self-esteem and body esteem: effects of gender, age, and weight

26
JOURNAL OF APPLIED DEVELOPMENTAL PSYCHOLOGY 17,321-346 (1996) Self-Esteem and Body Esteem: Effects of Gender, Age, and Weight BEVERLEY K. MENDELSON DONNA R. WHITE Concordia University MORTON J. MENDELSON McGill University Three studies explored how youngsters’ self-esteem and body esteem may be related to age, gender, and relative weight and to each other. Study 1 involved 379 8- to 12- vear-olds and Study 2. 85 13- to 15year-olds; Study 3 involved two samples (76 8- ii lo-year-olds; 85 11: to 13-year-aids) tested twice over 2 years. Self-&teem was assessed with Hatie& (1985b, 1988) Self-Perception Profile, which taps global self- worth and self-evaluations in specific domains. Body esteem was assessed with the Revised Body-Esteem Scale (Mendelson &White, 1993), which yielded two measures fBE-Aooearance and BE-Weight). Generally, self-esteem was not related to relative weight, but was associated with feelings ‘ibout one’s appearance. Body-esteem measures were inversely related to relative weight (although only BE-Weight was uniquely so) and were also associated with global self-worth. Self-esteem and body esteem were stable over 2 years. Overweight youngsters have been characterized as unhappy and maladjusted (Bruch, 1941). Their unhappiness is thought to be related to low self-esteem and poor body esteem. Although self-esteem and body esteem are terms used often- and sometimes interchangeably-to describe self-evaluations and feelings about the self, the constructs pose several measurement problems. Researchers initially considered self-esteem as either positive or negative self-evaluations across a variety of domains (e.g., Coopersmith, 1967; Piers & Harris, 1969) or as a global construct, independent of domain (Rosenberg, 1979). However, conceiv- Aspects of this research were part of a poster presented at the Canadian Psychological Association, Quebec City, Quebec, June 1992. We heartily thank the administrators, teachers, and students of the Baldwin-Cartier School Board for making this project possible, Heidi Zackon and Annick Bucholz for assistance with data collection, Rhonda Amsel for help in shaping this article, and Evelyn Schliecker and Annick Bucholz for comments on a draft of the article. The research was supported by Les Fonds pour la Formation des Chercheurs et Aide &la Recherche, Ministry of Education, Quebec. The Body-Esteem Scale for Adolescents and Adults (Mendelson, White, & Balfour, 1995). with expanded Weight and Attribution factors, is currently being tested. Correspondence and requests for reprints should be sent to Beverley K. Mendelson, Vanier College. 821 Blvd. Ste. Croix, Ville St. Laurent, QuCbec, Canada H4L 3X9. E-mail: <mmendel- [email protected]>. 321

Upload: beverley-k-mendelson

Post on 16-Sep-2016

215 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Self-esteem and body esteem: Effects of gender, age, and weight

JOURNAL OF APPLIED DEVELOPMENTAL PSYCHOLOGY 17,321-346 (1996)

Self-Esteem and Body Esteem: Effects of Gender, Age, and Weight

BEVERLEY K. MENDELSON

DONNA R. WHITE

Concordia University

MORTON J. MENDELSON

McGill University

Three studies explored how youngsters’ self-esteem and body esteem may be related to age, gender, and relative weight and to each other. Study 1 involved 379 8- to 12- vear-olds and Study 2. 85 13- to 15year-olds; Study 3 involved two samples (76 8- ii lo-year-olds; 85 11: to 13-year-aids) tested twice over 2 years. Self-&teem was assessed with Hatie& (1985b, 1988) Self-Perception Profile, which taps global self- worth and self-evaluations in specific domains. Body esteem was assessed with the

Revised Body-Esteem Scale (Mendelson &White, 1993), which yielded two measures fBE-Aooearance and BE-Weight). Generally, self-esteem was not related to relative weight, but was associated with feelings ‘ibout one’s appearance. Body-esteem measures were inversely related to relative weight (although only BE-Weight was uniquely so) and were also associated with global self-worth. Self-esteem and body esteem were stable over 2 years.

Overweight youngsters have been characterized as unhappy and maladjusted (Bruch, 1941). Their unhappiness is thought to be related to low self-esteem and poor body esteem. Although self-esteem and body esteem are terms used often- and sometimes interchangeably-to describe self-evaluations and feelings about the self, the constructs pose several measurement problems. Researchers initially considered self-esteem as either positive or negative self-evaluations across a variety of domains (e.g., Coopersmith, 1967; Piers & Harris, 1969) or as a global construct, independent of domain (Rosenberg, 1979). However, conceiv-

Aspects of this research were part of a poster presented at the Canadian Psychological Association,

Quebec City, Quebec, June 1992. We heartily thank the administrators, teachers, and students of the

Baldwin-Cartier School Board for making this project possible, Heidi Zackon and Annick Bucholz

for assistance with data collection, Rhonda Amsel for help in shaping this article, and Evelyn

Schliecker and Annick Bucholz for comments on a draft of the article. The research was supported by Les Fonds pour la Formation des Chercheurs et Aide & la Recherche, Ministry of Education, Quebec.

The Body-Esteem Scale for Adolescents and Adults (Mendelson, White, & Balfour, 1995). with expanded Weight and Attribution factors, is currently being tested.

Correspondence and requests for reprints should be sent to Beverley K. Mendelson, Vanier

College. 821 Blvd. Ste. Croix, Ville St. Laurent, QuCbec, Canada H4L 3X9. E-mail: <mmendel-

[email protected]>.

321

Page 2: Self-esteem and body esteem: Effects of gender, age, and weight

322 MENDELSON, WHITE, AND MENDELSON

ing self-esteem unidimensionally has not increased our understanding of the relation between self-esteem and evaluations of specific aspects of the self, like appearance. Other theories of self emphasize its multidimensionality and view body esteem as one of several important self-domains. Hatter (1982, 1983, 1985b, 1987, 1990) acknowledged that self-esteem includes a global component concerning the perception of one’s worth as a person; however, following James (1892), she suggested that it also reflects judgments of one’s competence or adequacy in particular domains-like school, the social sphere, athletics, and appearance-especially in domains in which success is valued.

The few existing studies of self-esteem and weight have relied on measures of global self-esteem (e.g., Rosenberg, 1979) or measures that combine self- evaluations across domains (e.g., Coopersmith, 1967; Piers & Harris, 1969). They either did not confirm expected weight-related deficits in self-esteem (Mendelson & White, 1985; Wadden, Foster, Brownell, &Finley, 1984; Schliecker, 1996), found only small, clinically insignificant deficits (Kaplan & Wadden, 1986), or found deficits only in extremely obese girls (Martin et al., 1988). In one study (Mendelson & White, 1985), weight-related deficits in self-esteem were modified by age and gender. Overweight adolescents were apt to have relatively low self-esteem, but overweight preadolescent children were not. Furthermore, boys and girls seemed to follow somewhat different developmental trajectories; being overweight adversely affected self-esteem during early adolescence in boys, but during later adolescence in girls. However, across-domain measures like those used in all of the studies could have masked possible weight-related differences in particular domains of self- esteem. Regardless, based on a meta-analysis, Friedman and Brownell (1995) concluded that, although self-esteem may be weakly related to obesity, the results are inconsistent across studies and additional research is necessary.

Self-evaluation of physical appearance, or body esteem, is the only specific domain of self-esteem that has been studied extensively in overweight individu- als, and the findings appear unequivocal. Overweight children, adolescents, and adults generally have lower body esteem than do their normal-weight peers (Allon, 1979; Hendry & Gillies, 1978; Mendelson & White, 1982, 1985; Stun- kard & Mendelson, 1967), and this is especially true for female individuals (Dwyer, Feldman, Seltzer, & Mayer, 1969; Gray, 1977; Mendelson & White, 1985; Miller, Coffman, & Linke, 1980). However, researchers have not yet investigated the effect of weight on other areas of self-competence identified by Hatter (1985a, 1985b, 1988). Mendelson and White (1985) hypothesized that being overweight may adversely affect self-evaluations in different domains for boys and girls; specifically, overweight boys may suffer low athletic self-esteem, and overweight girls, low self-esteem in social domains. Thus, one goal of this study was to clarify and extend existing findings by examining the influence of gender, age, and weight on global self-worth and on specific areas of self- esteem, especially in social and athletic domains.

Researchers have not yet considered the idea that body esteem may also be multidimensional. Rather, body esteem has been treated globally, whether mea-

Page 3: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 323

sured by self-report questionnaires (Gray, 1977; Hatter, 1985b, 1988; Mendelson & White, 1985), open-ended interviews (Allon, 1979), or reactions to body- related words (Secord & Jourard, 1953). Yet, there may be “domains” of body esteem. Feelings about one’s weight may be differentiated from feelings about one’s general appearance, and the embarrassment caused by social stigma at- tached to overweight (Allon, 1979) may be independent of other aspects of body esteem. There may also be important gender differences in the way individuals evaluate their bodies. Typically, women hold more stringent standards for their bodies than do men, and, if they fail to reach these high standards, their evalua- tions of their bodies might suffer.

