Abstract
Multiple measures of body dissatisfaction and behaviors associated with disordered eating were studied in 258 White girls, 223 White boys, 106 Black girls, and 82 Black boys. All participants were unpaid volunteers between the ages of 12 and 15 attending six middle schools in Delaware and Maryland. On two self-ideal figure drawing discrepancy measures and four attitude measures of body dissatisfaction, girls had greater dissatisfaction than boys, and Whites had greater body dissatisfaction than Blacks. Racial differences among girls were more common than among boys. Similar results were found for measures of behaviors associated with disordered eating. Gender and racial differences generally paralleled those reported in college populations but were considerably larger than those typically reported in older populations. The majority of Black girls and boys, like the majority of White girls and boys, were dissatisfied with the size of their bodies.
Body dissatisfaction has been generally recognized as an “essential precursor” to potentially life-threatening eating disorders such as anorexia nervosa or bulimia (Polivy & Herman, 2002, p. 192). Body dissatisfaction is also related to very common unhealthy attitudes and practices concerning eating, such as fad dieting, restricted eating, or binge eating. These behaviors are often described as disordered eating. Disordered eating is of considerable theoretical and practical importance because it is very common and often leads to guilt, impaired self-esteem, and a reduced quality of life (Johnson & Wardle, 2005; Sanftner, 2011).
The majority of research on body dissatisfaction and disordered eating has been done with White college women. This is not surprising because this population has a high level of body dissatisfaction and disordered eating, and it is readily accessible to researchers (Grogan, 2008; Thompson, Heinberg, Altabe, & Tantleff-Dunn, 1999). However, for many years this emphasis on advantaged White women, along with the belief that body dissatisfaction and disordered eating was uncommon and usually not problematic among children, early adolescents, racial and ethnic minorities, and males of all ages, discouraged research with other populations (Root, 1990).
Measuring Body Dissatisfaction
It is widely recognized that body dissatisfaction is a complex and multifaceted construct (e.g., Cash, Morrow, Hrabosky, & Perry, 2004; Thompson & van den Berg, 2002). Although the importance of using multiple measures of complex variables has been recognized for many years (e.g., Campbell & Fiske, 1959) and many questionnaire measures of body dissatisfaction have been developed for use with adults and children, most of the available research has emphasized dissatisfaction with body size (Gardner, 2001; Ricciardelli & McCabe, 2001). One of the most common methods of measuring size dissatisfaction has been a figure or silhouette scale, most often the Figure Rating Scale (Stunkard, Sorenson, & Schlusinger, 1983) or one of its many variations. These measures present the participant with drawings of their same-sex bodies that are ordered and numbered from very thin to extremely heavy. Body dissatisfaction is usually operationalized as the difference between the number of the drawing representing the participant’s actual body and the number of the drawing representing their ideal body. Much of the evidence of gender and racial or ethnic differences in body image is based primarily or exclusively on measures of dissatisfaction with body size obtained from a silhouette or figure rating scale (Grabe & Hyde, 2006; Ricciardelli & McCabe, 2001).
Although figure scales have made a major contribution to our understanding of body dissatisfaction, and they continue to be refined and improved (e.g., Swami, Salem, Furnham, & Tovée, 2008), they are not a substitute for multidimensional measures of body dissatisfaction. More important, there is substantial evidence that the existence, nature, and direction of gender and racial or ethnic differences in body dissatisfaction in adult populations vary across measures (Cash et al., 2004; Grabe & Hyde, 2006; Paxton, Neumark-Sztainer, Hannan, & Eisenberg, 2006; Roberts, Cash, Feingold, & Johnson, 2006). To the extent that this is true, it emphasizes the importance of using multiple measures of body dissatisfaction and the risks of generalizing results across measures.
Measuring Disordered Eating
Like body dissatisfaction, disordered eating is multidimensional. Two of the most widely studied aspects of disordered eating are restrained eating and an obsession with thinness. Restrained eating, an effort to control weight by restricting food intake, has been of theoretical interest since Herman and Mack (1975) demonstrated that it often leads to subsequent overeating. Restrained eating is now generally recognized as an important contributor to weight gain, depression, and reduced self-esteem (Johnson & Wardle, 2005; Sanftner, 2011). Concern with thinness, often described as an obsession or a drive for thinness, is of clinical importance because of its association with anorexia nervosa (Brumberg, 1997). It is also important because of its association with body dissatisfaction, low self-esteem, and a reduced quality of life (Hesse-Biber, 2007; Hesse-Biber, Leavy, Quinn, & Zoino, 2006). Both restrained eating and an excessive concern with thinness are usually measured with specialized questionnaires.
Gender Differences
Body dissatisfaction
Gender differences in body dissatisfaction have been recognized since the pioneering work of Secord and Jourard (Jourard & Secord, 1954; Secord & Jourard, 1953) and Calden, Lundy, and Schlafer (1959). This early work recognized three important themes that are still central to our understanding of gender differences in body dissatisfaction. First, women and girls have greater body dissatisfaction than men and boys. Second, dissatisfied women and girls almost always want to be thinner, but about half of dissatisfied men and boys want to be heavier. Third, men and boys want to increase the size of body features (e.g., upper arms, shoulders) associated with muscularity (Feingold & Mazzella, 1998; Ricciardelli & McCabe, 2001; Ricciardelli, McCabe, Williams, & Thompson, 2007).
