Normative Developmental Challenges and Dieting and Eating Disturbances in Middle

School Girls

Michael P. Levine Linda Smolak

Anne F. Moodey Melissa D. Shuman

Laura D. Hessen

(Accepted 28 June 1993)

Three predictions based on Levine and Smolak’s (1992) developmental model of diet- ing and eating disturbances were tested: ( I ) Changes in pubertal status andlor dating status increase the probability of nonpathological dieting in middle school girls; (2) concurrent change in pubertal status, dating status, and academic stress increases the probability of subclinical eating disturbances in girls with a slender body ideal; and (3) the co-occurrence of modeling cues and direct messages from peers andlor family about the importance of weight, shape, and dieting increases the probability of these effects. Three hundred eighty-two girls were asked about menarcheal status, dating status, and academic stress, as well as attitudes about shape, eating behavior, and perceptions of peer and family pressures for slenderness. Results confirmed several of the predictions, suggesting that the interaction among cumulative developmental changes in early adolescence, adherence to a slender body ideal, and sociocultural pressures for thinness may be useful in distinguishing middle schoolers at risk for subclinical eating disturbances from both girls who do not diet and girls whose dieting is or will be ”normative.” 0 1994 by John Wiley & Sons, Inc.

Early adolescence is one of the modal ages of onset €or body dissatisfaction, dieting, and eating problems (Crisp, 1980; Levine, 1987). Levine and Smolak (1992) have recently proposed a cumulative stressor model that emphasizes the contribution of normative developmental factors to individual differences in the severity of these problems.

Michael P. Levine, Ph.D., and Linda Smolak, Ph.D., are Associate Professors of Psychology at Kenyon College. Anne F. Moodey, B.A., Melissa D. Shuman, B.A., and Laura D. Hessen, B.A., are Kenyon graduates. Address reprint requests to Dr. Levine at Department of Psychology, Kenyon College, Gambier, OH 43022-9623.

lnternationallournal of Eating Disorders, Vol. 15, No. 1, 11-20 (1994) 0 1994 by John Wiley & Sons, Inc. CCC 0276-3478/94/010011-10

12 Levine et al.

The model postulates that three aspects of early adolescence-weight/fat gains asso- ciated with advanced pubertal status, the onset of dating, and the intensification of academic demands (achievement threat)-interact with a slender body ideal to deter- mine the onset of either nonpathological dieting or eating disturbances in adolescent girls (see Levine & Smolak, 1992; Smolak & Levine, in press, for a detailed review of relevant clinical and developmental literature). All three changes occur normatively dur- ing ages 11-14 years (Brooks-Gunn, 1987; Simmons & Blyth, 1987) and each appears to enhance the saIience of, and dissatisfaction with, appearance, body shape, and body fat (Gralen, Levine, Smolak, & Murnen, 1990; Richards, Boxer, Petersen, & Albrecht, 1990; Striegel-Moore, McAvay, & Rodin, 1986). Moreover, consistent with the pronounced gender difference in eating problems, girls are more likely than boys to experience the entry to middle school concurrently with puberty and the onset of dating, and to be more distressed by the accumulation of such stressors (Simmons, Burgeson, Carlton- Ford, & Blyth, 1987).

Based on Smolak and Levine’s model, four predictions were made. First, the onset of menstruation (indicating advanced pubertal status) or the onset of dating will increase body dissatisfaction and nonpathological dieting (Gralen et al., 1990). Second, the prob- ability of body dissatisfaction and dieting will be greater when these two developmental changes occur simultaneously (i.e., within the same year). Third, the co-occurrence of exposure to modeling cues and direct messages from peers and/or family about the importance of weight, shape, and dieting should increase the probability of dieting in connection with pubertal development and dating (Paxton et al., 1991; Pike & Rodin, 1991).

Fourth, the model predicts that those girls who begin menstruating and dating simul- taneously, and who have a slender body ideal, will be at risk for disordered eating attitudes and behaviors rather than simply nonpathological dieting. Their commitment to slenderness makes them more vulnerable to interpreting pubertal weight gains and the awakening of heterosexual interest in ways that promote the fear of fat and drive for thinness that underlie disturbed eating (Fairburn & Garner, 1988). The model also pre- dicts that the addition of academic threat concurrently with the simultaneous onset of menstruation and dating will push a middle school girl with a slender body ideal into disordered eating. Neither of these interactions is expected to be age specific, as the issue is concurrence of change, not timing (Smolak, Levine, & Gralen, 1993). Again, peer and family emphasis on slenderness is expected to intensify the predicted effects.

