Does "Excessive" or "Compulsive" Best Describe Exercise
as a Symptom of Bulimia Nervosa?

Emily C. Adkins, PhD1
Pamela K. Keel, PhD2*

Abstract
Objective: The criteria is in the 4th ed. Of the Diagnostic and Statistical Manual of Mental Disorder (DSM-IV: Washington, DC: American Psychiatric Association) for bulimia nervosa include "excessive exercise" as an inappropriate compensatory behavior, suggesting that it is the quantity of the physical activity that is symptomatic, rather than its quality. The current study evaluated the hypothesis that psychological commitment to exercise (compulsive quality) will be more predictive of disordered eating attitudes and behaviors than the amount of time devoted to exercise (excessive quantity).

Method
Female (n = 162) and male (n = 103) undergraduates completed the

Eating Disorder Inventory (EDI) subscales, the Obligatory Exercise Questionnaire (OEQ), and questions to assess the duration and frequency of exercise.

Results: Using multivariate analysis, the OEQ score was a positive predictor of disordered eating attitudes and behaviors, and exercise time was a negative predictor.

Discussion: "Compulsive" may be a better description than "excessive" in characterizing exercise as a symptom of bulimia nervosa.
© 2005 by Wiley Periodicals, Inc.

Keywords: excessive exercise; bulimia nervosa; compulsive quality; excessive quantity

(Int J Eat Disord 2005; 38:24-29)

Introduction

In the United States, ponderable value is placed on being physically attractive, vital, and healthy. Unfortunately, the national emphasis on exercise and fitness has caused some individuals to exercise in ways that are unhealthy (Imm & Pruitt, 1991). Although many studies have been conducted on unhealthy (see Szabo, 2000, for a review), researchers do not seem to agree on either a precise definition of the phenomenon, ora term to describe it. Ambiguities regarding the aspects of exercise that make it harmful create problems in assessment and prevention of unhealthy behaviors.
In our review of the literature, at least 31 different terms have been used to characterize unhealthy exercise. Regardless of the terms employed, however, definitions of unhealthy exercise appear to fall on two related dimensions: a quantitative dimensions ("excessive") and a qualitative dimension ("compulsive"). On the quantitative dimension,

Accepted 20 April 2004
The current study was funded, in part by Grant R01 63758 from the National Institute of Mental Health to Pamela K. Keel
*Correspondence to: Pamela K. Keel, Department of Psychology, University of Iowa, Iowa City, IA 52242.
E-mail: Pamela-keel@uiowa.edu
1Department of Psychology, Harvard University, Cambridge Massachusetts
2Department of Psychology, University of Iowa, Iowa City, Iowa Published online 21 June 2005 in Wiley InterScience
(www.interscience.wiley.com). DOI: 10.1002/eat.20140
© 2005 Wiley Periodicals, Inc.

On the quantitative dimension, exercise becomes excessive when its duration, frequency, or intensity exceeds what is required for physical health and increases the risk of physical injury (Davis & Fox, 1993); Davis, Fox, Cowles, Hastings, & Schawss, 1990). Excessive quantities of exercise have been associated with disturbed eating attitudes (Richert & Hummers, 1986) and weight preoccupation (Davis et al., 1990). On the qualitative dimension, exercise becomes compulsive when it is characterized by maintenance of a rigid exercise schedule, increasing priority over other activities to maintain the pattern of the exercise, detailed record keeping, and feelings of guilt and anxiety over missed exercise sessions (Wyatt, 1997), The compulsive quality of exercise has been associated with increased drive for thinness (Gulker, Laskis, & Kuba, 2001; Wyatt, 1997) and increased disordered eating attitudes and behaviors (Brehm & Steffen, 1998; Matheson & Crawford-Wright, 2000).
In the 4th ed. Of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association, 1994), the term "excessive" has been used to characterize exercise as a symptom of bulimia nervosa. This suggests that the quantity of the physical activity is symptomatic, rather than its quality, and this suggestion is reflected in the assessment of intense exercising using the Eating Disorders Examination (EDE; Fairburn & Cooper, 1993, p., 347). Although the dimensions of excessive

and compulsive are likely related to one another, the use of one dimension to define unhealthy exercise does not necessarily encapsulate the other. People who exercise excessively may do so because of a compulsive drive to exercise. However, it is possible for a person to have a high quantity of exercise but demonstrate flexibility concerning missed sessions and not experience anxiety over changes in exercise schedule. Similarly, it is possible for a person to compulsively follow an exercise regimen that does not consist of an excessive amount of exercise. Unfortunately, the likely association between dimensions may produce correlations with disordered eating that are largely carried by only one of the two aspects of unhealthy exercise.
To directly compare excessive and compulsive aspects of exercise, Seigel and Hetta (2001) administered a 24-item questionnaire to a random sample of 1,000 female Swedish young adults. The questionnaire was developed to assess exercise amount (frequency, duration, and intensity), obligatory exercising attitudes, body image, and eating attitudes. Participants were subsequently categorized as high-level exercisers (HiEx) if they exercised six or more times per week for an hour or more, and is having obligatory attitudes to exercise (ObEx) if they obtained greater than 95th percentile on th composite score of obligatory exercise items. High levels of exercise were associated with obligatory attitudes to exercise (Seigel & Hetta, 2001). However, only ObEx was associated with significant body image problems, recurrent weight-reducing attempts, binging, and postprandial impulses to vomit, whereas HiEx ws not. The researchers concluded that the symptoms and behaviors of eating disorders seem to be linked to ObEx rather than to the mere quantity of exercise.
Although studies have measured both the excessive quantity and the compulsive quality of exercise in their participants, comparisons between the relative degrees which these dimensions predict disordered eating attitudes and behaviors remain limited. To our knowledge, no study has examined the extent to which excessive and compulsive features of exercise are associated with disordered eating in multivariate analyses. This approach is needed, due to the positive association found between these dimensions (Seigel & Hetta, 2001). In addition, previous studies often do not control for whether exercise is motivated by a desire to control weight. Although exercise may become unhealthy in ways that are unrelated to eating disorder pathology, when considering the relationship between unhealthy exercise and eating disorders, weight control motivation is of central importance (McDonald & Thompson, 1992). Finally, studies restricted to female participants may not produce results that generalize to men with disordered eating patterns. Thus, the current study sought to evaluate the hypothesis that, among individuals who exercise for weight control, compulsive exercise will be more predictive of disordered eating than excessive exercise.