The largest review of gender differences in body dissatisfaction, a meta-analysis by Feingold and Mazzella (1998), examined 730 effect sizes obtained over a 50-year period. The authors concluded that gender differences were generally small, but almost all studies found that women had greater body dissatisfaction than men and that this difference increased over time. It appeared that men’s level of body dissatisfaction was reasonably stable over time and that the increased gender difference was primarily the results of increased body dissatisfaction in women. Similar results were reported by Cash (2002) but he, like Heatherton, Mahamedi, Striepe, Field, and Keel (1997) and Keel, Baxter, Heatherton, and Joiner (2007), also found evidence that in the 1990s gender differences began to decline. Cash, like Feingold and Mazzella, attributed this to reduced body dissatisfaction among women, but Keel et al. found evidence of both decreased dissatisfaction among women and increased dissatisfaction among men.
Although the finding that approximately 50% of the men who were dissatisfied with their weight wanted to be heavier and that this desired increase in weight represented an increase in muscularity was first reported by Jourard and Secord (1954), there was not much systematic interest in body dissatisfaction among men and boys, particularly dissatisfaction with muscularity, until the late 1990s (Pope, Phillips, & Olivardia, 2002). In recent years, this has become an important area of research (Grossbard, Neighbors, & Larimer, 2011; Jones & Morgan, 2010; Jung, Forbes, & Chan, 2010).
In their influential review, Ricciardelli and McCabe found that gender differences in children generally parallel those found in adults. Furthermore, girls have greater body dissatisfaction, particularly weight dissatisfaction, than boys. A large majority of girls want to lose weight, whereas boys want to lose weight associated with body fat and gain weight associated with muscle. As a consequence, on figure rating scales, boys are about as likely to desire to gain weight as to lose it. These gender differences appear at least as early as late childhood and appear to increase with age (Ricciardelli & McCabe, 2001).
Disordered eating
Gender differences in symptoms associated with disordered eating in adolescents also seem to parallel those found in adults. For example, early adolescent girls are 4 times more likely than boys to have symptoms associated with anorexia nervosa and 2.5 times more likely to have symptoms associated with bulimia (Beato-Fernández, Rodríguez-Cano, Pelayo-Delgado, & Calaf, 2007). Similarly, Strauss (1999) found that adolescent girls of normal weight are much more likely than boys of normal weight to describe themselves as overweight and were much more likely to be dieting.
Racial Differences in Adults
Body dissatisfaction
The important meta-analytic reviews by Grabe and Hyde (2006), Roberts et al. (2006), and Wildes, Emery, and Simons (2001) concluded that Black women had lower levels of body dissatisfaction than White women but emphasized that there was substantial variation in effect sizes across samples and measures. For example, Roberts et al. found that the mean difference between Black and White women, d = .28, (Cohen, 1988) was essentially identical to the value of d = .29 reported by Grabe and Hyde. However, in both reviews these values masked very large differences among studies and measures. Differences in types of measures appeared to be particularly important. Grabe and Hyde indicated that Black-White differences were found on questionnaire measures of body dissatisfaction but were often absent on silhouette or figure measures. Similarly, Roberts et al. found a difference of d = .32 for questionnaire measures of weight dissatisfaction and d = .33 for global body dissatisfaction but only a difference of d = .14 for silhouette measures of weight dissatisfaction. Roberts et al. also reported that over time there has been a decline in Black-White differences on measures of weight dissatisfaction but an increase in Black-White differences on measures of global body dissatisfaction. These results emphasize the importance of using both silhouette measures and questionnaire measures of body dissatisfaction. In addition, they indicate that considerable caution is necessary when generalizing the results from studies that have used only a single measure of body dissatisfaction or when comparing studies using different types of measures.
Disordered eating
The influential review by Crago, Shisslak, and Estes (1996) summarized the available knowledge on racial differences in disordered eating and concluded that Black women had a lower incidence of eating disorders than White women. Crago et al. also reported that Black women appeared to have less dieting, less restrained eating, and a lower drive for thinness than White women. Smolak and Striegel-Moore (2001) reached similar conclusions and provided a summary of risk factors associated with eating disorders in racial or ethnic minorities.
Effects of Age and Developmental Stages on Gender and Racial Differences
Because of the emphasis on studies of college women, relatively little information is available on children or early adolescents, and even fewer studies are available on racial and gender differences in these populations. The available data generally suggest that racial and gender differences in children align with results for college students and adults. For example, White girls have greater body dissatisfaction than White boys and gender differences among White children are greater than the gender difference for Black children (Adams et al., 2000; DeLeel, Hughes, Miller, Hipwell, & Theodore, 2009; Lawrence & Thelen, 1995). These studies also found that Black children prefer larger body sizes than White children and are less likely to be on diets. Consistent with the latter finding, Adams et al. (2000) reported that Black girls had fewer weight concerns than White girls. As is the case with adults, Striegel-Moore et al. (2000) found that the existence, size, and direction of the Black-White differences among pre- and early-adolescents were dependent on how body dissatisfaction is measured. Meta-analytic reviews have shown that differences among measures and the developmental stage of the participants influence the size and the direction of the difference in body dissatisfaction among Black and White girls and women (Roberts et al., 2006).