METHOD

Subjects

Questionnaires were administered to 382 girls attending middle school in a small, rural, Midwestern community. Almost all were Caucasians from middle class and work- ing class backgrounds. There were 146 sixth graders (87% of the population available), 118 seventh graders (83%), and 118 eighth graders (84%). Participation was voluntary, given parental consent. Only 2% refused. Most of the girls not tested were simply absent that day or otherwise unavailable.

As is often the case when lengthy questionnaires are administered to middle school students (e.g., Paxton et al., 1991), and when participants are informed of their right to skip items if they so wish, many girls (48%) had incomplete scores on one or more of the important variables. However, there was no evidence that this introduced a significant sampling bias.

Normative Developmental Challenges

Predictor Variables

13

Menarcheal Status Menarche is a salient event that girls can reliably report (Brooks-Gunn & Warren,

1988). It is also highly correlated with the pubertal "fat spurt" (Brooks-Gunn, 1987). Consequently, subjects were asked "Have you had your period yet?' and "If so, how old were you when you first had it?' Choices ranged from before 10 years to over 14 years, with 6-month intervals between 10 and 14 years.

Dating Status Subjects were asked if and at what age they had started dating. Emphasis was on the

girl's own appraisal of whether she was "dating," not on objective criteria, because increases in the salience of weight and shape and in dieting are assumed to be related' to individual interpretations about the relevance of appearance "now that I'm going out with boys ."

Simultaneous Change Subjects were categorized as having experienced (a) no changes (have not begun

either menstruation or dating); (b) no simultaneous changes (have begun menstruation, but not dating-or vice versa-or have begun both, but not within the same year); or (c) simultaneous changes (have begun dating and menstruating within the same year). The number of subjects in each of these categories, as well as the number of subjects in each of the menstrual and dating categories, is shown in Table 1.

Academic Threat Subjects completed a questionnaire concerning current academic stress (e.g., "I worry

that schoolwork will get to be too hard') and changes in academic stress relative to the previous academic year (e.g.{ "I worry about meeting the schoolwork demands of my new teachers"). All items were rated on a Likert scale from 1 (never) to 5 (very often or always). An 8-item scale was formed (alpha = .75), with higher scores (maximum = 40) reflecting greater academic threat.

Table 1. Menstrual and dating status (Ns) ~~

Grade

Variable 6 7 8 Total

Menstrual status Not begun Begun in the past year Begun over a year ago

Dating status Not begun Begun in the past year Begun over a year ago

Simultaneous change Not begun dating or menstruating No simultaneity Simultaneity

84 21 29

76 18 33

50 48 22

40 26 45

50 21 29

26 32 29

8 12 93

42 15 58

4 39 36

132 59

167

168 54

110

80 119 87

~

Note. Simultaneous change (simultaneity) means that the girl began menstruating and dating within a 12-month period.

14 Levine et al.

Family Messages Concerning Weight and Shape A 4-item scale of parental investment in daughter's slenderness was produced by

summing over the girl's report (on 5-point Likert scales) of how concerned each of her parents (considered separately) is that she might be or become too fat, as well as how important it is to each that she be thin (alpha = .82). A 3-item scale of mother's invest- ment in her own slenderness was produced by summing over the girl's report (6-point Likert scales) of how often her mother is on a diet, how important it is to her mother to be thin, and how important physical appearance in general is to her mother (alpha = .69).

Peer Messages Concerning Weight and Shape A 3-item scale called peer investment in dieting was formed by summing over the

girl's report (5-point Likert scales) of how many of her friends would like to be thinner, how many of her friends are on a diet, and how often she and her girl friends talk about weight, weight loss, and dieting (alpha = .73). A variable called peer modeling was also calculated by summing across a yes-no checklist of weight management techniques that the girl reported her girlfriends to be using.

Criterion Variables

Shape Dissatisfaction A visual scale consisting of nine figure drawings of a female body, ranging in equal

increments from quite thin (10) to quite heavy (90), was used to assess perception of current shape, ideal shape, and shape dissatisfaction (current-ideal) (Cohn et al., 1987).