Method

Participants

Participants (N = 265) were recruited from a university with flyers that offered entry in a $100 lottery or course credit for participation in a web-based survey. The mean (SD) age of the participants was 19.7 (1.5) years, and 162 (61%) participants were women. This group was 62.3% Caucasian, 12.5% Asian, 4.9% Hispanic, 4.9% Black, 0.4% Native American, 0.4% Hawaiian, and 14.2% biracial/other (0.8% missing). Participants ranged from varsity athletes to individuals who did not regularly did not exercise. The current study was approved by the human subjects committee at Harvard University.

Procedure
Participants visited the website for the study at their convenience. The website provided a brief description of the study, consent form, and instructions. Encryption software ensured privacy of participant information, and no record linked participant identities with survey data. The web-based survey contained items from three existing self-report instruments an a background questionnaire. Internal reliability statistics are reported in Table 1.

Disordered Eating Attitudes and Behaviors
The Drive for Thinness, Bulimia, and Body Dissatisfaction subscales of the Eating Disorder Inventory (EDI; Garner, Olmsted, & Polivy, 1983) were used, in combination, to tap a general construct that focuses on a concern with shape, weight, and eating that has demonstrated good psychometric properties (Welch, Hall, & Walkey, 1988).

Reasons for Exercise
The Reasons for Exercise Inventory (REI) assesses the importance of different motives for exercise (Silberstein, Mishkind, Streigel-Moore, Timko & Rodin, 1989). We used the factor structure suggested by Cash, Novy, and Grant (1994) based on the improved internal consistency associated with the resulting four subscales: Fitness/Health Management, Appearance/Weight Management, Stress/Mood Management, and Socializing. A median
split on the Appearance/Weight Management subscale score was used to categorize participants as appearance exercisers (n = 136; M = 5.57. SD = .66) or nonappearance exercisers (n = 129; M = 3.38, SD = .89).

Compulsive Exercise
The Obligatory Exercise Questionnaire (OEQ; Pasman & Thompson, 1988) measures compulsion to exercise. Although the OEQ contains two exercise quantity questions, all remaining items tap compulsion to exercise, and the OEQ has demonstrated excellent psychometric properties (Pasman & Thompson, 1988; Thompson & Pasman, 1991).

Exercise Activity and Descriptive Information
Demographic data, exercise frequency and duration, weight control practices, and history of an eating disorder were obtained through self-report. Likewise, body mass index (BMI) was calculated from self-reported height and weight, which provide reasonable accurate estimates of actual values (Stunkard & Albaum, 1981).

Statiscal Analyses
Data were downloaded from the web-server and and statistically analyzed using Statistical Package of Social Science (SPSS) software for Windows (Version 11.5; SPSS, Chicago, IL). The Kolmogorov-Smirnov test revealed that scores on most measures were not normally distributed. Thus, scores were transformed before use in parametric analyses. Associations between categorical variables were assessed by non parametric analyses. Hierarchical multiple regression analyses were computed to examine the predictive relationship of OEQ scores (compulsive quality) and average weekly exercise time (excessive quantity) for EDI composite scores in appearance and nonappearance exercisers. With EDI composite scores as the criterion variable, BMI and gender were controlled in an initial block, and OEQ score and average minutes of weekly exercise were entered into the regressions in a second block. Because of missing data, the number of participants and degrees of freedom vary slightly among analyses. Outliers were examined to ensure that they did not unduly influence

Table 1. Comparisons of high versus low appearance motivated exercisers

High Appearance-Motivated Nonappearance
Exercisers (n = 136) Exercisers (n = 129)
________________________ __________________

Variable (Alpha Reliability) M SD M SD t df p

Eating Disorder Inventory-2 subscales
Drive for Thinness (.904) 24.76 9.17 14.24 5.92 11.14 232.6 .000
Bulimia (.756) 17.28 6.90 12.68 3.94 6.70 216.9 .000
Body Dissatisfaction (.906) 34.67 11.12 24.91 9.11 7.73 251.3 .000
Disordered Eating Attitudes/Behaviors a (948) 76.68 23.53 51.33 16.03 10.16 237.4 .000

Reasons for Exercise Inventory subscales
Appearance/Weight Management (.881) 5.57 0.66 3.38 0.99 22.57 234.2 .000
Health/Fitness Management (.854) 4.97 1.14 5.00 1.07 -0.24 263.0 .807
Stress/Mood Management (.791) 4.19 1.53 4.15 1.38 .23 261.0 .822
Socializing (.617) 2.52 1.20 2.38 1.20 .91 258.0 .366

Obligatory exercise score (.898) 46.79 10.50 42.93 10.69 2.96 263.0 .003

Average exercise time (min/wk) 241.44 217.75 299.29 277.42 -1.87 238.9 .003

BMIb

Note: BMI = body mass index
a Composite score of drive For thnniss, Bulimia, and Body Dissaticsfaction EDI-2 subscales
B BMI = 703 x weight (kg) / height 2 (m).

