The role of social norms and friends’ influences on unhealthy weight-control behaviors among adolescent girls

ARTICLE IN PRESS

Social Science & Medicine 60 (2005) 1165–1173 www.elsevier.com/locate/socscimed

The role of social norms and friends’ influences on unhealthy weight-control behaviors among adolescent girls Marla E. Eisenberg, Dianne Neumark-Sztainer, Mary Story, Cheryl Perry Division of Epidemiology, University of Minnesota School of Public Health, 1300 Second Street SE, Ste. 300, Minneapolis, MN 55454, USA Available online 19 August 2004

Abstract Dieting is common among adolescent girls and may place them at risk of using unhealthy weight-control behaviors (UWCBs), such as self-induced vomiting, laxatives, diet pills, or fasting. Research has suggested that social factors, including friends and broader cultural norms, may be associated with UWCBs. The present study examines the relationship between the school-wide prevalence of current weight loss efforts among adolescent girls, friends’ dieting behavior, and UWCBs, and investigates differences in these associations across weight categories. Survey data were collected in 31 middle and high schools in ethnically and socio-economically diverse communities in Minnesota, USA. The response rate was 81.5%. Rates of UWCBs were compared across the spectrum of prevalence of trying to lose weight and friends’ involvement with dieting, using w2 analysis and multivariate logistic regression, controlling for demographic factors and clustering by school. Girls with higher body mass index (BMI) were more likely to engage in UWCBs than those of lower BMI. Multivariate models indicated that friends’ dieting behavior was significantly associated with UWCBs for average weight girls (OR=1.57, CI=1.40–1.77) and moderately overweight girls (OR=1.47, CI=1.19–1.82). The school-wide prevalence of trying to lose weight was significantly, albeit modestly, related to UWCBs for average weight girls (15th–85th percentile; OR=1.17, CI=1.01–1.36), and marginally associated for modestly overweight girls (85th–95th percentile; OR=1.21, CI=.97–1.50), even after controlling for friends’ dieting behaviors. The social influences examined here were not associated with UWCBs among underweight (o15th percentile) or overweight (495th percentile) girls. Findings suggest that social norms, particularly from within one’s peer group, but also at the larger school level may influence UWCBs, particularly for average weight girls. Implications for school-based interventions to reduce UWCBs are discussed. r 2004 Elsevier Ltd. All rights reserved. Keywords: Social norms; Peer influence; Dieting; Disordered eating; Adolescence; USA

Introduction Background

Corresponding author. Tel.: +1-612-626-8602; fax: +1612-626-7103 E-mail address: [email protected] (M.E. Eisenberg).

Dieting is a common occurrence among adolescent girls (Huon, 1994). According to the Youth Risk Behavior Surveillance System of the Centers for Disease Control (CDC) in the USA, almost two-thirds (59.4%) of adolescent girls report trying to lose weight in the past

0277-9536/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2004.06.055

ARTICLE IN PRESS 1166

M.E. Eisenberg et al. / Social Science & Medicine 60 (2005) 1165–1173

30 days (CDC, 2000). While many use healthy strategies, such as exercising (67.4%) or eating fewer calories or foods lower in fat (56.1%), unhealthy strategies, such as fasting (18.8%), using diet pills/powders/liquids (10.9%), and vomiting or taking laxatives (7.5%) were also common (CDC, 2000). Research has shown that adolescent dieters, particularly those practicing unhealthy weight-control behaviors (UWCBs), may be less likely to consume fruits and vegetables and adequate amounts of calcium rich foods, compared to non-dieters (Neumark-Sztainer, Story, Dixon, Resnick, & Blum, 1997). Further, adolescent dieters may actually experience excess weight gain or the onset of obesity (Stice, Cameron, Killen, Hayward, & Taylor, 1999). Dieting also increases girls’ risk for developing UWCBs, disordered eating (Drewnowski, Yee, Kurth, & Krahn, 1994; French & Jeffery, 1994; Patton, Johnson-Sabine, Wood, Mann, & Wakeling, 1990; Huon, 1994; Patton, Selzer, Coffey, Carlin, & Wolfe, 1999), and depression (Stice, Hayward, Cameron, Killen, & Taylor, 2000). One study found that dieters were eight times more likely to develop an eating disorder than non-dieters (Patton et al., 1990). More extreme weight-control methods such as induced vomiting, laxative abuse, and binge-purge cycling have also been associated with tooth decay, damage to the esophagus, digestive problems, amenorrhea, and cardiomyopathy (Herzog & Copeland, 1985; Ho, Dweik, & Cohen, 1998; Johnson & Whitaker, 1992). Given the multiple risks associated with UWCBs, the purpose of the current study is to examine elements of girls’ social environment and determine the extent to which they may be associated with such practices. Identifying modifiable social factors may suggest appropriate avenues for intervention and prevention of UWCBs. Social influences on weight-related issues Researchers have already begun to uncover certain social factors that may act on adolescent eating and weight-related behaviors, including direct persuasion (e.g. from parents), social norms and need to fit in with peers, and food availability at home, school, and away from home (Conner & Armitage, 2002; Stice, 1994; Story, Neumark-Sztainer, & French, 2002). Research has suggested that socio-cultural factors, such as a value on thinness, may be related to unhealthy weight-control and disordered eating behaviors (Levine, Smolak, & Hayden, 1994; Stice, 1994). In particular, exposure to media images of thin women have also been shown to contribute to poor body image and unhealthy weightcontrol practices (Field, Camargo, Taylor, Berkey, & Colditz, 1999; Field et al., 2001, 1999; Turner, Hamilton, Jacobs, Angood, & Dwyer, 1997). One of the most extensive segments of the literature on the social influences on adolescents’ weight-related

