happy; more happy; the same as now; less happy; much less happy." Items included happy, successful, healthy, good or bad looking, number of friends, number ...
Journal of Youth and Adolescence, Vol. 20, No. 3, 1991
Body Image Satisfaction, Dieting Beliefs, and Weight Loss Behaviors in Adolescent Girls and Boys Susan J. Paxton, 1,7 Eleanor H. Wertheim, 2 Kay Gibbons, 3 George I. Szmukler, 4 Lynne Hillier, 5 and Janice L. Petrovich 6
Received July 14, 1990;acceptedAugust 20, 1990
Body image and weight loss beliefs and behaviors were assessed in 341 female and 221 male high school students. Estimates of body dissatisfaction varied depending on the measurement strategy used. Despite having similar weight distributions around the expected norm, girls were significantly more dissatisfied with their bodies than boys. Body Mass lndex was positively related to body dissatisfaction in girls and boys, while higher exercise levels were related to higher body satisfaction in boys. Nearly two-thirds of girls and boys believed being thinner would have an impact on their lives, but the majority of girls 1Lecturer, La Trobe University, Melbourne, Victoria, Australia, 3083. Received Ph.D. from University of Tasmania. Main research interest in body image, weight loss behaviors, and eating disorders. 2Senior Lecturer, La Trobe University, Melbourne, Victoria, Australia, 3083. Received Ph.D. from University of Connecticut. Main research interest in body image, eating disorders and weight loss behaviors. 3Chief Dietitian, Royal Melbourne Hospital, Parkville, Melbourne, Victoria, Australia, 3052. Dietitian degree from Melbourne University, Melbourne. Main research interest in eating practices and clinical outcome. 4Consultant Psychiatrist, Royal Melbourne Hospital, Senior Associate, University of Melbourne, Parkville, Melbourne, Victoria, Australia, 3052. M.D. from University of Melbourne, D.P.M. from U.K. FRC Psych. Main research interests in eating disorders. 5Research Assistant, La Trobe University, Melbourne, Victoria, Australia, 3083. B.B.Sc.(Hons.) received from La Trobe University, Melbourne, B.A. received from University of Minnesota, Duluth. Main research interest in attitudes and attributions. 6Research Assistant at La Trobe University, Melbourne, Victoria, Australia, 3083. B.A. Received from University of Minnesota, Duluth. Main research interest in adolescence and eating disorders. 7"I"o whom reprint requests should be addressed at Department of Behavioural Health Sciences, La Trobe University, Bundoora, Victoria, Australia, 3083. 361 0047-2891/91/0600-0361506.50 9 1991 Plenum Publishing
Corporation
362
Paxton et al.
believed this would be positive while the majority of boys believed this would be negative. Thirteen percent of female subjects reported using one or more extreme weight loss behavior at least weekly. Beliefs regarding the effectiveness of different weight loss measures were assessed. Weight loss behaviors in this Australian sample appear similar to comparable U.S. samples.
INTRODUCTION Research in the past decade has pointed to a high prevalence of body dissatisfaction and weight loss behaviors among adolescents, especially girls (e.g., Crowther et al., 1985; Desmond et al., 1986; Eisele et al., 1986; Rosen and Gross, 1987). These findings are a cause for concern since poor body image and weight loss behaviors have been associated with disordered eating in female adolescent samples (Attie and Brooks-Gunn, 1989; Crowther et al., 1985; Grant and Fodor, 1986; Gross and Rosen, 1988; Johnson et aL, 1984). In reviewing the literature, however, it becomes clear that not all studies find similar proportions of subjects reporting body dissatisfaction. Eisele et al. (1986) found that 78% of 12-14-year-old girls would prefer to weigh less and only 14% reported they were satisfied with their current weight. A similar level (83%) of dissatisfaction was found by Storz and Greene (1983) in 14-18-year-old girls, while Desmond et al. (1986) reported that 43% of females and 9% of males classified as a normal weight described themselves as heavy. In contrast, Cohn et al. (1987) found no significant difference between actual and ideal figure in 10.5-15-year-old girls, although 38% had a thinner ideal than their current figure. The variation between the studies may reflect different subject characteristics, or alternatively, different methods of assessing body satisfaction. A first aim of the present study, therefore, is to examine body satisfaction using a variety of methods of measurement to examine whether different methods yield different results. In addition, since two studies (Cohn et al., 1987; Eisele et aL, 1986) indicate that body dissatisfaction may increase after puberty in girls (though not in boys), a variety of age levels will be examined to assess whether variations in results may be attributed to age differences. Although previous studies indicate considerable body weight dissatisfaction, especially in girls, they do not provide information regarding why adolescents want a different body size. Therefore, this study examined the perceived impact of being thinner. Further, since it appears that boys place greater value on physical effectiveness than appearance (Grant and Fodor, 1986), satisfaction with physical fitness and beliefs regarding the benefits of physical fitness were also examined.