RESULTS

Consistent with previous research, females had significantly higher composite EDI scores, the(260) = 7.34, p < .001, and lower BMI, the(176.35) = -4.35 p <.001, than males. However, there were no significant gender differences on any other measure. Reflecting this, there was no significant difference between men and women in the proportion of appearance versus nonappearance exercisers, x2(1,n=625) = .220, p=.639. There also were no significant differences between appearance and nonappearance exercisers in age, the(262) =.035, p=.972, or ethnicity, x2(6,n = 263) = 3.26, p = .776.
Table 1 presents comparisons between appearance and nonappearance exercisers on measures. Appearance exercisers had significantly higher composite EDI scores, OEQ scores, and BMI than nonappearance exercisers. However, there were no significant differences between groups on other motivations to exercise, the number of days per week spent exercising, x2(6,n = 263) = 10.52, p = .105; typical duration of exercise, x2 (5, n = 265) = 8.70, p = .122; or number of years engaged in regular exercise a x2(7,n = 265) = 2.87, p = .675. Overall, there was a trend-level difference between appearance and nonappearance exercisers on average exercise time (Table 1).
Table 2 summarizes weight control methods. Appearance exercisers were more likely to endorse use of dieting, extra exercise, fasting, vomiting, laxatives and diet pills than nonappearance exercisers. Appearance exercisers were also more likely to report a history of eating disorders, including anorexia nervosa and bulimia nervosa but not binge eating disorder.
Correlation matrices for appearance and nonappearance exercisers are presented in Table 3. For appearance exercisers, disordered eating was significantly and positively associated with Appearance/Weight Management. For nonappearance exercisers, disordered eating was associated only with Appearance/Weight Management. Thus the OE was not significantly related to disordered eating for nonappearance exercisers, and average exercise time was not significantly correlated with disordered eating for either appearance or nonappearance exercisers. However, exercise time had a strong, positive association with OE score and was positively correlated with Stress/Mood Management and Socializing in both groups. Both OE scores and average exercise time were positively associated with BAI for nonappearance exercisers. Finally, OE score was positively related to Stress/Mood
Management and Socializing in both groups. Both OE scores and average exercise time were positively associated with BAI for nonappearance exercisers. Finally OE score was positively related to Stress/Mood Management for both groups, and with Fitness/Health Management and Socializing for nonappearance exercisers.
Regression analyses for appearance and nonappearance exercisers, gender and BAI explained 36.5% of the variance in composite EDI scores. Female gender and BAI were both positive predictors. Block 2 accounted for an additional 9.3% of the explained variance, with BAI and gender maintaining unique contributions. The OE score was a positive predictor of composite EDI scores, whereas average weekly exercise time was a negative predictor. For nonappearance exercisers, gender and BAI were the only significant contributing variables, accounting for 21.6% of the explained variance in composite EDI scores. The addition of OE score and average weekly exercise time did not significantly explain the additional variance.
__________________

Table 2. Comparison of eating disorder history and weight control methods between exercise groups

High-Appearance-Motivated Exercisers Low Appearance-Motivated Exercisers
_______________________________ ________________________________

n % n % x2(1) p

Eating disorder history 21 16.0 5 4.0 10.15 .001
Anorexia nervosa 11 8.4 2 1.6 6.13 .013
Bulimia nervosa 13 9.9 2 1.6 8.03 .005
Binge eating
disorder 2 1.5 1 0.8 ____a .518
Dieting 102 75.0 40 31.08 51.51 .000
Extra exercise 102 75.0 56 43.4 27.44 .000
Fasting 42 30.9 10 7.8 22.46 .000
Vomiting 23 16.9 5 3.9 11.91 .001
Laxatives 15 11.0 1 0.8 12.27 .000
Diet pills 25 18.4 1 0.8 23.19 .000
Diuretics 5 3.7 0 0.0 4.83 .028

a Assumption for the chi-square test (that all expected counts should be greater than 5) was not met. The exact significance (two-tailed) of the Fisher’s exact test is given instead.

Table 3. Correlation matrices in high and low appearance- motivated exercisers

n 1 2 3 4 5 6 7 8

High appearance motivated
Disordered eating attitudes behaviors 135 __ .41** -.23* .00 _.11 .27* .06 .05
Appearance/weight management 136 __ -.04 -.04 _.04 .16 .05 _.00
Health/fitness management 136 ___ .45** .21 .22 .20 .04
Stress/mood Management 135 ___ .17 .41** .34** _.14
Socializing 132 ___ .19 .26** .22
Obligatory exercise score 136 ___ .64** _.03
Average weekly exercise time (min/wk) 135 ___ .01
BAI a 133 __

Low appearance motivated
Disordered eating attitudes behaviors 127 __ .31** _.14 .13 _.05 _.07 _.14 .01
Appearance/weight management 129 __ _.05 .04 .16 .02 _.19 .11
Health/fitness management 129 __ .44** .17 .33** .18 _.07
Stress/mood management 128 __ .26* .41** .26 _.02
Socializing 128 __ .30* .27* .05
Obligatory exercise score 129 __ .77** .24*
Average weekly exercise time (min/wk) 127 __ .35**
BAI 128 ___

a Note: BAI = body mass index.
* p < .01. **p < .001.

