Journal of Psychosomatic Research 52 (2002) 129 – 136

Dieting frequency among college females:

Association with disordered eating, body image, and
related psychological problems
Diann M. Ackarda,*, Jillian K. Crollb, Ann Kearney-Cookec
a
Westwood Lake Office Park II, 8421 Wayzata Boulevard, Suite 305 Golden Valley, MN 55426, USA
b
Division of Epidemiology, University of Minnesota, Minneapolis, MN, USA
c
Cincinnati Psychotherapy Institute, Cincinnati, OH, USA
Received 10 July 2000; accepted 26 June 2001

Abstract

Objective: To examine associations between dieting frequency perception, depression, exercise preoccupation, and feelings of
and eating disorder behaviors, body satisfaction, and related ineffectiveness and insecurity. Dieting frequency was inversely
factors. Method: Females (N = 345) whose average age and body associated with self-esteem, ideal body size, emotional regulation,
mass index (BMI) were 20.58 and 21.79, respectively, were and impulse control. Discussion: Independent of current BMI,
grouped into three categories of lifetime dieting frequency (never, frequency of dieting behaviors is strongly associated with negative
1 – 5 times, or 6 or more times) and matched on current BMI across emotions and problematic behaviors. As this study is correlational
categories. Results: Positive associations were found between in nature, future longitudinal studies should ascertain the sequence
dieting frequency and eating disorder symptoms and related of onset of these experiences. D 2002 Elsevier Science Inc. All
problems such as body dissatisfaction, current body size rights reserved.

Keywords: Body image; Depression; Dieting; Eating disorders; Self-esteem

Introduction There are issues of concern over the practice of

dieting. First, many of the individuals who diet are of
While the definition of ‘‘dieting’’ includes weight- normal weight for their height. In 1985, over 34.1 million
reduction efforts generally considered to be healthy (e.g., men and women dieters were found to be of normal
increased fruit and vegetable intake and decreased fat and weight, as compared to 22.6 million dieters who were
sugar intake), many individuals also consider ‘‘dieting’’ to overweight [4]. Studies have also indicated that approx-
include unhealthy weight-control behaviors such as fasting, imately 47% of normal-weight women and 32% of
skipping meals, and eliminating food groups [1]. Regard- normal-weight girls diet to lose weight [4– 6].
less of how one defines dieting, the lifetime prevalence of Second, many of the individuals who diet use
these weight-control behaviors in females is alarming. unhealthy weight-control behaviors such as fasting and
Results from a nationally representative population-based use of appetite suppressants, rather than healthier weight-
survey of high schools girls indicate that 58% had dieted loss practices such as reducing intake of fat and sweets
[2], and in a study of over 4000 adults, 75% of the women and increasing physical activity. For example, in a com-
indicated that they had been on a diet [3]. munity-based weight-gain prevention program, 22% of
women and 17% of men reported the use of at least
one unhealthy weight reduction behavior over the past
year [7]. Adolescents are also using unhealthy weight-
* Corresponding author. Tel.: +1-763-595-7294; fax: +1-763-595-
control measures. In a state-wide survey of 133,794 male
7293. and female adolescent students in Minnesota, 43% of girls
E-mail address: [email protected] (D.M. Ackard). and over 13% of boys in the 9th and 12th grades reported

0022-3999/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved.
PII: S 0 0 2 2 - 3 9 9 9 ( 0 1 ) 0 0 2 6 9 - 0
130 D.M. Ackard et al. / Journal of Psychosomatic Research 52 (2002) 129–136