behaviors addresses the role of peers and friends. Several researchers have demonstrated the important influence of friends, suggesting that the weight-related attitudes and behaviors among friendship groups may predict body image, dieting onset, chronic dieting, UWCBs, and eating disorder symptoms, even after controlling for various family, friend, and individual factors (Gibbs, 1986; Huon, Lim, & Gunewardene, 2000; Huon & Walton, 2000; Paxton, Schutz, Wertheim, & Muir, 1999; Pike, 1995). Several mechanisms have been proposed to explain these similarities, such as friends’ sharing information on weight-control strategies (Desmond, Prince, Gray, & O’Connell, 1986), friends participating in ‘‘fat-talk’’ (Nichter, 2000), modeling disordered eating behavior (e.g. teaching how to induce vomiting; Stice, 1994), or self-comparison to others in the clique (Wertheim, Paxton, Schutz, & Muir, 1997). Peer influence may also operate more broadly, in a larger social unit such as the school environment. Pressure to diet may come through direct contact in the form of weight-teasing, which is widespread—particularly among overweight individuals—and has been linked to UWCBs (Neumark-Sztainer et al., 2001). Wertheim et al. (1997) suggest a more indirect mechanism, such as the idealization of ‘‘popular’’ girls as thin and pretty, which may manifest itself in a desire to diet in order to emulate or fit in with this crowd, even in the absence of any overt persuasion from the popular girls. Social norms are another mechanism through which peer influence may operate without any direct contact. A high prevalence of a behavior among young people in a setting may send a subtle message that such behavior is accepted and indeed, expected, which may encourage adoption of that behavior throughout that social setting. Social norms of binge drinking (Lo & Globetti, 1995; Wechsler, Kuh, & Davenport, 1996) and binge eating (Crandall, 1988) have been tied to these behaviors among individuals in the relatively cohesive settings of college fraternities and sororities. Austin (2001) has also illustrated this relationship in middle schools, where the school-wide prevalence of dieting (without high-risk methods) was associated with the school-level prevalence of disordered eating behaviors. Social influences related to dieting may function differently across the weight spectrum. For example, overweight girls may be socially marginalized (Strauss & Pollack, 2003), may be frequent victims of weight-based teasing (Neumark-Sztainer et al., 2001), or may be motivated due to health concerns, all of which may impact their dieting behavior. The dieting behavior of underweight or average weight girls, by contrast, may be more directly influenced by the attitudes and behaviors of family or friends. However, the moderating effect of body weight in the relationship between socio-cultural influences and weight-related behaviors has not been well examined in the scientific literature (Stice, 1994).

ARTICLE IN PRESS M.E. Eisenberg et al. / Social Science & Medicine 60 (2005) 1165–1173

Although the existing literature documents the importance of social influences on girls’ dieting and UWCBs, to our knowledge no study has simultaneously examined the influences of friends and social norms. Such an investigation would allow us to parse out the contribution of proximal influences of friends and the more distal influences of the school environment. Distinguishing these influences may provide insight into strategies for reducing the use of unhealthy weightcontrol practices among girls. Thus, the purpose of the present study is to examine the relationship between two social factors: (a) the school-wide prevalence of girls’ trying to lose weight (a proxy for normative dieting behavior in the school environment) and (b) perceptions of friends’ dieting behavior (as modeling of dieting behavior), and associations with UWCBs in a large, community-based sample of adolescent girls. This study further addresses the question of whether these social factors operate consistently across weight categories of adolescent girls. We hypothesize that girls attending schools that have a high prevalence of girls trying to lose weight and who have many friends who diet will be more likely to engage in UWCBs than girls without these characteristics. Furthermore, we expect these social influences to be more salient for average weight and underweight girls than overweight girls.