Very importantly, Roberts et al. found that participant age influenced some racial differences but had no meaningful effect on others. Although age did not influence the Black-White difference in body size dissatisfaction based on silhouette measures, it had a profound effect on measures of global body dissatisfaction (Roberts et al., 2006, Figure 1). Among pre- and early-adolescents, Whites had less body dissatisfaction on questionnaire measures than Blacks, but this difference reversed around the time of puberty. Just as Grabe and Hyde (2006) found the largest Black-White differences among young adult women, Roberts et al. found the largest Black-White difference, d ≈ .40, around age 25. This indicates that both the size and the direction of Black-White differences on questionnaire measures of body dissatisfaction vary with the participants’ age. Consistent with the conclusions of Roberts et al., Paxton, Eisenberg, and Neumark-Sztainer (2006) found that Black-White differences on questionnaire measures of body dissatisfaction were not present in early adolescents, and DeLeel et al. (2009) found that 9- and 10-year-old White girls did not differ from similar group of overwhelming Black (90+%) minority girls on a figure rating scale measure of body dissatisfaction.
Just as is the case with adults, it appears that with early adolescents silhouette and questionnaire measures of Black-White differences in body dissatisfaction may produce different results (Roberts et al., 2006). This adds further emphasis to the importance of using both kinds of measures of body dissatisfaction.
Developmental Importance of Middle School
Middle-school years are developmentally crucial years in terms of physical, cognitive, social, and personal development. Given the magnitude and pervasiveness of the changes occurring in these years, it is not surprising that research shows they are associated with sharp declines in academic achievement, reduced self-esteem, and increases in depression (Seidman, Aber, & French, 2004). Although measureable differences in body dissatisfaction are present in boys and girls at the age 5 or 6 and increase throughout the grade school years, there are strong theoretical reasons to believe that body dissatisfaction will increase during middle school (Gardner, Friedman, & Jackson, 1999; Krahnstoever Davison, Markey, & Birch, 2003). During these years, social comparisons and concerns with how one is perceived by others (e.g., the imaginary audience) increase (Michael & Eccles, 2003). In addition, puberty is experienced by both genders during this developmental stage. During this time girls become more aware of the thin body ideal and more concerned with weight gain, whereas boys become more concerned with developing a larger, more muscular body. Girls would be expected to experience increased body dissatisfaction because of the unavoidable weight gain associated with puberty (Striegel-Moore et al., 2000). Boys, because of their relative delay in puberty, would be expected to become increasingly aware that their body falls short of the robust male body ideal (Jones & Crawford, 2005). In general, research supports these hypothetical gender differences but also indicates that puberty effects are greater for girls than for boys and are mediated by the co-occurrence of other stressful events, such as beginning middle school, or dating, and the presence of gender-dimorphic body ideals (Hayward & Sanborn, 2002; Michaud, Suris, & Deppen, 2006; Muris, Meesters, van de Blom, & Mayer, 2005; Smolak, Levine, & Gralen, 1993).
The Present Study
Unfortunately, the study of body dissatisfaction in middle-school-aged children has been limited, and with very few exceptions (e.g., Neumark-Sztainer et al., 2002), studies using multidimensional measures of racial and gender differences in this age group are not available. The limited available evidence, particularly the meta-analysis of Roberts et al. (2006) and the results of Paxton et al. (2006), suggests that levels of body dissatisfaction and the size of gender differences will be smaller in this age group than among later adolescents and young adults.
This study will investigate the influence of gender and race on both silhouette measures and multidimensional questionnaire measures of weight or body dissatisfaction and behaviors associated with disordered eating in Black and White early adolescent children. On the basis of previous research, particularly the meta-analyses by Roberts et al. (2006), Grabe and Hyde (2006), Wildes et al. (2001), and Feingold and Mazzella (1998), we developed the following hypotheses.
Hypothesis 1: Girls will have greater body size dissatisfaction and disordered eating behaviors than boys.
Hypothesis 1a (H1a): White girls will have greater body size dissatisfaction than White boys.
Hypothesis 1b (H1b): Black girls will have greater body size dissatisfaction than Black boys.
Hypothesis 1c (H1c): White and Black girls will report more behaviors associated with disordered eating than White and Black boys.
Hypothesis 2 (H2): Whites will have greater body size dissatisfaction and disordered eating behaviors than Blacks.
Hypothesis 2a (H2a): White girls will have greater body size dissatisfaction than Black girls.
Hypothesis 2b (H2b): White boys will have greater body size dissatisfaction than Black boys.
Hypothesis 2c (H2c): White girls will report more behaviors associated with disordered eating than Black girls.
Hypothesis 3: Desired body sizes will vary by race and gender.
Hypothesis 3a (H3a): Girls will desire a smaller body, but boys will be as likely to desire a larger body as a smaller body.
Hypothesis 3b (H3b): Black girls and boys will desire larger body sizes than White girls and boys.
