Addictive Behaviors 27 (2002) 561 – 573

Age of onset and body dissatisfaction in obesity

Jane Wardle*, Jo Waller, Emily Fox
ICRF Health Behaviour Unit, Department of Epidemiology and Public Health,
University College London, 2-16 Torrington Place, London WC1E 6BT, UK

Abstract

This study investigated variation in body dissatisfaction among obese women in relation to age of
onset of obesity, and evaluated the contribution of teasing about weight and self-esteem as possible
causes of higher body dissatisfaction in the early-onset group. One hundred and five obese women
(body mass index [BMI]  30), recruited from clinical settings in England, completed a mailed
questionnaire. The early-onset group (by age 16) had a higher BMI, greater body dissatisfaction, and
lower self-esteem. Predictably, they also reported more childhood teasing. Path analyses confirmed our
expectation that early onset of obesity has an adverse effect on body image, which is independent of
current BMI. Regression analyses failed to support the hypothesized mediating role of either childhood
teasing or lower self-esteem. The possible direction of causal effects between body dissatisfaction and
self-esteem is discussed, with the suggestion that early onset of obesity increases the risk of body
dissatisfaction, which in turn impairs self-esteem. D 2002 Elsevier Science Ltd. All rights reserved.

1. Introduction

Over the last few years, research investigating the psychological correlates of obesity has
begun to move away from comparisons between obese and normal weight groups towards a
‘second generation’ of studies (Friedman & Brownell, 1995) concerned with variation in
psychological well-being within obese groups. This approach acknowledges the heteroge-
neity of obesity and aims to examine the cognitive, emotional, and social factors, which can
increase the risk of psychological distress or eating disturbance among obese people.
Body image disturbance or body dissatisfaction has been one of the major areas of research in
the field of eating disorders (Thompson, 1995). Body image disturbance is widely acknowl-

* Corresponding author. Tel.: +44-20-7679-6642; fax: +44-20-7813-2848.

E-mail address: j.wardle@ucl.ac.uk (J. Wardle).

0306-4603/02/$ – see front matter D 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 3 0 6 - 4 6 0 3 ( 0 1 ) 0 0 1 9 3 - 9
562 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573

edged to be associated with increased psychological distress and to be a risk factor in the
aetiology of eating disturbance and disorders (Button, Loan, Davies, & Sonuga-Barke, 1997;
Ricciardelli, Tate, & Williams, 1997). While being prevalent in normal-weight groups
(especially adolescent girls), it has also been found to be widespread in obese populations
(e.g., Foster, Wadden, & Vogt, 1997; Smith, Thompson, Raczynski, & Hilner, 1999). More
importantly, however, body image varies significantly, even among the obese, with some obese
people having a reasonably positive view of their physique, while others are severely distressed
(Hill & Williams, 1998).
In community samples of adolescents or adults, BMI contributes significantly to body
dissatisfaction (Caldwell, Brownell, & Wilfley, 1997; Wardle & Marsland, 1990), and obese
groups tend to have higher body dissatisfaction than nonobese comparison groups (Sarwer,
Wadden, & Foster, 1998). However, associations have been inconsistent among the obese.
Positive associations between BMI and body dissatisfaction have been observed in some
studies (Akan & Grilo, 1995; Hill & Williams, 1998), but several clinical studies have found
no evidence for greater body dissatisfaction among those who are more obese (Foster et al.,
1997; Grilo, Wilfley, Brownell, & Rodin, 1994; Sarwer et al., 1998). Sarwer et al. (1998)
suggest that severity of body dissatisfaction among obese women is likely to be affected by
factors other than BMI, emphasising the importance of perceptions of overweight, rather than
objective BMI. In their weight loss study, Foster et al. (1997) comment that relatively small
reductions in weight might have a significant impact on body image, thereby reducing the
correlation between amount of weight loss and improvements in body image. It is also worth
noting that, in both these studies, approximately a third of participants were African
Americans. Given that the relationship between BMI and body image varies with ethnicity
(Altabe, 1996), this could affect the results. Grilo et al. (1994) used weight in correlational
analyses, rather than BMI, which might also affect the strength of the association.
While higher levels of obesity in adulthood may not necessarily cause body dissatisfaction,
there is evidence that becoming obese at an early age does increase vulnerability (Grilo et al.,
1994; Hill & Williams, 1998). Grilo et al. (1994) found that obese adults who had been obese
since childhood had substantially higher levels of body dissatisfaction than those who became
overweight as adults. Hill and Williams (1998) found that those women who were most obese
were most likely to have been overweight as children (before aged 10), and also to have
poorer body image. Their analyses do not directly examine the relationship between age of
onset and body dissatisfaction, but their data provide support for an association between early
onset of obesity and increased body dissatisfaction in adulthood.
One possible explanation for the link between age of onset and body image is that obesity
in children is likely to attract teasing and according to the ‘negative verbal commentary
hypothesis’ (Thompson, Coovert, Richards, Johnson, & Cattarin, 1995), childhood teasing
has an important influence on body image. Using covariate structure modeling in a sample of
adolescent girls, they inferred that the influence of weight status on body image is mediated
by weight-related teasing, and, in a longitudinal follow-up, they found that teasing at baseline
was independently predictive of body dissatisfaction at follow-up, providing strong support
for the influence of teasing on body image. Most studies looking at the impact of teasing have
focused on nonobese, adolescent populations (Schwartz, Phares, Tantleff Dunn, & Thomp-
 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573 563