Body Image Satisfaction
363
Closely related to body dissatisfaction, in girls particularly, are weight loss behaviors. At any one time a large number of adolescent girls are on diets, with estimates ranging from 25% (Leon et aL, 1989; Williams et al., 1986) to 63% (Rosen and Gross, 1987). In addition, studies report the frequent use of both moderate weight loss methods, such as exercising or reducing snacks, and more extreme techniques, such as fasting or vomiting (Crowther et aL, 1985; Lachenmeyer and Muni-Brander, 1988; Rosen and Gross, 1987). Very little is known, however, about beliefs regarding these methods. Storz and Greene (1983) asked high school girls to rate the "desirability" of certain weight loss methods. Unexpectedly, the methods rated as m o r e desirable were not necessarily the methods most used. A possible reason for this is that students might have understood "desirable" to mean healthy rather than effective and students may not see the healthy methods as the most effective ones. Therefore, the present paper specifically explored adolescent beliefs about the effectiveness of various weight loss methods. Finally, the current study extends the body image and dieting literature to a sample of Australian adolescents. In a comprehensive cross cultural study of adolescent self-image, Offer et al. (1988) found body image in Australian girls to be the lowest and in Australian boys the second lowest of the Western industrialized countries examined. Further research findings suggest that cultural factors such as race (Howat and Saxton, 1988; Rosen and Gross, 1987) and socioeconomic status (Lachenmeyer and MuniBrander, 1988; Lachenmeyer et al., 1988; Rosen and Gross, 1987) may be related to body image and dieting variables. However, most research to date has been conducted in the United States. It is important to ascertain the prevalence of body image dissatisfaction and dieting behaviors in other adolescent samples, in order to assess how generalizable U.S. results are to other western communities.
METHOD
Subjects Three hundred and forty-one female and 221 male high school students, from seven schools in the greater Melbourne area, participated in this study. The schools were selected to represent a range of geographic and socioeconomic status areas (Levels 1-5) according to the Ross Indicator (Ross, 1984). They included a private girls and private boys school, a girls public school, and four coeducational public schools. In each school three or four classes, each from a different year level, were assessed. In total there were
364
Paxton et al.
132, 155, 140, 114, and 21 students from Year 7, 8, 9, 10, and 11, respectively. The study was conducted in compulsory subject classes in order to minimize biasing toward any particular student type. The age of subjects ranged between 11 and 18 years with a mean of 14.0 years (SD = 1.28). The sample represented a range of ethnic backgrounds. The parents of about half the students were born in Australia, while parents of the other students were born in Italy (9%), Greece (8%), Great Britain (8%), other European countries (9%), Asian countries (6.5%), the Middle East (4.5%), other countries (2.5%), and no response given (3%). There were no significant differences between males and females on age (t = -1.93, df = 557, p = 0.54) or mother's education level (t = -.66, df = 400, p = .51), while boys' fathers were reported as slightly more educated than girls' fathers (t = -2.19, df = 386, p = .03). A week prior to the study, students were given parental consent forms. Eighty-five percent and 90% were returned from girls and boys, respectively, and these 562 students participated in the study. In fact, only 3.5% refused to participate and the rest were absent on the occasion of data collection.
Procedure
In two adjacent class periods, students completed a number of selfreport questionnaires. While one experimenter guided the class and answered questions, a second took height and weight measurements in a separate room. Questionnaires were all coded to increase confidentiality.
Measures of Body Image Attitudes Body image attitudes were assessed using the Body Dissatisfaction and Drive for Thinness subscales of the Eating Disorders Inventory (EDI; Garner et al., 1983) and the Body Figure Perception Questionnaire (Fallon and Rozin, 1985; Stunkard et aL, 1980). The latter test consists of two sets of figures (female if the subject is female, male if the subject is male) in which nine figures range from emaciated to obese. On the first set, subjects rate what they regard to be their current figure; on the second set, they rate the size they would like to be. The scale is numbered from 1 (smallest) to 17, with odd numbers associated with the figures and even numbers midway between figures. The discrepancy between current and ideal figure (current figure-ideal figure) represents body dissatisfaction.