Table 4. Regression analyses in high and low appearance-motivated exercisers

High Appearance-Motivated Low Appearance-Motivated
Exercisers (n = 136) Exercisers (n = 129)
_______________________ ________________________

Block 1: Control Block 2: Exercise Block 1: Control Block 2: Exercise
Variables Measures Variables Measures
_____________ ______________ _____________ ______________

ß t p ß t p ß t p ß t p

Control variables
Gendera .63 8.57 .000** .63 9.15 .000** .49 5.80 .000** .47 5.55 .000
Body mass index .26 3.45 .001* .27 3.92 .000*** .17 1.98 .05 .20 2.26 .03

Exercise measures
Average exercise time _.23 _2.74 .007* _.16 _1.23 .22
OE Score .40 4.67 .000** .08 .61 .54
R 2 .09 .01
R 2 .37 .46 .22 .23
Adjusted R 2 .36 .44 .20 .20
F_ .000*** 10.95 .000** .000** .87 .42
F 37.04 .000** 26.86 .000** 16.81 .000** 8.82 .000**

Note: OE = Obligatory Exercise Questionnaire
a 0 = Male; 1 = female
* p < .01. **pp < .001

DISCUSSION

The current study examined the relative contributions of the excessive quantity versus the compulsive quality of exercise to variance in disordered eating attitudes and behaviors. In a multivariate analysis, compulsion to exercise was a positive predictor of disordered eating symptoms whereas exercise time was a negative predictor in appearance exercisers. This supports characterizing the unhealthy exercise that is symptomatic of bulimia nervosa as compulsive rather than excessive and indicates a need to alter the way that exercise is assessed as a symptom of the disorder. Neither compulsive nor excessive aspects of exercise were predictive of disordered eating among nonappearance exercisers. Thus, compulsive exercise in and of itself is not indicative of eating pathology.

Precious findings of a positive association between exercise quantity and eating disorder symptoms (Davis et al., 1990; Richert & Hummes, 1986) may result from the positive association between exercise quantity and the compulsive quality of exercise. Because no other study has examined the association between exercise quality and disordered eating while controlling for exercise quality, it is difficult to determine whether or results could be consistent with previous findings. It is noteworthy that we did not find a direct positive correlation between exercise quantity and and disordered eating. However, this is consistent with previous research (Davis & Fox, 1993; Seigel & Hetta, 2001). Moreover, the inclusion of participants with a broad range of backgrounds, from individuals who do not exercise to varsity athletes may explain inconsistencies between our results and those form other studies.
Although hypotheses about the degree to which other reasons for exercise would be associated with disordered eating were not initially made, it was noteworthy that exercise for health and fitness was associated with less disordered eating, but a greater compulsion to exercise. This provides further support for earlier findings (McDonald & Thompson, 1992), suggesting that different motives for exercise may be differentially associated with eating and weight-related disturbance. This difference may reflect the fact that appearance and weight may be characterized as predominantly extrinsic motivations, whereas health and fitness are more intrinsic (Markland & Ingledew, 1997).
The current study possessed notable strengths. Use of a web-based questionnaire format standardized the procedure and made answers entirely anonymous, and anonymity has been associated with increased sensitivity in assessing potentially embarrassing behaviors (Keel, Crow, Davis, & Mitchell, 2002). The inclusion of men and women and broad range of exercisers also enhances the generalization of results. In addition, the large sample size provided adequate power for detecting relationships between variables, as did the use of reliable and well-validated measures. Yet, it is also important to recognize limitations of the study design. Results may not generalize to groups that differ from those in the current study in demographic features or ascertainment. In addition, results do not address the question of directionality. Do disordered eating behaviors and attitudes tend to lead those who exercise for appearance to do so compulsively? Or, do those exercising for such reasons tend to develop disordered eating behaviors and attitudes? Or, is there a compulsive personality trait that contributes both to developing compulsive exercise patterns and disturbed eating? Longitudinal research is required to address questions of chronological order.
Overall, the current study provided support for our hypothesis that the compulsory quality, rather than the excessive quantity, of exercise is a better characterization of exercise that is related to eating disorders. Although further research is necessary to replicate these findings, our results suggest a need to alter the way exercise is defined and assessed as a bulimic symptom. Specifically, when depicting exercise that serves as a compensatory method in bulimia nervosa, compulsive may be a more clinically useful characterization than excessive.

The authors acknowledge the contributions of the following individuals: Safia Jackson, Brent Eveleth, Bree Herne, Steve Staples, and Amy Wiseman, as well as financial support form the Department of Psychology, Harvard University.