son, 1999; Thompson, Coovert, et al., 1995; Thompson, Coovert, & Stormer, 1999;
Thompson & Heinberg, 1993), and it would be valuable to investigate the extent to which
the results generalize to the adult obese population. Grilo et al. (1994) found that reported
teasing about weight and appearance during childhood was associated with greater body
dissatisfaction in a sample of 40 patients attending a weight loss clinic. This association was
particularly strong among the small group (n=15) who reported being obese by the age of 18.
Childhood obesity might also impact on self-esteem, which in turn could compromise the
development of a positive body image. Although obese people do not consistently have lower
self-esteem than nonobese groups (e.g., Sarwer et al., 1998; Stunkard & Wadden, 1992),
several clinical studies have found that a more negative body image is associated with lower
self-esteem (Grilo et al., 1994; Sarwer et al., 1998). Grilo et al. (1994) found no significant
difference in self-esteem between their child- and adult-onset groups, but they did not
calculate overall correlations between age of onset and self-esteem score. Given their small
sample size, this lack of effect could be a matter of power.
The present study uses data from a sample of obese women drawn from a range of clinical
settings to investigate the relationship between age of onset of overweight, body dissatisfac-
tion, teasing, and self-esteem. Childhood-onset of weight problems was predicted to be
associated with higher body dissatisfaction, more childhood teasing, and lower self-esteem,
independently of differences in BMI. Multiple regression and path analysis were used: (i) to
gain a clearer picture of the causal relationships, (ii) to test the prediction that the association
between childhood-onset of obesity and poorer body image was mediated by weight-related
teasing in childhood, and (iii) to consider the nature of the association between body image
and self-esteem.

2. Methods

2.1. Participants

Two hundred and six obese women (BMI30) were contacted via a number of settings in
the UK. These consisted of two dietetics departments, an obesity clinic, and a general
practitioner’s surgery. Recruiting via a range of settings ensured a heterogeneous sample of
overweight people, some seeking treatment and some (from the GP list) not. Women recruited
in clinical settings varied as to their stage of treatment. This provided a more varied sample
than the generally homogenous populations often used in studies of this type.

2.2. Measures

Participants completed a five-page questionnaire booklet including measures of body image,

teasing, self-esteem, and weight. Self-esteem was assessed using the Rosenberg Self-Esteem
Inventory (Rosenberg, 1979), a widely used 10-item scale of general self-esteem. A higher
score on the scale indicates lower self-esteem on this scale. Childhood teasing was measured
using the six-item weight/size teasing factor from the Perception of Teasing Scale (POTS)
564 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573