Methods Data for this study come from Project EAT (Eating Among Teens), a comprehensive study of eating patterns and weight concerns among adolescents. Thirty-one public middle and senior high schools in ethnically and socio-economically diverse communities in the urban and suburban school districts of the Minneapolis/St. Paul metropolitan area participated in the study. A 221-item survey assessing a range of socio-economic, personal, and behavioral factors relevant to dietary intake and weight status was completed by students during regular class periods, and height and weight were measured by trained research staff in a private area of the school. Study procedures were approved by the University of Minnesota Human Subjects’ Committee and by the research boards of the participating school districts. The response rate for student participation was 81.5%. The main reasons for non-participation were absenteeism and failure to return consent forms. Further details of the study design and data collection methods are given elsewhere (Neumark-Sztainer et al., 2001; NeumarkSztainer, Story, Hannan, Perry, & Irving, 2002) Measures The survey was designed by Project EAT investigators. Most items were drawn from existing adolescent

1167

surveys; some were developed to meet the specific needs of the study. The survey went through multiple revisions based on input from experts in the field, an adolescent advisory board, and 68 7th and 10th grade students (at different schools in the same school districts as the final survey) who pre-tested the survey. Follow-up pilot testing was conducted with 161 students over a two-week interval to assess psychometric properties of questions, problems with missing data, potential for bias, and consistency of responses. Minor changes to the survey were made based on feedback from pilot tests. The school-wide dieting norm was calculated as the prevalence of girls at each school who indicated that they were ‘‘currently trying to lose weight’’. Two schools with a small number of respondents (n ¼ 10) were dropped from the sample, leaving 29 schools. Schoolwide dieting prevalence ranged from 21.1% to 60.0% of girls who reported currently trying to lose weight, with a mean of 42.2%. This distribution of school prevalence was divided into quartiles, and all girls in each school were assigned the prevalence level of their school. Friends’ dieting was assessed with one item ‘‘Many of my friends diet to lose weight or keep from gaining weight’’ which had response options of ‘‘not at all’’, ‘‘a little bit’’, ‘‘somewhat’’, ‘‘very much’’, and ‘‘I don’t know’’. Girls who responded ‘‘I don’t know’’ were not included in analyses using the friends’ dieting variable. The most frequent weight loss methods described in the research literature were chosen for inclusion in this survey. UWCBs were measured with a series of items which read ‘‘Have you done any of the following things in order to lose weight or keep from gaining weight during the past year?’’ (yes/no for each method). Items included (1) took diet pills; (2) made myself vomit; (3) used laxatives; (4) used diuretics; (5) fasted; (6) ate very little food; (7) used food substitute (powder/special drink); (8) skipped meals; and (9) smoked more cigarettes. These items were modified from previous large-scale surveys of adolescents (Grunbaum et al., 2002; Neumark-Sztainer, Butler, & Palti, 1995), and had a Cronbach’s a of .70 for girls in the present study. For the majority of specific weight-control behaviors, test–retest k’s ranged from .50 to .68; however, lower values were found for laxatives (.29) and food substitutes (.44). Students indicating they had used at least one of the extremely unhealthy behaviors (i.e. vomiting, laxatives, diuretics, smoking) or at least two of the less extreme behaviors (i.e. diet pills, fasted, ate little food, food substitute, skipped meals) were classified as using UWCBs. This criterion is more stringent than what has been used previously (Neumark-Sztainer et al., 2001; Neumark-Sztainer, Story, Hannan, & Moe, 2002), and was selected in order to reduce the possibility of confounding typical dieting behavior with the use of unhealthy methods.

ARTICLE IN PRESS 1168

M.E. Eisenberg et al. / Social Science & Medicine 60 (2005) 1165–1173

Race/ethnicity was assessed with one item: ‘‘Do you think of yourself as (1) white, (2) black or AfricanAmerican, (3) Hispanic or Latino, (4) Asian-American, (5) Hawaiian or Pacific Islander, or (6) American Indian or Native American.’’ Due to small numbers of some racial/ethnic groups in some body mass index (BMI) categories, respondents were grouped as white and nonwhite for multivariate analysis. Five levels of socioeconomic status (SES) (low to high) were based on the highest educational level completed by either parent for most respondents. Where this information was missing (n ¼ 1058), eligibility for public assistance, eligibility for free or reduced cost school meals, and parental employment status were used to infer SES (Breiman, Friedman, Olshen, & Stone, 1984; Neumark-Sztainer et al., 2002). School level was defined as middle school (grades 7–8) vs. high school (grades 9–12). BMI was determined by anthropometric measures of height and weight, using standardized equipment and procedures. Height without shoes was measured with a portable stadiometer, using the Frankfort Plane technique, and was recorded to the nearest .1 cm. Weight was measured in street clothes without heavy outerwear or shoes. Weight was assessed with a portable digital scale, which had been calibrated with a certified 50-lb weight, and was recorded to the nearest .5 lb. Because 11% of girls were missing observed height and weight data, self-reported height and weight were substituted in these cases to reduce the missing proportion to 3%, as the self-reported and observed measures were highly correlated (r ¼ :85; po:001). BMI was calculated according to the formula: weight (kg)/height (m)2. Gender- and age-specific cut points for underweight (o15th percentile), average weight (15th–85th percentile), moderately overweight (85th–95th percentile), and overweight (495th percentile) were based on data from the Centers for Disease Control and Prevention (Kuczmarski et al., 2000). Data analysis w2 tests of significance were used to detect differences in the proportion of girls reporting UWCBs at schools in each quartile of prevalence of girls’ trying to lose weight, and at each level of friends’ dieting. Mixed multivariate logistic regression models were used to examine simultaneously the relationship between the two social factors (school prevalence of trying to lose weight, and friends’ dieting) and UWCBs, and then these two social factors and UWCBs, controlling for grade level, race and SES. The use of mixed models accounted for the intraclass correlation of students within schools, resulting from the sampling design. All analyses were stratified by BMI category. The SAS 8.12 macro-GLIMMIX was used, with specifications appropriate to a dichotomous dependent variable (SAS Institute, Cary, NC).

