Weight Preoccupation as a Function of Observed Physical Attractiveness: Ethnic Differences Among Normal-Weight Adolescent Females

Transcription

1 Weight Preoccupation as a Function of Observed Physical Attractiveness: Ethnic Differences Among Normal-Weight Adolescent Females Natalie Colabianchi, 1 PHD, Carolyn E. Ievers-Landis, 2 PHD, and Elaine A. Borawski, 1 PHD 1 Department of Epidemiology and Biostatistics, Case Western Reserve University School of Medicine, and 2 Division of Behavioral Pediatrics and Psychology, Rainbow Babies and Children s Hospital, Case Western Reserve University School of Medicine Objective To examine the association between observer ratings of physical attractiveness and weight preoccupation for female adolescents, and to explore any ethnic differences between Caucasian, African-American, and Hispanic females. Methods Normal-weight female adolescents who had participated in the National Longitudinal Study of Adolescent Health in-home Wave II survey were included (n = 4,324). Physical attractiveness ratings were made in vivo by interviewers. Using logistic regression models stratified by ethnicity, the associations between observer-rated attractiveness and weight preoccupation were examined after controlling for demographics, measured body mass index (BMI) and psychosocial factors. Results Caucasian female adolescents perceived as being more attractive reported significantly greater weight preoccupation compared with those rated as being less attractive. Observed attractiveness did not relate to weight preoccupation among African-American or Hispanic youth when controlling for other factors. Conclusions For Caucasian female adolescents, being perceived by others as more attractive may be a risk factor for disordered eating. Key words attractiveness; dieting; disordered eating; racial differences; weight preoccupation. Excessive concern about weight is prevalent among adolescents and is worthy of both research and clinical attention because of its association with problematic eating behaviors (Duncan, Ritter, Dornbusch, Gross, & Carlsmith, 1985; Neumark-Sztainer & Hannan, 2000). Eating disorders, such as anorexia, bulimia, and excessive dieting, can lead to severe and potentially long-lasting health and mental health consequences and are ranked as the third most common chronic illness for young women (Abrams & Stormer, 2002). Adolescents who are dieting for weight loss even though they are of normal or low body weight have been characterized as inappropriate dieters (Pesa, 1999). Little is known about which factors predispose adolescent females to be inappropriate dieters. The identification of such factors is important for better understanding of which individuals are most at risk and why they have adopted this behavior. One risk factor of excessive weight preoccupation may be physical attractiveness. The desire to be more physically attractive according to societal standards has been linked to weight preoccupation among preadolescents and adolescents. Frequency of reading fashion magazines among older girls (5th through 12th graders) was associated with more frequent dieting and exercising to lose weight because of a magazine article (Field et al., 1999). Likewise, both boys and girls in a large national sample (9 14 years of age) who reported making a considerable effort to look like same-sex figures in the media were more likely to develop weight concerns and become constant All correspondence concerning this article should be addressed to Carolyn E. Ievers-Landis, PhD, Division of Behavioral Pediatrics and Psychology, Rainbow Babies and Children s Hospital. Case Western Reserve University School of Medicine, Euclid Avenue, Cleveland, Ohio carolyn.landis@uhhs.com. Journal of Pediatric Psychology 31(8) pp , 2006 doi: /jpepsy/jsj091 Advance Access publication December 21, 2005 Journal of Pediatric Psychology vol. 31 no. 8 The Author Published by Oxford University Press on behalf of the Society of Pediatric Psychology. All rights reserved. For permissions, please journals.permissions@oxfordjournals.org

2 804 Colabianchi, Ievers-Landis, and Borawski dieters (Field et al., 2001). In another study, female adolescents reading of women s beauty and fashion magazines was related to the use of pathogenic dieting methods of restricting calories to 1,200 or less per day or taking diet pills (Thomsen, Weber, & Brown, 2002). Although numerous advantages have been identified for individuals who are physically attractive (Buss, 1994; Feingold, 1992), recent studies have pinpointed one disadvantage a potentially greater risk for eating disorders (Davis, Claridge, & Fox, 2000; Davis, Shuster, Dionne, & Claridge, 2001). This research has demonstrated that Caucasian females who are perceived by others as being more attractive report greater concerns about their body weight (Davis et al., 2000, 2001). Davis and colleagues reported that interviewer ratings of the facial attractiveness of Caucasian college-aged women were positively associated with weight preoccupation of these women after controlling for body weight and neurotic perfectionism, which is the inability to receive satisfaction from almost any endeavor. These findings were interpreted from the social psychological phenomena of expectancy model (Miller & Turnbull, 1986) and the self-fulfilling prophecy effect (Zebrowitz, 1997), that is, women who are perceived as being more attractive act according to how society expects them to act. Another possibility is that individuals perceived as being more attractive may rely on their appearance to a greater degree for their self-worth in comparison with those who are perceived as being less attractive. Whether or not the Davis and colleagues findings generalize to other ethnic groups is not yet known. No research could be located that addresses whether or not physical attractiveness is a risk factor for eating disorders in other ethnic groups, such as among