Eating Attitudes in Urban and Suburban Adolescents

Martin Fisher Doris Pastore

Marcie Schneider Cynthia Pegler

Barbara Napolitano

(Accepted 18 August 1993)

We administered the Eating Attitudes Test (EAT), Rosenberg Self-Esteem Scale, and Spielberger Trait Anxiety Inventory to two populations of high school students. The first group, 268 suburban females (mean age 16.2 years), completed their question- naires in May 1988. The second population, 389 females and 281 males (mean age 16.0 years) in a city school with 92% black or Hispanic students, completed their question- naires in February 1990 and were also measured for height and weight as part of a health screening. Scores of 21 or higher on the EAT-26 were achieved by 17.5% of the subur- ban females, 15.0% of the urban females, and 6.0% of the urban males. Significantly more suburban females (63%) considered themselves overweight, compared with both urban females (35%) and males (19%),yet only 14% of suburban females were calcu- lated to be >lo% over ideal body weight, compared with 45% of urban females and 39% of urban males. Contrary to expectations, self-esteem was higher and anxiety lower in the urban students than the suburban students; self-esteem and anxiety were each significantly correlated with higher EAT scores in both populations, but believing oneself overweight was correlated with higher EAT scores in only the suburban stu- dents. These data indicate that abnormal eating attitudes are present among both urban and suburban students but with important differences in their manifestations and implications. 0 1994 by lohn Wiley & Sons, lnc.

Martin Fisher, M.D., is Chief of the Division of Adolescent Medicine, North Shore University Hospital-Cornell University Medical College. Doris Pastore, M.D., i s Attending Physician in the Division of Adolescent Med- icine, Mount Sinai Medical Center. Marcie Schneider, M.D., i s Associate Chief, Division of Adolescent Med- icine, North Shore University Hospital-Cornell University Medical College. Cynthia Pegler, M.D., i s in private practice of Adolescent Medicine and Clinical Instructor in Pediatrics at Cornell University Medical College. Barbara Napolitano, M.A., is Biostatistician in the Department of Research, North Shore University Hospital- Cornell University Medical College. Correspondence should be addressed to Martin Fisher, M. D., Division of Adolescent Medicine, North Shore University Hospitai, 300 Community Drive, Manhasset, N Y 11030.

This paper was presented at The North American Scientific Symposium on Eating Disorders in Adoles- cence, December 6, 1991, Seattle, Washington.

lnternationallournal of Eating Disorders, Vol. 16, No. 1, 67-74 (1994) 0 1994 by John Wiley & Sons, Inc. CCC 0276-3478194101 0067-08

68 Fisher et al.

Those who treat eating disorders have concentrated most efforts on patients with overt cases of anorexia or bulimia nervosa. Yet it became increasingly clear during the 1980s that large numbers of adolescents in nonclinical settings also have abnormal eating attitudes and weight concerns that may require attention. In the 1990s, we are discov- ering more about the prevalence of those abnormal attitudes in various populations and how strongly they are associated with other psychological and psychosocial issues. We are also learning to distinguish attitudes that are truly problematic from those that may be an artifact of the tests we employ.

During the past few years, our group in New York has had the opportunity to study these issues by administering the Eating Attitudes Test (EAT) of Garner and Garfinkel (1979,1980) to adolescents in two very different settings. The results of our studies have enabled us to compare and contrast the abnormal eating and weight attitudes of urban and suburban adolescents, as well as females and males, and to evaluate the association of these attitudes with both perceived and actual weight status, and with the psycho- logical issues of self-esteem and anxiety.

METHODS

Procedures

Two studies are presented in this report. The first study was performed in the gym classes of a suburban high school in May 1988. It included 268 females and was pub- lished in the July 1991 issue of the Iuurnal of Adolescent HeaZth (Fisher, Schneider, Pegler, & Napolitano, 1991). The second took place in the gym classes of an urban high school in February 1989. This study included 399 females and 281 males and was presented at the Society for Adolescent Medicine Annual Meeting in March 1991 (Pastore, Fisher, & Friedman, 1991). The adolescents in both studies provided demographic data, described their attitudes about their weight, and completed the EAT, Rosenberg Self-Esteem Scale, and Spielberger Trait Anxiety Inventory. In the suburban study, percent ideal body weight was calculated based on the students’ reported weights and heights; in the urban study, actual weights and heights were obtained as part of a complete health screening. Although this report combines the two studies in order to compare the data, it is clear that there are several differences between how the studies were performed. Also worth noting is that one important group, suburban males, is not included.

Instruments

The questionnaires completed by the students in both studies included the following:

1. The students were asked to report their age, grade, weight, and height on a cover sheet and were asked to indicate whether they considered themselves normal weight, overweight, or underweight.