Weight Management Behavior Subjects indicated, on 5-point Likert scales, how often within the past 6 months they

had used each of seven methods in a conscious attempt to lose weight: not eating as much at meals or snacks, cutting out or reducing sweets, skipping meals, fasting for a day or longer, exercising, keeping careful track of what one is eating, and following a special diet. Responses were recoded as "never" or "rarely" = 0, "sometimes" = 1, and "often" or "always" = 2, and then summed.

Disturbed Eating The Child Eating Attitudes Test (ChEAT; Maloney, McGuire, Daniels, & Specker,

1989) addresses a variety of attitudes and behaviors associated with anorexia nervosa and bulimia nervosa. Its 25 items are adapted from, and scored like, the Eating Attitudes Test (EAT; Garfinkel & Garner, 1982). Evidence of adequate reliability and validity has been presented by Maloney, McGuire, and Daniels (1988) and Maloney et al. (1989).

RESULTS

Grade Effects

All variables except those related to menarcheal and dating status were submitted to one-way analyses of variance (ANOVAs) as a function of grade. Only three between- grade differences were statistically significant and they are not relevant to the present

Normative Developmental Challenges 15

study. The noteworthy feature here is the absence of significant grade differences in weight management, shape dissatisfaction, and disturbed eating. Therefore, subsequent analyses were conducted on the entire sample, independent of grade.

Recent Onset of Menstruation and Dating

The criterion measures were moderately intercorrelated, TS = + -31 to .42, all ps < .001. Consequently, separate multivariate t tests were used to examine the effects of menstrual and dating status. A 2 X 2 multivariate analysis of variance (MANOVA) was not used because missing data left the n for the factorial MANOVA at 83 fewer than the n for the dating t test and 41 fewer than the n for the menstrual status f test. Girls who had begun dating or menstruating more than a year ago were omitted from the t-test analyses.

Recent Menstruation There was a significant effect for recent menstruation, Hotelling’s T(3,145) = 2.59, p

= .05. Univariate analyses indicated that, as predicted, girls who had begun menstru- ating in the past year used more weight management methods than did girls who had not yet begun menstruating, M s = 3.90 versus 2.30, F(1,147) = 7.74, p < .01. The proportion of variance accounted for is 4%. Contrary to prediction, there was no sig- nificant between-group difference in shape dissatisfaction. There was also no significant difference between these two groups in disturbed eating.

Recent Dating This analysis also yielded a significant effect, Hotelling’s T(3,179) = 4.32, p < .01.

Univariate analyses indicated that, as predicted, girls who had begun dating in the past year engaged in more weight management than did girls who had not yet begun dating, M s = 4.44 versus 2.96, F(1,181) = 7.61, p < .01. The proportion of variance accounted for is 3.5%. Girls who had begun dating also had higher ChEAT scores, M s = 16.22 versus 10.62, F(1,181) = 10.53, p < .001, proportion of variance accounted for = 5.3%. Recent dating versus not yet dating had no significant effect on shape dissatisfaction.

Correlates of Dieting, Shape Dissatisfaction, and Disturbed Eating

Weight Management A multiple regression analysis was conducted in which simultaneous change was

entered first, followed by ideal shape. Forced entry of these two variables was followed by a stepwise entry of variables that have a more general and equal influence within the model of nonpathological dieting (Levine & Smolak, 1992): peer investment in dieting, peer modeling, mother’s investment in her own slenderness, and parental investment in daughter’s slenderness. A square root transformation of the weight management vari- able, performed to correct a heteroskadascity problem, served as the dependent vari- able. This regression was statistically significant, adjusted R2 = .14, F(3,200) = 12.42, p < .0001. The three significant predictors were simultaneous change (p = .IS), mother’s investment in her own slenderness (p = .23), and peer modeling (p = .16) (all ts significant at p < .Ol).

16 Levine et al.

Shape Dissatisfaction After ideal body shape, a component of shape dissatisfaction, was eliminated from the

list of predictors, the same regression analysis as that for weight management was conducted for shape dissatisfaction. This regression was also statistically significant, although again only a small amount of variance was explained, adjusted R2 = .09, F(3,217) = 7.87, p < .301.

Disturbed Eating In this regression analysis the following hypothesized determinants of, disturbed eat-

ing were entered as a block: simultaneous change X ideal shape, simultaneous change x academic threat, and simultaneous change x academic threat x ideal shape. This was followed by stepwise entry of the social learning influences. In addition, quadratic terms for each predictor were inserted to test the possibility that the relationship between cumulative stressors and a pathological outcome involves a threshold and is therefore nonlinear. To correct a heteroskadascity problem, the square root of ChEAT score served as the dependent variable.