African- American or Hispanic females. African-American women have been found to have lower levels of body dissatisfaction and less internalization of the thin body ideal than Caucasian women, both of which are risk factors for the development of an eating disorder (e.g., Miller et al., 2000; Russell & Cox, 2003; Wilfley et al., 1996). In addition, African-American young women are less likely to overestimate their weight than Caucasian young women (Miller et al., 2000). The abovementioned findings have received some support in studies of adolescent African-American girls as well (White, Kohlmaier, Varnado-Sullivan, & Williamson, 2003). Hispanic college women engage in less restrained eating in general than Caucasian women despite weight and shape concerns (Arriaza & Mann, 2001). Taken together, this literature suggests that the relationship between physical attractiveness and weight preoccupation would be less strong for African-American and Hispanic females than for Caucasian adolescent females. Many other risk and protective factors have been associated with weight preoccupation during adolescence. For example, in a sample of 12,105 normalweight girls and boys in grades 7 12, inappropriate dieting (i.e., trying to lose weight despite low or normal body weight) was the most strongly related to lower selfesteem, followed by depressed mood (Pesa, 1999). Family connectedness did not differentiate between normalweight dieters and nondieters. In a subsequently published study of 6,728 girls and boys of all weight ranges in grades 5 12, self-reported binge-purge cycling ( Have you ever binged and purged or not? ) was significantly related to depressive symptoms and self-esteem (Neumark-Sztainer & Hannan, 2000). Likewise, in a sample of 81,247 9th- to 12th-grade girls and boys of all weight ranges, disordered eating, defined as either self-reported unhealthy weight control behaviors (i.e., fasting or skipping meals, smoking cigarettes, using diet pills, vomiting, or using laxatives for weight control) or binge eating, was significantly related to emotional well-being (mood, stress, sadness, hopelessness, etc.), self-esteem, and family connectedness (Croll, Neumark-Sztainer, Story, & Ireland, 2002). Ethnic differences in risk and protective factors related to disordered eating behaviors were only reported in the Croll et al. (2002) study and were most evident among females. In stratified analyses of African Americans and Hispanics, emotional wellbeing was related to disordered eating among both African- American and Hispanic females, whereas self-esteem was related to disordered eating only among Hispanic females. Family connectedness was not related to disordered eating for African-American nor Hispanic females. In summary, our primary goal was to determine whether Davis and colleagues research on the link between attractiveness and weight preoccupation, based on Caucasian college-aged women (Davis et al., 2000, 2001), would be evident in other ethnic groups and a younger population. We tested whether objectively rated physical attractiveness would be positively related to weight preoccupation in a large-scale national sample of normal weight female adolescents after controlling for other factors found to be related to unhealthy eating behaviors (e.g., depressed mood, low self-esteem, impaired family connectedness) (Croll et al., 2002; Neumark-Sztainer & Hannan, 2000; Pesa, 1999). We hypothesized that attractiveness and weight preoccupation would be significantly associated across all ethnicities after controlling for other factors, with the strongest associations found in Caucasian adolescent females.

3 Weight Preoccupation and Attractiveness 805 Method Participants The data used for this report are from the National Longitudinal Study of Adolescent Health (Add Health). Add Health is a nationally representative sample of adolescents in grades Add Health is comprised of an inschool survey, an in-home extensive survey on a random sample of the adolescents in the schools that were sampled (three waves of data), a parental interview for those who completed the in-home survey, as well as data on the schools and neighborhoods where these adolescents were living. Additional details on the sampling methods and design can be found elsewhere (Resnick et al., 1997). The use of this data for secondary data analyses was approved by the investigators Institutional Review Board. This report uses the in-home Wave II survey (n = 14,738) as well as the corresponding parental data set for parental education. Adolescents who did not have a sample weight assigned to them because of various reasons (e.g., they were selected as a pair such as a twin or half-sibling, but their pair was not interviewed, they did not have a sample identifier, or their school identification number was missing) were excluded from the analyses (n = 1,168). Since these analyses focus on normalweight female adolescents, adolescents with a measured body mass index (BMI) of greater or equal to the 85th percentile or below the 5th percentile for their age and gender were excluded from the analyses (n = 4,324) as were male participants (n = 4,449). These BMI cut-offs are consistent with the Center of Disease Control s (CDCs) definition of normal weight ( dnpa/bmi/bmi-for-age.htm). Furthermore, as all analyses were stratified by ethnicity, those not categorized as Caucasian, African American, or Hispanic were excluded from the sample (n = 473). The resulting final sample included female adolescents in grades 8 12 with normal BMIs (n = 4,324). Measures Outcome Measures Four different binary outcome measures were used to describe weight preoccupation: whether the participant was trying to lose weight; whether the participant had dieted or exercised to lose weight or to keep from gaining weight in the past seven days; whether the participant had engaged in inappropriate dieting behaviors (i.e., self-induced vomiting, taking diet pills or