______________

References

American Psychiatric Association. (1994). Diagnostic and statistical
Manual of mental disorders (4th ed.). Washington, DC:
Author.
Brehm, B.J. & Steffen, J.J. (1998). Relation between obligatory
exercise and eating disorders. American Journal of Health Be-
havior. 22, 108-119.
Cash, T.F., Novy, P.L., & Gran, J.R. (1994). Why do women ex-
excise? Factor analysis and further validation of the Reasons for
Exercise Inventory. Perceptual and Motor Skills, 78,539-544.
Davis, C., & Fox, J. (19930. Excessive exercise and weight pre-
occupation in women. Addictive Behaviors, 18, 201-211.
Davis, C., Fox, J., Cowles, M. Hastings, P., & Schwass K. (1990).
The functional role of exercise in women. Journal of
Psychosomatic Research, 34, 563-574.
Fairburn, C.G., & Cooper, Z. (1993). The Eating Disorder
Examination (12th ed.) In C.G. Fiarurn & G.T. Wilson
(Eds.), Binge eating: Nature, assessment, and treatment
(Pp. 317-360). New York; Guilford Press.
Garner, D.M., Olmsted, 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, 15-34/
Gulker, M.G., Laskis, T.A., & Kuba, S.A. (2001). Do excessive
exercisers have a higher rate of obsessive-compulsive
symptomatology? Psychology Health and Medicine, 6, 387-
376.
Imm, P.S., & Pruitt, J. (1991). Body shape satisfaction in female
exercisers and nonexercisors. Women and Health, 17 87-96.
Keel, P.K., Crow, S.J., Davis, T.L., & Mitchell, J.E. (2002). Assess-
ment of eating disorders: Comparison of interview and question-
naire data from a long-term follow-up study of bulimia nervosa.
Journal of Psychosomatic Research, 53, 1043-1047.
Markland, D., & Ingledew, D.K. (1997). The measurement of
exercise motives: Factorial validity and invariance across
gender of a revised Exercise Motivations Inventory. British
Journal of Health Psychology, 2, 361-376.
Matheson, H., & Crawford-Wright, A. (2000). An examination of
eating disorder profiles in student obligatory and non-
obligatory exercisers. Journal of Sport Behavior,23, 42-50.
McDonald, K. & Thompson, J.K. (1988). Body image and eating
disturbance, body image dissatisfaction, and reasons for
exercising: Gender differences and correlational findings
International Journal of Eating Disorders, 11,289-292.
Pasman, L. & Thompson, J.K. (1988). Body image and eating
disturbance in obligatory runners, obligatory weight lifters,
and sedentary individuals. International Journal of Eating
Disorders, 7, 759-769.
Richert, A., & Hummers, J. (1986). Patterns of physical activity
in college students at possible risk for eating disorders
International Journal of Eating Disorders, 5, 757-763.
Seigel, K., & Hetta, J. (2001). Exercise and eating disorder
symptoms among young females. Eating and Weight
Disorders: EWD, 6, 32-39.
Silberstein, L.R., Mishkind, M.S., Striegel-Moore, R.H., Timko,
& Rodin, J. (1989). Men and their bodies: A comparison
of homosexual and heterosexual men. Psychosomatic
Medicine, 51, 337-346.
Stunkard, A.J., & Albaum, J.M. (1981). The accuracy of self-
reported weights. The American Journal of Clinical Nutrition,
34, 1593-1599.
Szabo, A. (2000). Physical activity and psychological dysfunction.
In S.J.H. Biddle, K.R. Fox, & S.H. Boutcher (Eds.). Physical
activity and psychological well-being (Chapter 7). New York
Routledge.
Thompson, J.K., & Pasman, L. (1991). The obligatory exercise
questionnaire. Behavior Therapist, 14, 137.
Welch, G., Hall, A., & Walkey, F., (1988). The factor structure of
Eating Disorders Inventory. Journal of Clinical Psychology, 44
51-56.
Wyatt, L.M., (1997). Obsessive-compulsiveness and disordered
eating in obligatory and non-obligatory exercisers. Unpublished
doctoral dissertation, California School of Professional
Psychology, Los Angeles.


Family Meals: One Deterrent to Disordered Eating?

Regular, structured and harmonious family meals with all family members preset may play a role in preventing unhealthy weight-control behaviors among teens, according to the results of a recent study
(J Adolesc Health 2004; 35:350).
Many factors can contribute to unhealthy weight-control practices and other disordered eating patterns among teenagers, according to Dr. Dianne Neumark-Sztainer and colleagues at the University of Minnesota, Minneapolis. These include societal pressures to be thin, parents’ attitudes toward weight, family relationships, peer dieting practices, and perceptions of body image. However, the structure and atmosphere of meals at home are important as well, according to the researchers.
After surveying 4,476 ethnically diverse adolescents from public middle and senior high schools participating in the Project EAT (Eating Among Teens) study, and gathering anthropometric data as well, the Minnesota researchers found several patterns of disordered eating related to family meals. Adolescent girls who reported more frequent family meals, high priority of family meals, a positive atmosphere at meals, and more structured meals had less risk for engaging in unhealthy weight control behaviors and chronic dieting. The atmosphere at family meals was also inversely related with binge eating. Girls who ate three to four famiy meals per week had approximately one-third the risk of extreme weight control practices compared to girls who did not eat any family meals in the past week.
Among teenage boys, the frequency of family meals and all measures of the environment of family meals were inversely associated with extreme weight-control behaviors. The priority given to family meals and the atmosphere of the meals were inversely associated with less extreme unhealthy weight-control
behaviors. Any association with binge eating was inconsistent.
Family meal patterns were inversely associated with chronic dieting among the girls but not among the boys. Gender differences may have been a factor, owing to increased sensitivity of girls to family nuances, differential interpretation of the term "dieting" used in this study, and or/lower frequencies of boys reporting chronic dieting than girls, according to the authors.