Hatter’s ( 1985b; 1988) subscale for appearance is limited to six items on the child version of the test and to five items on the adolescent version. Consequently, the subscale may not assess an adequate range of issues related to body esteem. Moreover, weight is not mentioned on the adolescent version, and is mentioned on only one item of the child version, but that item also mentions height (i.e., “happy with their height and weight”). The Revised Body-Esteem Scale (Mendelson & White, 1993) includes a variety of items that reflect physical appearance, weight, and possibly social attributions. Thus, our goals regarding body esteem were to explore the factor structure of the Revised Body-Esteem Scale and to clarify how body esteem is related to gender, age, and weight. We wondered particularly about possible age effects on body esteem, because age contrasts failed to reach signifi- cance in a previous study (Mendelson & White, 1985).

Hatter’s (1993) formulation of self-esteem raises questions about how global self-worth is related to domain-specific dimensions. Hatter found high and robust correlations between global self-worth and self-evaluations of physical appearance across the lifespan, which suggests that global self-worth and body esteem should be related. Although body esteem may be related to factors besides relative weight, it may also incorporate a weight component. If so, however, it cannot simply be assumed that global self-worth is related to “weight-esteem.” Indeed, global self- worth may not be related to relative weight and feelings about weight as much as is often assumed. The Revised Body-Esteem Scale (Mendelson & White, 1993)-a multidimensional measure that is more extensive than Hatter’s physical appearance subscale-was used to help clarify how global self-worth may be uniquely related to different aspects of body esteem, particularly to feelings about general appearance and to feelings about weight.

Mendelson and White (1985) found a relation between body esteem and an across-domain measure of self-esteem, at least in adolescents. Moreover, their cross-sectional study indicated body-esteem deficits in overweight children as young as 8 to 10 years old, but self-esteem deficits only in overweight youngsters who had reached early adolescence. On the basis of these findings, they hypothe- sized the following time-lagged relation between body esteem and self-esteem: Obesity causes youngsters to feel bad about their weight; such feelings translate into low body esteem; in turn, low body esteem erodes self-esteem globally and in a variety of domains, including the social sphere and athletics.

Page 4: Self-esteem and body esteem: Effects of gender, age, and weight

324 MENDELSON, WHITE, AND MENDELSON

Zumpf and Hatter (1989, cited in Hatter, 1993) attempted to address the direction of influence between feelings about appearance and global self-worth. They asked 13- to 15-year-old adolescents whether one’s appearance influences one’s global self-worth or vice versa. Half the adolescents said that body esteem affected global self-worth, whereas half said the reverse. It is possible that there are individual differences in the direction of influence between self-esteem and body esteem. However, it is also possible that individuals-especially early adolescents-are not fully aware of the direction of influence and may respond based on other factors.

Thus, another goal of this study was to clarify how global self-worth and body esteem may be interrelated. Using cross-sectional data, we examined whether or not body esteem improves prediction of global self-worth, after allowing for the effects of gender, age, and weight. Similarly, we examined whether global self- worth improves prediction of aspects of body esteem, after allowing for the effects of gender, age, and weight. Moreover, longitudinal data were examined to explore the question of direction of influence.

In summary, this research asked how multidimensional constructs of self- and body esteem may be related to gender, age, and weight and how global self- worth and body esteem may be related to each other. Multidimensional aspects of self-esteem were measured with Harter’s (1985a, 1988) Self-Perception Profile. Potentially multidimensional aspects of body esteem were measured with the Revised Body-Esteem Scale (Mendelson & White, 1993). Three studies are reported: Study 1 involved a large sample of children representing a wide weight range; Study 2 involved adolescents; Study 3 examined possible longitudinal effects related to self- and body esteem.

STUDY 1: SELF-ESTEEM AND BODY ESTEEM IN SCHOOL CHILDREN

Mendelson and White (1985) did not find a significant relation in school children between weight and an across-domain measure of self-esteem. However, their sample was relatively small, which reduced the power of their statistical tests. In contrast, they did find weight-related deficits in children’s body esteem, but they did not examine possible components of body esteem. Study 1, which explored how children’s self-esteem and body esteem may be related to gender, age, and weight and to each other, extended Mendelson and White’s research in three important ways: It involved a large sample of children; it included both domain-specific and global measures of self-esteem; it examined possible components of body esteem.

METHOD

Participants The participants were drawn from two, English-speaking, publicly funded, Cath- olic elementary schools in predominantly White, middle-class suburbs of Mon-

Page 5: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 325

treal. The sample included 379 children (220 girls and 159 boys) between 8.5 and 12.5 years old (M = 10.4 years).

Each child’s relative weight was assessed (at the end of the study) with the Baldwin-Wood gender-appropriate weight-for-height-for-age norms intended for international use (Jeliffe, 1966). Relative weight was defined as 100% times actual weight divided by appropriate weight for height and age. The children ranged in weight from 74% (underweight) to 190% (overweight) (M = 106%). Relative weight was not related to gender, t(377) = 1.36, p > .05 (girls: M = 106%; SD =

15%; boys: M = 108%; SD = 17%), or to age, r(377) = .lO, p > .05.

Procedures Participants were tested individually at school for about 30 min by one of two experimenters. Although expensive, individual testing was used for three rea- sons: It ensured that all children fully understood the format and instructions of the questionnaires; it reduced possible influences (on responses or on willingness to be weighed and measured) of social comparison or monitoring that might occur in group testing; it provided complete privacy during weighing and mea- suring. The participants completed the Self-Perception Profile for Children (Hatter, 1985b) and the Revised Body-Esteem Scale (Mendelson & White, 1993).

&$-Esteem. The Self-Perception Profile for Children is a 36-item measure of self-esteem with good psychometric properties (Hatter, 1985a). The measure consists of six 6-item subscales: Global Self-Worth (SE-Global), Scholastic Competence (SE-School), Social Acceptance (SE-Social), Athletic Competence (SE-Athletic), Behavioral Conduct (SE-Conduct), and Physical Appearance (SE- Appearance, which was used here only for validation of the Revised Body- Esteem Scale). Participants first determine which of two contrasting items per- tains to them (e.g., “Some kids find it hard to make friends” vs. “Other kids find it’s pretty easy to make friends”); they then assess whether the choice is “really true” or “sort of true” for them. Thus, each item can be scored on a 4-point scale ranging from 1 (really negative) to 4 (really positive). Subscale scores were obtained by computing the mean response for the items. Cronbach alpha coeffi- cients computed for this sample indicated adequate internal consistency for all six subscales (SE-Global: .72; SE-School: .77; SE-Social: .72; SE-Athletic: .78; SE-Conduct: .73; SE-Appearance: .78).

Body Esteem. The Revised Body-Esteem Scale (Mendelson & White, 1993) was used to assess children’s attitudes and feelings about their body and appear- ance. The 20 yes-no items (Table I), which all have obvious face validity, tap evaluations of one’s appearance and body or attributions of evaluations about one’s appearance to others. The test has very good split-half reliability, r(95) =

.85, p < .Ol. A factor analysis with oblique rotation was used to determine if the Revised

Page 6: Self-esteem and body esteem: Effects of gender, age, and weight

326 MENDELSON, WHITE, AND MENDELSON

TABLE 1 Factor Loadings ( > .40) for Body-Esteem Items

Factor

1 2 3

No. Item Appearance Weight Attribution

3 16 6

15 20 11

1 17 12 14 18 8

9 10 7 5

13 2 4

19

I’m pretty happy about the way I look I often wish I looked like someone else I like what I see when I look in the mirror I’m looking as nice as I’d like to I worry about the way I look I wish I looked better I like what I look like in pictures My looks upset me I often feel ashamed of how I look I think I have a good body I’m as nice looking as most people There are lots of things I’d change about my looks if I could I’m proud of my body I really like what I weight I wish I were thinner My weight makes me happy Other people make fun of the way I look Kids my own age like my looks Most people have a nicer body than I do My parents like my looks

.67

.66

.65

.62

.61

.60

.59

.57

.56

.54

.53

.52

.50 .83 .81 .74

.82

.66

Note. Items 4 and 19 were dropped from the factor analysis (see explanation in text).

Body-Esteem Scale tapped more than one construct. Oblique rotation was used because it was assumed that aspects of body esteem would be intercorrelated. The analysis yielded a four-factor solution, but it was rejected because only one item had a loading greater than .40 on the fourth factor (“my parents like my looks”). The analysis was recomputed without the item, yielding a three-factor solution; however, one item (“most people have a nicer body than I do”) did not load greater than .40 on any factor, so it was dropped, and a recomputed analysis yielded the three-factor solution shown in Table 1, which shows the factor loadings ( > .40) for the retained items.