(Thompson, Cattarin, Fowler, & Fisher, 1995). This is a retrospective questionnaire measuring
frequency of weight-related teasing in childhood. Body dissatisfaction was assessed with a
composite scale derived for the study. Stunkard, Sonrenson, and Schlusinger’s (1983) Figure
Rating Scales (FRS) were used to find out which of nine figures participants: (a) felt like most of
the time, (b) believed that they actually looked like, (c) would most like to look like, and
(d) would describe as overweight. Results from the FRS have been shown to be reliable
(Thompson & Altabe, 1991). The discrepancy between the figure they felt like most of the time
and the one they would most like to look like (a–c; self-ideal discrepancy) formed part of our
measure of body dissatisfaction. A discrepancy score was also calculated between the slimmest
figure, which was described as overweight, and the figure they felt like most of the time (a–d;
self-overweight discrepancy). The figure which participants believed looked most like them (b)
was also included. Added to this score was the rating on the item ‘‘I feel satisfied with the shape
of my body,’’ from the Body Dissatisfaction Scale of the Eating Disorder Inventory (Garner,
Olmstead, & Polivy, 1983). This is rated on a six-point scale from always, very often, or often
(0) to sometimes (1), rarely (2), and never (3). Finally, the score from a self-categorisation of
weight was added, scored from 1 to 5, from very underweight to very overweight (Wardle &
Foley, 1989). The resultant Body Dissatisfaction Scale had good internal reliability (Cronbach’s
a=.78) for this sample, with item-total correlation between .46 and .75. Scores could range
from 1 to 40. As an indication of convergent validity, the correlation between body
dissatisfaction scores and Body Image Avoidance Questionnaire (Rosen, Srebnik, Saltzberg,
& Wendt, 1991) were calculated on the full sample (including women whose BMI was below
30). The correlation was high (r=.57, P<.001) suggesting that the body dissatisfaction score
had reasonable convergent validity.
Participants were asked at what age they first became overweight, and reported their current
height and weight. BMI was calculated using Quetlet’s index (weight [kg]/height [m2]). Data
from participants with a reported BMI of <30 (n=33) were excluded from analyses.
Demographic measures included age, ethnic origin, marital status, number of children,
level of education, and occupation/partner’s occupation. Participants were also asked whether
one of their parents was overweight.

2.3. Procedure

New referrals to the two dietetics departments, and overweight patients from a GP list,
were written to, enclosing a copy of the questionnaire to be returned by post. A reminder was
sent after 2 weeks if the questionnaire had not been received. Attendees at the obesity clinic
were approached in person and invited to participate. Questionnaires were either completed at
the clinic or returned by post.

3. Results

A total of 105 women with BMI of 30 or over returned completed questionnaires, a

response rate of 51%. BMI ranged from 30.0 to 62.6 (mean=37.5±6.3). Participants were
 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573 565

mostly middle-aged, married, and white, with the majority having nonmanual occupations.
The demographic characteristics of the sample as a whole, and subdivided into those with
early onset (up to age 16) and adult onset (17 and over) of overweight are shown in Table 1.
Women in the early-onset group were significantly younger at the time of our assessment
[t(103)= 3.61, P<.001], they were more likely to be single than married [c2(2)=7.87,
P=.02], to have fewer children [c2(3)=19.7, P<.001], and to have more overweight parents

Table 1
Demographic characteristics of the whole sample and child- and adult-onset groups
Whole sample Child-onset Adult-onset
(n = 105) (n = 54) (n = 51)
Characteristic n % n % n %
Age in years (mean, 95% CI) 44.4 [42.0 to 46.9] 40.4 [36.9 to 43.8] 48.8 [45.7 to 51.8]
Marital status
Married 76 73 36 67 40 78
Single 14 13 12 22 2 4
Divorced/widowed 15 14 6 11 9 18

Number of children
0 28 27 23 42 5 10
1 19 18 12 22 7 14
2 27 26 9 17 18 35
3 or more 31 30 10 19 21 41

Educational level
No qualifications 37 35 18 33 19 37
GCSE/O levels 20 19 15 28 5 10
A level 8 8 3 6 5 10
Trade certificate/diploma 28 27 10 18 18 35
Degree/postgraduate degree 12 11 8 15 4 8

Occupation
Nonmanual 60 57 30 56 30 59
Manual 44 43 24 44 21 41

Ethnic origin
White 100 95 51 94 49 96
Black 4 4 2 4 2 4
Asian 1 1 1 2 0 0

Parents overweight
None 26 25 9 17 17 33
Mother only 33 31 15 27 18 35
Father only 16 15 9 17 7 14
Both 30 29 21 39 9 18
566 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573