Method
Participants and Procedure
Early adolescent boys and girls were recruited from six middle schools in Delaware and Maryland. All were public schools in suburban communities with populations that ranged from 15,000 to 71,000. The majority of students represent middle- or upper-middle-class families. Based on ZIP codes and the 2000 census data, median family income in these communities ranged from 113% to 269% (median = 159%) of the national median. In all schools the largest racial groups were White (39% to 67%, median = 59%) and Black (21% to 34%, median = 31%). Because adequate samples of other racial or ethnic groups could not be obtained, the study was limited to White and Black students. Following approval from the principal investigator’s Institutional Review Board, local and regional schools were contacted by the principal investigator to obtain permission from principals, other relevant administrators, and classroom teachers. Consent forms approved by the IRB were distributed to parents/guardians of students to obtain parental permission for participation in this study. A total of 800 consent forms were distributed to the parents of Black and White students and 669 forms indicating consent were returned (87% participation rate). After parental consent was obtained, assent forms were distributed to students to ensure that they were voluntarily participating in the study and to inform them of their right to decline to participate at any time. Teachers distributed anonymous questionnaires in the classroom during regular school hours. Although none of the students were compensated for their participation, the schools received small honorariums for their assistance with sampling. It usually took the participants less than 25 minutes to complete the materials.
The final sample consisted of 223 White boys and 258 White girls and 82 Black boys and 106 Black girls. All participants attended Grades 6 through 9 and were between 11 and 15 years of age. The vast majority (93%) of the sample participants were 12, 13, or 14 years of age. A 2 x 2 ANOVA (analysis of variance) on participant age indicated a significant main effect of gender, F(1, 667) = 16.22, p < .001, pη2 = .024, and a significant Gender x Race interaction, F(1, 667) = 5.15, p < .05, pη2 = .008. The results indicated that the boys (White:
BMI was computed from self-reported height and weight using the standard formula of weight (kg)/height2 (m). A 2 X 2 ANOVA on BMI indicated a significant main effect of race, F(1, 640) = 18.40, p < .001, pη2 = .028. The body size of the Black sample (boys:
Measures
Three different groups of measures were administered to both boys and girls: figure measures of body dissatisfaction, questionnaire measures of body dissatisfaction, and measures of eating behaviors. Unless noted below, all measures were answered on a 5-point scale anchored by 1 = strongly disagree and 5 = strongly agree.
Figure measure of body dissatisfaction
The Contour Drawing Rating Scale (CDRS: Thompson & Gray, 1995) is a pictorial instrument consisting of nine line drawings of figures ranging from very thin to very obese. As originally developed, the figures were numbered from 1 (very slender) to 9 (very heavy). Wertheim, Paxton, and Tilgner (2004) reported that the scale had satisfactory test-retest reliability with adolescent samples. We followed their practice of renumbering the figures from 1 to 18, thus allowing the participants to choose a value between two of the drawings. The participants were shown the CDRS figures for their gender and asked to identify the figure representing their present body size, their desired body size, the body sizes desired by their same and other gender peers, and the body size they thought their parents would like them to be. Discrepancies between responses to these questions were used as measures of body dissatisfaction. Because the results for the various discrepancy scores were nearly identical, to simplify the analyses only the results for the actual ideal body discrepancy will be reported.
Questionnaire measures of body dissatisfaction
The Appearance Evaluation Scale (AES) is a 7-item measure of global body satisfaction from the Multidimensional Body-Self Relations Questionnaire (Cash, 2000). Sample items are “I like my looks just the way they are,” and “I like the way my clothes fit me.” Coefficient alphas for the scale were White: boys = .83, girls = .82, and Black: boys = .80, girls = .79.
The Body-Esteem Scale for Adolescents and Adults (BESAA; Mendelson, Mendelson, & White, 2001) consists of 23 statements reflecting affective evaluations of body characteristics. The BESAA has three factor-analytically derived measures. The Appearance scale measures general feelings about appearance. Sample items are “I like what I look like in pictures,” and “I like what I see when I look in the mirror.” The Weight scale measures general feelings about weight. Sample items are “I am satisfied with my weight,” and “My weight makes me unhappy” (reverse scored). The Attribution scale measures perceptions of others’ evaluations about one’s body and appearance. Sample items are “Other people consider me good looking” and “I’m as nice looking as most people.” Items are answered on a 5-point scale anchored as never to always. In each case, higher scores indicate greater body satisfaction. Coefficient alphas for the BESAA measures were Appearance: White, boys = .84, girls = .89; Black, boys = .84, girls = .91; Weight: White, boys = .87, girls = .92; Black, boys = .81, girls = .90; Attribution: White, boys = .75, girls = .77; Black, boys = .71, girls = .70.
Measures of behaviors associated with disordered eating
Participants completed the 7-item Drive for Thinness Scale (DFTS) from the Eating Disorder Inventory (EDI; Garner, Olmstead, & Polivy, 1983). Sample items are “I am terrified of gaining weight, “and “I think about dieting.” High scores on this measure indicate a drive for thinness. Coefficient alphas for the scale were White, boys = .85, girls = .87; Black, boys = .80, girls = .81.