Results Characteristics of the sample Characteristics of the student sample are shown in Table 1. Approximately two-thirds of the girls were in high school, and almost half (45.7%) were white. Girls were fairly evenly distributed across the five SES categories. The majority (62.8%) of girls were of normal weight, 19.9% were moderately overweight and an additional 12.5% were overweight. Many girls (44.9%) reported engaging in UWCBs. Likewise, many reported their friends were either somewhat (26.4%) or very much (15.8) involved in dieting to lose or not gain weight. Almost one-third (29.2%) went to a school in the highest quartile of prevalence of girls’ trying to lose weight, and one-third (34.2%) went to a school in the second highest quartile. More than one-quarter of girls were in these categories due to a higher prevalence of Table 1 Characteristics of student sample (n ¼ 2337; 29 schools) Characteristic

N

%

School level Junior/middle High school

798 1510

34.6 65.4

Race/ethnicity White African-American Hispanic Asian Native American Mixed/other

1051 461 117 473 90 107

45.7 20.1 5.1 20.6 3.9 4.7

455 431 575 484 303

20.2 19.2 25.6 21.5 13.5

109 1420 449 283 1028

4.8 62.8 19.9 12.5 44.9

Friends diet to lose/not gain weight Not at all A little bit Somewhat Very much Do not know

529 552 492 294 336

28.3 29.6 26.4 15.8 16.4

School prevalence of dieting Highest quartile Medium high Medium low Lowest quartile

682 800 461 394

29.2 34.2 19.7 16.9

SES Low Low-middle Middle High-middle High BMI Underweight (o15th percentile) Average weight (15th–85th percentile) Moderately overweight (85th–95th percentile) Overweight (495th percentile) Unhealthy weight-control practices

ARTICLE IN PRESS M.E. Eisenberg et al. / Social Science & Medicine 60 (2005) 1165–1173

trying to lose weight in larger schools. Overweight girls (60.1%) were more likely to engage in UWCBs than moderately overweight (54.0%), average weight (40.2%), or underweight (21.3%) girls (w2 ¼ 78:0; po:001).

Associations between school-wide prevalence of girls’ trying to lose weight, friends’ dieting, and unhealthy weight control School-wide prevalence of girls’ trying to lose weight and friends’ dieting had a low but statistically significant Spearman correlation (r ¼ :11; po:001). The proportion of girls practicing UWCBs at each level of school prevalence and friends’ dieting is shown in Table 2, across BMI categories. School-wide prevalence of girls’ trying to lose weight and friends’ dieting were significantly related to UWCBs among average weight and moderately overweight girls. Among average weight girls, 27.9% of those in the lowest quartile of school prevalence engaged in UWCBs, whereas 43.7% of those in the highest quartile reported UWCBs (w2 ¼ 17:9; po:001). Among moderately overweight girls, the relationship was less linear but still showed a trend in which girls at schools with a higher prevalence of trying to lose weight were more likely to use UWCBs (w2 ¼ 17:4; po:001). Friends’ dieting behavior was similarly related to UWCBs: 28.7% of average weight girls whose friends were ‘‘not at all’’ involved with dieting reported using UWCBs, while more than twice that proportion (59.5%) of girls whose friends were ‘‘very much’’ involved with dieting used UWCBs (w2 ¼ 68:4; po:001). Forty-three percent of moderately overweight girls whose friends did not diet reported UWCBs, compared to 70.2% of moderately overweight girls whose friends were involved in dieting (w2 ¼ 14:8;

1169

po:01). Neither social factor was significantly related to UWCBs among underweight nor very overweight girls. An unstratified multivariate mixed model which mutually adjusted for both social influence variables was run. In this model, friends’ dieting was significantly associated with UWCBs (OR=1.49, CI=1.35–1.65), and school-wide prevalence of trying to lose weight was marginally associated with UWCBs (OR=1.13, CI=1.00–1.29). This model was re-run controlling for personal covariates (school level, race/ethnicity, SES, and BMI), as well as clustering of individuals by school. The relationship of friends’ dieting and UWCBs remained unchanged; school-wide prevalence, however, was no longer associated with UWCBs. In multivariate mixed models stratified by BMI, which mutually adjusted for both social influence variables (Table 3), friends’ dieting was significantly associated with UWCBs for average weight girls (OR=1.57, CI=1.40–1.77) and moderately overweight girls (OR=1.47, CI=1.19–1.82). School-wide prevalence of trying to lose weight was significantly associated with UWCBs for average weight girls (OR=1.17, CI=1.01–1.36) and marginally (po:10) associated with UWCBs for moderately overweight girls (OR=1.21, CI=.97–1.50). Thus, average weight girls in each prevalence quartile of schools have odds of using UWCBs which are 17% greater than the odds for girls in the next lower quartile of schools. These stratified models were re-run controlling for personal covariates, and are shown in Table 4. Friends’ dieting behavior remained significantly associated with UWCBs for average weight girls (OR=1.58, CI=1.40–1.79) and moderately overweight girls (OR=1.48, CI=1.18–1.84), such that greater involvement of friends in dieting was related to a greater likelihood of using UWCBs. The school-wide prevalence of trying to lose weight was marginally (po:10) related