[c2(2)=7.20, P=.03]. The groups showed similar patterns of occupational class, level of
education, and ethnic origin.
Scores on anthropometric and psychological variables for these two groups and the
sample as a whole are shown in Table 2. Women who were overweight by the age of 16
tended to have a higher BMI [t(103)=2.91, P=.004], greater body dissatisfaction [t(103)=3.06,
P=.003], lower self-esteem [t(103)=2.81, P=.006], and report more teasing during childhood
[Mann–Whitney U=301, P<.001]. The early-onset group was also younger at the time of
assessment, indicating that their higher weight was not simply a result of being overweight
for longer.
Bivariate correlations between the psychological and anthropometric variables for the
whole sample are shown in Table 3. As only 60% of participants reported ever having been
teased (partly because the POTS only asks about teasing in childhood), the distribution of
POTS scores was severely skewed. For this reason, POTS scores were grouped into tertiles,
and the three-level variable was used in analyses of the full sample. Higher body
dissatisfaction was associated with higher BMI, earlier age of onset, more childhood teasing,
and lower self-esteem. Earlier age of onset was also associated with more teasing and lower
self-esteem, indicating that both teasing and self-esteem could play a mediating role in the
relationship between age of onset and body dissatisfaction.
The possible mediating effect of teasing was explored further by comparing two regression
models (see Models 1 and 2, Table 4). In Model 2, body dissatisfaction was regressed onto
self-esteem, age, BMI, and age of onset of obesity, these variables together explaining 44% of
the variance. This gave a b value for age of onset of .22, with 3.4% of the variance in body
dissatisfaction being explained by age of onset (sr 2 =.034). The addition of teasing to the
model (see Model 1) only very slightly reduced the unique variance explained by age of onset
(sr 2 =.023) and had no effect on the b value, indicating that, in the group as a whole, teasing
was not mediating the effect of age of onset.
The early-onset group was examined separately to see if the teasing effect would
emerge more clearly when the analyses were restricted to the group who were more likely
to have been teased. In these analyses, the POTS score is used in full, since the

Table 2
Anthropometric and psychological characteristics of child- and adult-onset groups (mean and 95% CI)
Whole group (n = 105) Child-onset (n=54) Adult-onset (n = 51)
Age (years) 44.4 [42.0 to 46.9] 40.4 [36.9 to 43.8] 48.8 [45.7 to 51.8]
Height (m) 1.64 [1.62 to 1.65] 1.64 [1.62 to 1.66] 1.64 [1.62 to 1.66]
Weight (kg) 100.7 [97.1 to 104.3] 105.3 [99.8 to 110.8] 95.8 [91.4 to 100.2]
BMI 37.5 [36.3 to 38.7] 39.2 [37.3 to 41.0] 35.7 [34.3 to 37.2]
Body dissatisfaction 25.7 [24.9 to 26.5] 26.8 [25.8 to 27.8] 24.5 [23.5 to 25.6]
Report childhood teasing (n, %) 63 (60%) 49 (91%) 14 (27%)
POTS score for teasing 13.0 [11.5 to 14.5] 17.9 [15.9 to 19.9] 7.8 [6.6 to 8.9]
RSE self-esteem scorea 24.1 [23.1 to 25.0] 25.4 [24.7 to 26.7] 22.7 [21.3 to 24.0]
a
Higher score indicates lower self-esteem.
 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573 567

Table 3
Correlations between variables for the whole group (n = 105)
Body Age of onset Teasing (POTS
dissatisfaction BMI Self-esteem of overweight tertile)
Body dissatisfaction
BMI .38 ( P < .001)
Self-esteem scorea .58 ( P < .001) .15 (n.s.)
Age of onset .39 ( P < .001) .34 ( P < .001) .25 ( P = .01)
of overweight
Teasing .31 ( P = .001) .34 ( P < .001) .21 ( P = .03) .65 ( P < .001)
(POTS tertile)
Age .18 (n.s.) .11 (n.s.) .26 ( P = .007) .46 ( P < .001) .37 ( P < .001)
a
Higher score indicates lower self-esteem.

distribution is not so skewed. The mean age of onset of overweight in this group was
8.8 years (± 4.1). Bivariate correlations between variables are shown in Table 5.
Unexpectedly, even in the subgroup who were overweight by age 16, earlier age of onset
was still significantly associated with both body dissatisfaction and teasing. The asso-
ciation between age of onset and teasing persisted, but the association with self-esteem
was no longer significant in the early-onset sample, although in magnitude it was only