fasting or skipping meals to control or lose weight [8]. Method

Diet pills or speed were used by 2% of boys and over
8% of girls in the 9th and 12th grades, and 2% of all Participants
youth surveyed reported using laxatives in an attempt to
control or lose weight. Participants in the current study represent a subsample
Third, there is debate as to the accuracy of restraint of 560 female university students from a larger study [20].
theory (the assumption that dietary restraint is etiolog- Within this original sample, differences were seen by BMI
ically significant for the development of eating disor- categories [c2 (df = 8) = 47.93, P < .00001]; approximately
ders). Supporting restraint theory, results from a 43% of those who were underweight (BMI < 19), 68% of
prospective study of 1010 adolescent girls indicate that those who were normal weight (BMI between 19 –24), and
among the 34.6% of the overall study population who 87% of those who were overweight (BMI > 24) had dieted.
indicated dieting at baseline, 21% of these girls Because of these differences in dieting frequency across
developed eating disorders by the time of the 12-month BMI, 345 females were selected from the larger sample by
follow-up [9]. In contrast, recent research by Lowe using frequency-matching procedures on BMI for each
[10,11] raises question as to the etiological relationship dieting category. Participants were grouped into one of
between dieting and binge-eating. No significant rela- three categories based upon their self-reported lifetime
tionship between dieting and binge-eating was found in dieting frequency [none, 1 –5 times, or 6 or more times]
either study. Consequently, the authors conclude that and were matched across category groups on BMI within
factors other than dieting maintain binge-eating in one unit of kg/m2, resulting in 115 females per frequency
bulimic individuals [11]. of dieting category.
While dieting efforts among average weight or under- The average age of participants who participated in the
weight individuals are most often considered negative [5], current study was 20.58 years (S.D. = 3.01, range 18 to
dieting efforts among overweight individuals have been 48). Self-report height and weight means were 6500
viewed more positively [1]. For example, Wing and (S.D. = 2.63 in.) and 132 lb (S.D. = 20.00), respectively,
Jeffery [12] found that among overweight individuals, corresponding to an average BMI of 21.79 (range 15.84 to
modest weight reductions (10 – 15% of initial body 33.78, S.D. = 3.03). Most participants were single (56%) or
weight) achieved by the combination of calorie reduction in a monogamous relationship (40%), Caucasian (78%) or
and exercise implementation were associated with positive Black (14%), and either Catholic (43%) or Protestant (24%).
changes in cardiovascular risk factors. Furthermore,
results from a study of over 14,000 respondents to a Procedure
popular magazine survey suggest that the frequency of
dieting is less important than the individual’s perception Female students were recruited by ‘advertisements’ for
of being a weight cycler in its association with psycho- the study. Interested students were instructed to sign-up for
logical problems such as low self-esteem and body selected study dates and times on sheets posted in a hallway
satisfaction [13]. However, detrimental consequences of outside of an introductory psychology classroom. They
dieting have also been documented, including later onset signed up for the study to receive extra credit for introduct-
of obesity among female adolescents [14], as well as ory undergraduate courses, namely psychology and wom-
socioenvironmental, psychological, and other physical en’s studies courses. At their study date and time,
effects [15]. participants were seated in groups in a classroom with a
Despite a large body of literature on emotional and maximum of 25 seats at least 3 ft apart and were provided
behavioral factors associated with dieting, we are with a brief overview of the study. Participants provided
unaware of any studies that have looked at these their written consent and were told that they were free to
associations independent of body mass index (BMI). It withdraw from the study at any time. Survey packets,
has been well established that BMI is strongly associated randomly coded without identifying participant’s name,
with dieting frequency [16] as well as disordered eating were distributed to each person. Participants were asked to
and body image dissatisfaction [13,17 – 19]. Conse- refrain from discussing the study with other university
quently, BMI has been statistically controlled in the students until the end of the school year to prevent selection
current study in order to obtain a more accurate under- bias and avoid influencing others’ responses. Data were
standing of the specific relationship between dieting and collected in compliance with the Institutional Review Board
disordered eating, body image dissatisfaction, and other at the university where the study was conducted.
psychopathology. The purpose of the present study was
to evaluate the relationship between lifetime dieting Instruments
frequency and disordered eating behaviors and character-
istics, body image satisfaction, depression, self-esteem, Dieting frequency
and other psychological variables, after controlling for Lifetime dieting frequency was assessed with the
current BMI. question ‘‘Have you ever dieted to lose weight.’’ The
 D.M. Ackard et al. / Journal of Psychosomatic Research 52 (2002) 129–136 131