The participants completed the 10-item Restrained Eating Scale (RES) from the Dutch Eating Behavior Inventory (Van Strien, Frijters, Bergers, & Defares, 1986). Items were answered on a 5-point scale anchored by never (1) and always (5). A “does not apply to me” alternative was also available. Sample items are “Do you try to eat less at mealtimes than you would like to eat?” and “Do you deliberately eat foods that are slimming?” The scale was scored by summing the items and computing the mean. High scores on this measure indicate restrained eating. Coefficient alphas for the scale were: White, boys = .91, girls = .91; Black, boys = .87, girls = .86.
Results
Because of missing data, Ns varied from 592 to 669. Most of the missing data were from the information (height and weight) needed to calculate BMI. Correlations among all of the variables are shown in Table 1. Correlations for Blacks appear above the diagonal, and correlations for Whites appear below the diagonal. Correlations for boys appear above the correlations for girls. Because a large number of correlations were computed, alpha was set at .01. Inspection of the table indicates that, as expected, all of the measures of body dissatisfaction and disordered eating were related to each other.
Correlations and n for All Variables by Ethnicity (Blacks Above Diagonal) and Gender (Boys Top Line).
Note: BMI = body mass index, Body 12 = self-ideal body on Contour Drawing Rating Scale (Thompson & Gray, 1995), |Body 12| = absolute value of self-ideal body on Contour Drawing Rating Scale (Thompson & Gray, 1995), AES = Appearance Evaluation Scale (Cash, 2000); BESAA = Body Esteem Scale for Adolescents and Adults (Mendelson, Mendelson, & White, 2001), Drive for Thinness Scale (Garner, Olmstead, & Polivy, 1983); Restrained Eating Scale (Van Strien, Frijters, Bergers, & Defares, 1986).
p < .01.
Also as expected, BMI was correlated with all of the measures of body dissatisfaction and measures of disordered eating. Because groups differed in BMI and BMI was correlated with every other measure, group comparisons would be confounded. For this reason, all comparisons were made using separate univariate analysis of covariance (ANCOVA) with BMI as covariate. Post hoc comparisons between adjacent means used Holm’s (1979) sequential Bonferroni test to maintain alpha = .05 with a family size of 4.
Size and Weight Dissatisfaction on the Figure Measure
Discrepancy measures (actual-ideal) of dissatisfaction with body size and weight on the CDRS can be used to measure two related but different variables. If the primary interest is in the direction of body dissatisfaction, the arithmetic value of the discrepancy is used. With this scoring, positive scores indicate a desire for a smaller body size. If the primary interest is in total dissatisfaction, the absolute value of the discrepancy is used. For this method of scoring, the larger the score, the greater the desire for a different (i.e., either smaller or larger) body size.
Directional body dissatisfaction
The results of a 2 (race) x 2 (gender) ANCOVA using the arithmetic value of the actual-ideal discrepancy score are shown in Table 2. All of the means were positive, indicating that all four groups desired smaller bodies. The significant main effects of race and gender were qualified by a significant interaction. This interaction indicated that the racial difference was present only for girls (d = .53) and a gender difference was present only in the White sample (d = .65). Both of these differences fell in the moderate range. The finding that White girls had greater body dissatisfaction than White boys confirmed H1a. The finding that White girls had greater body dissatisfaction than Black girls confirmed H2a. Although the difference between Black boys and Black girls and the difference between White boys and Black boys were in the direction predicted by H1b and H2b, the differences were small and insignificant.
Adjusted Means, Standard Errors, Cohen’s d Between Adjacent Means, and ANCOVAs (BMI as Covariate) for Figure Measures of Body Dissatisfaction.
Note: ANCOVA = analysis of covariance; BMI = body mass index; CDRS = Contour Drawing Rating Scale (Thompson & Gray, 1995).
Difference between adjacent means significant at alpha = .05 using Holm’s (1979) sequential Bonferroni test with a family size of 4.
Chi-square goodness-of-fit tests were used to compare the proportion of participants desiring a change in body size as a function of race and gender. Neither White, χ2(N = 216, df = 2) = 1.76, nor Black boys, χ2(N = 81, df = 2) = 2.00, showed a significant preference for direction of change. That is, both White and Black boys were as likely to want a larger body as they were to want a smaller body. However, parallel tests found that both White, χ2(N = 255, df = 2) = 112.73, p < .001, and Black, χ2(N = 108, df = 2) = 20.39, p < .01, girls showed a significant preference for a smaller body size. These results confirmed H3a. More important, a majority of both boys (White = 63%, Black = 63%) and girls (White = 78%, Black = 75%) were dissatisfied with their body size.
Nondirectional body dissatisfaction
The results of a 2 (race) x 2 (gender) ANCOVA using the absolute value of the ideal-actual discrepancy score are shown in Table 2. The main effects of race and gender were significant, but the interaction was not. As predicted by H1a and H1b, the main effect of gender indicated girls had greater body dissatisfaction than boys. As predicted by H2a and H2b, the main effect of race indicated that Whites had greater body dissatisfaction than Blacks.
