes me want to avoid close contact with others who are sick. 16. This article encourages me to wash my hands frequently. 17. Ive felt pressure from this article to get adequate rest. Behavioral Intentions (from pilot study) 18. Reading this article makes me want to increase my level of physical activity. 19. Reading this article makes me want to start a weight loss diet.
111 Appendix I: Message Rating Form You will be asked to rate your agreement with ma ny statements about the article you just read. Some of the questions will seem very relevant to what you just read and some will not. Please read each of the following items carefully and indicate the number that best reflects your agreement with the statement to the best of your ability. Definitely Disagree Mostly Disagree Neither Agre e Nor Disagree Mostly Agree Definitely Agree 1 2 3 4 5 Statement Level of Agreement 1. This article is convincing. 2. This article is effective. 3. This article is applicable to me. 4. This article is easy to understand. 5. This article is credible. 6. This article is influential to me.
112 Appendix J: Eating Di sorder Inventory-2 Body Dissatisfaction subscale: 1 2 3 4 5 6 Always Usually Often Sometimes Rarely Never Always .Never 1. I think that my stomach is too big. 2. I think that my thi ghs are too large. 3. I think that my stomach is just the right size. 4. I feel satisfied with the shape of my body. 5. I like the shape of my buttocks. 6. I think my hips are too big. 7. I think that my thighs are just the right size. 8. I think that my butto cks are too large. 9. I think that my hips ar e just the right size. Drive For Thinness subscale: 1. I eat sweets and carbohydrates without feeling nervous. 2. I think about dieting. 3. I feel extremely guilty after overeating. 4. I am terrified of gaining weight. 5. I am preoccupied with a desire to be thin. 6. If I gain a pound, I worry I will keep gaining.
113 Appendix K: SATAQ-3 You will be asked to rate your agreement with ma ny statements about the article you just read. Some of the questions will seem very relevant to what you just read and some will not. Please read each of the following items carefully and indicate the number that best reflects your agreement with the statement to the best of your ability. Definitely Disagree Mostly Disagree Neither Agree Nor Disagree Mostly Agree Definitely Agree 1 2 3 4 5 1. Ive felt pressure from TV or magazines to lose weight. 2. I would like my body to look like the people who are on TV. 3. I compare my body to the bodies of TV and movie stars. 4. TV commercials are an important source of information about fashion and being attractive. 5. Ive felt pressure from TV or magazines to look pretty. 6. I would like my body to look lik e the models who appear in magazines. 7. I compare my appearance to the appearance of TV and movie stars. 8. Ive felt pressure from TV or magazines to be thin. 9. I would like my body to look like the people who are in movies. 10. I compare my body to the bodies of people who appear in magazines. 11. Ive felt pressure from TV or magazines to have a perfect body 12. I wish I looked like the models in music videos. 13. I compare my appearance to the appearance of people in magazines. 14. Ive felt pressure from TV or magazines to diet. 15. I wish I looked as athletic as the people in magazines. 16. I compare my body to that of people in good shape. 17. Ive felt pressure from TV or magazines to exercise. 18. I wish I looked as athletic as sports stars. 19. I compare my body to that of people who are athletic. 20. Ive felt pressure from TV or magazines to change my appearance. 21. I try to look like the people on TV. 22. I try to look like the people in music videos. 23. I try to look like sports athletes.