original response categories were: Never dieted; Yes, I Body mass index
have dieted 1– 5 times; 6 – 10 times; 11– 15 times; and Current BMI was calculated by asking participants to
more than 15 times. The original five categories were self-report their height and weight and using these values
collapsed into three categories (Never, 1– 5 times, and 6 in the standard BMI formula [weight in kilograms divided
or more times) for the current study in order to obtain a by squared height in meters]. Ideal BMI values were
more even distribution and allow for easier matching calculated using each participant’s response to the question
across dieting categories. ‘‘Indicate your ideal weight in pounds (what you would
like to weigh)’’ as the weight value within the formula.
Eating disorders
The Eating Disorders Inventory-2 (EDI-2) is a reliable Depression
and valid 91-item multidimensional self-report instrument The Center for Epidemiological Studies—Depression
that assesses characteristics of eating disorders and related scale (CES-D) is a reliable and valid 20-item self-report scale
psychological concerns [21]. On a 6-point Likert scale measuring symptoms of depression in the general population
(‘untransformed scores’), individuals indicate how often [25]. On a 4-point Likert scale (from 0 – 3), individuals
they engage in the queried characteristics. These scores indicate how often they experience different symptoms of
retain the integrity of the data and reduce skew [22]. EDI-2 depression. Higher scores indicate worse symptomatology.
scores are often weighted from 0– 3 (‘transformed scores’)
for ease of clinical interpretation when comparing to EDI Self-esteem
norms, with higher scores indicating worse symptomato- The Rosenberg Self-Esteem Scale [26] is a 10-item
logy. The Bulimia scale measures cognitive and behavioral 5-point Likert scale questionnaire that assesses level of
aspects of binge-eating. The Ineffectiveness scale includes self-esteem with established reliability and validity [27].
statements about inadequacy and lack of control over life, Higher scores represent higher self-esteem.
and the Perfectionism scale includes items that convey the
need for superiority of personal achievements. The Inter- Affect regulation
personal Distrust scale measures feelings of distrust and The Trait Meta-Mood Scale [28] is a 30-item self-
alienation concerning close relationships. Interoceptive report scale designed to assess relatively stable individual
Awareness relates to one’s ability to recognize and respond differences in people’s tendency to attend to their moods
to emotions, and Maturity Fears assesses concerns about and emotions (Attention subscale), discriminate clearly
growing older and the desire to remain in the security of among them (Clarity subscale), and regulate them
childhood. The Asceticism scale includes items about (Repair subscale).
spiritual ideals such as self-restraint and control of bodily
urges, while the Impulse Regulation scale measures the Exercise
tendency toward a range of impulsive behaviors such as The Obligatory Exercise Questionnaire [29] is a
self-destructiveness and substance abuse. The Social Insec- 20-item questionnaire with documented reliability and
urity scale is designed to measure tension and disappoint- validity that measures general physical activity. On a 4-
ment in social relationships. point Likert scale, individuals report how often they
The Bulimia Test-Revised (BULIT-R) is a 28-item experience each exercise-related situation. Higher scores
instrument measuring symptoms of bulimia with estab- indicate more excessive physical activity. Three factors
lished reliability and validity [23]. Individuals indicate have been identified from the Obligatory Exercise Ques-
which behaviors they engage in and to what extent on a tionnaire with sample statements from each factor noted in
5-point Likert scale. Their overall score is a sum of their parentheses: Emotional Element of Exercise (‘‘When I
responses for 28 of the items, with higher scores indicating don’t exercise, I feel guilty’’), Exercise Frequency and
worse symptomatology. Intensity (‘‘I engage in physical exercise on a daily basis’’),
and Exercise Preoccupation (‘‘When I miss an exercise
Body image session, I feel concerned about my body possibly getting
Body image was assessed with several individual items out of shape’’) [30].
and two scales from the EDI-2. The EDI-2 Body Dissat-
isfaction scale assesses dissatisfaction with overall shape Statistical analyses
and size, and the Drive for Thinness scale captures
information about the individual’s pursuit of thinness. Data were analyzed using SPSS for Macintosh, Version
The Body Image Assessment instrument assesses per- 6.1 [31]. Categorical demographic variables were examined
ception of current and ideal body shape with established with chi-square analyses while continuous demographic
reliability and validity [24]. Participants select one female variables were examined with ANOVAs across dieting
body silhouette perceived to represent her current figure or frequency groups. Continuous scores on dependent varia-
body shape, and one silhouette for the preferred figure or bles were examined with means and were compared using
body shape; figures range from 1 (thinnest) to 9 (heaviest). ANOVA across lifetime dieting frequency categories. For all
132 D.M. Ackard et al. / Journal of Psychosomatic Research 52 (2002) 129–136