Binge eating
The patterns of association between family meals were somewhat different for binge eating. Among girls, the strongest correlate of binge eating was the atmosphere at family meals (this factor did not, however, remain statistically significant after taking into account overall family connectedness). The findings suggest that girls may respond to a negative familial environment through binge eating. Alternatively, girls who are already binge eating may perceive the family environment as more negative. The same pattern was not reported among boys.

Clinical applications
Findings from this study suggest that family meals have an important part to play in preventing unhealthy weight-control behaviors among teenagers. The authors suggest that healthcare providers working with youth and their parents take the time to discuss family meal patterns and explore realistic, real-world strategies for improving the environment of family meals.

(Taken from Eating Disorders Review May/June 2005)

Regular Article

Media-Portrayed Idealized Images, Body Shame, and Appearance Anxiety

Fiona Monro, B Psychology (Hons)Abstract:
Gail Huon, PhD*

ABSTRACT
Objective: This study was designed to determine the effects of media-portrayed idealized images on young women’s body shame and appearance anxiety, and to establish whether the effects depend on advertisement type and on participant self-objectification.

Method: Participants were 39 email university students. Twenty- four magazine advertisements comprised 12 body-related and 12 non-body related products, one half of each with, and the other one half without, idealized images. Pre-exposure and post exposure body shame and appearance anxiety measures were recorded.

Results: Appearance anxiety increased after viewing advertisements featuring idealized images. There was also a significant interaction between self-objectification level and idealized body (presence vs absence). No differences emerged for body-related compared with non-body related

product advertisements. The only result for body shame was a main effect for time. Participants; body shame increased after exposure to idealized images, irrespective of advertisement type.

Discussion: Although our findings revealed that media-portrayed idealized images detrimentally affect the body image of young women, they highlight the individual differences in vulnerability and the different effects for different components of body image. These results are discussed in terms of their implications for the prevention and early intervention of body image. © 2005 by Wiley Periodicals, Inc.

Keywords: appearance anxiety; body image; body shame; idealized images; self-objectification

Introduction

Western women are encouraged to achieve an idealized body shape that is tubular and thin (Groesz, Levine, & Murnen, 2002; Guillen & Barr, 1994. Media-portrayed images, especially those presented in the context of advertisements for dieting and weight-altering products, promote the idea that body shape and size are flexible, and that achieving the thin ideal is realtively easy (Brownell, 1991). So, the mass media are generally agreed to be an influential source of images and messages about the idealized body that women and girls are expected to strive for (e.g., Anderson & DiDomenico, 1992; Botta, 1999;

Accepted 21 June 2004
School of Psychology, University of New Sough Wales, Sydney, Australia
*Correspondence to: Professor Gail Huon, MD, School of Psychology, University of New South Wales, Sydney, NSW, 2052,Australia. E-mail: G.Huon@unsw.edu.au
Published online 21 June 2005 in Wiley InterScience (www.interscience.wiley.com. DO1: 10,1002/eat20153
© 2005 Wiley Periodicals, Inc.

Botta, 1999; Field, Camargo, Taylor, Berkey, & Colditz, 1999).
Some researchers have found strong positive correlations between the frequency with which adolescent girls read magazine dieting articles and their desire to change their body weight and shape (e.g., Field et al., 1999; Utter, Newmark-Sztainer, Wall & Story, 2003). Others have shown that television viewing is associated with body dissatisfaction (Botta, 1999; Field et al., 2001; Harrison & Cantor, 1997), although the correlation appears to hold only for specific program and article types. Tiggemann and Pickering (1996) found, for example, that the viewing of serials and soap operas was associated with body dissatisfactions, and watching music video clips was related to drive for thinness. However, overall television viewing was not related to body dissatisfaction, perceived weight , or drive for thinness. In other words, there is some evidence that exposure to media-portrayed idealized images is associate with negative feelings about the body, and with the desire to alter its weight and shape. From correlational studies, however, we cannot be sure about the direction of the effect.
Some researchers have, therefore, compared the effects of
exposure to thin ideal media images with a control condition in which no idealized body was present. For example, Turner, Hamilton, Jacobs, Angood, and Hovde (1997) showed that after viewing magazine images , body dissatisfaction was greater in the experimental than in the control condition, and Lavine, Sweeney, and Wagner (1999) showed a similar result for viewing television images. However, neither study included baseline measures to establish the impact of exposure with some certainty. In contrast, Cattarin, Thompson, Thomas, and Williams (2000) and Pinhas, Toner, Ali, Garfinkel,and Stuckless (1999) measured female students’ body dissatisfaction and mood both before and after exposure to idealized magazine or television images. Those authors found that the increase in body dissatisfaction and negative mood from before to after viewing was greater in the experimental than in the control group.
Yet, it is unlikely that all young women are affected by idealized images to the same degree. Women who are more likely to be vulnerable are those whose attention is focused on appearance. According to Fredrickson and Roberts (1997), some women are more likely than others to see the self from the point of view of an outsider, and regard their body as an object other people’s gaze. The authors coined the term self-objectification to characterize the tendency to value appearance over and above ability or any other attributes (Fredrickson & Roberts, 1997). Such focus on appearance is believe to result in increased body shame and appearance anxiety, especially in conditions where the body is subjected to scrutiny (Frederickson & Roberts, 1997; McKinley & Hyde, 1996). Exposure to media-portrayed idealized images focuses attention on the body, and encourages scrutiny of the acceptability of its shape and size (Kilbourne, 1994; Myers & Biocca, 1992).
The focus of attention on the idealized images in some advertisements is direct. Thin idealized bodies are used to persuade women to buy products that help to improve the appearance of their body. In others, it is indirect. Idealized body images are also used to persuade people to buy non-body related products. An important research question is this: Are the effects of the direct and indirect focus on idealized bodies the same? Idealized images associated with products that are designed alter the body, for example, through dieting or exercise, focus attention explicitly on its unacceptability. They might be expected, therefore, to affect feelings of body shame and appearance anxiety to a greater extent than advertisements that simply exploit the use of idealized body images for non-body-related products.
This study was concerned with the effects on young women’s body shame and appearance anxiety of exposure to media-portrayed idealized images. We sought to determine the effects of exposure not only to advertisements with and without the presence of an idealized body, but importantly, of exposure to idealized bodies in the contexts of body-related and non-body-related advertisements. We also believed it important to establish whether the effects would be different for women who differ in their level of self-objectification.
We predicted that, when compared with images that do not feature idealized bodies, exposure to idealized bodies would result in increased body shame and appearance anxiety. We also expected that the effect would be more pronounced for body-related products, and for high self-objectifiers.