Table 2 Unhealthy weight-control behaviors (%), by social influences, stratified by BMI category Underweight

Average weight

Moderately overweight

Overweight

School prevalence of trying to lose weight 1 (lowest quartile) 2 3 4 (highest quartile) w2

n ¼ 108 23.1 31.8 20.0 13.3 2.7, NS

n ¼ 1399 27.9 41.4 42.4 43.7 17.9, po.001

n ¼ 443 35.0 57.7 63.3 52.1 17.4, po.001

n ¼ 281 70.5 47.5 63.1 60.9 6.3, po.10

Friends diet Not at all Little bit Somewhat Very much w2

n ¼ 89 15.4 16.7 13.6 40.0 2.1a, NS

n ¼ 1147 28.7 33.8 52.2 59.5 68.4, po.001

n ¼ 356 42.9 50.0 63.1 70.2 14.8, po.01

n ¼ 222 57.7 53.3 59.3 68.6 2.8, NS

a 2

w test may be invalid due to small cell sizes.

ARTICLE IN PRESS 1170

M.E. Eisenberg et al. / Social Science & Medicine 60 (2005) 1165–1173

Table 3 Odds ratios for unhealthy weight-control behaviors, adjusted for two social factors Characteristic

Underweight n ¼ 89

Average weight n ¼ 1147

Moderately overweight n ¼ 356

Overweight n ¼ 222

School prevalence of trying to lose weighta Friends dieta

.82 (.49, 1.39) 1.29 (.69, 2.40)

(1.01, 1.36) (1.40, 1.77)

1.21 (.97, 1.50) 1.47 (1.19, 1.82)

.91 (.70, 1.18) 1.18 (.92, 1.52)

Bold text indicates significance at .05 level. a Note: Odds ratios are for each category compared to the next lower category, e.g. highest vs. next highest quartile of school prevalence, friends diet somewhat vs. little bit.

Table 4 Odds ratios for unhealthy weight-control behaviors, mutually adjusted for all variables Characteristic

Social factors School prevalence of trying to lose weighta Friends dieta Personal characteristics High school (vs. middle) White (vs. non-White) SES categorya

Underweight

Average weight

Moderately overweight

Overweight

n ¼ 87

n ¼ 1119

n ¼ 346

n ¼ 211

.87 (.46, 1.66) 1.02 (.50, 2.07)

1.12 (.99, 1.27) (1.40, 1.79)

1.11 (.88, 1.39) (1.18, 1.84)

.81 (.62, 1.08) 1.22 (.93, 1.59)

.75 (.17, 3.23) (2.53, 117.64) (.21, .76)

(1.39, 2.52) .97 (.73, 1.29) (.80, .98)

1.62 (.98, 2.69) 1.08 (.66, 1.77) .79 (.65, .95)

1.31 (.73, 2.34) 1.62 (.89, 2.96) .95 (.75, 1.20)

Bold text indicates significance at .05 level. a Note: Odds ratios are for each category compared to the next lower category, e.g. highest vs. next highest quartile of school prevalence, friends diet somewhat vs. little bit, and high vs. high-middle SES.

to UWCBs for average weight girls (OR=1.12, CI=.99–1.39).

Discussion Results from this study show that social influences— perceptions of friends’ dieting and, to a lesser extent, the prevalence of trying to lose weight throughout a school—are associated with UWCBs for a large group of adolescent girls. The relationship of the school-wide prevalence of trying to lose weight and UWCBs was maintained for average weight girls and moderately overweight girls (at a marginally significant level) after friends’ dieting was included in multivariate models. This suggests that social norms may influence UWCBs above and beyond one’s immediate circle of friends, particularly for average weight girls. These findings add some support to prior studies describing the ‘‘social contagion’’ of dieting and unhealthy weight control (Austin, 2001; Crandall, 1988), and extend previous findings to the population of high school girls. Similar work suggests that social norms of cigarette smoking are related to adolescent smoking regardless of more proximal influences (Eisenberg & Forster, 2003). School and community norms for adolescent drinking were also significantly related to the prevalence of 8th