Table 4
Regression models showing the effects of age of onset of obesity, BMI, and age on body dissatisfaction, with and
without teasing and self-esteem
b sr 2 P
Model 1 (all variables)
Self-esteem .51 .24 < .001 F [5,104] = 8.84, adjusted R2 = .43
Age .08 .005 .33
BMI .23 .045 .005
Age of onset of obesity .22 .023 .04
Teasing tertiles .01 < .001 .93

Model 2 (excluding teasing)

Self-esteem .51 .24 < .001 F [4,104] = 8.75, adjusted R2 = .44
Age .08 .005 .33
BMI .23 .048 .004
Age of onset of obesity .22 .034 .01

Model 3 (excluding self-esteem)

Age .01 < .001 .89 F [4,104] = 7.07, adjusted R2 = .19
BMI .27 .064 .005
Age of onset of obesity .27 .036 .035
Teasing tertiles .04 < .001 .76
568 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573

Table 5
Correlations between variables for the childhood-onset group (n = 54)
Body Age of onset of Teasing (POTS
dissatisfaction BMI Self-esteem overweight score)
Body dissatisfaction
BMI 0.39 ( P = .003)
Self-esteem scorea .57 ( P < .001) .16 (n.s.)
Age of onset of overweight .40 ( P = .003) .31 ( P = .02) .21 (n.s.)
Teasing (POTS score) .17 (n.s.) .23 (n.s.) .22 (n.s.) .47 ( P < .001)
Age .16 (n.s.) .06 (n.s.) .23 (n.s.) .11 (n.s.) 0.25 (n.s.)
a
Higher score indicates lower self-esteem.

slightly lower than in the full sample. By and large, the size of the correlations was very
similar to the results from the whole group. The strongest change was in the association
between teasing and body dissatisfaction, which was reduced in size and no longer
significant. It therefore appears that although women with early onset of obesity have a
greater risk of being teased, higher levels of teasing within this early-onset group do not,
in themselves, promote greater body dissatisfaction.
In the light of the observation that the earlier onset group had lower self-esteem as well
as higher body dissatisfaction, the data were examined to see if self-esteem played a
mediating role between age of onset and body dissatisfaction or whether it seemed more
likely that body dissatisfaction was influencing self-esteem (see Models 1 and 3, Table 4).
When body dissatisfaction was regressed onto BMI, age, age of onset of obesity, and
teasing (see Model 3), the total amount of explained variance in body dissatisfaction was
19%, with the independent variance explained by age of onset being 3.6% (b = .27).
Adding self-esteem to the model (Model 1) increased the adjusted R2 to .44, but only
decreased the variance explained by age of onset to 2.3% (b = .22). This indicates that
self-esteem independently accounts for a large proportion of the variance in body
dissatisfaction, but does not support the hypothesis that it plays a role in mediating the
effect of age of onset of weight problems.
Given that so many of the variables were significantly correlated, we carried out path
analyses to help understand the relationships between them. Age was included in all the
analyses because older participants tended to have higher self-esteem (b = .18, P =.05), but
is not illustrated in the model because there were no other direct or indirect effects. The
following multiple regression equations were evaluated to create the path diagram: (i) BMI
was regressed onto age and age of onset; (ii) teasing was regressed onto age, BMI, and age of
onset; (iii) body dissatisfaction was regressed onto age, BMI, age of onset, teasing, and self-
esteem; and (iv) self-esteem was regressed onto age, BMI, age of onset, teasing, and body
dissatisfaction. The results of these analyses are illustrated in the path diagram in Fig. 1. It can
be seen that although early onset of obesity is associated with more teasing, teasing has no
effect on body dissatisfaction. Body dissatisfaction is directly affected by BMI, age of onset,
and self-esteem, although it is unclear from our analyses whether the relationship between
 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573 569

Fig. 1. Path analysis for the whole sample (n = 105). (Dotted lines indicate associations which do not reach
statistical significance).

body dissatisfaction and self-esteem is bidirectional (hence, the double-headed arrow between
these two variables).