114 Appendix L: Dutch Eating Behavior Questionnaire-Restraint Scale Please indicate the best response to des cribe your usual behavior/behavior over the last two weeks : Never Seldom Sometimes Often Always Dieting Intentions 1. Did you eat less than you normally would to lose weight? 2. Did you try to eat less at mealtimes than you would like to eat?. 3. How often did you refuse food or drink because you were concerned about your weight? 4. Did you watch exactly what you ate? 5. Did you deliberately eat foods that were slimming? 6. If you ate too much, did you eat less than usual the next day? 7. Did you deliberately eat less in order not to become heavier? 8. How often did you try not to eat between meals because you were watching your weight? 9. How often in the evenings did you try not to eat because you were watching your weight? 10. Did you take into account your weight in deciding what to eat? 1. Do you plan to eat less than you normally would to lose weight? 2. Do you plan to eat less at mealtimes than you would like to eat? 3. Do you plan to refuse food or drink to lose weight? 4. Do you plan to watch exactly what you eat? 5. Do you plan to deliberatel y eat foods that are slimming? 6. If you overeat one day, do you plan to eat less than usual the next day? 7. Do you plan to deliberately eat less in order to not become heavier? 8. Do you plan to try to not eat between meals because you plan on watching your weight? 9. Do you plan to eat less in the evenings to control your weight? 10. Do you plan to take your weight into account when deciding what to eat?
115 Appendix M: Eating Disorder Examination-Questionnaire Please circle the response that descri bes your behavior over the past week : On how many days during the past week ... 1. Have you felt fat? . . . . . . . . . . . . . . 2. Have you had a definite fear that you might gain weight or become fat?. . . . . . . . . . . Over the past week ... Not at all Slightly Moderately Extremely 3. Has your weight influenced how you think about (judge) yourself as a person? . . . . . . . .0 1 2 3 4 5 6 4. Has your shape influenced how you think about (judge) yourself as a person? . . . . . . . .0 1 2 3 4 5 6 1. During the past week have there been times when you felt you have eaten what other people would regard as an unusually large amount of food given the circumstances? 6. During the times when you ate an unusually large amount of food, did you experience a loss of control i.e. feel you couldn't stop eating or control what or how much you were eating? 7. How many times during the past week have you eaten an unusually large amount of food and experienced a loss of control ?____________ (please write in number or indicate zero) 8. During the past week have you had other times where you felt you uncontrollably ate a large amount of food, but the amount eaten would not have been considered large by most people? 9. How many times during the past week have you have uncontrollably eaten a large amount of food that others might not consider large?________________ (please write in number or indicate zero) 10. How many times during the past week have you made yourself sick in order to prevent weight gain or counteract th e effects of eating?________________ (write in number or indicate zero) 11. How many times during the past week have you used laxatives or diuretics in order to prevent weight gain or counteract th e effects of eating?__________ (write in number or indicate zero) 12. How many times during the past week have you engaged in excessive exercise specifically for the purpose of c ounteracting overeating episodes?_______________ (write in number or indicate zero)
116 Appendix N: Modified Eating Di sorder Examination-Questionnaire Intentions Strongly Disagree Somewhat Disagree Neither Agree Nor Disagree Somewhat Agree Strongly Agree 1. I plan to make myself sick in order to prevent weight gain or counteract the effects of eating. 1 2 3 4 5 2. I plan to use laxatives or diuretics in order to prevent weight gain or counteract the effects of eating. 1 2 3 4 5 3. I plan to vigorously exercise for an hour or more in order to prevent weight gain or counteract the effects of eating. 1 2 3 4 5 4. I plan to use diet pills in order to prevent weight gain or help me lose weight. 1 2 3 4 5 5. I plan to smoke cigarettes in order to prevent weight gain or help me lose weight. 1 2 3 4 5 6. I plan to skip meals in order to prevent weight gain or help me lose weight. 1 2 3 4 5
117 Appendix O: Eating Disorder Inventory-3 Referral Form In the past 3 months, how often have you Never Once a month or less 2-3 times per month Once a week 2-6 times per week Once a day or more 1. Gone on eating binges (eating a large amount of food while feeling out of control)? 0 1 2 3 4 5 2. Made yourself sick (vomited) to control your weight? 0 1 2 3 4 5 3. Used laxatives to control your weight or shape? 0 1 2 3 4 5 4. Exercised 60 minutes or more to control your weight? 0 1 2 3 4 5 5. In the past 6 months, have you lost 20 pounds or more? 0 1 2 3 4 5