grade drinking in a study of 20 communities in northeastern Minnesota (Roski et al., 1997). Unhealthy weight control may therefore be part of a set of highrisk behaviors for which young people may be motivated by normative behavior among their peers. This study is also consistent with previous findings that dieting behaviors of friends are associated with weight-related behaviors in individuals (Paxton et al., 1999; Pike, 1995). Paxton et al. (1999) found that girls within friendship groups tend to have similar dieting behaviors. They further found that the girls’ dieting behaviors were related to the types of conversations that the girls had about weight and dieting within their friendship groups such as making body comparisons. The present study extends previous findings by showing that friends’ weight-related attitudes and behaviors may have most relevance for girls whose weight does not deviate greatly from the norm, i.e. for average and moderately overweight girls, but is less relevant for those who are underweight or very overweight. The finding that school-wide prevalence of trying to lose weight was only marginally significant in the fully adjusted models may speak to the importance of personal characteristics (such as grade level and SES) in influencing UWCBs, or may simply result from decreased power due to additional covariates in those models. Other factors may explain the lack of significant

ARTICLE IN PRESS M.E. Eisenberg et al. / Social Science & Medicine 60 (2005) 1165–1173

findings for underweight and very overweight girls. Although UWCBs were common among overweight girls, these behaviors appeared to be unrelated to the two social influences addressed here. BMI may be the salient factor for these girls rather than friends or broader peer norms. Because overweight girls face consistent and pervasive media pressure to be thin, the specific impact of their friends may be harder to discern. Similarly, underweight girls may feel little social pressure to engage in unhealthy weight control if they perceive that they already conform to the ‘‘thin-ideal’’ of contemporary American culture (Stice, 1994).

1171

more, the measure used in this study reflects a participant’s perception of her friends’ dieting behavior, as linked data directly from friends’ was not available. It is possible, therefore, that a participant’s own weightrelated attitudes or behaviors may have colored her understanding of her friends’ dieting behavior. Finally, no information is available on students who did not participate in this study, and they may differ systematically from participants on items of interest. However, this concern is minimized due to the high response rate achieved here (81.5%). Implications

Limitations and strengths The present study had a number of strengths that contribute to our ability to draw conclusions from the findings. A large number of schools (n ¼ 29) were included in the sample; this yielded adequate variance of prevalence of girls’ trying to lose weight and permitted meaningful multilevel analysis. This research also used a large, school-based sample of ethnically and socio-economically diverse students. The findings, therefore, are generalizable to a broad population of girls. This study also used a relatively robust measure of unhealthy weight control, assessing nine distinct behaviors, which is an advantage over previous populationbased studies of youth (Grunbaum et al., 2002; Neumark-Sztainer, Story, Resnick, & Blum, 1998). Finally, to our knowledge, this is the first study to examine social influences on weight-control behaviors, stratifying by BMI category. Certain limitations must be considered in the interpretation of this research. First, the primary independent variables used in this analysis were each derived from a single survey item. Future research on social influences should employ a more comprehensive measure of social norms and friends’ dieting behaviors. Second, because the measures of school-wide dieting and friends’ dieting were quite general, they may have assessed both healthy and unhealthy weight-control strategies. Thus, the relationship between healthy dieting strategies (both school-wide and among friends) and UWCBs cannot be specifically identified here. In addition, since frequency of UWCBs was not assessed, girls may have been classified as using UWCBs even if their use was isolated and not part of their typical weight-control strategy. This may have contributed to marginal or non-significant results reported here. This study is also based on cross-sectional data; therefore, a causal relationship between social influences and UWCBs cannot be inferred. Particularly in the arena of friends’ dieting, it is possible that girls self-select into friendship groups with other girls who share pre-existing dieting behaviors, or that a girl’s unhealthy weightcontrol practices influence her friends to diet. Further-

This research suggests the possible influential role of social norms above and beyond the role of friends’ attitudes and behaviors. Future research should build on these findings by examining multiple measures of social norms and friends’ dieting, as well as additional social influences which might play a part in UWCBs, such as pressure from family members or media. Longitudinal studies are also warranted, to determine the causal nature of these associations. Austin (2001) applies Rose’s (1992) prevention paradigm to UWCBs by pointing out that the largest number of cases of unhealthy weight control arise from dieting among normal weight girls (those in the middle of the BMI curve), rather than among overweight girls (those typically considered at high risk). Thus, a reduction in the prevalence (a ‘‘shift in the curve’’, to use Rose’s language) of dieting behavior among normal weight girls would result in a reduction in cases of UWCBs. In keeping with this framework, our findings suggest that interventions aimed at changing school norms of dieting may have the potential for influencing UWCBs among individual girls. School-based programs have the advantage of involving a large number of young people, which is an essential step toward changing social norms in this population. Although healthy weight gain prevention methods are appropriate for weight management, school administrators, teachers, school nurses, and other personnel should work toward a decreased emphasis on the ‘‘thin-ideal’’ and dieting as a normative behavior among girls of a normal, healthy weight. Educational programs which teach skills to critically evaluate cultural messages about femininity and beauty may be a useful in reducing body dissatisfaction, dieting, and UWCBs. Such programs should also include an emphasis on healthy eating and physical activity for overall health. As dieting behavior and body dissatisfaction have been noted in girls in elementary school and preschool (Davison & Birch, 2001; Davison, Markey, & Birch, 2000; Robinson, Chang, Haydel, & Killen, 2001), initiating such programs at a young age may be advisable in order to prevent the development of UWCBs in adolescence.