Self and ideal body size
The results of a 2 (race) x 2 (gender) ANCOVA on the number of the figure the participants selected as representative of their actual body size are shown in Table 3. The main effects of race and gender interaction were not significant. The significant effect for gender indicated that White boys selected a larger figure than White girls and Black boys selected a larger figure than Black girls.
Adjusted Means, Standard Errors, Cohen’s d Between Adjacent Means, and ANCOVAs (BMI as Covariate) for Self and Ideal CDRS Figure.
Note: ANCOVA = analysis of covariance; BMI = body mass index; CDRS = Contour Drawing Rating Scale (Thompson & Gray, 1995).
Difference between adjacent means significant at alpha = .05 using Holm’s (1979) sequential Bonferroni test with a family size of 4.
The results of a 2 (race) x 2 (gender) ANCOVA on the number of the figure the participants selected as representative of their ideal body size are also shown in Table 3. The two significant main effects were qualified by a significant interaction. The interaction indicated that Black girls chose a larger ideal body size than White girls, but the ideal body size of Black and White boys did not differ. The results for girls confirmed H3b, but the results for boys did not.
Questionnaire Measures of Body Satisfaction
Global body satisfaction
The AES provides a measure of global body satisfaction. The results of a 2 (race) x 2 (gender) ANCOVA are shown in Table 4. Although the effect of gender was not significant, the effect of race and the interaction were significant. This interaction indicated that Blacks were more satisfied than Whites and this difference was larger for girls (d = .70) than for boys (d = .39).
Adjusted Means, Standard Errors, Cohen’s d Between Adjacent Means, and ANCOVAs (BMI as Covariate) for Questionnaire Measures of Body Dissatisfaction.
Note: ANCOVA = analysis of covariance; BMI = body mass index; AES = Appearance Evaluation Scale (Cash, 2000); BESAA = Body Esteem Scale for Adolescents and Adults (Mendelson, Mendelson, & White, 2001).
Difference between adjacent means significant at alpha = .05 using Holm’s (1979) sequential Bonferroni test with a family size of 4.
Body esteem measures
The BESAA provides three measures of body esteem. Separate 2 (race) x 2 (gender) ANCOVAs were computed for each measure.
On the BESAA Appearance scale, a measure of general feelings about one’s appearance, the significant main effects of race and gender were qualified by a significant interaction indicating that Black girls and White boys had more favorable feelings about their appearance than White girls. These results confirmed H1a and H2a.
On the BESAA Weight scale, a measure of satisfaction with one’s weight, the significant effect of race was qualified by a significant interaction indicating that Black girls had greater weight satisfaction than White girls and White boys had greater weight satisfaction than White girls. The racial difference between girls (d = .53), which was consistent with H2, fell at the lower extreme of the medium range and the gender difference for Whites (d = .35) fell in the small range. These results confirmed H1a and H1b.
On the BESAA Attribution scale, a measure of the perception of how others evaluate one’s body and appearance, only the effect of race was significant. This result indicated that Black boys and girls felt that others had a more favorable perception of their appearance than did White boys and girls. For both boys (d = .55) and girls (d = .68) effect sizes fell in the medium range. These results confirmed H2a and H2b.
Summary of Body Satisfaction Measures
On each of the four body satisfaction measures Black girls scored higher than White girls. This offers strong support for H2a. Means were in the same direction for Black boys and White boys, but the differences were smaller and were not significant for two of the four comparisons. These results offered limited support for H2b. As predicted by H1a, on three of the four comparisons of body satisfaction, White boys had greater body satisfaction than White girls. However, the parallel H1b, predicting that similar differences would be found for Black boys and girls received no support.
Measures of Behaviors Associated With Disordered Eating
The results of separate 2 (race) x 2 (gender) univariate ANCOVAs on scores from the DFTS and the RES are shown in Table 5. On the DFTS, a measure of fear of weight gain, the effect of race was not significant. The significant effect of gender was qualified by a significant interaction indicating that White girls had a greater fear of gaining weight than White boys did. These results supported H1c and H2c.
Adjusted Means, Standard Errors, Cohen’s d Between Adjacent Means, and ANCOVAs (BMI as Covariate) for Measures of Disordered Eating.
Note: ANCOVA = analysis of covariance; BMI = body mass index; Drive for Thinness Scale (Garner, Olmstead, & Polivy, 1983); RES = Restrained Eating Scale (Van Strien, Frijters, Bergers, & Defares, 1986).
Difference between adjacent means significant at alpha = .05 using Holm’s (1979) sequential Bonferroni test with a family size of 4.
On the RES, a measure of the frequency of behaviors intended to reduce weight, the effect of race was not significant. The significant effect of gender was qualified by a significant interaction indicating that White girls engaged in more behaviors intended to reduce weight than White boys. These results supported H1c and H2c.