Body Image Satisfaction
365
Eight items measuring perceived impact of being thinner were included. Subjects indicated from five choices how aspects of their lives would change if they were thinner, e.g., "If you were thinner you would be: Much more happy; more happy; the same as now; less happy; much less happy." Items included happy, successful, healthy, good or bad looking, number of friends, number of dates, how intelligent one appears and how easily one gets what one wants. A similar 8 items assessed perceived impact of physical fitness with the sentence stem "If you were physically fitter . . . . " A 4-item Satisfaction with Fitness Scale was constructed including how satisfied subjects were with "your ability to perform normal physical activities," "your sporting ability," "how physically strong you are," and "your level of physical fitness." The reliability of this scale was assessed on a preliminary sample of 60 nursing students and adolescents, and was satisfactory, with an alpha of .83. In the current sample the alpha levels were .85 for females and .84 for males. All corrected item total correlations were greater than .20. Previous to actually weighing and measuring the subjects, subjects were asked to estimate their height and weight, and to rate their weight on a 5-point scale from very underweight to very overweight. They were also asked to estimate how many hours exercise they performed in a week.
Measures of Weight Loss Behaviors An Eating and Diet Questionnaire, constructed for this study, assessed dieting history, sources of information on weight loss methods, and peer and parental pressure to diet. In addition, subjects were given a list of weight loss strategies and asked to rate them according to whether they believed each method helps a lot, helps a little, or does not help to lose weight. Students were also given the option to indicate if they did not know of the particular method described. Finally, demographic information including parent's birthplace, educational level, and employment was collected.
RESULTS In the following analyses, an alpha level of .01 was used. Rather than including only subjects with complete data (which would have significantly decreased sample size and possibly biased the data), all possible subjects were used in each analysis and dfs are included. The only exception is that estimates of current vs. ideal figure, impact of thinness and fitness, and estimates of over- or underweight were made for subjects who had corn-
366
Paxton et al.
plete data whenever those measures were used comparatively. In t tests, pooled variance estimates were used, except where the df has a decimal point indicating that separate variance t's were calculated. All probability levels are two tailed.
Body Assessment Estimates When asked to estimate their weight, 29.1% and 12.3% of girls and boys respectively left this question blank, while 27.1% of girls and 12.3% of boys omitted an estimate of height. In both cases the difference between girls and boys was significant (sex by weight omission ~2 = 20.78, df = 1, p < .00005; height omission, Z2 = 16.58, df = 1, p < .00005). Of girls, 17.4% actually refused to be weighed, and 3.8% refused to be measured for height. This contrasts with 1.4% of boys refusing to be weighed and .4% refusing to be measured (sex by weight refusal 22 = 33.1, df = 1, p < .00005; height refusal 22 = 4.9, df = 1, p = .03). Boys were quite accurate in their estimates of weight and height, as indicated by correlations between self-report and actual weight and height of r = .94 and r = .90, respectively. Girls were slightly less accurate with correlations between self-report and actual weight of r = .87 and actual height of r = .79. Girls, but not boys, significantly underestimated their height (girls: mean actual height = 164.0 cm, mean self-report height = 159.22 cm, t = 3.54, df = 242, p < .0005; boys: mean actual height = 167.0 cm, mean self-report = 166.5, t = 1.67, df = 192, p = .10), and their weight (girls: mean actual weight = 52.71 kg, mean self-report = 51.51 kg, t = 3.47, df = 210, p < .001; boys: mean actual weight = 57.85 kg, mean selfreport = 57.6 kg, t = .95, df = 192, p = .34).
Attitude Towards Weight The data indicated considerable variation between girls and boys on their assessment as to whether they were very underweight (girls, .6%; boys, 1.4%), underweight (girls, 4.9%; boys, 9.7%), a good weight (girls, 47.9%; boys, 70.4%), overweight (girls, 43.6%; boys 18.1%), or very overweight (girls, 3.1%, boys, .5%), and a sex by perceived weight distribution was nonrandom (~2 = 46.5, df = 4, p < .00005). From height and weight measures the Body Mass Index (BMI = weight (kg)/height 2 (m); Keys et al., 1972) was calculated. This was adjusted for normal changes in body proportion with age in the manner described by Coates et al. (1980) and a body mass percentile determined.
Very underweight Underweight Good weight Overweight Very overweight
Self-rating
5.6 50.0 38.9 5.6 0
Under (n = 18) .5 3.3 65.0 30.1 1.1
Normal (n = 183) 0 0 28.6 71.4 0
Over (n = 35)
Girls
0 0 8.6 74.3 17.1
Very over (n = 35) 11.1 44.4 44.4 0 0
Under (n = 9) 1.4 11.6 80.1 6.8 0
Normal (n = 146)
0 0 61.3 35.5 3.2
Over (n = 31)
Boys
0 0 42.9 57.1 0
Very over (n = 28)
Table I. The Percentage of Girls and Boys Who Are Under the Norm (_.120%) Using BMI Criteria, Who Believe Themselves To Be Very Underweight, Underweight, a Good Weight, Overweight, or Very Overweight
o
O ,.