The first factor, BE-Appearance (general appearance and body) consisted of 13 items and accounted for 33% of the total variance. The 13 items were summed to form a BE-Appearance score, which had high internal consistency: Cron- bath’s (Y = .87; item-to-total rs(377) = .34 to .63, ps < .Ol. BE-Appearance

Page 7: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 327

was highly correlated with the SE-Appearance subscale of the Self-Perception Profile, r(377) = .73, p < .Ol, which indicates good construct validity.

The second factor, BE-Weight (satisfaction with weight) consisted of three items and accounted for 8% of the total variance. The three items were summed to form a BE-Weight score, which had adequate internal consistency: Cronbach’s (Y = .77; item-to-total rs(377) = .56 to .69, ps < .Ol . BE-Weight was correlated similarly with BE-Appearance, r(377) = .47, p < .Ol, and with SE-Appearance, r(377) = .49, p < .Ol. Although the body-esteem subscales covaried moder- ately, they were considered independently for three reasons: They represented different factors in the factor analysis; their correlation coefficient was lower than their Cronbach’s alpha coefficients; they tapped conceptually distinct constructs.

The third factor, BE-Attribution (attribution of positive evaluations to others) accounted for 7% of the total variance. However, it consisted of only two items, so BE-Attribution scores were not analyzed further.’

RESULTS

Analyses were conducted to examine how self-esteem may be related to gender, age, and relative weight, how body esteem may be associated with these predic- tors, and, finally, how body esteem and self-esteem may covary with each other.

Self-Esteem Tables 2a through 2e present means and Tables 3a through 3e present correlations with relative weight for the self-esteem subscales as a function of age and gender. The means for this sample were, at most, only marginally higher than the means obtained by Hatter (1985b) for two samples in Colorado.

To assess how children’s self-esteem may be related to gender, age, and relative weight, each self-esteem subscale was subjected to a separate hierarchi- cal multiple regression. The three main effects (gender, age, and relative weight) were entered as a set into the regression equation first, followed by the three two- way interactions (Gender X Age, Gender X Relative Weight, and Age X Rela- tive Weight). The Gender X Age X Relative Weight interaction was entered last,

‘Conducting a factor analysis of dichotomous data is admittedly problematic, even with 379

participants. Nonetheless, the two subscales that were retained for further analyses (BE-Appearance

and BE-Weight) have obvious face validity. BE-Appearance includes only items that were unrelated to weight and that tap respondents’ own evaluations (i.e., not judgments of others’ evaluations). BE-

Weight includes only the weight-related items. Moreover, support for separate factors of body esteem has subsequently been obtained with an

expanded measure, the Body-Esteem Scale for Adolescents and Adults (Mendelson, White, & Bal- four, 1995). The revised scale consists of 30 items and uses a 5-point Likert format. Factor analysis of

data from 325 late adolescents and young adults confirmed three subscales: BE-Appearance (12

items; Cronbach o = .92), BE-Weight (9 items; Cronbach OL = .93), and BE-Attribution (9 items;

Cronbach 01 = .77).

Page 8: Self-esteem and body esteem: Effects of gender, age, and weight

328 MENDELSON, WHITE, AND MENDELSON

TABLE 2 Means (and Standard Deviations) of Self- and Body-Esteem Measures

for Girls and Boys in Study 1

Age Groupa

Gender 8-9 10-11 12-13 Total

(a) SE-Global Girls Boys Total

(b) SE-School Girls Boys Total

(c) SE-Social Girls Boys Total

(d) SE-Athletic Girls Boys Total

(e) SE-Conduct Girls Boys Total

(f) BE-Appearance Girls Boys Total

(g) BE-Weight Girls Boys Total

3.4 (.6) 3.4 (.5) 3.4 (.5)

3.1 (.6) 2.9 (.7) 3.0 (.6)

3.0 (.7) 2.8 (.7) 2.9 (.7)

2.9 (.7) 3.0 (.6) ?.9 (.6)

3.3 (.5) 3.0 (.5) 3.2 (.6)

10.6 (3.5) 11.4 (1.7) 10.9 (2.9)

2.2 (1.1) 2.7 (.7) 2.4 (1.0)

3.2 (.6) 3.2 (.6) 3.2 (.6)

2.9 (.6) 2.8 (.7) 2.9 (.6)

2.8 (.7) 3.0 (.7) 2.9 (.7)

2.7 (.7) 2.9 (.8) 2.8 (.7)

3.0 (.5) 2.8 (.6) 2.9 (.6)

9.5 (3.7) 10.7 (2.9) 9.9 (3.4)

1.9 (1.2) 2.5 (.9) 2.1 (1.1)

3.3 (.6) 3.4 (.5) 3.4 (5)

3.0 i.7) 3.0 (.6) 3.0 (.6)

3.1 (.6) 3.2 (.6) 3.1 (.6)

2.8 (.7) 3.2 (.7) 3.0 (.7)

3.1 (.6) 2.9 (.6) 3.0 (.6)

8.9 (3.5) 11.2 (2.2) 10.0 (3.2)

1.7 (1.2) 2.0 (1.1) 1.9 (1.2)

3.3 (.6) 3.4 (.5) 3.3 (.6)

3.0 (.6) 2.9 (.6) 3.0 (.6)

2.9 (.7) 3.0 i.7) 3.0 (.7)

2.8 (.7) 3.0 (.7) 2.9 (.7)

3.2 (.6) 2.9 (.6) 3.1 (.6)

9.7 (3.6) 11.2 (2.2) 10.3 (3.2)

1.9 (1.2) 2.4 (.9) 2.1 (1.1)

aFor the sake of tabling the data, the sample was divided into three groups: 8- to g-year olds (M = 8.9; Range = 7.9-9.4; ngir, = 84; nboy = 58); lo- to 1 l-year olds (M = 10.4; Range = 9.5-l 1.4; ngir, = 73; nboy = 43); 12-to 13-year olds (M = 12.0; Range = 11.5-13.4; ngir, = 63; nboy = 58).

but was not significant here or in any subsequent analyses, so it will not be mentioned again.

Predictors are reported here and in subsequent analyses only if, at their step of entry, the overall F for the equation was significant (p < .0.5) and their part correlations were both significant (p < .05) and greater than .1414 (i.e., only if

Page 9: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 329

they accounted uniquely for at least 2% of the variance). There were few predic- tors of the self-esteem measures. Gender accounted for a significant amount of variance in two subscales. Boys scored higher than girls on SE-Athletic (3.0 vs.

2.8), rPati (373) = .18, p < .Ol, tending to claim more than girls that, for example, they “do very well at all kinds of sports.” However, boys scored lower on SE-Conduct (2.9 vs. 3.2), r,,,(373) = - .21, p < .Ol, which included items like “do the right thing.”

TABLE 3 Correlations Between Relative Weight and Each Self- and Body-Esteem Measure

for Girls and Boys in Study 1

Age Group”

Gender 8-9 IO-II 12-13 Total

(a) SE-Global Girls Boys Total

(b) SE-School Girls Boys Total

(c) SE-Social Girls Boys Total

(d) SE-Athletic Girls Boys Total

(e) SE-Conduct Girls Boys Total

(f) BE-Appearance Girls Boys Total

(g) BE-Weight Girls Boys Tota I

-67 .03 .I2 -.I8 .Ol -.04

-.I0 -.03 .13 .06 .04 .Ol

-.lO .I8 .05

-.08 .I1 .oo

-.08 -.ll .04 .oo

-.Ol -.07

-.13 .12

-.Ol

-.15 -.12 -.15 .13 .02 .07

-.02 -.06 -.Ol

-.ll -.I1 -.I2

-.I3 -.09 -.05 -.06 -.I0 .07

-.05 -.03 -.03

.06 -.08 -.33** -.lO

.14 .I6 .20 .16*

.I1 .oo -.08 .oo

-.37*x -.18** -.40x* -.21*x -.25** -.16**

-.I1 .I0

-.07

-.I2 -.19 -.13

-.62** -.47*x -.49** -.52** -.50** -.46**

-.50x* -.36** -&I** -.53*x - .48** -.37*x

aSee footnote in Table 2. *p < .05. **p < .Ol.

Page 10: Self-esteem and body esteem: Effects of gender, age, and weight

330 MENDELSON, WHITE, AND MENDELSON

With one exception, neither relative weight, nor interactions with relative weight accounted for a significant amount of variance in the self-esteem measures. The Gender X Relative Weight interaction was significant for SE-Conduct, r,,,,(373) = .15, p < .Ol. As can be seen from the correlations in Table 3e, self-evaluation of behavioral conduct was negatively associated with weight for girls (significantly so only for older girls), whereas it was positively associated with weight in boys (significantly so for the overall sample). Contrary to Mendelson and White’s (1985) hypothesis, overweight girls did not have deficits in the social domain (Table 3c), nor overweight boys in the athletic domain (Table 3d), at any age.