118 Appendix P: Multidimensional Health Behavior Inventory Directions : The following statements describe a br oad range of health-related actions or behaviors that you may or may no t do. Read each behavior statement and circle the number following each stat ement that tells how often you usually do th is behavior/plan to/how often over the past two weeks have you: NEVER RARELY SOMETIMES OFTEN ALWAYS 1. Take time for rela xation every day. 2. Limit red meat in your diet every day. 3. Limit fat in your diet every day. 4. Eat red meat more than two ti mes a week. 5. Eat fewer calories to lose weight. 6. Eat at least one serving or more of red meat on most days (include beef, pork, ham, bacon, lamb, live r, and lunch meat no t made from poultry). 7. Limit sugar in yo ur diet every day. 8. Eat non-fat or lo w-fat dairy products. 9. Do something good for yourself every day. 10. Choose foods with whole grains every day. 11. Check your cholesterol level at least once a year. 12. Seek health information. 13. Get adequate sleep every day. 14. Check your blood pressure at least twice a year. 15. Read food and medicine labels before purchasing or consuming the product. 16. Question your health care provi der or seek a second opinion. 17. Maintain a first aid kit. 18. Get 7-8 hours sleep every day. 19. Participate in recreat ional physical activities at least twice a week. 20. Limit salt in your diet every day. 22. Limit intake of "sweets" in your diet. 23. Do stretching exercises every day. 24. Eat 2-3 servings of vegetables daily. 25. Obtain a regular health ch eck-up when you are not sick. 26. Control stress in your life. 27. Exercise vigorously for at least 20 minutes 3 times a week. 28. Keep daily stress levels low. 29. Increase your physical activity to lose weight. 30. Run, jog, or swim for exercise at least 3 times per week. 31. Discuss health concerns with health resource person. 32. Eat 2-3 servings of fruit per day. 33. Eat at least one or more servings of the following items every day: chips, candy bars, cake, doughnut s, pastries, muffins, cook ies,ice cream, pudding, chocolate.
119 Appendix Q: Pilot Stud y Attention Check Items Directions: Based on the article you ha ve just read, please circle True or False for each question. Pro-Dieting, Obesity Prevention Message: 1. Only 15% of adults in the United Stat es are overweight or obese. 2. There is little that can be done to prevent overweight and obesity. 3. Losing weight and/or maintaining a low body weight is very impor tant in preventing overweight and obesity. 4. An individual is categorized as obese if he or she has a BMI of 30 or above. 5. People who are currently thin do not have to worry about preventing overweight or obesity. Anti-Dieting, Eating Disorder Prevention Message 1. Only 1% of girls and women have disordered eating. 2. There is little that can be done to prevent disordered eating. 3. Stopping dieting is very im portant in the prevention of disordered eating. 4. Symptoms of disordered eating include overeating, restricting food intake, and compulsive exercising. 5. Genetics do not strongly influence body weight, size, and shape. No-Dieting, Flu Prevention Control Message 1. About 80% of U.S. residents get the flu each year. 2. There is little that can be done to prevent the flu. 3. Getting vaccinated is very importa nt in the prevention of the flu. 4. Symptoms of the flu include fever, headach e, tiredness, sore throat, nasal congestion, and body aches. 5. Healthy people do not have to worry about complications from the flu.
120 Appendix R: Main Stud y Attention Check Items Directions: Based on the article you ha ve just read, please circle True or False for each question. Pro-Dieting Condition: 1. 40% of adults in the United States are overweight or obese. 2. There is little that can be done to prevent overweight and obesity. 3. Most Americans should lose some fat, even those in the upper end of the average range. 4. An individual is categorized as obese if he or she has a BMI of 30 or above. 5. Thin people not have to worry a bout weight gain. Anti-Dieting Condition: 1. Approximately 1-2% of girls and wome n have disordered eating. 2. There is little that can be done to prevent disordered eating. 3. Symptoms of disordered eating are ofte n considered normal in our culture. 4. Dieting is unnecessary for weight control. 5. Genetics do not strongly influence body weight, size, and shape. No-Dieting Control Condition: 1. About 80% of U.S. residents get the flu each year. 2. Gastro-intestinal symptoms are very common in adults who have the flu. 3. Getting vaccinated is very importa nt in the prevention of the flu. 4. Three anti-viral drugs are available for use in preventing the flu. 5. Healthy people do not have to worry about complications from the flu.