laxatives) in the past 7 days to lose weight; and whether the participant felt that she was overweight (i.e., described herself as slightly overweight or very overweight). Independent Variables The main independent variable was attractiveness, which was assessed in vivo by the interviewers who conducted the 2-hour at-home interviews. The attractiveness question was on a 5-point scale ranging from very attractive to very unattractive. Since few adolescents were rated as being very unattractive or unattractive, these two categories were combined, resulting in a 4- point scale. Covariates in the model included demographic factors (age and socioeconomic status), BMI, and psychosocial factors (i.e., depressed mood, self-esteem, and family connectedness). Age was measured as a continuous variable. Socioeconomic status was measured by parental education (highest level of either mother or father). Education was determined from the parental report when available; otherwise, it was taken from the adolescent report. Adolescents with a parent with a college degree were compared with those adolescents whose parents had less than a college degree. BMI was calculated using a SAS program provided by the CDC ( based on the 2000 CDC growth charts (SAS software, Version 9.1). The percentiles take into account both gender and age. The interviewer collected the height and weight data used for these calculations. Three psychosocial constructs depressed mood, self-esteem, and family connectedness were included in the model as covariates based on prior research linking them to disordered eating behaviors (Croll et al., 2002; Neumark-Sztainer & Hannan, 2000; Pesa, 1999). For comparability, depressed mood was measured by the same two scales (made up of 13 items) that were utilized by Pesa (1999). The first scale of nine items includes symptoms of negative affect over the past 7 days, such as feeling as if you couldn t shake the blues even with help from family and friends. The second scale contains four positive affect items (e.g., you enjoy life ) that were reverse coded and will be referred to hereafter as the lack of positive affect depressed mood scale. Both of these scales are on a 4-point Likert scale ranging from 0 to 3, where 0 means never or rarely in the past 7 days and 3 means most of the time or all of the time (alpha =.84 and.72, respectively). The self-esteem scale contains five items, such as you have a lot to be proud of and you feel socially accepted, on a 5-point Likert scale which was reversed coded such that 1 indicates strongly disagree and 5 indicates strongly agree. The original scale described by

4 806 Colabianchi, Ievers-Landis, and Borawski Pesa (1999) included six items; however, one question was deleted because it was already included in the lack of positive affect scale measuring depressed mood (alpha =.83). Finally, the family connectedness scale contains four items (e.g., you and your family have fun together and your family pays attention to you ), each on a 5-point scale ranging from 1 meaning not at all to 5 meaning very much (alpha =.73). All four psychosocial scales were scored by calculating the average score, with higher scores representing higher endorsement of the construct. Analyses All of the analyses were stratified by ethnicity, as the association between observer-rated attractiveness and weight preoccupation was expected to be stronger for Caucasian girls compared with African-American and Hispanic girls. Race/ethnicity was coded as Caucasian non-hispanic, African American non-hispanic, and Hispanic. Other ethnicities were not included because of insufficient sample size. In this sample, there were 2,591 normal-weight non-hispanic Caucasian females, 963 normal-weight non-hispanic African American females, and 770 normal-weight Hispanic females. Henceforth in this article, the ethnic groups will be called Caucasian, African American, and Hispanic. A series of unconditional logistic regression models were conducted for each of the four outcomes across each of the three ethnicities. Analyses were performed using Stata Version 7 to control for the complex survey design (Stata Corporation, 2002). All independent variables were entered simultaneously and kept in the model regardless of significance. Table I. Descriptives of Psychosocial Characteristics and Body Mass Index (BMI) Results Descriptive Statistics The mean age was 16.4 years of age. The average attractiveness score was 1.8 (on a scale from 0 to 3, where 3 meant very attractive ). The average interviewermeasured BMI score was 20.8, although significant differences existed between ethnicities (Table I). Caucasians had statistically significantly lower BMIs (20.6) compared with African Americans (21.1) and Hispanics (21.1). On average, self-esteem was high, with a mean of 4.1 on a 5-point scale. Each ethnicity was significantly different from the others with African Americans having the highest self-esteem (4.3), followed by Caucasians (4.1), and then Hispanics (4.0). Family connectedness scores were moderate with an average of 3.73 on a 5- point scale and no significant differences by ethnicity. On both the depressed mood scales, scores were generally low indicating low depression. On the negative affect scale, on average the adolescents scored 0.45, although there were significant differences between Caucasians (0.42) and African Americans and Hispanics (0.55 and 0.53, respectively). There was a similar trend in the lack of positive affect scale, although scores were generally higher. Each ethnicity was significantly different from the other, with Caucasians having the lowest scores for lack of positive affect (0.90), followed by African Americans (1.03), and then Hispanics (1.19). More than half of this sample of normal-weight adolescent girls (55.2%) reported dieting or exercising to lose weight or to keep from gaining weight (Table II). Greater numbers of Caucasian adolescents reported Overall (n = 4,324) [M (SD)] Caucasian (n = 2,591) [M (SD)] African American (n = 963) [M (SD)] Hispanic (n = 770) [M (SD)] BMI a 20.8 (0.06) 20.6 (0.06) 21.1 (0.10) 21.1 (0.16) Self-esteem b 4.10 (0.01) 4.09 (0.02) 4.27 (0.03) 3.99 (0.04) Family connectedness 3.73 (0.02) 3.73 (0.02) 3.69 (0.04) 3.77 (0.06) Negative affect a 0.45 (0.01) 0.42 (0.01) 0.55 (0.03) 0.53 (0.03) Lack of positive affect b 0.95 (0.02) 0.90 (0.02) 1.03 (0.04) 1.19 (0.05) Percentages represent weighted percentages. a Caucasians significantly different than African Americans and Hispanics at p <.01. b All ethnicities significantly different from one another at p <.05. Table II. Descriptives of Outcome Variables Overall [n (%)] Caucasian [n (%)] African American [n (%)] Hispanic [n (%)] Overall χ 2 p value Dieting or exercising to lose weight 2,313 (55.2) 1,511 (58.5) 403 (42.6) 399 (49.4) 61.0 <.01 Trying to lose weight 1,591 (35.4) 984 (36.4) 273 (26.9) 334 (39.3) Think they are overweight 1,100 (25.4) 690 (26.2) 180 (19.4) 230 (28.1) Inappropriate dieting behaviors 71 (1.8) 54 (2.1) 9 (0.8) 8 (1.2) Percentages represent weighted percentages.

5 Weight Preoccupation and Attractiveness 807 dieting or exercising in the past 7 days to lose weight or remain the same weight (58.5%) as compared with African Americans (42.6%) and Hispanics (49.4%). Overall, over a third (35.4%) of this sample was trying to lose weight. Hispanic females were the most likely to be trying to lose weight (39.3%), followed by Caucasians (36.4%), and then African Americans (26.9%). Interestingly, a lesser percentage of adolescents considered themselves to be overweight (25.4%). Fewer African American adolescents felt that they were overweight (19.4%) as compared with Caucasians (26.2%) and Hispanics (28.1%). Finally, only a small percentage (1.8%) of adolescents reported engaging in inappropriate dieting behaviors over the past 7 days. More Caucasian adolescents reported engaging in inappropriate dieting behaviors (2.1%) versus 1.2% of Hispanics and 0.8% of African Americans. Owing to the small number of adolescents with this outcome, analyses for this outcome are considered to be exploratory. Unconditional Logistic Regressions Unconditional logistic regressions were run for each of the four outcomes separately for Caucasian, African American, and Hispanic female adolescents. Each regression controlled for demographics, psychosocial factors, and interviewer-measured BMI. Trying to Lose Weight As hypothesized, in this sample of normal-weight adolescents, Caucasian females who were rated by interviewers as being more physically attractive were significantly more likely to be trying to lose weight compared with those rated as being less attractive (OR = 1.29, CI = ) after controlling for demographics, psychosocial factors, and measured BMI (Table III). The association between attractiveness and trying to lose weight was not present in either African American or Hispanic normal-weight females after controlling for other factors (OR = 1.01, CI = and OR = 0.93, CI = , respectively). Dieting and Exercising to Lose Weight or to Keep from Gaining Weight As hypothesized, when controlling for other potential factors, Caucasian female adolescents of normal weight were significantly more likely to be dieting or exercising to lose weight or to keep from gaining weight if they were rated by an observer as being more attractive (OR = 1.28, CI = ) compared with those who were rated as being less attractive. No association was found for African-American or Hispanic adolescents between observer ratings of attractiveness and dieting or exercising to lose weight or to keep from gaining weight when controlling for other factors (OR =.93, CI = and OR = 1.07, CI = for African-American and Hispanic females, respectively). Thinking they were Overweight Contrary to our expectations, no association was found for any ethnicity between observers ratings of attractiveness and whether the girls believed they were overweight when controlling for demographics, psychosocial factors, and measured BMI. The odds ratios were 0.98 for Caucasian adolescents (CI = ), 0.81 for African-American adolescents (CI = ), and 0.99 for Hispanics (CI = ). Engaging in Inappropriate Dieting Behaviors Also contrary to our predictions, no associations were found for any ethnicity between observer-rated attractiveness and engaging in inappropriate dieting behaviors (i.e., self-induced vomiting, taking diet pills or laxatives) in the past 7 days to lose weight after controlling for other factors. Owing to the small percentage of our sample that engaged in these behaviors (1.8%), the associations for the demographics and psychosocial covariates are not described. Relationships of Covariates with Weight Preoccupation To summarize the relationships of the covariates with the various indicators of weight preoccupation among the three ethnic groups, findings are reviewed by each category namely demographics, measured BMI, and psychosocial factors (Table III). Age was inversely related to two different weight preoccupation measures for Caucasian and African-American adolescents. Specifically, older Caucasian adolescents were less likely to report exercising or dieting to lose weight or to keep from gaining weight than were younger Caucasian adolescents. Similarly, older African-American adolescents were less likely to report that they were trying to lose weight than younger African-American adolescents. Surprisingly, after controlling for all other variables, socioeconomic status was not significantly associated with any outcome. Socioeconomic status was marginally associated with reported dieting or exercising to lose weight or to keep from gaining weight in Caucasian female adolescents (p =.08). BMI had a significant positive association with trying to lose weight, dieting or exercising to lose weight or to prevent weight gain, and an increased risk of believing they were overweight for girls of all three ethnicities.