Body Esteem Tables 2f and 2g present means and Tables 3f and 3g present correlations with relative weight for the two body-esteem measures as a function of age and gender. To examine how body esteem may be related to a child’s gender, age, and relative weight, each body-esteem measure was subjected to the same hier- archical multiple regression as described earlier.

Gender and age accounted for significant amounts of variance in both body- esteem measures (Table 4). Boys scored higher than girls on BE-Appearance (11.2 vs. 9.7), tending to agree with items such as “I like what I see when I look in the mirror”; theyalsoscored higheronBE-Weight (2.4~~. 1.9), tending todisagree with items like “my weight makes me unhappy.“Olderchildren tended to have low scores on BE-Appearance-although age accounted for less than 2% of its variance-and on BE-Weight; thus older children were apt to agree with items like “I wish I looked better” and to disagree with items such as “I really like what I weigh.”

The Age X Gender interaction was significant for BE-Appearance. As can be

TABLE 4 Regression Equations for Body-Esteem Measures in Study 1 (N = 375)

Predictors

(a) BE-Appearance Gender

Age Relative Weight Gender x Age Gender x Relative Weight Age x Relative Weight

(b) BE-Weight Gender

Age Relative Weight Gender x Age Gender x Relative Weight Age x Relative Weight

r

.22 -.I5 -.I6

.20 -.20 - .46

Part,

.24** -.14** -.16**

.13**

.06 -.09

.24** -.17** -.46**

.Ol

.08

.Ol

AR2 R2

.10**

.03* .13**

.29**

.Ol .30**

*p < .05. **p < .Ol.

Page 11: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 331

seen in Table 2f, the sex difference, favoring boys, appeared greatest for 12- to 13- year-olds, t( 119) = 4.36,~ < .Ol, althoughitwasalsosignificantfor lo- to 1 l-year- olds, t( 114) = 2.02,~ < .05, andmarginallysofor S- to9-year-olds, t( 140) = 1.95, p < .06. The apparent increase in the size of the sex difference presumably reflected a significant negative correlation between age and BE-Appearance for girls, ~(2 18) ZZ -.23, p < .Ol, but not for boys, r(157) = -.03, p > .05.

Last, relative weight accounted for variance in both body-esteem measures. With increasing weight, children were apt to have lower scores on BE- Appearance (Table 3f) and, especially, BE-Weight (Table 3g). As indicated by partial correlations, BE-Weight was associated with relative weight, independent of BE-Appearance, r(376) = -.44, p < .Ol, whereas BE-Appearance was not associated with relative weight, independent of BE-Weight, r(376) = .07, p >

.05; that is, only BE-Weight was uniquely related to weight.

Links Between Self-Esteem and Body Esteem The final goal of Study 1 was to explore possible associations between children’s feelings about their bodies and their feelings about themselves. We first examined the importance of the two body-esteem scores in accounting for variance in the self-esteem measures over and above that accounted for by gender, age, relative weight, and their interactions. Each self-esteem subscale was subjected to a hierarchical multiple regression, with subsets of predictors entered in the follow- ing order: gender, age, and relative weight; three two-way interactions of gender, age, and relative weight; and the two body-esteem scores. The two-way interac- tions of the body-esteem scores with gender, age, and relative weight were entered in a last step, but they did not account for a significant amount of variance here or in any subsequent analyses, so they will not be mentioned further.

BE-Appearance accounted for variance in all the self-esteem subscales, part rs(368): SE-Global: .46, p < .Ol; SE-School: .25, p < .Ol; SE-Social: .30, p -=c

.Ol; SE-Athletic: .19, p < .Ol; SE-Conduct: .18, p < .Ol. Children who felt good about their appearance also tended to evaluate highly their global self-worth and their competence in other domains. Interestingly, feelings about one’s weight, or interactions with feelings about weight, did not further improve pre- diction of global self-worth or self-esteem in any domain.

The next analyses determined if variance in body esteem-over and above that accounted for by gender, age, relative weight, and their interactions-may be accounted for by variance in global self-worth or by variance in self- evaluations in specific domains. The body-esteem subscales were separately subjected to hierarchical multiple regressions paralleling those just described. SE-Global accounted for a significant amount of variance both in BE- Appearance, r,,,(367) = .35, p < .Ol, and in BE-Weight, r,,(367) = .16, p <

.Ol . Independent of other predictors, children with high global self-worth tended to feel good particularly about their appearance but also about their weight. However, feelings of competence in other domains did not further improve prediction of either body-esteem measure.

Page 12: Self-esteem and body esteem: Effects of gender, age, and weight

332 MENDELSON, WHITE, AND MENDELSON

Summary

Self-Esteem. SE-Athletic was higher for boys than for girls, but SE-Conduct was lower for boys. In addition, SE-Conduct was associated with weight pos- itively for boys, but negatively for girls, particularly those between 12 and 13 years old. How children evaluated their global self-worth and their competence in four domains was also associated with how they felt about their appearance, but not with how they felt about their weight.

Body Esteem. Boys had higher body esteem than did girls, but the difference for BE-Appearance increased with age, due to an age-related decline in girls’ feelings about their appearance. There was also an age-related decline in BE- Weight for both boys and girls. Heavier children were apt to have lower scores on both body-esteem measures, but especially, and uniquely, on BE-Weight. How children felt about their appearance and, less so, how they felt about their weight, was also related to how they evaluated their global self-worth.

The data from this sample do not support clinical intuitions that relative weight affects body esteem, which, in turn, affects self-esteem. Only one mea- sure of self-esteem, SE-Conduct, covaried with relative weight, and in that case, the relation was different for boys and girls. Moreover, independent of other predictors, BE-Appearance, not BE-Weight, predicted self-esteem. Finally, al- though SE-Global predicted BE-Weight, it did so independent of weight.

STUDY 2: SELF-ESTEEM AND BODY ESTEEM IN ADOLESCENTS

Study 1 suggested that self-esteem is unrelated either to weight or to feelings about weight, at least in children younger than 13 years old. However, the finding for SE-Conduct hinted that weight may affect self-esteem at older ages, at least for girls, which, along with previous research (Mendelson & White, 1985), raises the possibility that self-esteem and weight may be related in adoles- cents older than those in Study 1. Study 1 did not support Mendelson and White’s (1985) hypothesis that overweight boys have low athletic self-esteem and over- weight girls, low self-evaluation in the social domain; yet the question remains whether such deficits may occur in older youngsters. Therefore, the goal of Study 2 was to address our questions concerning self-esteem, body esteem, gender, age, and weight in a sample of adolescents.

METHOD

Participants The participants were drawn from two, English-speaking, publicly funded, Catho- lic high schools that were fed by the elementary schools of participants in Study 1.

Page 13: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 333

The sample included 85 youngsters (48 girls and 37 boys) from 13.3 to 15.4 years old (M = 14.0), who had participated 2 years earlier in Study 1. In one of the high schools, it was also possible to test 3 1 youngsters (23 girls and 8 boys) between 14.3 and 15.5 years (M = 14.8) who participated for the first time. There were proportionally somewhat more girls among new participants, but the difference was not significant x2(1, N = 116) = 2.30, p > .05. The retested and new participants did differ in age, t( 114) = 9.76, p < .Ol. Even so, the subsamples were combined, yielding a sample of 116 adolescents (7 1 girls and 45 boys) from 13.3 to 15.5 years old (M = 14.2 years).

Each youngster’s relative weight was assessed, as in Study 1, at the end of data collection. The children ranged in weight from 78% to 188% (M = 108%). Relative weight did not differ significantly between girls (M = 106%, SD = 20%) and boys (M = ill%, SD = 20%), t(l14) = 1.48, p > .05, did not differ for retested (M = 108%, SD = 21%) and new (M = 107%, SD = 18%) participants, t(ll4)= -.23,p>.05,andwasnotassociatedwithage,r(ll4)= .06,p>.O5.

Procedures Participants were seen individually, either at school or at home, for about 30 min by one of two experimenters. Individual testing was used for the reasons previ- ously noted and to maintain comparability both between studies and, within Study 2, between participants tested at home and those tested at school. The participants completed the Self-Perception Profile for Adolescents (Hatter, 1985b) and the Revised Body-Esteem Scale (Mendelson & White, 1993).