ARTICLE IN PRESS 1172

M.E. Eisenberg et al. / Social Science & Medicine 60 (2005) 1165–1173

The development of curricula, policies, and programs aimed at shifting the social norm away from unnecessary dieting is warranted. Further research is needed to test these components in a variety of school settings in order to identify successful strategies for changing social norms in a school context.

Acknowledgements This study was supported by grant #MCJ-270834 (D. Neumark-Sztainer, principal investigator) from the Maternal and Child Health Bureau (Title V, Social Security Act), Health Resources and Service Administration, US Department of Health and Human Services

References Austin, S. B. (2001). Population-based prevention of eating disorders: an application of the rose prevention model. Preventive Medicine, 32, 268–283. Breiman, L., Friedman, J., Olshen, R., & Stone, C. (1984). Classification and regression trees. Belmont, CA: Wadsworth International Group. Centers for Disease Control and Prevention. (2000). CDC surveillance summaries. MMWR, 49, 1–5. http:// www.cdc.gov/mmwr/preview/mmwrhtml/ss4905a1.htm. Conner, M., & Armitage, C. J. (2002). The social psychology of food. Philadelphia: Open University Press. Crandall, C. S. (1988). Social contagion of binge eating. Journal of Personality and Social Psychology, 55(4), 588–598. Davison, K. K., & Birch, L. L. (2001). Weight status, parent reaction, and self-concept in five-year-old girls. Pediatrics, 107(1), 46–53. Davison, K. K., Markey, C. N., & Birch, L. L. (2000). Etiology of body dissatisfaction and weight concerns among 5-yearold girls. Appetite, 35(2), 143–151. Desmond, S. M., Prince, J. H., Gray, N., & O’Connell, J. K. (1986). The etiology of adolescents’ perceptions of their weight. Journal of Youth and Adolescence, 15(6), 461–474. Drewnowski, A., Yee, D. K., Kurth, C. L., & Krahn, D. D. (1994). Eating pathology and DSM-III-R bulimia nervosa: a continuum of behavior. American Journal of Psychiatry, 151(8), 1217–1219. Eisenberg, M. E., & Forster, J. L. (2003). Adolescent smoking behavior: measures of social norms. American Journal of Preventive Medicine, 25(2), 122–128. Field, A. E., Camargo, C. A., Jr., Taylor, C. B., Berkey, C. S., & Colditz, G. A. (1999). Relation of peer and media influences to the development of purging behaviors among preadolescent and adolescent girls. Archives of Pediatrics and Adolescent Medicine, 153(11), 1184–1189. 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(1), 54–60.

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(3), E36. French, S. A., & Jeffery, R. W. (1994). Consequences of dieting to lose weight: effects on physical and mental health. Health Psychology, 13(3), 195–212. Gibbs, R. E. (1986). Social factors in exaggerated eating behavior among high school students. International Journal of Eating Disorders, 15(6), 1103–1107. Grunbaum, J. A., Kann, L., Kinchen, S. A., Williams, B., Ross, J. G., Lowry, R., & Kolbe, L. (2002). Youth risk behavior surveillance—United States, 2001. MMWR Surveillance Summaries, 51(SS04), 1–64. Herzog, D. B., & Copeland, P. M. (1985). Eating disorders. New England Journal of Medicine, 313, 295–303. Ho, P. C., Dweik, R., & Cohen, M. C. (1998). Rapidly reversible cardiomyopathy association with chronic ipecac ingestion. Clinical Cardiology, 21(10), 780–783. Huon, G. F. (1994). Dieting, binge eating, and some of their correlates among secondary school girls. International Journal of Eating Disorders, 15(2), 159–164. Huon, G., Lim, J., & Gunewardene, A. (2000). Social influences and female adolescent dieting. Journal of Adolescence, 23, 229–232. Huon, G. F., & Walton, C. J. (2000). Initiation of dieting among adolescent females. International Journal of Eating Disorders, 18, 226–230. Johnson, J., & Whitaker, A. H. (1992). Adolescent smoking, weight changes, and binge-purge behavior: associations with secondary amenorrhea. American Journal of Public Health, 82(1), 47–54. Kuczmarski, R. J., Ogden, C. D., Grummer-Strawn, L. M., Flegal, K. M., Guo, S. S., Wei, R., Mei, Z., Curtin, L. R., Roche, A. F., & Johnson, C. L. (2000). CDC growth charts: United States. Advance Data, 314, 1–27. Levine, M. P., Smolak, L., & Hayden, H. (1994). The relation of sociocultural factors to eating attitudes and behaviors among middle school girls. Journal of Early Adolescence, 14(4), 471–490. Lo, C. C., & Globetti, G. (1995). The facilitating and enhancing roles Greek associations play in college drinking. International Journal of the Addictions, 30(10), 1311–1322. Neumark-Sztainer, D., Butler, R., & Palti, H. (1995). Eating disturbances among adolescent girls: evaluation of a schoolbased primary prevention program. Journal of Nutrition Education, 27(1), 24. Neumark-Sztainer, D., Falkner, N., Story, M., Perry, C., Hannan, P. J., & Mulert, S. (2001). Weight-teasing among adolescents: correlations with weight-status and disordered eating behaviors. International Journal of Obesity, 26, 123–131. Neumark-Sztainer, D., Story, M., Dixon, L. B., Resnick, M., & Blum, R. (1997). Correlates of inadequate consumption of dairy products among adolescents. Journal of Nutrition Education, 19, 12–20. Neumark-Sztainer, D., Story, M., Hannan, P. J., Perry, C. L., & Irving, L. M. (2002). Weight-related concerns and behaviors among overweight and nonoverweight adolescents: implications for preventing weight-related disorders.