4. Discussion

Our findings confirm that considerable variation in body dissatisfaction can be observed in
a sample of obese women drawn from a range of clinical settings. This highlights the
importance of investigating why some women are able to maintain a more positive self-image
despite falling short of the cultural norms for a slim appearance, while others experience
considerable dissatisfaction. In our sample, variation in BMI explained part of the variation in
body dissatisfaction, as has been found in some previous studies (Akan & Grilo, 1995), but
there was still a good deal of unexplained variance.
One of the aims of the study was to take forward Grilo et al.’s (1994) observation that
childhood-onset obesity is associated with higher levels of body dissatisfaction, and to test
the hypothesis that this is because there is more childhood teasing within this group (Grilo
et al., 1994; Thompson, Coovert, et al., 1995) and teasing compromises body image. In
the present sample, approximately 50% of women reported being overweight by the age of
16, which we defined as the child-onset group. This group had a higher BMI and a
stronger family history of obesity, consistent with the idea that familial, as opposed to
sporadic, obesity might start at an earlier age. The BMI difference was not simply due to
having been overweight for longer, since women in the child-onset group were in fact
younger. This pattern of findings would make sense if there is a vulnerable subgroup in
the population whose weight problems begin earlier and develop faster, making them
likely to seek professional help at an earlier age. It should be noted, however, that using a
self-report measure of age of onset of obesity poses problems of reliability. Although self-
570 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573

report is normally used, future research might consider using medical records or
alternative methods to corroborate these kinds of retrospective data, making findings
more reliable.
The negative verbal commentary (or teasing) hypothesis proposes that early onset of
obesity increases the risk of teasing during the vulnerable years when body image is being
formed and this partly explains why the early-onset group have higher body dissatisfaction. In
the present sample, bivariate associations showed that earlier onset was indeed associated
with more teasing and that teasing was associated with higher body dissatisfaction — thereby
meeting the criteria for mediation. However, in a multiple regression analysis, the proportion
of the variance in body dissatisfaction ‘explained’ by age of onset was only very slightly
reduced by the addition of teasing to the model, indicating that teasing is not likely to mediate
the effect of age of onset.
As teasing was relatively rare among those who were not overweight in childhood
(thereby producing a skewed distribution for teasing), we also examined the association
between age of onset, teasing, and body dissatisfaction specifically among the early-onset
group. Even, within this group who could all potentially have been teased about their
weight as children, those who were more teased were no more dissatisfied than those who
were less teased. Different results have been reported from normal-weight adolescents
(Thompson, Coovert, et al., 1995), but firstly the extent of teasing was probably much
higher among our early-onset obese group (direct comparisons of scores are not possible),
and, secondly, our sample were being assessed in adulthood, and it is possible that weight-
related teasing in childhood influences body image in adolescence or early adulthood but
the effect does not persist into later adulthood. Longitudinal research would be needed to
track the effect of childhood teasing over time. It is less clear why our results should be
different from those reported by Grilo et al. (1994) who also used a clinically obese
sample and found a correlation of .74 between body dissatisfaction and reported teasing in
their 15 childhood-onset cases. However, they used a different measure of body image and
an earlier version of the teasing questionnaire, as well as having a small sample. In
addition, their correlation analyses did not actually test the mediating role of teasing in
linking age of onset with body dissatisfaction, but clearly the role of teasing deserves
further investigation.
Interestingly, in our sample, even among those whose weight problems had begun by 16,
the earlier the onset, the worse their body image (see Table 5), suggesting that there must be
some other potent factor that makes being fat early in childhood have a particularly damaging
and persistent effect on body image.
Self-esteem was also considered as a factor that might mediate the effects of childhood
obesity. The childhood obese group had lower self-esteem, and lower self-esteem was also
associated with a poorer body image. However, regression analyses indicated that the effect
of age of onset on body dissatisfaction was direct, rather than being mediated by self-esteem.
Self-esteem was found to have a large independent effect, explaining a greater proportion of
the variance in body dissatisfaction than any of the other variables measured.
There has been a certain amount of speculation about the direction of causal effects
between self-esteem and body dissatisfaction — does body dissatisfaction adversely affect
 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573 571