An equation with four significant predictors accounted for a moderate portion of the variation in ChEAT scores, adjusted R2 = .27, F(4,205) = 20.50, p < .0001. The four significant predictor variables were: (simultaneous change x academic threat)’ (p = .50), (simultaneous change x academic threat x ideal shape)2 (p = - .28), the quadratic of parental investment in daughter’s slenderness (p = .24), and the quadratic of peer investment in dieting (p = .24) (all ts significant at p < .02). The p for the interaction involving ideal shape is negative because greater eating disturbance is related to a more slender ideal shape.

Thus, as predicted, girls who had experienced two changes simultaneously and who reported greater levels of academic pressure had the highest levels of disturbed eating. More important, the predicted interaction among the three developmental stressors and slender body ideal also contributed significantly to the variance in ChEAT scores.

Correlates of Weight Management versus Correlates of Disordered Eating

To test the model’s ability to distinguish the correlates of nonpathological weight management from the correlates of disturbed eating, a stepwise discriminant function analysis was conducted on three groups of girls. Nondieters (n = 61; 19% of the total sample without missing data) were defined as those who had a weight management score of 0 and a ChEAT score 6 7 . A ChEAT score of 7 is below the 50th percentile of a distribution of EAT scores obtained from 675 high school girls (Rosen, Silberg, & Gross, 1988), and represents an average of a slightly symptomatic response on only 7 (28%) of the 25 ChEAT items. Nonpathological dieters (n = 145; 44%) had a weight management score between 1 and 6 (an average of less than ”sometimes” on all 7 weight management items) and a ChEAT score 625 (an average of ”slightly symptomatic” on all the ChEAT items; below 90th percentile of the high school sample; Rosen et al., 1988). Pathological dieters ( n = 12; 5%) had a weight management score 3 7 and a ChEAT score a26. The following variables were entered into the stepwise analysis: simultaneous change, si- multaneous change x ideal shape, simultaneous change x academic threat X ideal shape, peer modeling cues, peer investment in dieting, mother’s investment in her own thinness, and parental investment in their daughter’s slenderness.

Using maximization Mahalanbois’ differences between groups as the criterion for the stepwise analysis, one significant discriminant function was identified ~’(10) = 48.65, p

Normative Developmental Challenges 17

< .0001. The five variables constituting this function, as well as their standardized canonical correlation values, are simultaneous change X academic threat (1.96), simul- taneous change x academic threat x ideal shape ( - 1.55), peer modeling cues (.05), peer investment in dieting (0.32), and parental investment in daughter's slenderness (0.51). The Wilks value of the function was 0.68, that is, 32% of the variation in group assign- ment was explained. Overall, a statistically significant 54% of the cases were correctly classified (see Table 2), ~'(4) = 33.46, p < .001.

DISC USSlON

In general the results of the present study suggest that Levine and Smolak's (1992; Smolak et al., in press) model, which focuses on the interaction among cumulative developmental changes in early adolescence, adherence to a slender body ideal, and sociocultural pressures for thinness, may be useful in distinguishing middle school girls at risk for subclinical eating disturbances from both girls who do not diet and girls whose dieting is or will be "normative."

Correlates of Nonpathological Dieting

Recent Onset of Menstruation or Dating As predicted, girls who had recently begun menstruating and girls who had recently

begun dating engaged in more extensive and frequent weight management efforts than did girls who had not begun. However, considered separately, each variable explained only about 4% of the variation in weight management scores. Furthermore, across grades the recent onset of menstruation or dating was not associated with greater levels of shape dissatisfaction. Thus, this portion of the model was only minimally supported.

Concurrent Onset of Menstruation and Dating As predicted, girls who had begun menstruating and dating within the same year

tended to engage in significantly more weight management behavior than girls who had undergone those changes sequentially or had undergone only one of the changes. Synchrony in the two normative changes of early adolescence appears to compound

Table 2. Discriminant function classification

Classification Group

Nonpathological Pathological Actual Group N Nondieters Dieters Dieters x2 (2)

~

Nondieters 39 30 8 1 35.2 (77%) (21%) (2%) p < .001

Nonpathological Dieters 93 36 40 17 9.7 (39%) (43%) (18%) p < .01

Pathological Dieters 9 1 2 6 4.7 (11%) (22%) (67%) p < .10

Note. Variables constituting the discriminant function are simultaneous change, simultaneous change X ideal shape, simultaneous change X academic threat X ideal shape, peer modeling, peer investment in dieting, mother's investment in her own thinness, and parental investment in their daughter's slenderness. Chi-square statistics are based on a null assumption of random and equal distribution within each row.