Method

Overview and Design

This experiment was concerned with the effects of exposure to idealized body magazine images on young women’s body shame and appearance anxiety. The focus was on whether high self-objectifiers would respond differently than low self-objectifiers. Further, it was examined whether the effects would be different for body-related and non-body-related product advertising images. Thus, the experiment involved a 2 x (2) x (2) factorial design. The first factor was self-objectification level (high vs. low), the second factor was advertisement type (body-related vs. non-body related), and the third was idealized body (present vs. absent). The dependent measure tested in the analysis of variance (ANOVA) were body shame and appearance post-pre difference scores.

Participants

Thirty-nine women 17-37 years old participated in this experiment. They included psychology students (n =28) who participated in exchange for course credit, and other college students who volunteered to be involved to learn about psychological research (n = 11)

Materials

The materials used in this study included pretest measures, experimental stimuli, and dependent measures.

Pretest Measures. Questionnaires were used to provide baseline measures for restraint status, appearance anxiety, body shame, self-objectification, self-monitoring, and exposure to magazines. The Restraint Questionnaire (RQ; Herman & Polivy, 1975) was used to assess the extent to which participants are attitudinally and behaviorally concerned with their weight and dieting. The 12-item Social Physique/Appearance Anxiety Scale (SPAS; Hart, Leary & Rejeski, 1989), which assesses how anxious participants feel when their bodies are observed or evaluated by others, was used to measure appearance anxiety in this experiment. Participants responded to each question on a 5-point scale from definitely disagree to definitely agree. This scale has high reliability (? = .90) and good construct validity ® =.50) (Hart et al., 1989). A total score was calculated by summing responses. Total scores range from 12 to 60, with high scores indicating a high level of appearance anxiety. The eight-item Body Shame subscale of the Objectified Body Consciousness Scale (BS-OBCS; McKinley & Hyde, 1996) indicates how ashamed participants are of the their body, and the extent to which participants have internalized the cultural thin ideal. Scores on the scale range from 8-14, with high scores indicating a higher level of body shame. The eight-item Surveillance subscale of the Objectified Body Consciousness Scale (S-OBCS; McKinley & Hyde, 1996) assesses the extent to which an individual monitors her appearance and views her body as if an observer. Possible scores on this scale range form 8-40. High levels of self-monitoring (surveillance) manifest as high scores on this scale. The Self-Objectification Questionnaire (SOQ; Fredrickson, Roberts, Noll, Quinn, & Twenge, 1998) was used to assess self-objectification in this study. This scale requires participants to rank a series of statements regarding the importance that they place on appearance and abilities on a scale from 0 to 9. A self-objectification score is obtained b subtracting the ranks assigned to ability statements from the ranks assigned to appearance statements so that possible scores on this scale range from -25 to 25.
Pretest scales to provide baseline measures were also obtained for appearance anxiety, body shame, and self-monitoring using visual analog scales (VAS).

Experimental Stimuli. Twenty-four images were selected from popular magazines (e.g., Cleo, Cosmopolitan, Men’s Health, Who Weekly). Twelve images were advertisements for body-related products, and 12 were advertisements for non- body-related products. Within each of these advertisement types, one half of the advertisements featured idealized bodies. The other one half were produced with using computer manipulation via Adobe Photoshop (version 7.0; Adobe Systems, San Diego, CA), that is, they were the same, without the body in the image.

Dependent Measures. Experimental measures consisted of VAS for self-monitoring, appearance anxiety, and body shame. These were the same as the pretest measure VAS. Body shame and appearance anxiety difference scores were obtained by subtracting the initial VAS score on that measure from the postexposure score.
A free recall memory test of the advertisements that the participants had viewed was used to ensure processing of the images and enhance the cover story.