Table 1
Means, standard deviations, and ANOVA results for dieting frequency with eating disorder behaviors and characteristics (N = 345)
Dieting frequency
Eating disorder behaviors and characteristics Never, M (S.D.) 1 – 5 times, M (S.D.) > 6 times, M (S.D.) df F value P value
EDI-2
Asceticism 3.22 (1.91) 3.77 (2.33) 5.70 (4.02) (2,340) 22.96 .0000a
Bulimia 0.36 (0.87) 1.17 (2.27) 2.90 (3.48) (2,342) 32.48 .0000a,b
Interpersonal distrust 2.32 (2.63) 2.23 (3.23) 3.40 (3.72) (2,340) 4.67 .0100a,*
Impulse regulation 1.62 (2.95) 2.84 (3.67) 5.41 (5.91) (2,342) 22.63 .0000a
Ineffectiveness 1.61 (3.58) 2.03 (3.02) 5.41 (6.32) (2,342) 24.17 .0000a
Interoceptive awareness 1.23 (2.32) 2.68 (3.92) 5.88 (5.81) (2,341) 35.85 .0000a,b
Maturity fears 2.67 (3.10) 3.19 (3.42) 4.20 (4.20) (2,342) 5.37 .0051c,*
Perfectionism 5.80 (4.04) 6.69 (4.04) 7.31 (4.63) (2,340) 3.67 .0265c,*
Social insecurity 2.75 (3.52) 2.70 (2.82) 4.92 (4.06) (2,342) 15.12 .0000a
BULIT-R 39.31 (9.18) 47.85 (13.68) 72.27 (21.13) (2,342) 140.57 .0000a,b
Means and standard deviations reported for the EDI-2 are based upon transformed scores.
a
Post hoc (Turkey’s test) analyses indicate frequent (  6 times) dieting group significantly different ( P .01) from never dieted and moderate dieting
(1 – 5 times) groups.
b
Post hoc (Tukey’s test) analyses indicate never dieted group significantly different ( P .01) from moderate (1 – 5 times) and frequent dieting (  6 times)
groups.
c
Post hoc (Turkey’s test) and analyses indicate never dieted group significantly different ( P .01) from frequent dieting (  6 times) group.
* Analyses using untransformed scores for the EDI-2 resulted in the following significance levels different from those listed above: Interpersonal Distrust
( P = .0175), Maturity Fears ( P = .0142), and the Perfectionism ( P = .0048).

group comparisons, Tukey’s multiple-comparison tests were status, race, and religion. Individuals who were married or
conducted post hoc to assess differences between pairs of separated/divorced were more likely than single or partnered
groups at P .01. Results using the transformed scores of individuals to diet: 100% (n = 10) of married or separated/
the EDI-2 scales are reported in the tables, and any different divorced (n = 3) individuals reported dieting, compared to
results using untransformed scores for the EDI-2 are 67% (n = 93) of partnered and 64% (n = 124) of single
reported as notes in the table. Post hoc ANCOVAs, control- individuals [c2(6) = 17.35, P < .01]. African American
ling for current body image figure or age in separate (34.7%, n = 17) and Hispanic (50%, n = 1) individuals were
analyses, were conducted to assess if the associations would least likely to report dieting when compared to Caucasian
remain significant. (71.1%, n = 192) and Asian (76.5%, n = 13) individuals
[c2(8) = 33.89, P < .00005]. All individuals who identified
themselves as Jewish (100%, n = 5) reported dieting 6 or
Results more times, compared to 36.5% (n = 54) of Catholic, 26.5%
(n = 22) of Protestant, 35.0% (n = 21) of other religions, and
Demographic differences among dieting frequency 26.5% (n = 13) of individuals with no religious affiliation
groups were assessed with chi-square analyses. Although [c2 (8) = 26.64, P < .001]. No differences were found
analyzed with small subsample sizes, results indicated between frequency of dieting categories and current age or
differences between dieting frequency groups on marital age at onset of puberty.

Table 2
Means, standard deviations, and ANOVA results for dieting frequency with body image (N = 345)
Dieting frequency
Body image measures Never, M (S.D.) 1 – 5 times, M (S.D.) >6 times, M (S.D.) df F value P value
EDI-2
Body dissatisfaction 5.65 (6.53) 10.87 (7.01) 16.88 (7.95) (2,341) 70.18 .0000a,b
Drive for thinness 1.12 (1.93) 4.65 (4.86) 10.83 (5.99) (2,342) 131.63 .0000a,b
Ideal BMI 20.66 (2.14) 19.96 (1.74) 19.31 (1.51) (2,342) 15.93 .0000a,b
Body image assessment: current figure 3.38 (1.09) 3.56 (0.95) 3.90 (0.99) (2,342) 7.94 .0004a
Body image assessment: preferred figure 2.93 (0.75) 2.54 (0.76) 2.37 (0.75) (2,342) 16.93 .0000b
Means and standard deviations reported for the EDI-2 are based upon transformed scores.
a
Post hoc (Tukey’s test) analyses indicate frequent (  6 times) dieting group significantly different ( P .01) from never dieted and moderate dieting
(1 – 5 times) groups.
b
Post hoc (Tukey’s test) analyses indicate never dieted group significantly different ( P .01) from moderate (1 – 5 times) and frequent dieting
(  6 times) groups.
 D.M. Ackard et al. / Journal of Psychosomatic Research 52 (2002) 129–136 133