SepEsteem. The Self-Perception Profile for Adolescents is a 45-item measure of self-esteem with good psychometric properties (Hatter, 1988). The measure consists of nine 5-item subscales: Global Self-Worth (SE-Global), Scholastic Competence (SE-School), Social Acceptance (SE-Social), Close Friendship (SE- Friendship), Romantic Appeal (SE-Romantic), Conduct/Morality (SE-Conduct), Athletic Competence (SE-Athletic), Job Competence (SE-Job), and Physical Ap- pearance (SE-Appearance, which again was used only for validation of the Revised Body-Esteem Scale). The questionnaire uses the same two-option, four- choice format described above. Subscale scores were obtained by computing the mean response for the items. Cronbach alpha coefficients were computed for the present sample. There was poor internal consistency for SE-Job (.41), which was dropped from further analyses. There was marginal internal consistency for SE- Romantic (.65) and SE-Conduct (.68, when one item was dropped). The internal consistency was adequate to very good for the remaining six subscales (SE- Global: .84; SE-School: .83; SE-Social: .81; SE-Friendship: .72; SE-Athletic: .90; SE-Appearance: .86).

Body Esteem. The Revised Body-Esteem Scale (Mendelson & White, 1993) was used to assess adolescents’ feelings and attitudes toward their appearance.

Page 14: Self-esteem and body esteem: Effects of gender, age, and weight

334 MENDELSON, WHITE, AND MENDELSON

Because the sample was not large enough to support a factor analysis, BE- Appearance and BE-Weight were computed as in Study 1. BE-Appearance had high internal consistency (Cronbach’s (Y = .85; item-to-total rs(l14) = .31 to .73, ps < .Ol); it was highly correlated with SE-Appearance, r(l14) = .65, p < .Ol, providing further evidence of construct validity. BE-Weight had adequate internal consistency (Cronbach’s (Y = .74; item-to-total TS( 114) = .5 1 to .65, ps < .Ol); it was correlated similarly with BE-Appearance, r(l14) = .52, p < .Ol, and with SE-Appearance, r(l16) = .57, p < .Ol.

RESULTS

The first analyses examined if retested participants differed significantly-in ways other than age-from those who participated for the first time. Analyses were then undertaken to address how self-esteem may be related to gender, age, and relative weight, how body esteem may be related to these predictors, and how body esteem and self-esteem may covary with each other.

Retested Versus Newly Tested Participants Initial analyses were conducted to verify that the retested and new participants scored similarly on the self-esteem and body-esteem measures (Table 5), which would justify combining their data in a single sample. A MANCOVA was ap- plied separately to the seven self-esteem measures and to the two body-esteem measures; gender and group (retested and new) were between-subjects factors, and relative weight was the covariate. Relevant here, neither group nor the Group X Gender interaction accounted for significant amounts of variance multi- variately or univariately (ps > .05). Therefore, subsequent analyses were con- ducted without distinguishing between retested and new participants.

TABLE 5 Means (and Standard Deviations) of Predictor and Criterion Variables for Retested

(n = 81) and New (n = 35) Participants in Study 2

Group

Variable

SE-Global SE-School SE-Social SE-Friendship SE-Romantic SE-Athletic SE-Conduct BE-Appearance BE-Weight

Retested New F(l, 111)

3.2 (.6) 3.1 (.7) 3.3 (.5) 3.4 (.5) 2.5 (.5) 2.9 (.6) 3.0 (.6) 8.9 (3.3) 1.8 (1.3)

3.0 (.6) 1.37

3.1 (.6) .05

3.1 (.6) 1.73

3.5 (.6) .I0

2.3 (.5) .92 2.7 (.6) .I7

2.9 (.4) 1.64 7.6 (3.3) 1.55 1.7 (1.3) .07

Page 15: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 335

Self-Esteem The next analyses assessed how adolescents’ self-esteem may be related to gen- der, age, and relative weight. Each self-esteem subscale was subjected to the hierarchical multiple regression described in the parallel section of Study 1. Again, there were very few predictors of self-esteem. Boys scored higher than girls on both SE-Athletic (3.1 vs. 2.7), r,,,(112) = .30, p < .Ol, and SE- Romantic (2.7 vs. 2.4), r,,,(112) = .29, p < .Ol. Thus, boys tended to rate themselves higher than did girls on items like “do very well at all kinds of sports” and “people their own age will be romantically attracted to them.”

With only one exception, relative weight and interactions with relative weight did not account for variance in the self-esteem measures. The Age X Relative Weight interaction was, however, significant for SE-Global, r,,,.J 109) = - .19. Global self-worth and relative weight were positively, but not significantly, corre- latedforadolescentsyoungerthan themedianage( 14.2years), r(57) = .17,p > .05, but negatively correlated for those older than the median age, r(55) = - .28, p < .Ol . Contrary to Mendelson and White’s (1985) hypothesis, overweight adolescent girls did not have deficits in social domains (SE-Social, SE-Romantic, or SE- Friend) nor overweight adolescent boys in the athletic domain (SE-Athletic).

Body Esteem To examine how adolescents’ body esteem may be related to gender, age, and relative weight, each body-esteem measure was subjected to the same hierarchi- cal multiple regression. Boys scored higher than did girls on BE-Appearance (9.8 vs. 7.8), rpan( 112) = .30, p < .Ol, and on BE-Weight (2.1 vs. 1.6) rpert( 112) = .23, p < .Ol . Older adolescents tended to evaluate their appearance more negatively than did younger adolescents, r,,,(ll2) = -.24, p < .Ol. Finally, with increasing weight, adolescents were apt to have lower scores on BE-Appearance, rpart( 112) = - .3 1, p < .Ol, and even more so, on BE-Weight, r,,,(l12) = - .46, p < .Ol. Again, BE-Weight was associated with relative weight, independent of BE-Appearance, Ypartial( 113) = - .44, p < .Ol, but BE- Appearance was not associated with relative weight, independent of BE-Weight, Ye.,.& 113) = -05, p > .05; that is, only BE-Weight was uniquely related to weight.

Links Between Self-Esteem and Body Esteem The next analyses addressed the degree to which the two body-esteem scores accounted for variance in self-esteem over and above that accounted for by gender, age, relative weight, and their interactions. Each self-esteem subscale was subjected to the same hierarchical multiple regression described in the paral- lel section of Study 1.

BE-Appearance accounted for variance in three self-esteem subscales, part rs(108): SE = Global: .52, p < .Ol; SE-Social: .19, p < .05; and SE-Athletic: .25, p < .O 1. Adolescents who felt good about their appearance also tended to

Page 16: Self-esteem and body esteem: Effects of gender, age, and weight

336 MENDELSON, WHITE, AND MENDELSON

evaluate highly their global self-worth and their competence in social and athletic domains. In addition, BE-Weight accounted for variance in SE-Romantic, r,,,,(107) = .18, p < .05; that is, adolescents’ who felt good about their weight also highly evaluated their own romantic appeal. However, like the results ob- tained with children, feelings about weight, did not predict global self-worth or self-esteem in other domains.

To examine the importance of self-esteem in accounting for variance in body esteem, each body-esteem subscale was subjected to a hierarchical multiple regression with subsets of predictor variables entered in the following order: gender, age, and relative weight; three two-way interactions of sex, age, and relative weight; the eight self-esteem scores. Given the number of participants, the two-way interaction terms involving the self-esteem subscales and gender, age, or relative weight, were not entered into the equations.

Beyond the effects of gender, age, relative weight, and their interactions already discussed, SE-Global accounted for a significant amount of variance in BE-Appearance, r,,,(102) = .42, p < .Ol. Over and above other predictors, adolescents with high global self-worth tended to feel good about their appear- ance. Moreover, SE-Romantic accounted for a significant amount of variance in BE-Weight, r,,,,.J 102) = .20, p < .Ol. Over and above other predictors, adoles- cents with a high evaluation of their romantic appeal tended to feel good about their weight.

Summary

Self-Esteem. As in Study 1, boys evaluated their athletic competence higher than did girls. Interestingly, they also evaluated their romantic appeal higher. Also as in Study 1, adolescents’ evaluations of their global self-worth and of their competence in scholastic, social, and athletic domains, were associated with positive evaluations of their appearance. However, unlike the case with children, adolescents’ evaluations of their conduct were unrelated to body esteem.

Also corroborating the results in Study 1, adolescents’ self-esteem was gener- ally unrelated to their weight or to their feelings about their weight. However, global self-worth and weight covaried negatively for adolescents between 14.2 and 15.5 years old, a finding that partially replicates results obtained in early adolescent boys and in adolescent girls (Mendelson & White, 1985), but not in a large group of 15- to 17-year-olds (Schliecker, 1996).

Body Esteem. Again, boys had higher body esteem than did girls. With age, both boys and girls tended to evaluate their appearance, but not their weight, more poorly. Heavier adolescents were apt to have lower opinions about their appearance and, especially, about their weight. Independent of these effects, adolescents with high global self-worth were apt to feel good about their appear- ance, and those who felt good about their romantic appeal were apt to regard their weight positively.