self-esteem or does low self-esteem increase the risk of body dissatisfaction? The data from
the present study are cross-sectional, so causal pathways can only be hypothesized.
Nevertheless, the pattern of results linking age of onset, body dissatisfaction, and self-
esteem could be considered to be informative about the likely direction of effects between
body dissatisfaction and self-esteem, and they suggest that early obesity adversely affects
body image, which in turn impairs self-esteem. Of course, this observation is not
inconsistent with there also being a tendency for those who have lower self-esteem to
perceive themselves as more overweight, hence, the bidirectional link between body
dissatisfaction and self-esteem in the path diagram. Given the cross-sectional nature of
the study, one cannot draw firm conclusions about the direction of causality between self-
esteem and body dissatisfaction, but the results are consistent with the idea that interventions
aimed at reducing body dissatisfaction would improve self-esteem as some authors have
suggested (e.g., Rosen, Orosan, & Reiter, 1995), but, equally, that fostering self-esteem
might improve body image (Foster et al., 1997).
As the present sample were all obese adults, any conclusions about the effect of
childhood obesity on adult self-esteem are necessarily restricted to those who remain
overweight as adults. It could well be interesting to examine self-esteem in the normal-
weight adult population in relation to childhood weight problems. While persistence of
obesity from childhood though to adulthood is a general trend, 50–75% of overweight
adolescents do not become obese adults (Must & Strauss, 1999). Nevertheless, this group
may be vulnerable as a consequence of their early experiences with this highly visible and
stigmatized condition.
The generalisability of these results is limited in a number of other ways. As with much
research in this area, a mainly clinical sample was used, although attempts were made to
recruit participants from a variety of sources. Participants were all women, were mostly white,
married, and tended to be from higher social groups. Given that ethnic differences in the
relationship between body image and self-esteem have been noted (e.g., Akan & Grilo, 1995;
Brugman et al., 1997) and that body dissatisfaction has different patterns for women than men
(e.g., Wade & Cooper, 1999), it would be unwise to assume that our results are generalisable
to other groups. Also of concern is the relatively low response rate. Those women choosing to
participate in a study such as this could conceivably have different characteristics from those
refusing, which again limits the generalisability of our findings.
These results add to the growing body of knowledge regarding variations in psycho-
logical well-being among obese people. They confirm the heterogeneity of obesity and
reiterate the fact that poor body image and low self-esteem are not the inevitable
consequences of having a high BMI. What seems clear is that overweight in childhood
is a risk factor for poorer body image and self-esteem. If the development of overweight in
children cannot be prevented, it will be important to do all that is possible to foster
positive body image and high self-esteem in this vulnerable group. This may create a
certain conflict when, in the interests of weight control, the overweight child has attention
drawn to his or her appearance. There is certainly a need for approaches to managing
childhood obesity that are as sensitive to the psychological, as to the physical, aspects of
the disorder.
572 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573