Procedure
This experiment was conducted in small groups. Participation took approximately 1 hr. This study was conducted in two sessions. Baseline data were collected in the initial session, and the experimental manipulation was conducted in the second session.
In the initial session, the experiment was presented to participants as an experiment investigating the effects of personality variables on the emotional experience of, and memory for, magazine advertisements, and participants completed consent forms and pretest measures as described earlier in the text.
In the second session, participants viewed 24 magazine advertisements, four blocks each of six idealized body present body-related product advertisements, idealized body absent non-body related product advertisements, and idealized body absent non-body-related product advertisements. Their order was counterbalanced. Participants were instructed that they would have 20 s to look at each advertisement, and that they should study the advertisements carefully and take in as much information as possible. At the end of each block of advertisements, participants were required to complete VAS assessing appearances anxiety and body shame. After viewing all the advertisements, and completing all the VAS, participants were given the free recall memory test. Finally participants were asked what they believed the aims of the study were, and then fully debriefed about the experiment. Their age, weight, and height details were also recorded at this time.

Results

Participant Characteristics

Participants’ self-objectification level was determined according to their score on the SOQ (Fredrickson et al., 1998). Two participants were excluded because of incomplete data. Thus, the final sample of participants was 37 women comprising 18 low self-objectifiers and
19 high self-objectifiers. The mean self-objectification scores for the two and high self-objectifying groups were -10.88 and 8.26, respectively. The difference between those means was significant (F = 49.6, F c = F0.05;1.36 = 4.113). The means for age, weight, height, body mass index (BMI), number of magazines read, and time spent reading magazines for both high and low self-objectifying groups are presented in Table 1. Table 1 shows that there were no other significant difference between the low and high self-objectifiers.

Effects of Idealized Body Exposure
The post-pre exposure difference scores were calculated for all participants, for all experimental conditions. Table 2 presents the means and standard deviations for those scores, separately for low and high self-objectifiers.

Body Shame. Table 2 shows that participants experienced less body shame after exposure to images without idealized bodies. This was the case for both non-body related product and body-related product advertisements. However after exposure to advertisements containing idealized bodies (non-body related and body-related product advertisements) their body shame was greater. This pattern of results occurred for both the high and low self-objectifiers. Thus, the main effect for idealized body condition was significant (F=1.35=5.77; Fc = F0.05; 1.35 = 4.12). However, the self-objectification main effect, the advertisement type main effect, and the interactions were not significant.

Appearance Anxiety. There was a significant main effect for idealized body condition (F1.35 = 14.368; Fc = F 0.05; 1.35 = 4,12). That is, irrespective of self-objectification level or advertisement type, participants experience less appearance anxiety after exposure to advertisements without an idealized body,and greater appearance anxiety after viewing advertisements featuring an idealized body.
The salient pattern of results in Table 2, however, is the significant two-way interaction between self-objectification level and idealized body condition (presence vs. absence) (F = 5.83; F0.0512;1.35 =5.85). To facilitate interpretation, this interaction is represented graphically in Figure 1.
The difference in appearance anxiety between body present and body absent conditions was greater for high self-objectifiers than for low self-objectifiers. High self-objectifiers experienced markedly greater appearance anxiety in the body present condition than the body absent condition. , whereas low self-objectifiers’ appearance anxiety was only slightly higher in the body present condition than in the body absent condition. There were no other significant interaction effects.

Table 1. Means and standard deviations for age, weight, height, BMI, number of magazines read, and time spent reading magazines, showing the similarities between the low and the high self-objectification groups

Self-Objectification Level

Total (n = 37) Low (n =18) High (n = 19)
-------------------- ----------------- ------------------
M SD M SD M SD F P

Age 19.68 3.23 20.22 4.49 19.16 1.07 1.01 .32
Weight 58.84 9.36 60.53 10.22 57.24 8.42 1.15 .29
Height 1.65 0.09 1.67 0.10 1.63 0.08 1.73 .20
BMI 21.61 2.68 21.68 2.75 21.54 2.69 0.27 .87
Magnoa 2.24 1.40 2.33 1.81 2.16 0.90 0.14 .71
Magexpb 3.43 2.78 2.78 2.70 4.05 2.80 1.99 .17

Note: BMI - body mass index
aNumber of magazines read per month
bHours per month spent reading magazines

Table 2. Means and standard deviations of body shame and appearance anxiety, after exposure to each of the magazine advertisement conditions, for low and high self-objectifying women

Low Self-Objectifiers (n = 18) High Self-Objectifiers (n = 19)

Idealized Body Absent Idealized Body Present Idealized Body Absent Idealized Body Present
------------------------------ -------------------------------- ------------------------------- --------------------------------

Nonbody Ads Body Ads Nonbody Ads Body Ads Nonbody Ads Body Ads Nonbody Ads Body Ads
------------------- -------------- ------------------ ------------- ------------------ ------------- ------------------- --------------

M SD M SD M SD M SD M SD M SD M SD M SD

Body Shamea -3.08 19.75 -1.12 18.37 1.52 24.26 2.77 21.94 -5.03 19.47 6.16 23.59 8.03 17.01 7.02 19.33
Appearance Anxietyb -9.78 21.62 -4.75 21.97 -0.15 22.88 -4.90 29.85 -5.89 27.51 -0.05 24.89 9.45 27.12 11.76 21.16

Note: Ads = Advertisements
aBody shame deviation score (post-pre).
BAppearance anxiety deviation score (post-pre)

Figure 1. Significant interaction between self-objectification level and idealized body for appearance anxiety difference scores. Diamonds = low self-objectification; squares = high self-objectification.