6 808 Colabianchi, Ievers-Landis, and Borawski Table III. Odd Ratios (OR) and Confidence Intervals (CI) for Unconditional Logistic Regressions Stratified by Ethnicity Caucasian female adolescent African-American female adolescent Hispanic female adolescent OR (95% CI) SE t p value OR (95% CI) SE t p value OR (95% CI) SE t p value Trying to lose weight Attractiveness 1.29 ( ) < ( ) ( ) Body mass index 1.62 ( ) < ( ) < ( ) <.01 Age 0.95 ( ) ( ) ( ) Socioeconomic status a 1.01 ( ) ( ) ( ) Self-esteem 0.49 ( ) < ( ) ( ) <.01 Negative affect 1.55 ( ) ( ) ( ) Lack of positive affect 1.03 ( ) ( ) ( ) Family connectedness 0.96 ( ) ( ) ( ) Exercising or dieting to lose weight or to keep from gaining weight Attractiveness 1.28 ( ) < ( ) ( ) Body mass index 1.35 ( ) < ( ) < ( ) <.01 Age 0.90 ( ) ( ) ( ) Socioeconomic status a 1.24 ( ) ( ) ( ) Self-esteem 0.66 ( ) < ( ) ( ) Negative affect 0.92 ( ) ( ) ( ) Lack of positive affect 0.87 ( ) ( ) ( ) Family connectedness 1.12 ( ) ( ) ( ) Believe that they are overweight Attractiveness 0.98 ( ) ( ) ( ) Body mass index 1.80 ( ) < ( ) < ( ) <.01 Age 0.93 ( ) ( ) ( ) Socioeconomic status a 1.12 ( ) ( ) ( ) Self-esteem 0.48 ( ) < ( ) ( ) <.01 Negative affect 1.16 ( ) ( ) ( ) Lack of positive affect 1.04 ( ) ( ) ( ) Family connectedness 0.86 ( ) ( ) ( ) a At least one parent with college education versus less than college education.

7 Weight Preoccupation and Attractiveness 809 Self-esteem offered a significant protective effect for trying to lose weight and believing they were overweight for all three ethnic groups. Additionally, higher scores on the self-esteem scale offered a significant protective effect for Caucasians only for dieting and exercising to lose weight or to keep from gaining weight. Family connectedness was related to weight preoccupation only among Hispanic girls. For this ethnic group, higher scores on the family connectedness scale were associated with a decreased likelihood of dieting or exercising to lose weight or to keep from gaining weight. Last, the negative affect scale of the depressed mood measure was significantly related to weight preoccupation among Caucasians and Hispanic adolescents. Higher levels of negative affect (i.e., more depressed mood) were significantly associated with an increased likelihood of trying to lose weight for Caucasian and Hispanic girls and with dieting or exercising to lose weight or to keep from gaining weight for Hispanic girls. In contrast, higher scores on the lack of positive affect scale (i.e., more depressed mood) was significantly related to a decreased likelihood of dieting or exercising to lose weight or to keep from gaining weight for Hispanic girls. Analyses Controlling for Any Rater Effects Because there were 274 interviewers who rated the physical attractiveness of the 4,324 adolescents, additional analyses were run to control for the fact that the same interviewer would rate multiple adolescents (resulting in nonindependence). Robust variance estimators were calculated to control for the clustered data within an interviewer. The odds ratios in these analyses between ratings of physical attractiveness and each of the four outcomes were nearly identical across all ethnicities as the earlier analyses, and the same associations were found. Owing to software limitations, this analysis controlled for interviewer effect but did not take into account the complex survey design. In summary, based on these analyses, it is believed that the nonindependent ratings within an interviewer did not significantly alter any of the findings of relationships between objective physical attractiveness and weight preoccupation among females from all three ethnic groups. Discussion Supporting our hypotheses, non-hispanic Caucasian female adolescents who were normal weight and rated by observers as being more physically attractive were more likely to be (a) trying to lose weight and (b) dieting or exercising to lose weight or to keep from gaining weight compared with those who were rated as being less attractive after controlling for demographics, psychosocial factors, and interviewer-measured BMI. Caucasian females who were rated as being very attractive were more than twice as likely (OR = 2.2) to be trying to lose weight and to be dieting or exercising to lose weight or to keep from gaining weight (OR = 2.1) compared with those who were rated as the least attractive. These findings are consistent with Davis and colleagues research on facial attractiveness relating to weight preoccupation among primarily Caucasian college-aged women (Davis et al., 2000; 2001). The finding that adolescent girls who are perceived as being more attractive are more preoccupied by weight concerns is not surprising. These girls are likely to have been praised from a young age for their looks; consequently, their physical appearance and being considered attractive by others may have become an important part of their self-identity. To maintain this self-identity, these girls may feel more pressure to maintain a socially preferred ideal weight, because thinness and attractiveness are closely tied. These findings suggest that physical beauty may indeed function as a risk factor