Table 3
Means, standard deviations, and ANOVA results for dieting frequency with depression, self-esteem, and affect regulation (N = 345)
Dieting frequency
Psychological measures Never, M (S.D.) 1 – 5 times, M (S.D.) > 6 times, M (S.D.) df F value P value
CES-depression scale 12.35 (9.86) 15.19 (9.32) 22.46 (13.07) (2,328) 25.52 .0000a
Rosenberg self-esteem scale 32.02 (5.11) 31.21 (4.58) 27.63 (5.43) (2,342) 24.50 .0000a
Trait Meta-Mood Scale
Full scale 65.31 (12.94) 67.12 (15.28) 74.30 (17.18) (2,342) 11.19 .0000a
Attention Index 26.33 (5.21) 27.32 (6.91) 28.14 (7.50) (2,342) 2.16 n.s.
Clarity Index 26.02 (7.64) 26.77 (8.02) 30.60 (9.49) (2,342) 9.79 .0001a
Repair Index 12.97 (4.43) 13.03 (4.59) 15.56 (4.91) (2,342) 11.64 .0000a
a
Post hoc (Tukey’s test) analyses indicate frequent (  6 times) dieting group significantly different ( P .01) from never dieted and moderate dieting
(1 – 5 times) groups.

Dieting frequency and eating disorder behaviors Dieting frequency and depression, self-esteem, and
and characteristics affect regulation

Associations between dieting frequency and eating dis- Dieting frequency was positively associated with psy-
order behaviors and characteristics are shown in Table 1. chological problems such as depression as measured by the
Dieting frequency was associated with the BULIT-R and all CES-D, and affect regulation problems such as more dif-
core EDI-2 subscales: dieting frequency was positively ficulty discriminating among moods, and less confidence in
associated with symptoms of bulimia, asceticism, interper- ability to regulate moods and emotions as measured by the
sonal distrust, difficulties with impulse regulation, intero- Trait Meta-Mood Scale (see Table 3). Furthermore, fre-
ceptive awareness, maturity fears, and feelings of quency of dieting was inversely associated with self-esteem
ineffectiveness, perfectionism, and social insecurity. The as assessed by the Rosenberg Self-Esteem Scale.
higher the number of times that an individual had dieted,
the more the person appeared to struggle with eating Dieting frequency and exercise
disorder behaviors and characteristics.
The full scale and three factors of the Obligatory Exer-
Dieting frequency and body image cise Questionnaire were associated with dieting frequency
(see Table 4). Higher frequencies of dieting behavior were
Significant differences were found across categories on positively associated with greater emotional attachment to
all of the body image variables (see Table 2). Dieting and preoccupation with exercise, and frequency and intens-
frequency was positively associated with the Body Dis- ity of exercise.
satisfaction and Drive for Thinness scales of the EDI-2,
and inversely associated with ideal BMI values. Fre- Post hoc analyses
quency of dieting was also associated with current and
preferred body figure images as measured by the Body Despite accounting for current BMI in all analyses, there
Image Assessment instrument. Despite being matched on were differences among dieting groups on current body
current BMI, dieting frequency was positively associated image figure, with those with a higher frequency of dieting
with perception of current body size, and inversely seeing themselves as significantly larger than those with less
associated with ideal body size. frequent or no dieting. Consequently, post hoc analyses

Table 4
Means, standard deviations, and ANOVA results for dieting frequency with exercise (N = 345)
Dieting frequency
Exercise measure Never, M (S.D.) 1 – 5 times, M (S.D.) >6 times, M (S.D.) df F value P value
Obligatory exercise questionnaire
Full scale 16.38 (6.74) 21.89 (7.13) 25.84 (8.72) (2,342) 45.18 .0000a,b
Emotional element of exercise 1.89 (1.92) 4.07 (2.64) 5.78 (2.96) (2,339) 66.69 .0000a,b
Exercise frequency and intensity 2.85 (2.58) 5.30 (2.54) 5.86 (2.86) (2,342) 17.46 .0000b
Exercise preoccupation 0.86 (1.15) 1.23 (1.11) 1.73 (1.20) (2,342) 16.44 .0000a,b
a
Post hoc (Tukey’s test) analyses indicate frequent (  6 times) dieting group significantly different ( P .01) from never dieted and moderate dieting
(1 – 5 times) groups.
b
Post hoc (Tukey’s test) analyses indicate never dieted group significantly different ( P .01) from moderate (1 – 5 times) and frequent dieting
(  6 times) groups.
134 D.M. Ackard et al. / Journal of Psychosomatic Research 52 (2002) 129–136