ARTICLE IN PRESS M.E. Eisenberg et al. / Social Science & Medicine 60 (2005) 1165–1173 Archives of Pediatrics and Adolescent Medicine, 156, 171–178. Neumark-Sztainer, D., Story, M., Hannan, P., & Moe, J. (2002). Overweight status and eating patterns among adolescents: where do youth stand in comparison to the Healthy People 2010 Objectives? American Journal of Public Health, 92(5), 844–851. Neumark-Sztainer, D., Story, M., Resnick, M., & Blum, R. (1998). Lessons learned about adolescent nutrition from the Minnesota Adolescent Health Survey. Journal of the American Dietetic Association, 98, 1449–1456. Nichter, M. (2000). Fat talk: what girls and their parents say about dieting. Cambridge, MA: Harvard University Press. Patton, G. C., Johnson-Sabine, E., Wood, K., Mann, A. H., & Wakeling, A. (1990). Abnormal eating attitudes in London school girls—a prospective epidemiologic study: outcome at twelve month follow-up. Psychological Medicine, 20, 383–394. Patton, G. C., Selzer, R., Coffey, C., Carlin, J. B., & Wolfe, R. (1999). Onset of adolescent eating disorders: population based cohort study over 3 years. British Medical Journal, 318(7186), 765–768. Paxton, S. J., Schutz, H. K., Wertheim, E. H., & Muir, S. L. (1999). Friendship clique and peer influences on body image concerns, dietary restraint, extreme weight-loss behaviors and binge eating in adolescent girls. Journal of Abnormal Psychology, 108(2), 255–266. Pike, K. M. (1995). Bulimic symptomatology in high school girls. Psychology of Women Quarterly, 19, 373–396. Robinson, T. N., Chang, J. Y., Haydel, K. F., & Killen, J. D. (2001). Overweight concerns and body dissatisfaction among third grade children: the impacts of ethnicity and socioeconomic status. The Journal of Pediatrics, 138(2), 181–187. Rose, G. (1992). The strategy of preventive medicine. New York: Oxford University Press.

1173

Roski, J., Perry, C. L., McGovern, P. G., Williams, C. L., Farbakhsh, K., & Veblen-Mortenson, S. (1997). School and community influences on adolescent alcohol and drug use. Health Education Research: Theory & Practice, 12(2), 255–266. Stice, E. (1994). Review of the evidence for a sociocultural model of Bulimia Nervosa and exploration of the mechanisms of action. Clinical Psychology Review, 14(7), 633–661. Stice, E., Cameron, R. P., Killen, J. D., Hayward, C., & Taylor, C. B. (1999). Naturalistic weight-reduction efforts prospectively predict growth in relative weight and onset of obesity among female adolescents. Journal of Consulting and Clinical Psychology, 67(6), 967–974. Stice, E., Hayward, C., Cameron, R. P., Killen, J. D., & Taylor, C. B. (2000). Body-image and eating disturbances predict onset of depression among female adolescents: a longitudinal study. Journal of Abnormal Psychology, 109(3), 438–444. Story, M., Neumark-Sztainer, D., & French, S. (2002). Individual and environmental influences on adolescent eating behaviors. Journal of the American Dietetic Association, 102(Suppl. 3), S40–S51. Strauss, R. S., & Pollack, H. A. (2003). Social marginalization of overweight children. Archives of Pediatrics and Adolescent Medicine, 157, 746–752. Turner, S. L., Hamilton, H., Jacobs, M., Angood, L. M., & Dwyer, D. H. (1997). The influence of fashion magazines on the body image satisfaction of college women: an exploratory analysis. Adolescence, 32(127), 603–614. Wechsler, H., Kuh, G., & Davenport, A. E. (1996). Fraternities, sororities and binge drinking: results from a national study of American colleges. National Association Student Personnel Administrators, 33(4), 260–279. Wertheim, E. H., Paxton, S. J., Schutz, H. K., & Muir, S. L. (1997). Why do adolescent girls watch their weight? An interview study examining sociocultural pressures to be thin. Journal of Psychosomatic Research, 42(4), 345–355.