Page 17: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 337

STUDY 3: SELF-ESTEEM AND BODY ESTEEM: LONGITUDINAL EFFECTS

Studies 1 and 2 established that body esteem was concurrently associated with relative weight in children and adolescents. There was some indication as well, that self-esteem may be related to weight or feelings about weight, at least in older children and adolescents. However, being overweight may have a delayed influence on self-esteem (Mendelson & White, 1985), which would only be apparent in a longitudinal study. The goal of Study 3 was to ascertain if self- esteem was predicted by relative weight or body esteem 2 years earlier. In addition, the study examined how body esteem may be influenced by weight and self-esteem 2 years earlier.

METHOD

Participants

Sample I: Child-Child. The first sample included 76 children in Study 1 (46 girls and 30 boys), who participated when they were 8.2 and 10.2 years old (A4 = 8.9) and who were retested approximately 2 years later when they were between 10.2 and 11.9 years old (M = 10.8). The children’s relative weight ranged from 83% to 158% (M = 107%) at Time 1 and from 82% to 156% (M = 108%) at Time 2; children’s relative weights were highly correlated across 2 years, r(74) = .89, p < .Ol.

Sample 2: Child-Adolescent. The second sample included the 85 youngsters (48 girls and 37 boys) who participated in Study 1, when they were between 11.2 and 13.0 years old (M = 11.9), and, 2 years later, in Study 2, when they were between 13.3 and 15.4 years old (M = 14.0). The children’s relative weights ranged from 83% to 190% (M = 108%) at Time 1 and from 78% to 188% (A4 = 108%) at Time 2; the youngsters’ relative weights were also highly correlated across 2 years, r(83) = .88, p < .Ol.

Procedures and Measures Sample 1 received the procedures and measures from Study 1 both times. Sample 2 received the procedures and measures from Study 1 at Time 1 and those from Study 2 at Time 2.

RESULTS

The first analyses examined test-retest correlations and differences in the crite- rion variables. Subsequent analyses examined if self-esteem at retest was related to weight and body esteem 2 years earlier and if body esteem at retest was related to weight and self-esteem 2 years earlier.

Page 18: Self-esteem and body esteem: Effects of gender, age, and weight

338 MENDELSON, WHITE, AND MENDELSON

Longitudinal Comparisons MANCOVAs were applied to the data-one MANCOVA for five self-esteem measures and one for two body-esteem measures-separately for Sample 1 and Sample 2. Gender and time (Time 1 and Time 2) were used as between-subjects factors, and relative weight was the covariate. Table 6 presents the observed means at Time 1 and Time 2 for the self-esteem and body-esteem scores, the results of univariate F tests for time, and the correlations across times. Relevant in this section, the analyses yielded a significant multivariate effect of time for the self-esteem data for Sample 1, F(5,69) = 4.43, p < .Ol, and Sample 2, F(5, 78) = 3.11, p < .05, and for the body-esteem data for Sample 2, F(2, 8 1) = 3.33, p < .05. SE-Social increased over the 2 years of childhood (Table 6a) and over 2 years from childhood to adolescence (Table 6b). That is, children’s feel- ings about their social competence seemed to improve over time. The only other change during childhood involved SE-Conduct (Table 6a); children’s evaluations of their behavioral conduct declined over 2 years. The only other change from childhood to adolescence involved BE-Appearance (Table 6b); youngster’s eval- uations of their appearance declined over that 2-year period. With one exception, the correlations for the self-esteem and body-esteem scores indicated moderate stability across 2 years during childhood and from childhood to adolescence. The

TABLE 6 Means (and Standard Deviations) at Each Testing Time for Study 3

Testing

Variable Time 1 Time 2 F r

(a) Sample 1: Child-Child (N = 76) SE-Global 3.5 (.5) 3.4 (4) .48 .33** SE-School 3.1 i.7) 2.9 (.7) 2.77 .54** SE-Social 3.0 (.7) 3.2 (.6) 9.65** .27** SE-Athletic 3.0 l.7) 2.9 l.7) .12 .39** SE-Conduct 3.2 (.6) 3.1 (.6) 4.18* .63** BE-Appearance 11.4 (2.3) 11.0 (2.7) 1.53 .34** BE-Weight 2.3 (1.0) 2.2 (1.2) 3.84 .50**

(b) Sample 2: Child-Adolescent (N = 85) SE-Global 3.3 (5) 3.2 (6) 1.71 &I** SE-School 3.0 (.6) 3.1 (.7) 2.93 .55** SE-Social 3.1 (.5) 3.3 (5) 5.25* .28** SE-Athletic 2.9 (.7) 2.9 (.8) .14 .63** SE-Conduct 3.0 (.5) 3.0 (.6) .19 .21 BE-Appearance 9.8 (3.1) 8.9 (3.3) 6.73* .54** BE-Weiaht 1.9 (1.1) 1.8 (1.3) .40 .57**

Notes. Fs refer to tests of differences between testing times; rs refer to the cor- relations across testing times.

*p < .05. **p < .Ol.

Page 19: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 339

TABLE 7 Partial Correlations Between Time-2 Criterion Variables and Time-l Predictors

Variables, Controlling for Time-2 Predictors

Time-l Predictor (Controlling for Time-2 Predictor)

Relative Weight BE-Appearance BE-Weight SE-Global

(a) Sample 1: Child-Child (N = 76) SE-Global .06 .19 -.I8 SE-School -.06 .06 -.07 SE-Social .09 .oo -.06 SE-Athletic .09 .07 -.oo SE-Conduct .02 .09 -.I3 BE-Appearance -.31*x -.I0 BE-Weight -.20 -.I8

(b) Sample 2: Child-Adolescent (N = 65) SE-Global -.22* .23* .32** SE-School -.03 -.02 .23* SE-Social -.09 -.ll -.04 SE-Athletic .02 .06 .08 SE-Conduct -.14 -.07 .02 SE-Romantic -.02 .I0 -.05 SE-Friend -.14 -.18 -.12 BE-Appearance .oo .03 BE-Weight -.Ol -.12

exception involved SE-Conduct for Sample 2, which is not surprising given that SE-Conduct was the behavioral conduct subscale (Harter 1985b) for children, but the conduct/morality subscale (Harter, 1988) for adolescents.

Longitudinal Prediction of Body Esteem and Self-Esteem The next analyses examined if self-esteem at Time 2 was predicted by weight or body esteem at Time 1 and, similarly, if body esteem at Time 2 was predicted by weight and self-esteem at Time 1. Table 7 shows the partial correlations between Time-2 criterion variables and Time- 1 predictors, controlling for Time-2 predic- tors (e.g., between SE-Global at Time 2 with relative weight at Time 1, control- ling for relative weight at Time 2). In such analyses, it is essential to partial out the Time-2 predictor to eliminate the possibility that an apparent prediction across time is due to a concurrent correlation and stability in the predictor. In general, Time-l predictors did not account for variance in Time-2 criterion variables for either sample.

For Sample 1 (Table 7a), BE-Appearance correlated with Time-l relative weight. Interestingly, Time-2 BE-Appearance covaried more with Time-l rela- tive weight, r(74) = -.38, p < .Ol, than with Time-2 relative weight, r(74) =

Page 20: Self-esteem and body esteem: Effects of gender, age, and weight

340 MENDELSON, WHITE, AND MENDELSON

- .28, p < .Ol . Moreover, after controlling for Time-l relative weight, Time-2 BE-Appearance did not covary with Time-2 relative weight, r(73) = .15, p > .05; however, after controlling for Time-2 relative weight, Time-2 BE- Appearance still covaried with Time-l relative weight, r(73) = - .3 1, p ==c .Ol .

That is, children’s feelings about their body were predicted better by their weight 2 years earlier than by their current weight.

For Sample 2 (Table 7b), SE-Global appeared to be predicted by all three Time- 1 variables. To explore this further, SE-Global,,,,,,,,,, was regressed using a stepwise procedure on SE-Global,,i,, and on the six predictor variables (i.e., relative weight, BE-Appearance, and BE-Weight at childhood and adolescence). Only SE-GlobalChild, r,,,(82) = .43, p < .Ol, and BE-Appearance,,,o,esCe,,, r,,,(82) = .3 1, p < .Ol , were entered into the equation, accounting for 37% of the variance. None of the Time-l variables significantly improved prediction. Similarly, the apparent prediction of SE-Schooladolescent by BE-Weigh&, was eliminated if SE-School,,i,, was entered into the equation first. Thus, no evi- dence was found to support the view that weight or body esteem predicted any aspect of self-esteem 2 years later; there was also no evidence found to demon- strate that weight or self-esteem predicted either aspect of body esteem.

DISCUSSION

The studies reported here extended previous work on the relation between self- esteem and weight by using a domain-specific measure of self-esteem; they also corroborated earlier findings of sex and age effects on self-esteem (Harter, 1985b). As well, the research added to the literature on body esteem by identify- ing two factors of Mendelson and White’s (1985) Revised Body-Esteem Scale and by specifying age-related differences in body esteem. Most important, the studies clarified the relation between self-esteem and body esteem.