18 Levine et al.

slightly the effects of each on dieting behavior. As hypothesized, synchronous girls tended to have higher shape dissatisfaction scores, but the small amount of variance accounted for indicates that the mediators of the effects of synchronous changes on short- and long-term dieting are clearly subjects for further research.

Sociocultural Influences As predicted, a girl's reports of her own exposure to peer dieting techniques and of

her mother's investment in her own slenderness were both significant independent correlates of nonpathological dieting. This suggests that (a) maternal and peer concerns with weight and shape serve as important modeling cues for a young adolescent girl's weight management efforts, and/or (b) girls who are committed to weight management are more attuned to similar behavior in peers. From a social learning perspective these implications seem intuitively obvious, yet we could find no other empirical evidence with which to support or refute the modeling hypothesis. Reports of significant corre- lations between mother's and daughter's scores on measures of dieting and disordered eating (e.g., Pike & Rodin, 1991) are inherently ambiguous, because they do not permit a discrimination between modeling influences, direct contingencies of reward and pun- ishment in regard to weight control, and direct communication of information.

Correlates of Eating Disturbances

Synchronous Developmental Changes The heart of our model is the hypothesis that a middle school girl's commitment to a

slender ideal may be translated into disturbed eating attitudes and behaviors by the various ways in which synchronous occurrence of pubertal changes in physical devel- opment, dating status, and academic threat focus considerable attention on the signif- icance of fat deposition and body weight for attractiveness and achievement (Levine & Smolak, 1992; Smolak & Levine, in press; Smolak et al., 1993). In general the results of the regression analysis support this theory and suggest a complex threshold phenom- enon wherein the addition of academic threat to the simultaneity of advanced pubertal status and dating onset is associated with a large increase in risk for disturbed eating over the effect of the simultaneity alone. The results of the discriminant function analysis suggest that the confluence of these normative changes and a slender body ideal may distinguish risk for eating disturbances from less serious forms of body image concerns and weight management. If this vulnerability-cumulative stressor effect can be repli- cated within a longitudinal analysis, it would have important implications for primary prevention of the eating disorders.

Sociocu I tu ral Influences Perceived parental pressure to be slender and peer investment in dieting also ac-

counted for a proportion of variance in disturbed eating. Moreover, increments in these variables at the high end of the scale were associated with accelerated increments in disturbed eating. These two variables, along with exposure to peer modeling cues, also contributed to the equation that discriminated pathological from less extreme dieters. Interestingly, these two variables were not independent correlates of nonpathological dieting. Whether girls with higher ChEAT scores actually receive more pressure (Pike & Rodin, 1991) or, because of a cognitive schema, simply perceive more pressure (Cooper & Fairburn, 1993) is an issue for future research.

Normative Developmental Challenges 19

The present study clearly suggests that, in its present form, Levine and Smolak's (1992) model does not do a good job of explaining either body dissatisfaction or non- pathological dieting. However, the results of the discriminant function analysis provide very hopeful evidence for the ability of the model to achieve one of its major goals, namely to discriminate between eating disordered and non-eating disordered girls. Nev- ertheless, the present study constitutes only tentative support for the model. Loss of subjects because of incomplete data is certainly reason for caution in interpretation, as is the absence of body mass indices, which could not be entered into the regression analyses because so few girls reported their height and/or weight. In addition, refine- ments are needed in the measurement of key variables such as pubertal status, change in academic pressures, and schematic "attitudes" about the meaning and importance of weight and shape. Finally, the shortcomings of cross-sectional research must also be acknowledged (Smolak & Levine, in press). The measure of concurrent change here was simply whether the girl had begun menstruating and dating within the same year. Thus, we do not know when some of the girls in the simultaneous change group began to manifest disturbed attitudes and behaviors in regard to weight, shape, and eating. Analysis of a longitudinal sample indicates that the crucial interaction between the developmental stressors and a slender body ideal results in fairly immediate and sig- nificant increases in disturbed eating (Smolak et al., 1993).