121 Appendix S: Distraction Task Now, Id like you to take about 5-10 minutes to think about vacation destinations you have learned about through the media but ha ve never been to. After giving it some thought, Id like you to imagine your top 5 vaca tion destinations you have read about or heard about through the media but that you have not yet been to. Please take your time with this. Id like you to think about these places a nd visualize yourself on vacation in each of them. What would it be like? What woul d you be doing there? What would you see? What sensations would you feel? To help you with this exercise, Id like you to write your top 5 vacation destinations in the space below. Id also like you to provide a brief description of each destination, what you would like to do and see there, and how the media has describe d this destination. Travel Destination Activities/Sights/Feelings There Media Description 1. 2. 3. 4. 5.
About the Author Megan Roehrig received a Bachelor of Arts degree in Psychology from the University of Michigan in 1999, and a Mast ers of Arts degree in Clinical Psychology from the University of South Florida in 2003. She completed her pre-doctoral internship in clinical psychology at the University of Chicago Medical Center in 2007 and will begin a post-doctoral research associate position at Yale Univ ersity School of Medicine in September of 2007. Her research is focused on risk factors, treatment, and prevention of eating disorders and obesity, bariatric surgery outcome, and body image assessment. She has co-authored several peer-reviewed jour nal articles and book chap ters in this field.
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001925256
007 cr mnu|||uuuuu
008 080130s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002170
An experimental study of pro-dieting and anti-dieting psychoeducational messages :
b effects on immediate and short-term psychological functioning and weight control practices in college women
h [electronic resource] /
by Megan Roehrig.
[Tampa, Fla.] :
University of South Florida,
ABSTRACT: While dieting is relatively normative in our society, it is controversial within the fields of eating disorders and obesity. Dieting for weight loss has been touted by the obesity prevention field as a solution to the growing obesity epidemic, yet a body of research in the eating disorders field has also implicated it in the etiology and maintenance of eating pathology. Thus, a divergence in approaches toward dieting has emerged, with both prodieting and anti-dieting messages being recommended. Little is known, however, about the impact of these two types of messages on immediate and short-term psychological functioning and weight control intentions and behaviors. The current study sought to explore this gap in the extant literature by conducting an experimental study that evaluated the two messages. Undergraduate women (N=139) were randomly assigned to either a pro-dieting, anti-dieting, or no-dieting (control) message condition.Psychological functioning and weight control variables were assessed at baseline, posttest, and a two-week follow-up. Results indicated that the pro-dieting message resulted in significantly greater post-test perceived pressure to lose weight, dieting intentions, and thin-ideal internalization intentions while the anti-dieting message yielded significantly lower post-test bulimic intentions. Healthy eating behavior significantly increased from baseline to follow-up in the pro-dieting condition while there were no changes in the other two conditions. Post-test perceived pressure was found to fully mediate the relationship between diet message and post-test dieting, bulimic, thin-ideal internalization, and healthy eating intentions as well as follow-up healthy eating behavior. Trait thin-ideal internalization levels moderated the relationship between diet message and post-test perceived pressure and thin-ideal internalization intentions.Exploratory analyses revealed that overweight participants in the pro-dieting condition increased significantly from pre to post-test on state body dissatisfaction and had the highest level of post-test perceived pressure compared to all other groups. Nonoverweight participants in the pro-dieting condition also had significantly greater posttest perceived pressure to lose weight than both weight status groups in the other two conditions. Findings are discussed in the context of the prevention goals of the obesity and eating disorders fields. Limitations of the study and directions for future research are offered.
Dissertation (Ph.D.)--University of South Florida, 2007.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 121 pages.
Advisor: J. Kevin Thompson, Ph.D.
t USF Electronic Theses and Dissertations.