(ANCOVAs) were conducted for all associations, control- The results from the current study emphasize the strong
ling for current body image figure. All previously reported and potentially detrimental association between dieting
results remained significant. frequency and problematic behaviors and characteristics
Participants were not matched on age, but the age of and psychological distress among this group of young
participants in the current study ranged from 18 to 48. To females. Stice and Agras [32] in their investigation of
evaluate if qualitative differences on the dependent variables bulimic subtypes concluded that the combination of dieting
by dieting frequency were accounted for by age, ANCOVAs and negative affect seems to indicate a more severe variant
with age as the covariate were completed on all associations. of bulimia. It is possible that those individuals who have
Results indicate that all analyses remained significant even dieted more frequently and who report greater symptoms
after controlling for age. of psychopathology are prone to the development of eating
disordered behaviors and characteristics. Alternatively, fre-
quent dieting in a young normal-weight group of females
Discussion may be part of a general pattern of eating disorder
symptomatology, perhaps more indicative of some under-
Results from the current study provide robust, consistent lying depressive qualities and problematic attitudes related
evidence that regardless of BMI, dieting frequency among to eating.
young normal-weight college females is associated with The clinical implications of dieting behavior are com-
more symptoms of and greater severity of eating disorder plicated. Dieting, if defined as eating healthfully and
behaviors and emotional distress. Dieting frequency was engaging in moderate exercise, may be the best approach
positively associated with number and severity of eating to weight loss and/or weight maintenance in overweight
disorder symptoms, and body dissatisfaction coupled with individuals, with positive effects on physical and mental
an emphasis on outward appearance. Despite controlling for health [1,12]. Therefore, we cannot simplistically conclude
current BMI, dieting frequency was positively associated that all dieting is bad. However, individuals in the current
with perception of current body size and inversely associ- study were young lean women, and should be discouraged
ated with ideal body size. Post hoc analyses controlling for from dieting. Furthermore, individuals of any weight and
current body perception or age remained significant; there- age should be dissuaded from engaging in unhealthy
fore associations are not related to age differences among weight-control behaviors such as fasting, skipping meals,
diet groups nor related to the higher body figures perceived using laxatives or diuretics, engaging in self-induced
by the more frequent dieters. Frequency of dieting was also vomiting, or using extreme intensities or frequencies of
related to affective disturbances such as depression, low exercise. More specific recommendations regarding the
self-esteem, difficulty discriminating between and regulat- healthfulness of dieting behaviors need to be evaluated
ing moods, maturity fears, and feelings of ineffectiveness, on an individual level.
perfectionism, and insecurity. In addition, dieting frequency
was associated with relationship issues such as distrust in Strengths and limitations
interpersonal relationships, and behavioral problems such as
exercise ‘addiction’ and preoccupation, and a need for self- This study included a number of important factors that
control compounded by poor impulse control skills. add to the utility of the results. A major strength of this
While the current study does control for current BMI by study was the use of a weight-matched sample to control for
matching subjects across dieting categories, it cannot the effect of current BMI. Furthermore, the use of psycho-
address the body composition (muscle weight, fat weight, metrically sound standardized instruments to assess these
bone density, etc.) of the individual. Consequently, individ- constructs allows for comparison of this sample to others in
uals with the same BMI value are likely to be quite diverse future research. Finally, a nonclinical sample of female
in terms of shape, bias regarding the accuracy of their self- college students allows readers to understand a broader
report weight, and body composition. This may account for spectrum of behaviors than if surveying only those students
why, despite matching on BMI, the group of women who who present at a weight-loss clinic or an eating disorder
reported more frequent dieting also reported perceiving their treatment facility. In addition, this sample can provide a
current figure as larger. glimpse of some of the experiences related to dieting and
Results from the current study are consistent with others psychological health that the college population of females
that have found many dieters to be of normal weight [4– 6]. may be experiencing.
Although our sample ranged in BMI from an underweight Nonetheless, certain limitations should be taken into
15.84 to an overweight 33.78, the modal BMI was 20.67, account. First, the authors acknowledge that results from
clearly within the range typically recognized as normal- the current study are correlational in nature, and therefore
weight. The fact that a young group of women with a mean causation cannot be determined. It is possible that enga-
BMI within the normal weight range diet is not as alarming ging in the behavior of dieting may lead to the later onset
as understanding that many of the young women from this of problems. However, it is also possible that certain
sample have dieted as often as 6 times. characteristics (dissatisfaction with body shape and size,
 D.M. Ackard et al. / Journal of Psychosomatic Research 52 (2002) 129–136 135