Self-Esteem There were generally few predictors of global self-esteem or of self-evaluations in any of the domains. Gender affected three subscales. As in a previous study (Harter, 1985b), boys had higher perceived athletic competence than girls, both in childhood and adolescence. This gender difference is consistent with gender roles, because boys presumably participate in sports, value athletic competence, and receive praise for athletic ability more so than girls. Thus, boys may have been more competent athletically or may have simply provided more socially desirable responses to athletic items to enhance themselves in a domain deemed important for males. Regardless, gender accounted for only a small amount of the variance in the athletic subscale; clearly factors unrelated to gender roles influence an individual’s self-evaluations of athletic competence.

As in a previous study (Harter, 1985b), school-aged boys evaluated their behavioral conduct lower than did girls. Again, this difference is role consistent.

Page 21: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 341

Although boys may have been less inclined than girls to respond to conduct items in a socially desirable way, boys do seem to conform less to behavioral stan- dards. Males are generally more aggressive than females in every society, and gender differences are found early in life (Maccoby & Jacklin, 1974). Biological factors may contribute to higher levels of aggression, but parental reinforcement of sex-stereotyped toy choices and behavior may also affect behavioral conduct, particularly in boys.

Finally, adolescent boys saw themselves as more romantically attractive than did adolescent girls. Because there were no measures of dating or romantic involvement, it was not possible to determine if the boys’ self-evaluations were reality based. However, it seems unlikely so. Adolescent boys may simply have lower expectations, or feel less pressured, to be romantically involved than adolescent girls of the same age, and their lower standards could result in higher self-evaluations. Alternatively, adolescent boys may attribute romantic unin- volvement to a lack of desire on their own part, whereas adolescent girls may attribute it to a lack of romantic appeal. Further research is necessary to examine these factors and to explore the role of physical maturity as a predictor of, or mediating variable for, self- and body esteem.

With only two exceptions (discussed later), neither relative weight nor inter- actions with relative weight predicted any of the self-esteem measures. Thus, no support was obtained for Mendelson and White’s (1985) suggestion that being overweight may adversely affect boys’ self-esteem in the athletic domain (SE- Athletic) and girls self-esteem in the social domain (SE-Social, SE-Friendship, and SE-Romantic). Future research, however, may include independent informa- tion about athletic competence and social functioning to test the possibility that self-evaluations in these domains may reflect actual competence more than weight, even if actual competence were related to weight.

In childhood, heavy girls, particularly between 12 and 13 years old, tended to evaluate their behavioral conduct negatively. Perhaps, heavier girls see being overweight as a sign of eating, which they consider an inappropriate behavior. In contrast, heavy boys were apt to evaluate their behavioral conduct somewhat positively. This Weight X Gender interaction is consistent with some of our research on adolescents’ perception of family functioning. Although the heaviest girls rated their families lowest on cohesion, expressiveness, and participation in family decisions, obese and overweight boys did not differ from normal boys on any dimension of family functioning; rather, underweight boys approximated the heaviest girls in ratings of the family (Mendelson, White, & Schliecker, 1995). Overweight boys apparently do not feel that they are violating social or behav- ioral norms by being heavy; indeed, they may see themselves as conforming to those norms.

In adolescence, at least between about 14 and 15.5 years, global self-worth was negatively related to weight. Based on a meta-analysis, Friedman and Brownell ( 1995) concluded that the relationship between self-esteem and weight

Page 22: Self-esteem and body esteem: Effects of gender, age, and weight

342 MENDELSON, WHITE, AND MENDELSON

in adolescents was weak at best, and that the effect was inconsistent across studies. Our results-a relatively weak correlation in a limited age range- certainly do not resolve the issue.

Nonetheless, the equivocal results in the area of self-esteem and weight high- light the need to address some methodological and conceptual issues in future research. Comprehensive studies of youngsters over a wide range of age and weight from ethnically diverse backgrounds may provide more conclusive re- sults. Attention should be paid to the choice of a self-esteem measure depending on the theoretical thrust of the study. Perhaps the construct of global self-esteem should be reassessed for the overweight population. How important is global self-esteem as compared to other domains of self-esteem? Do some overweight individuals link global self-esteem to their physical appearance whereas others do not?

The existing large-scale studies of self-esteem and weight may also be com- plemented by idiographic research. Studies may explore the different ways that individuals interpret their own weight (e.g., as relevant to the social sphere or as a hindrance to athletics) or the importance that individuals attribute to appearance or weight in their everyday lives. Such clarifications may modify the relation between self-esteem and weight in particular domains.

Body Esteem Body esteem does not appear to be unidimensional. There were at least two distinct factors of the Revised Body-Esteem Scale (Mendelson & White, 1993): general feelings about appearance and satisfaction with weight. The long-term stability of the subscales-independent of gender, age, weight, and their interactions-attest to their robustness. Although the body-esteem subscales covaried somewhat, they were treated as separate measures for the reasons noted in Study 1; the approach was validated by the fact that the two subscales were differentially related to other variables. Thus, despite possible preconceptions that concerns about weight are necessarily related to concerns about appearance, the two subscales were sufficiently independent to broaden our understanding of body esteem and of its relation to self-esteem. Most conceptualizations and measures of body esteem focus on appearance, but satisfaction with weight may be particularly important in understanding overweight youngsters’ perception of their bodies. Thus, future research on weight and body esteem should use a measure with an expanded weight scale (e.g., Mendelson, White, & Balfour, 1995).

Girls had lower opinions of their appearance and weight than did boys both in childhood and adolescence, which corroborates earlier research (Harter, 1985b; Mendelson & White, 1985), at least for appearance. Moreover, the gender differ- ence in school-aged children’s opinions about appearance increased with age, because of an age-related decline in girls’ self-evaluations. Starting early in life, girls in North America may incorporate unattainable stereotypes about an ideal

Page 23: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 343

appearance and weight and may consequently develop a general dislike of their

own body. In previous research (Mendelson & White, 1985), the association between

body esteem and age was equivocal. The effect of age was clarified in this article, presumably because we separately analyzed the two body-esteem factors. In childhood, as children grew older, their feelings about their appearance and weight declined. In adolescence, older adolescents tended to evaluate their ap- pearance, but not their weight, more negatively. These results may reflect age- related increases in self-awareness; youngsters may become more realistic about their own looks and more aware of ideal standards. However, feelings about weight may become less important as other domains of self-esteem increase in importance.

The results concerning body esteem and weight were clear. With increasing weight, youngsters were apt to have poorer opinions about their appearance and weight. These results are consistent with the ideas that youngsters are sensitive to societal norms for appearance and weight and that overweight youngsters learn to dislike their looks. However, being overweight may influence body esteem more specifically than previously thought. Within the weight range studied, relative weight uniquely predicted only feelings about weight. That is, feelings about appearance were not associated with weight, independent of feelings about weight.

Links Between Body Esteem and Self-Esteem Youngsters who had positive feelings about their appearance-although not nec- essarily about their weight-tended to have high global self-worth. This is consistent with Harter’s (1993) finding that the correlations between perceived appearance and global self-worth are high and robust across the lifespan even in situations where other domains should take precedence. However, of particular interest is the finding that youngsters’ feelings about their weight did not uniquely predict their self-esteem, with the exception that adolescents who felt good about their weight positively evaluated their own romantic appeal. Con- versely, children with high global self-worth tended to have positive feelings about their appearance and about their weight. Adolescents with high global self- worth tended to feel good about their appearance; those with a high evaluation of their romantic appeal tended to feel good about their weight. These results again confirm the close interdependence of global self-worth and various dimensions of body esteem.

It should be noted that there was no evidence that weight or body esteem predicted any aspect of self-esteem 2 years later; there was also no evidence found to demonstrate that self-esteem predicted either aspect of body esteem. That is, we were unable to determine any causal links between self-esteem and body esteem over time. Future research should address this question in a more comprehensive longitudinal study. Interestingly, children’s feelings about their

Page 24: Self-esteem and body esteem: Effects of gender, age, and weight

344 MENDELSON, WHITE, AND MENDELSON

body were predicted better by their weight 2 years earlier than by their current weight. Perhaps knowing or judging one’s current weight has a delayed effect on feelings about one’s appearance.

The research clarifies the basic difference between feelings about physical appearance and feelings about weight; most importantly, the results reinforce the belief that, regardless of actual weight, global self-worth is related to positive feelings about appearance, and not necessarily to feelings about weight. Hatter (1985b) has indicated that global self-worth reflects the extent to which one is successful or competent in domains that one considers important. However, self- evaluation of physical appearance may be related to global self-worth whether or not individuals claim that physical appearance is important (Hatter, 1993). Un- like competence in other domains, physical appearance can always be judged by the self and others; even weight, within reasonably wide limits, may not be as salient as overall appearance. Regardless, future research should address how importance attributed distinctly to appearance or to weight interact with self- evaluations of appearance or weight to predict global self-worth.