among Caucasian adolescent females for weight preoccupation and therefore potentially for the development of eating disorders. A review that found consistent support for a thin-ideal internalization as a risk factor for eating pathology supports this notion (Stice, 2002). The differential findings by ethnicity may be supported by previous research. Research shows that the association between a thin physique and perceptions of physical attractiveness is much stronger in Caucasian populations compared with minority populations, as African-American women report less weight dissatisfaction and a heavier and less narrowly defined body ideal than Caucasian women (Striegel-Moore & Smolak, 1996). Further supporting this notion, a recent focus group found that African-American and Hispanic women contest ideologies defining thinness as inherently beautiful and instead espouse a body ethic of self-acceptance and nurturance that rejects mainstream cultural pressures to reshape bodies to approximate aesthetic ideals (Rubin, Fitts, & Becker, 2003). Contrary to our hypothesis, normal-weight female adolescents who were rated as being more attractive were not more likely to think they were overweight. This was true in all three ethnic groups. This means that girls who are perceived by others as being more attractive are not more likely to (falsely) believe that they are overweight compared with girls who are perceived as being less attractive. Also contrary to our expectations,

8 810 Colabianchi, Ievers-Landis, and Borawski objective ratings of attractiveness were not associated with whether girls reported they were likely to engage in inappropriate dieting behaviors. This finding may be a function of the very small number of normal weight adolescents that engaged in this behavior (n = 71) and/or using a nationally representative sample rather than a clinical sample of adolescents who engage in disordered eating practices. It is noteworthy that the prevalence of inappropriate dieting behaviors in this sample is lower than the prevalence rates seen in other studies. Prevalence estimates of disordered eating among adolescent females from other population-based surveys varies greatly from 13% in all weight ranges reporting that they have ever engaged in binge-purge cycling (Neumark-Sztainer & Hannan, 2000) to 56% in all weight ranges reporting a wide variety of disordered eating behaviors (e.g., binge eating, fasting or skipping meals, using diet pills, vomiting, laxatives, or smoking cigarettes to lose weight) in the past year (Croll et al., 2002). The lower percentage of adolescent females in our sample reporting inappropriate dieting behaviors (i.e., self-induced vomiting, taking diet pills or laxatives) may be because of a variety of factors including the specificity of the behaviors, the short-time frame (i.e., within the past week), and the normal-weight sample. Rates of disordered eating such as binge-purge cycling and self-induced vomiting have tended to be highest among overweight youth, which are not included in this sample (Neumark-Sztainer, Story, Falkner, Beuhring, & Resnick, 1999). Last, the weight preoccupation measures in this study are taken from the in-home portion of the Add Health study, which is administered by an interviewer. Prevalence estimates of disordered eating from interviewer-assisted surveys have been found to be substantially lower than prevalence estimates from self-administered surveys in adolescents (French et al., 1998). How the low prevalence of reported behavior affects the generalizability of the null findings is not known. Furthermore, the measure of inappropriate dieting behaviors is not a direct indication of an eating disorder. A diagnosis of an eating disorder is not available from these data. Interestingly, only 60% of those who reported engaging in inappropriate dieting behaviors also reported that they thought they were overweight, which would be a hallmark indication of an eating disorder. A greater proportion (87%) of these adolescents reported that they were trying to lose weight. That BMI was consistently related to weight preoccupation, even in a sample of normal-weight adolescent females, is not surprising. A study with an ethnically diverse sample of college women also found that concerns about body weight and body shape varied according to the individual s weight, that is, the more one weighs, the higher the concern (Arriaza & Mann, 2001). In their study, ethnic differences in eating disorder symptoms of concern about body weight and shape disappeared after controlling for BMI with few exceptions (i.e., restrained eating remained more common among Caucasians). Thus, regardless of level of attractiveness, females on the upper end of normal weight may be more likely to engage in attempts to lose weight or maintain their weight. The associations between psychosocial factors (selfesteem, depressed mood, and family connectedness) and weight preoccupation are generally consistent with other studies of dieting behaviors and disordered eating (Croll et al., 2002; Neumark-Sztainer & Hannan, 2000; Pesa, 1999; Stice, 2002). The findings of ethnic differences on risk and protective factors may be compared with