References

Akan, G. E., & Grilo, C. M. (1995). Sociocultural influences on eating attitudes and behaviors, body image, and
psychological functioning: a comparison of African – American, Asian – America, and Caucasian college wom-
en. International Journal of Eating Disorders, 18, 181 – 187.
Altabe, M. N. (1996). Issues in the assessment and treatment of body image disturbance in culturally diverse
populations. In J. K. Thompson (Ed.), Body image, eating disorders and obesity ( pp. 129 – 147). Wash-
ington: APA.
Brugman, E., Meulmeester, J. F., Spee-van der Wekke, A., Beuker, R. J., Zaadstra, B. M., Radder, J. J., & Verloove-
Vanhorick, P. S. (1997). Dieting, weight and health in adolescents in the Netherlands. International Journal of
Obesity, 21, 54 – 60.
Button, E. J., Loan, P., Davies, J., & Sonuga-Barke, E. J. S. (1997). Self-esteem, eating problems, and psychological
well-being in a cohort of schoolgirls aged 15 – 16: a questionnaire and interview study. International Journal of
Eating Disorders, 21, 39 – 47.
Caldwell, M. B., Brownell, K. D., & Wilfley, D. E. (1997). Relationship of weight, body dissatisfaction, and self-
esteem in African American and white female dieters. International Journal of Eating Disorders, 22, 127 – 130.
Foster, G. D., Wadden, T. A., & Vogt, R. A. (1997). Body image in obese women before, during, and after weight
loss treatment. Health Psychology, 16, 226 – 229.
Friedman, M. A., & Brownell, K. D. (1995). Psychological correlates of obesity: moving to the next research
generation. Psychological Bulletin, 117, 3 – 20.
Garner, D. M., Olmsted, M. A., & Polivy, J. (1983). Development and validation of a multi-dimensional eating
disorder inventory for anorexia nervosa and bulimia. International Journal of Eating Disorders, 2, 15 – 34.
Grilo, C. M., Wilfley, D. E., Brownell, K. D., & Rodin, J. (1994). Teasing, body image and self-esteem in a
clinical sample of obese women. Addictive Behaviors, 19, 443 – 450.
Hill, A. J., & Williams, J. (1998). Psychological health in a non-clinical sample of obese women. International
Journal of Obesity, 22, 578 – 583.
Must, A., & Strauss, R. S. (1999). Risks and consequences of childhood and adolescent obesity. International
Journal of Obesity, 23 (Suppl. 2), S2 – S11.
Ricciardelli, L. A., Tate, D., & Williams, R. J. (1997). Body dissatisfaction as a mediator of the relationship
between dietary restraint and bulimic eating patterns. Appetite, 29, 43 – 54.
Rosen, J. C., Orosan, P., & Reiter, J. (1995). Cognitive behavior therapy for negative body image in obese women.
Behavior Therapy, 26, 25 – 42.
Rosen, J. C., Srebnik, K., Saltzberg, E., & Wendt, S. (1991). Development of a body image avoidance ques-
tionnaire. Psychological Assessment, 3 (1), 32 – 37.
Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.
Sarwer, D. B., Wadden, T. A., & Foster, G. D. (1998). Assessment of body image dissatisfaction in obese women:
specificity, severity, and clinical significance. Journal of Consulting and Clinical Psychology, 66, 651 – 654.
Schwartz, D. J., Phares, V., Tantleff Dunn, S., & Thompson, J. K. (1999). Body image, psychological functioning,
and parental feedback regarding physical appearance. International Journal of Eating Disorders, 25, 339 – 343.
Smith, D. E., Thompson, J. K., Raczynski, J. M., & Hilner, J. E. (1999). Body image among men and women in a
biracial cohort: the CARDIA study. International Journal of Eating Disorders, 25, 71 – 82.
Stunkard, A. J., Sonrenson, T., & Schlusinger, F. (1983). Use of the Danish adoption register for the study of
obesity in thinness. In S. Kety, L. P. Rowland, R. L. Sidman, & S. E. Matthysse (Eds.), The genetics of
neurological and psychiatric disorders ( pp. 115 – 120). New York: Raven.
Stunkard, A. J., & Wadden, T. A. (1992). Psychological aspects of severe obesity. American Journal of Clinical
Nutrition, 55, 524S – 532S.
Thompson, J. K. (1995). Assessment of body image. In D. B. Allison (Ed.), Handbook of assessment methods for
eating behaviors and weight-related problems ( pp. 119 – 144). Sage Publications.
Thompson, J. K., & Altabe, M. N. (1991). Psychometric qualities of the Figure Rating Scale. International
Journal of Eating Disorders, 10, 615 – 619.
 J. Wardle et al. / Addictive Behaviors 27 (2002) 561–573 573

Thompson, J. K, Cattarin, J., Fowler, B., & Fisher, E. (1995). The Perception of Teasing Scale (POTS): a revision
and extension of the Physical Appearance Related Teasing Scale (PARTS). Journal of Personality Assessment,
64, 146 – 157.
Thompson, J. K., Coovert, M. C., Richards, K. J., Johnson, S., & Cattarin, J. (1995). Development of body image,
eating disturbance, and general psychological functioning in female adolescents: covariance structure modeling
and longitudinal investigations. International Journal of Eating Disorders, 18, 221 – 236.
Thompson, J. K., Coovert, M. C., & Stormer, S. M. (1999). Body image, social comparison, and eating dis-
turbance: a covariance structure modeling investigation. International Journal of Eating Disorders, 26, 43 – 51.
Thompson, J. K., & Heinberg, L. (1993). Preliminary test of two hypotheses of body image disturbance. Interna-
tional Journal of Eating Disorders, 14, 59 – 63.
Wade, T. J., & Cooper, M. (1999). Sex differences in the links between attractiveness, self-esteem and the body.
Personality and Individual Differences, 27, 1047 – 1056.
Wardle, J., & Foley, E. (1989). Body image: stability and sensitivity of body satisfaction and body size estimation.
International Journal of Eating Disorders, 8, 55 – 62.
Wardle, J., & Marsland, L. (1990). Adolescent concerns about weight and eating: a social – developmental
perspective. Journal of Psychosomatic Research, 34, 377 – 391.