depression, low self-esteem) place some individuals at Institute (NHLBI) Grant No. T32HL07328 (DMA) and
greater risk for engaging in dieting, and potentially for by the Adolescent Health Training Program (Maternal and
developing an eating disorder. Second, dieting was not Child Health Bureau, HRSA) Grant #5T71MC0000622
objectively defined. Future research should allow partic- (JKC). The authors would like to thank Carol B.
ipants to specify the types, duration, and outcomes of Peterson, PhD, Michael D. Resnick, PhD, and several
their dieting behaviors. Some participants may have anonymous reviewers for their excellent comments on
dieted by making moderate decreases in fat intake and drafts of this manuscript.
increases in fruit and vegetable intake, generally regarded
as health-promoting behaviors [33]. Others may have
engaged in more extreme weight-reduction efforts such References
as fasting and meal skipping that can have a negative
impact on emotional and physical health [15]. One [1] Brownell KD, Rodin J. The dieting maelstrom: is it possible and
previous study found significant differences in self- advisable to lose weight? Am Psychol 1994;49:781 – 91.
[2] Schoen C, Davis K, Collins KS, Greenberg L, Des Roches C, Abrams
reported dieting and energy intake depending on the M. The commonwealth fund survey of the health of adolescent girls.
phrasing of the question [34]; for women, questions New York (NY): Commonwealth Fund, 1997.
asking about current dieting were associated with the [3] Jeffery RW, Adlis SA, Forster JL. Prevalence of dieting among work-
lowest caloric intake, followed by questions assessing ing men and women: the healthy worker project. Health Psychol
1991;10:274 – 81.
decreased calories over the past year, regular dieting
[4] Stephenson MG, Levy AS, Sass NL, McGarvey WE. 1985 NHIS
over past year, high restraint, ever dieted, and doing findings: nutrition knowledge and baseline data for the weight loss
anything to lose weight. It will be important in future objectives. Public Health Rep 1987;102:61 – 7.
research to understand the qualitative differences of diet- [5] Biener L, Heaton A. Women dieters of normal weight: their motives,
ing practices by using reliable and valid dieting assess- goals, and risks. Am J Public Health 1995;85:714 – 7.
ment tools. Finally, the use of a nonrepresentative sample [6] Moses N, Banilivy M, Lifshitz F. Fear of obesity among adolescent
girls. Pediatrics 1989;83:393 – 8.
of young, predominantly normal-weight college women [7] Neumark-Sztainer D, Sherwood NE, French SA, Jeffery RW. Weight
from psychology and women’s studies courses limits the control behaviors among adult men and women: cause for concern?
generalizability of the findings. Obes Res 1999;7:179 – 88.
[8] Minnesota Department of Children, Families and Learning. Minnesota
Directions for future research student survey: 1998 statewide tables. St. Paul, MN, 1998.
[9] Patton GC, Johnson-Sabine E, Wood K, Mann AH, Wakeling A.
Abnormal eating attitudes in London school girls—a prospective epi-
Future research should continue to control for the effects demiologic study: outcome at twelve month follow-up. Psychol Med
of BMI while evaluating the qualitative aspects of dieting 1990;20:383 – 94.
(type of diet, duration of diet, success or failure of the [10] Lowe MR, Gleaves DH, DiSimone-Weiss RT, Furgueson C,
weight-reduction effort, etc.) on outcome variables. This Gayda CA, Kolsky PA, Neal-Walden T, Nelsen LA, McKinney
S. Restraint, dieting, and the continuum model of bulimia nervosa.
approach would address some of the limitations of the J Abnorm Psychol 1996;105:508 – 17.
current study and build upon extant literature. In addition, [11] Lowe MR, Gleaves DH, Murphy-Eberenz KP. On the relation of
researchers should evaluate the possibility that other expe- dieting and bingeing in bulimia nervosa. J Abnorm Psychol 1998;
riences such as weight-related teasing or rejection from 207:263 – 71.
thinner peers add to the negative consequences of dieting [12] Wing RR, Jeffery RW. Effect of modest weight loss on changes in
cardiovascular risk factors: are there differences between men and
found in the current study. women or between weight loss and maintenance? Int J Obes
In conclusion, results from the current study provide 1995;19:67 – 73.
strong evidence for the association between frequent diet- [13] Friedman MA, Wilfley DE, Pike KM, Striegel-Moore RH, Rodin J.
ing and disordered eating concerns and behaviors, body The relationship between weight and psychological functioning
among adolescent girls. Obes Res 1995;3:57 – 62.
image dissatisfaction, and other psychological problems in
[14] Stice E, Cameron RP, Killen JD, Hayward C, Taylor CB. Naturalistic
a sample of young, normal-weight college females. Diet- weight-reduction efforts prospectively predict growth in relative
ing behaviors among females may be descriptive of a weight and onset of obesity among female adolescents. J Consult Clin
broader pattern of disordered eating behaviors not neces- Psychol 1999;67:967 – 74.
sarily meeting clinical criteria. Future research should [15] French SA, Jeffery RW. Consequences of dieting to lose weight:
explore these relationships longitudinally and with more effects on physical and mental health. Health Psychol 1994;13:
195 – 212.
qualitative measures. [16] French SA, Perry CL, Leon GR, Fulkerson JA. Dieting behaviors and
weight change history in female adolescents. Health Psychol
1995;14:548 – 55.
Acknowledgments [17] Friedman MA, Schwartz MB, Brownell KB. Differential relation of
psychological functioning with the history and experience of weight
cycling. J Consult Clin Psychol 1998;66:646 – 50.
Preparation of this article was supported in part by the [18] Garner DM, Garner MV, Van Egeren LF. Body dissatisfaction ad-
Training Program in Behavioral Aspects of Cardiovascular justed for weight: the body illusion index. Int J Eating Disord
Disease from the National Heart, Lung, and Blood 1992;12:263 – 71.
136 D.M. Ackard et al. / Journal of Psychosomatic Research 52 (2002) 129–136