the findings of Croll et al. (2002) in a multiethnic sample of 81,247 9th 12th graders who completed the 1998 Minnesota Student Survey. A notable difference is that the Croll et al. study focused on disordered eating behaviors (e.g., fasting or skipping meals, taking diet pills, taking laxatives, vomiting, or smoking cigarettes, binge eating) rather than on weight preoccupation. While this study found that family connectedness was a protective factor for weight preoccupation only for Hispanic females, Croll et al. found this relationship for disordered eating behaviors amongst the full sample but did not find a significant association in Hispanic female adolescents in stratified models. In addition, this study found depressive symptomatology to be a risk factor for weight preoccupation only for Caucasian and Hispanic females, whereas the Croll et al. study found emotional well-being to be a factor for disordered eating in these groups as well as for African-American girls. Finally, in this study, self-esteem was a protective factor against weight preoccupation for all three ethnic groups, whereas self-esteem was not a significant factor for protecting against disordered eating among African-American girls in the Croll et al. study. Owing to these inconsistencies, further exploration of ethnic group differences in risk and protective factors for weight preoccupation and disordered eating behaviors is warranted. Limitations to be considered when interpreting these findings are as follows. First, and most importantly, as only one interviewer rated the attractiveness of each girl, the percentage of agreement or kappas between raters is not available. Furthermore, attractiveness is measured through one variable with a limited

9 Weight Preoccupation and Attractiveness 811 range. Additionally, there is no documentation that the interviewers were trained on how to reliably assess attractiveness. Although the interviewers were different races and ethnicities, there is no information about the race or ethnicity of the interviewer for any one respondent to examine systematic bias. However, post hoc cluster analyses that controlled for an interviewer effect yielded similar results; thus, it is unlikely that interviewers had any substantial biases against any one ethnic group or that they rated attractiveness significantly differently from each other. Second, questions about dieting and exercising behaviors (with the exception of the inappropriate dieting behaviors questions) were not detailed enough to allow for a determination of whether the girls used healthy and/or appropriate means to maintain their weight or lose a small amount of weight. Thus, whether weight preoccupation is indeed the appropriate term for girls who reported attempting weight control techniques is not truly known. However, as a larger percentage of girls reported engaging in weight control techniques than reported believing they were overweight lends some credence to the possibility that these girls were indeed inappropriate dieters (Pesa, 1999). Third, the cross-sectional nature of the data prohibits any claims toward causality. Whether psychosocial factors or BMI led to greater or less weight preoccupation or whether dieting/exercise behaviors led to changes in body fat and symptoms of depression, self-esteem, and so on cannot be inferred. Fourth, since the number of minority females in our sample was not as large as the number of Caucasians, there was not equal power to detect differences in all ethnic groups, although the sample of each ethnicity was reasonably large. Finally, measures of psychosocial functioning were self-report and are, therefore, subject to reporting biases. However, attractiveness and BMI were based on interviewers observations and measurement, respectively. Our findings have important clinical implications for the psychosocial and medical care of female adolescents. Psychologists and medical personnel who have a clinical caseload comprised of adolescent and young adult females need to be aware that those who are perceived as being more attractive by others, regardless of how they perceive themselves, might feel unduly pressured to fit into society s current ideal body type. These young women are at greater risk for engaging in dietary and physical activity habits that may be detrimental to their physical health to pursue this ideal. Regular counseling may be warranted regarding recommended methods for achieving and maintaining a healthy body weight. Potentially, beneficial techniques for reducing weight preoccupation among normal-weight females include recognition and treatment of depressive symptomatology and promotion of positive self-esteem. Health care professionals must face the challenge of maintaining a balance between discouraging weight preoccupation in normal-weight adolescents while providing them with health promotion messages to support prevention of overweight. Acknowledgments This research uses data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris, and funded by a grant P01-HD31921 from the National Institute of Child Health and Human Development, with cooperative funding from 17 other agencies. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. Persons interested in obtaining data files from Add Health should contact Add Health, Carolina Population Center, 123 W. Franklin Street, Chapel Hill, NC (addhealth@unc.edu). Received March 16, 2005; revisions received August 1, 2005 and October 15, 2005; accepted November 12, 2005 References Abrams, L. S., & Stormer, C. C. (2002). Sociocultural variations in the body image perceptions of urban adolescent females. Journal of Youth and Adolescence, 31, Arriaza, C. A., & Mann, T. (2001). Ethnic differences in eating disorder symptoms among college students: The confounding role of body mass index. Journal of American College Health, 49, Buss, D. M. (1994). The evolution of desire. New York: Basic Books. Croll, J., Neumark-Sztainer, D., Story, M., & Ireland, M. (2002). Prevalence and risk and protective factors related to disordered eating behaviors among adolescents: Relationship to gender and ethnicity. Journal of Adolescent Health, 31, Davis, C., Claridge, G., & Fox, J. (2000). Not just a pretty face: Physical attractiveness and perfectionism in the risk for eating disorders. International Journal of Eating Disorders, 27, Davis, C., Shuster, B., Dionne, M., & Claridge, G. (2001). Do you see what I see?: Facial attractiveness and weight preoccupation in college women. Journal of Social and Clinical Psychology, 20,

10 812 Colabianchi, Ievers-Landis, and Borawski Duncan, P. D., Ritter, P. L., Dornbusch, S. M., Gross, R. T., & Carlsmith, J. M. (1985). The effects of pubertal timing on body image, school behavior, and deviance. Journal of Youth and Adolescence, 14, Feingold, A. (1992). Good-looking people are not what we think. Psychological Bulletin, 111, Field, A. E., Camargo, C. A., Jr., Taylor, C. B., Berkey, C. S., Roberts, S. B., & Colditz, G. A. (2001). Peer, parent, and media influences on the development of weight concerns and frequent dieting among preadolescent and adolescent girls and boys. Pediatrics, 107, Field, A. E., Cheung, L., Wolf, A. M., Herzog, D. B., Gortmaker, S. L., & Colditz, G. A. (1999). Exposure to the mass media and weight concerns among girls. Pediatrics, 103, E36. French, S. A., Peterson, C. B., Story, M., Anderson, N., Mussell, M. P., & Mitchell, J. E. (1998). Agreement between survey and interview measures of weight control practices in adolescents. International Journal of Eating Disorders, 23, Miller, K. J., Gleaves, D. H., Hirsch, T. G., Green, B. A., Snow, A. C., & Corbett, C. C. (2000). Comparisons of body image dimensions by race/ethnicity and gender in a University population. International Journal of Eating Disorders, 27, Miller, D. T., & Turnbull, W. (1986). Expectancies and interpersonal processes. Annual Review of Psychology, 37, Neumark-Sztainer, D., & Hannan, P. J. (2000). Weightrelated behaviors among adolescent girls and boys: Results from a national survey. Archives of Pediatrics and Adolescent Medicine, 154, Neumark-Sztainer, D., Story, M., Falkner, N., Beuhring, T., & Resnick, M. (1999). Sociodemographic and personal characteristics of adolescents engaged in weight loss and weight/muscle gain behaviors: Who is doing what? Preventive Medicine, 28, Pesa, J. (1999). Psychosocial factors associated with dieting behaviors among female adolescents. Journal of School Health, 69, Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris, K. M., Jones, J., et al. (1997). Protecting adolescents from harm. Findings from the National Longitudinal Study on Adolescent Health. Journal of the American Medical Association, 278, Rubin, L. R., Fitts, M. L., & Becker, A. E. (2003). Whatever feels good in my soul : Body ethics and aesthetics among African American and Latina women. Culture, Medicine, and Psychiatry, 27, Russell, W. D., & Cox, R. H. (2003). Social physique anxiety, body dissatisfaction, and self-esteem in college females of differing exercise frequency, perceived weight discrepancy, and race. Journal of Sport Behavior, 26, Stata Corporation. (2002). Stata (Version 7.0). College Station, TX: Stata Corporation. Stice, E. (2002). Risk and maintenance factors for eating pathology: A meta-analytic review. Psychological Bulletin, 128, Striegel-Moore, R. H., & Smolak, L. (1996). The role of race in the development of eating disorders. In L. Smolak, M. P. Leveine, & R. H. Striegel-Moore (Eds.), The developmental psychopathology of eating disorders (pp ). Machway, NJ: Erlbaum. Thomsen, S. R., Weber, M. M., & Brown, L. B. (2002). The relationship between reading beauty and fashion magazines and the use of pathogenic dieting methods among adolescent females. Adolescence, 37, White, M. A., Kohlmaier, J. R., Varnado-Sullivan, P., & Williamson, D. A. (2003). Racial/ethnic differences in weight concerns: Protective and risk factors for the development of eating disorders and obesity among adolescent females. Eating and Weight Disorders, 8, Wilfley, D. E., Schreiber, G. B., Pike, K. M., Striegel- Moore, R. H., Wright, D. J., & Rodin, J. (1996). Eating disturbance and body image: A comparison of a community sample of adult Black and White women. International Journal of Eating Disorders, 20, Zebrowitz, L. A. (1997). Reading faces. Window to the soul?. Boulder, CO: Westview Press.