Clinical Implications The fact that positive feelings about appearance, but not feelings about weight, are related to global self-worth may have clinical implications for weight- reduction programs. Traditional clinical models have claimed that overweight individuals have low self-esteem, a message that is certainly not supported by previous research (Friedman & Brownell, 1995) or by the findings here. None- theless, clinicians have concluded that one way to raise self-esteem is through weight reduction, and, given the pervasiveness of this idea, parents of over- weight youngsters may adopt a similar strategy. The thinking is that weight reduction would improve satisfaction with weight, which would, in turn, influ- ence global self-worth-a model that was not supported by our data, because satisfaction with weight was not independently associated with global self-worth. Also, given the high failure and recidivism rates of most weight-reduction pro- grams, emphasizing weight reduction as a panacea for improving self-esteem seems at best short-sighted, at worst dishonest, and certainly frustrating for most individuals. Moreover, parents who try to force a weight-reduction program on their children, especially adolescents, may compromise family relationships or their children’s sense of independence and perhaps thereby undermine self- esteem.

What appears unequivocal is that even overweight individuals who have good feelings about their appearance also have high global self-worth. Therefore, we suggest that clinicians investigate the feasibility that they or parents may enhance self-esteem in overweight youngsters by focusing on improving feelings about appearance. A number of strategies for doing so, independent of weight reduc- tion, may prove useful: changing personal grooming, incorporating aerobic exer- cise into daily routines, training to be more assertive in the face of stereotyping

Page 25: Self-esteem and body esteem: Effects of gender, age, and weight

SELF-ESTEEM AND BODY ESTEEM 345

about body image, and changing notions about “ideal weight” to accommodate a “healthy weight” for body type and age. To this end, parents may make lifestyle changes of their own and carefully examine the attitudes that they explicitly or implicitly transmit to their children.

However, the traditional treatment model may still be preserved, providing that weight loss were only one of several treatment goals. That is, the suggestion to enhance feelings about appearance could be integrated into a behavioral or cognitive-behavioral treatment model. Even though weight is not uniquely re- lated to individuals’ feelings about appearance, increased attention to one’s ap- pearance, which can be incidental to weight loss, might still enhance self- evaluations of appearance. If a successful diet does enhance feelings about general appearance, weight reduction may improve global self-worth. Regard- less, methods of enhancing self-esteem should emphasize improving general feelings about appearance, rather than concentrating solely on weight reduction or on satisfaction with weight.

It should also be noted that the model presented in this article is based on group results and consequently may not be valid in all situations. More generally, the model may be appropriate only for mildly to moderately overweight individu- als. Clinicians may have to use a different model to help severely obese individu- als realistically live with their weight condition, a situation that may not change rapidly or significantly. The attendant problems of low self-esteem in this popu- lation may necessitate other solutions. Certainly, clinical judgment is necessary to evaluate the suitability of any program for a particular individual.

REFERENCES

Allon, N. (1979). Self-perceptions of the stigma of overweight in relationship to weight losing patterns. American Journal of Clinical Nutrition, 32. 470-480.

Bruch, H. (1941). Obesity in childhood and personality development. American Journal of Orfho- psychiatry, I I, 467-474.

Coopersmith, S. (1967). The antecedenrs of self-esteem. San Francisco, CA: Freeman.

Dwyer, J.R., Feldman, J.J., Seltzer, C.C., & Mayer, J. (1969). Body image in adolescents: Attitudes

toward weight and perception of appearances. Journal of Nutrition Education, I, 14- 19. Friedman, M.A., & Brownell, K.D. (1995). Psychological correlates of obesity: Moving to the next

research generation. Psychological Bulletin, 117. 3-20. Gray, H. (1977). Social aspects of body image: Perception of normalcy of weight and affect of

college undergraduates. Perceptual and Motor Skills. 45. 1035-1040. Hatter, S. (1982). The perceived competence scale for children. Child Development. 53. 87-97. Hatter, S. (1983). Developmental perspectives on the self-system. In P.H. Mussen (Series Ed.) &

E.M. Hetherington (Vol. Ed.), Handbook of child psychology: Vol. 4. Socialization, person- ality and social development (pp. 275-386). New York: Wiley.

Hatter, S. (1985a). Competence as a dimension of self-evaluation: Toward a comprehensive model of

self-worth. In R. Leahy (Ed.), The development qf the self (pp. 55-122). New York: Aca-

demic Hatter, S. (1985b). Manualfor the self-perception profile for children. Denver: University of Denver

Press.

Page 26: Self-esteem and body esteem: Effects of gender, age, and weight

346 MENDELSON, WHITE, AND MENDELSON

Hatter, S. (1987). The determinants and mediational role of global self-worth in children. In N. Eisenberg (Ed.), Contemporary issues in developmental psychology (pp. 219-242). New

York: Wiley.

Harter, S. (1988). Manual for the self-perception profile for Adolescents. Denver: University of

Denver Press.

Harter, S. (1990). Developmental differences in the nature of self-representations: Implications for the understanding, assessment, and treatment of maladaptive behavior. Cognitive Therapy and Research, 14, 113-142.

Harter, S. (1993). Causes and consequences of low self-esteem in children and adolescents. In R.F. Baumeister (Ed.), Self-esteem: The puzzle of low self-regard (pp. 87-116). New York: Ple-

num.

Hendry, L.B., & Gillies, P. (1978). Body type, body-esteem, school and leisure: A study of over- weight, average and underweight adolescents. Journal of Youth and Adolescence, 7, 181-

195.

James, W. (1892). Psychology: The briefer course. New York: Henry Holt.

Jeliffe, D. (1966). The assessment of the nutritional status of the community. World Health Organiza- tion Monographs Series, 53, l-271.

Kaplan, K.M., & Wadden, T.A. (1986). Childhood obesity and self-esteem. Journal of Pediatrics, 109, 367-370.

Maccoby, E.E., & Jacklin, C.N. (1974). The psychology of sex dtferences. Stanford, CA: Stanford

University Press.

Martin, S., Housley, K., McCoy, H., Greenhouse, P., Stigger, F., Kenney, M.A., Shoffner, S., Fu, V., Korslund, M., Ercanli-Huffman, F.G., Carter, E., Chopin, L., Hegsted, M., Clark, A.J., Disney, G., Moak, S., Wakefield, T., & Stallings, S. (1988). Self-esteem of adolescent girls

as related to weight. Perceptual and Motor Skills, 67, 879-884. Mendelson, B.K., & White, D.R. (1982). Relation between body-esteem and self-esteem of obese

and normal children. Perceptual and Motor Skills, 54, 899-905. Mendelson, B.K., & White, D.R. (1985). Development of self-body-esteem in overweight young-

sters. Developmental Psychology, 21, 90-96. Mendelson, B.K., & White, D.R. (1993). Manual for the Body-Esteem Scale for Children. Un-

published manuscript, Concordia University, Montreal, Quebec.

Mendelson, B.K., White, D.R., & Balfour, L. (1995). The Body-Esteem Scale for Adolescents and Adults. Unpublished manuscript, Concordia University, Montreal, Quebec.

Mendelson, B.K., White, D.R., & Schliecker, E. (1995). Adolescents’ weight, sex, and family functioning. International Journal of Eating Disorders, 17, 73-79.

Miller, T.M., Coffman, J.G., & Linke, R.A. (1980). Survey on body image, weight and diet of college students. Journal of the American Dietetic Association, 77, 561-566.

Piers, E.V., & Harris, D.B. (1969). Manual for the Piers-Harris children’s self-concept scale. Los

Angeles: Western Psychological Services. Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.

Schliecker, E. (1996). Eating patterns, self-perceptions, and interpersonal relations in adolescent boys and girls. Unpublished doctoral dissertation, Concordia University, Montreal, Quebec.

Secord, P.F., & Jourard, S.M. (1953). The appraisal of body-cathexis: Body cathexis and self. Journal of Consulting Psychology, 17, 343-347.

Stunkard, A., & Mendelson, M. (1967). Obesity and body image: I. Characteristics of disturbances

in the body image of some obese persons. American Journal of Psychiatry, 123, 1296- 1300. Wadden, T.A., Foster, G.D., Brownell, K.D., & Finley, E. (1984). Self-concept in obese and

normal-weight children. Journal of Consulting and Clinical Psychology, 52. 1104- 1105. Zumpf, C.L., & Hatter, S. (April, 1989). Mirror, mirror on the wall: The relationship between

appearance and self-worth in adolescent males and females. Paper presented at the Society

for Research in Child Development, Kansas City, MO.