Discussion
Body Dissatisfaction
Significant main effects of race were found on both of the figure measures of body dissatisfaction from the CDRS and all four questionnaire measures of body satisfaction among girls. In each of these cases, the effect of race was primarily the result of less body dissatisfaction and the greater body satisfaction among Black girls. These results parallel results from the studies of older adolescent and adult women reviewed by Grabe and Hyde (2006), Roberts et al. (2006), and Wildes et al. (2001). Although there was no racial difference among boys on the two figure measures and on two of the questionnaire measures, Black boys had greater body satisfaction than White boys on the AES and BESAA Attribution measures. This indicates that most of the time racial differences among boys were not present. When they were present they were limited to differences in global satisfaction with appearance and were considerably smaller than parallel differences for girls. As with the results with girls, the results for racial differences among boys were generally consistent with studies using older populations (e.g., DiGioacchino, Sargent, & Topping, 2001).
Comparisons with published data
The three major meta-analytic reviews of racial differences (Grabe & Hyde, 2006; Roberts et al., 2006; Wildes et al., 2001) found that most studies have reported small mean effect sizes (d < .30) for Black-White differences among adolescent and adult women. However, our effect sizes for girls ranged from d = .53 to d = .70 with a median size of d = .68. Although our effect sizes appear to be larger than those reported in reviews, there are two factors that complicate the interpretation of this apparent difference. First, the reviewers all commented on the large range of effect sizes reported in the literature. Our values are about twice as large as the mean values in the reviews, but they are within the upper extreme of the values reported in the literature. Second, consistent with the conclusions of the reviewers, we found that racial differences, particularly between girls, were smaller on figure measures of body size dissatisfaction than on questionnaire measures. Because most of our comparisons were based on questionnaire measures, and the most common measure used in most studies of racial differences in body dissatisfaction have been self-ideal discrepancy scores from figure drawing scales, this might have contributed to the larger effect sizes found in our study. Nevertheless, our results suggest the clear possibility that racial differences among middle-school girls may be larger than parallel differences among older adolescent and young adult women. This possibility requires future investigation.
Similarly, the size of the gender differences we observed, which appear to be larger than those observed in studies of older adolescents and adults, seems surprising. This is because the limited information on body dissatisfaction among children suggests that (a) it is less severe and (b) gender differences are smaller than among older adolescents and adults. Most research shows that body dissatisfaction increases over the course of adolescence and peaks in college-aged populations. The review by Robert et al. (2006) shows a clear inverted U relationship between questionnaire measures of body dissatisfaction and age. This function begins in early adolescence, peaks among college women, and then declines until middle age. Based on available research, we could predict that body dissatisfaction in middle-school children would be expected to be less severe, and gender differences would be expected to be smaller than in late adolescent and college-age groups. However, our effect sizes, particularly from White girls and boys, suggest this is not always the case. The apparent discrepancy between our results and the results of similar studies with older groups indicates that, like racial differences, gender differences in body dissatisfaction among middle-school children requires additional investigation.
Why Black girls have greater body satisfaction
One common explanation for the finding of greater body satisfaction among Black girls and women, as compared to White girls and women, involves differences in their body ideals. The White body ideal typically emphasizes an extremely slender body, whereas most of the research shows that Black girls and women prefer larger bodies than White girls and women (Nollen et al., 2006). On the CDRS we found a similar racial difference in body ideals with Black girls preferring a larger body size (d = .47) than White girls. Unlike many studies, we used BMI to control for actual body size in this comparison. As a consequence, this difference is not an artifact of our finding that Black girls had larger bodies.
It seems likely that Black-White differences in body dissatisfaction go far beyond preferences for different body sizes. In an important anthropological study of 14- to 17-year-old adolescents, Parker et al. (1995) found large Black-White differences in perceptions of beauty and ideal body shape. They found that White girls had narrow and rigid beauty concepts that emphasized external appearance, particularly a very slender body. This narrow beauty ideal was strengthened by competition between girls and criticisms of their weight or appearance by other girls and their families. This narrow emphasis on achieving a “perfect” body, particularly perfect weight, almost inevitability lead to unfavorable comparisons between their own body and the ideal body. Given this emphasis, it is not surprising that when Parker et al. asked their White sample to describe a “perfect” girl, their description focused on her weight and appearance.
In contrast, Parker et al. (1995) asked the same question to their Black sample, they usually began by describing the “perfect” girl’s personality. Black girls were not indifferent to external appearance, but emphasized the importance of an “inner beauty” that conveyed self-confidence and pride. As a reflection of these values, the majority of their Black participants defined beauty in terms of “attitude and personality when you deal with others” (p. 108). When they evaluated a person’s external appearance they were more concerned with her personal “style” and “making what you’ve got work for you” than with her weight (p. 108). Parker also found that Black girls received more positive feedback about their appearance from their friends and family than did White girls. This support from others, particularly their mothers and other girls, stood in contrast with the often critical feedback received by White girls.
The work by Parker et al. (1996) and others (e.g., Root, 1990) makes it clear that Black-White differences in weight dissatisfaction are only a surface manifestation of much large differences in the meaning of beauty and perceptions of the body. The research emphasis on differences in weight dissatisfaction may serve to minimize the significance of these much broader issues. The importance of these issues began to be more evident when results from questionnaire measures that do not reflect weight dissatisfaction are examined. For example, we found that Black girls had greater body satisfaction on the AES and BESAA Appearance and Attribution scales. The results from these measures, which do not contain questions on weight, would appear to reflect the influence of the more flexible and individual perceptions of beauty found among Black girls. It is important that future research begins to emphasize the study of variables other than weight.