[19] Shisslak CM, Crago M, McKnight KM, Estes LS, Gray N, Parnaby [28] Salovey P, Mayer JD, Goldman SL, Turvey C, Palfai TP. Emotional
OG. Potential risk factors associated with weight control behaviors in attention, clarity, and repair: exploring emotional intelligence using
elementary and middle school girls. J Psychosom Res 1998;301 – 13. the Trait Meta-Mood Scale. In: Pennebaker JW, editor. Emotion, dis-
[20] Ackard DM. Development of bulimia nervosa: familial and psycho- closure, and health. Washington (DC): American Psychological Asso-
logical pathways. Unpublished doctoral dissertation, University of ciation, 1995. pp. 125 – 54.
Cincinnati, Cincinnati, OH, 1998. [29] Pasman L, Thompson JK. Body image and eating disturbance in
[21] Garner DM. Eating disorder inventory: 2. Professional manual. Odes- obligatory runners, obligatory weightlifters, and sedentary individuals.
sa (FL): Psychological Assessment Resources, 1991. Int J Eating Disord 1998;7:759 – 69.
[22] Schoemaher C, van Strien T, can der Staak C. Validation of the Eating [30] Steffen JJ, Brehm BJ. The dimensions of obligatory exercise. Eating
Disorders Inventory in a nonclinical population using transformed and Disord: J Treat Prev 1999;7:219 – 26.
untransformed responses. Int J Eating Disord 1994;15:387 – 93. [31] SPSS. SPSS 6.1 base system user’s guide, Part 1 & 2 Macintosh
[23] Thelen MH, Farmer J, Wonderlich S, Smith M. A revision of the version. Chicago (IL): SPSS, 1994.
Bulimia Test: the BULIT-R. Psychol Assess 1991;3:119 – 24. [32] Stice E, Agras WS. Subtyping bulimic women along dietary restraint
[24] Williamson DA, Davis CJ, Bennett SJ, Goreczny AJ, Gleaves DH. and negative affect dimensions. J Consult Clin Psychol 1999;67:
Development of a simple procedure for assessing body image distur- 460 – 9.
bances. Behav Assess 1989;11:433 – 46. [33] US Department of Health and Human Services. People healthy 2000:
[25] Radloff LS. The CES-D scale: a self-report depression scale for re- national health promotion and disease prevention objectives—full re-
search in the general population. Appl Psychol Meas 1977;1: port, with commentary. Washington (DC): US Government Printing
385 – 401. Office, 1991.
[26] Rosenberg M. Society and the adolescent self-image. Princeton (NJ): [34] Neumark-Sztainer D, Jeffery RW, French SA. Self-reported dieting:
Princeton Univ. Press, 1965. how should we ask? What does it mean? Associations between
[27] Fleming JS, Courtney BE. The dimensionality of self-esteem: hier- dieting and reported energy intake. Int J Eating Disord 1997;22:
archical facet model for revised measurement scales. J Pers Soc Psy- 437 – 49.
chol 1984;404 – 42.