REFERENCES

Brooks-Gunn, J. (1987). Pubertal processes and girls' psychological adaptation. In R. M. Lerner & T. T. Foch

Cohn, L. D., Adler, N. E., Irwin, C. E., Millstein, S. G., Klesges, S., & Stone, G. (1987). Body-figure prefer-

Cooper, M. J., & Fairburn, C. G. (1993). Demographic and clinical correlates of selective information process-

Crisp, A. H. (1980). Anorexia n m s u : Let me be. Orlando, I% Grune & Stratton. Fairburn, C., & Garner, D. M. (1988). Diagnostic criteria for anorexia nervosa and bulimia nervosa: The

importance of attitudes to weight and shape. In D. M. Garner & P. E. Garfinkel (Eds.), Diagnostic issues in anorexia nervosa and bulimia nervosa (pp. 3655). New York: Brunner/Mazel.

Garfinkel, P. E., & Gamer, D. M. (1982). Anorexia n m s a : A multi-dimensional perspective. New York Brunner/ Mazel.

Gralen, S. J., Levine, M. P., Smolak, L., & Murnen, S. K. (1990). Dieting and disordered eating during early and middle adolescence: Do the influences remain the same? International Journal of Eating Disorders, 9,

Levine, M. P. (1987). Student eating disorders: Anorexia nmosu and bulimia. Washington, D C National Education Association.

Levine, M. P., & Smolak, L. (1992). Toward a developmental psychopathology of eating disorders: The example of early adolescence. In J. H. Crowther, S. E. Hobfoll, M. A. P. Stephens, & D. L. Tennenbaum (Eds.), The etiology of bulimia: The individual and familial context (pp. 5940). Washington, DC: Taylor & Francis.

Maloney, M. J., McGuire, J., & Daniels, S. R. (1988). Reliability testing of a children's version of the Eating Attitudes Test. Iournal of the American Academy of Child and Adolescent Psychiatry, 5, 541-543.

Maloney, M. J., McGuire, J., Daniels, S. R., & Specker, B. (1989). Dieting behavior and eating attitudes in children. Pediatrics, 84, 482-489.

Paxton, S. J., Wertheim, E. H., Gibbons, K., Szmukler, G. I., Hillier, L., & Petrovich, J. L. (1991). Body image satisfaction, dieting beliefs, and weight loss behaviors in adolescent girls and boys. Journal of Youth and Adolescence, 20, 361-379.

Pike, K., & Rodin, J. (1991). Mothers, daughters, and disordered eating. Journal of Abnormal Psychology, 100, 198-204.

kchards, M. H., Boxer, A. M., Petersen, A. C., & Albrecht, R. (1990). Relation of weight to body image in pubertal girls and boys from two communities. Developmental Psychology, 26, 313-321.

Rosen, 1. C., Silberg, N. T., & Gross, J. (1988). Eating Attitudes Test and Eating Disorders Inventory: Norms for adolescent girls and boys. Journal of Consulting and Clinical Psychology, 56, 305-308.

(Eds.), Biological-psychosocial interactions in early adolescence (pp. 12%153). Hillsdale, NJ: Erlbaum.

ences in male and female adolescents. Journal of Abnormal Psychology, 96, 276279.

ing in patients with bulimia nervoa. Znternationul Journal of Eating Disorders, 13, 10%116.

501-512.

20 Levine et al.

Simmons, R. G., & Blyth, D. A. (1987). Moving into adolescence: The impact of pubertal change and school context. Hawthorne, NJ: Aldine.

Simmons, R. G., Burgeson, R., Carlton-Ford, S., & Blyth, D. A. (1987). The impact of cumulative change in early adolescence. Child Development, 58, 1220-1234.

Smolak, L., & Levine, M. P. (in press). Critical issues in the developmental psychopathology of eating dis- orders. In L. Alexander & B. Lumsden (Eds.), Understanding eating disorders. Washington, DC: Taylor & Francis.

Smolak, t., Levine, M. P., & Gralen, S. J . (1993). The impact of puberty and dating on eating problems among middle school girls. ]ournu1 of Youth and Adolescence, 22, 355368.

Striegel-Moore, R., McAvay, G., & Rodin, J. (1986). Psychological and behavioral correlates of feeling fat in women. lnternational Journal of Eating Disorders, 5, 935-947.