Importance of absolute levels of dissatisfaction
The consistent finding that Black girls have lower body dissatisfaction than White girls should not lead to the conclusion that body dissatisfaction is not an important problem for Black girls. Similarly, the finding that boys have less body dissatisfaction than girls does not mean that it is not an important issue for boys. Levels of body dissatisfaction among boys and Black girls may not be as high as among White girls, but when 63% of boys and 75% of Black girls are dissatisfied with their bodies, it is clearly a cause for concern. These results are particularly important because most of the literature on body dissatisfaction in children and adolescents, like the research with adults, emphasizes group differences in body dissatisfaction and there is relatively little discussion of the nearly universal finding that a majority of people, irrespective of their gender, race, and age, are dissatisfied with their bodies (e.g., Grogan, 2008). Our finding that about 75% of middle-school girls are dissatisfied with their body size, with 64% of White girls and 54% of Black girls wanting to have smaller bodies, indicates that body dissatisfaction and weight concerns are well established in both races by the middle-school years. In addition, it is important to note that the proportion of both Black and White middle-school girls who are dissatisfied with their body size is very similar to studies using college populations (e.g., Forbes & Frederick, 2008).
Attitudes and Behaviors Associated With Disordered Eating
We found that attitudes and behaviors associated with disordered eating were more common in girls than in boys and more common among White girls than among Black girls. These differences parallel the differences often reported in older populations. Although effect sizes for racial differences on attitudes and behaviors associated with disordered eating were small, this finding is consistent with recent evidence that Black-White differences in the incidence of behaviors associated with disordered eating (e.g., Franko, Becker, Thomas, & Herzog, 2007), like the incidence of clinical eating disorders (Brown, Cachelin, & Dohm, 2009), are smaller than they were once thought to be. Effect sizes for gender differences, particularly between White girls and boys, were considerably larger than parallel racial differences. These results, like our results from measures of body satisfaction, indicate that gender and racial differences in attitudes and behaviors that are known to be important risk factors for eating pathologies and impaired self esteem are well established in middle-school-aged children.
Limitations
We used the same measures of body dissatisfaction for boys and girls, but these measures were not sensitive to gender differences in the nature of body dissatisfaction. Because of the strong emphasis on the thin body ideal, most girls and women are dissatisfied with their weight and want to be thinner. However, the male body ideal involves both low body fat and high muscularity. Consequently, boys and men want to lose weight associated with fat, but gain weight associated with muscle (Pope et al., 2002). However, none of our measures accessed dissatisfaction with muscularity, and it was impossible to determine how many of the boys who wanted a larger body did so because of dissatisfaction, not with their weight, but with their muscularity. Even with this limitation, we did find racial differences among boys on two measures reflecting global body satisfaction. In both of these cases, Black boys had greater body satisfaction than White boys. The absence of a measure of muscularity does not mean that our assessment of male body dissatisfaction was incorrect or invalid, but it does mean that it was incomplete. For a more complete assessment of body dissatisfaction, it is important that future studies, even in preadolescent boys, include measures of concern with muscularity.
Because we did not control for socioeconomic class, it may appear that our results are potentially confounded. Although it has often been asserted that body dissatisfaction, like eating disorders, is much more common in upper-class populations, in their influential meta-analysis, Wildes et al. flatly stated, “This simply isn’t true” (2001, p. 541). Subsequent work has supported this conclusion (e.g., DeLeel et al., 2009). It is possible that controls for socioeconomic class may have been necessary at one time (Brumberg, 1997), but there is strong evidence to indicate that it is no longer necessary. As a consequence, it is most unlikely that our results reflect socioeconomic differences among our groups.
It may appear that another limitation is the lack of controls for age of puberty, particularly in our sample of girls. However, this limitation is much more apparent than real. As a group, Black girls experience puberty sooner than White girls (Posner, 2006; Wu, Mendola, & Buck, 2002), and early puberty is often associated with increased body dissatisfaction (Heinberg, 2001; Michaud et al., 2006). To the extent that this is true, Black girls would be expected to have greater body dissatisfaction than White girls. However, we found consistent evidence that Black girls had less body dissatisfaction than White girls. Such a result would appear to be very unlikely if the presumed earlier puberty of Black girls had meaningful impact on our results.
Summary and Conclusions
Our results indicate that racial and gender differences in levels of body dissatisfaction in middle-school children parallel the effects typically found among older adolescents and young adults. Contrary to studies of younger children suggesting that racial and gender differences in body dissatisfaction would be lower in early adolescents than in older populations, we found that they were at least equal to, and possibly greater than, differences reported from older populations. We also found that gender and racial differences in attitudes and behaviors associated with disordered eating among middle-school children were very similar to those found in older populations. Taken as a whole, our results indicate that racial and gender differences in body dissatisfaction and attitudes and behaviors associated with disordered eating are well established by the middle-school years. This indicates that efforts to identify and modify the origins of these differences, like efforts to modify body dissatisfaction in all groups, need to be focused on younger, probably much younger, populations. Additional investigations of body dissatisfaction in middle-school children are clearly indicated.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the University of Delaware Research Foundation grant.