2. Each student completed the EAT. As described by Garfinkel and Garner (1979,1980), this self-report instrument measures a broad range of symptoms characteristic of anorexia nervosa and bulimia. It includes questions about eating attitudes and eating behaviors, has been widely tested for both reli- ability and validity, and can be administered in either a 26-item (EAT-26) or 40-item (EAT-40) version. Scores above 30-32 on the EAT-40 or 21 on the

Eating Attitudes 69

EAT-26 are generally considered to be consistent with the diagnosis of an eating disorder. Factor analysis of the EAT may be used to subdivide the scores into three factors (Garner, Olmstead, Bohr & Garfinkel, 1982): Factor I relates to an avoidance of fattening foods and preoccupation with being thin- ner, Factor I1 consists of items reflecting thoughts about bulimia and food preoccupation, and Factor 111 includes items related to self-control of eating. The urban students completed the EAT-26 version, whereas the suburban students completed the EAT-40 version. For the purposes of this report, the suburban scores were calculated for the 26 items of the EAT-26, eliminating the answers to the other 14 questions.

3. Self-esteem was evaluated in both studies using the Rosenberg Self-Esteem Scale (Rosenberg, 1965, 1979). This 10-item scale, developed in 1962 for as- sessing high school students, has shown excellent validity and reliability in use with adolescents. It is scored using a 1-6 rating scale, with higher scores indicating lower self-esteem.

4. The trait version of the State-Trait Anxiety Inventory for Children, developed by Charles Spielberger (1973), was administered as the final questionnaire. This 20-item scale is scored on a 20-60 point system, with high scores indi- cating higher levels of trait anxiety. This test has been well validated in junior high school students, and also has been employed with high school students.

Subjects

Three groups of students participated in the two studies: (1) suburban females, (2) urban females, and (3) urban males. Mean ages of the three groups were 16.2 years for the suburban females and 16.0 years for both the urban females and urban males. Fifty-nine percent of the suburban students were in grades 11-12, while 45% of the urban students were in grades 11-12, reflecting the fact that some of the urban students were below grade level. Although students’ race was not asked directly in either study, school data indicate that the suburban school was 94% white whereas the urban school was 89% black or Hispanic.

Data Analysis

All data in both studies were coded and computerized with statistical analyses per- formed using the SAS package (SAS Institute Inc., Cary, North Carolina). Percent ideal body weight was calculated using a generally acceptable estimate for adolescent females of 100 lb for 60 in. and adding or subtracting 5 Ib for every in. over or under 60 in.; this estimate has been found to closely match those available through standard growth charts and tables (Baldwin & Wood, 1961; Hamill et al., 1979). Two-sample f tests from survey statistics were used to compare the study sample to literature controls for the EAT, Rosenberg, and Spielberger tests. Correlations between pairs of variables were evaluated using the Spearman rank correlation coefficient, the nonparametric correlation coefficient most suitable for ordinal data (Hollander & Wolfe, 1973). The Kruskal-Wallis test (a nonparametric counterpart to analysis of variance) was used to determine differ- ences among three patient groupings (i.e., suburban females, urban females, and urban males), and pairwise Mann-Whitney tests (a nonparametric counterpart to the two sam- ples t test) were used to determine in which pairs the differences existed.

70

RESULTS

EAT

Fisher et al.

Scores on the EAT-26 are presented in Table 1. Although 17.5% of suburban females scored 21 or higher on the EAT (suggestive of a possible eating disorder), 15% of urban females and only 6% of urban males achieved such a score. When the means of 12.7 for the suburban females, 10.1 for the urban females, and 8.3 for the urban males were analyzed by the Kruskal-Wallis test, a significant difference ( p < .OOOl) was detected; the Mann-Whitney test found differences between suburban females and urban females (p < .0017) and between urban females and males ( p < .OOl) . In evaluating the factor scores of the EAT, it is noted that the three groups had significantly different scores on ”pre- occupation with losing weight” (Kruskal-Wallis p < .0001), but not ”bulimia and food preoccupation” or “self control of eating.” Thus, although large numbers of suburban females achieved high scores on the EAT, with special emphasis on preoccupation with losing weight, fewer but still many, urban girls had high EAT scores, but with less overall preoccupation with losing weight. Fewer males had high EAT scores or weight preoccupation.

Weight Status

Perceived and actual weight status are also presented in Table 1. As noted, 10% of the suburban females reported being underweight, 27% said they were normal weight, and

Table 1. suburban adolescents

Eating attitudes, weight, self-esteem, and anxiety among urban and

Suburban Females Urban Females Urban Males ( n = 268) (n = 399) ( n = 281)

~ ~~ ~~ ~~

EAT-26 scores Score 2 2 1 17.5% 15.0% 6.0% Mean score 12.7 10.1 8.3 F, (preoccupatin with losing weight) 8.5 6.0 4.4 F, (bulimia and food preoccupation) 1.4 1.1 1.1

Underweight 10% 13% 18%

F, (self-control of eating) 2.7 3.0 2.7 Perceived weight

Normal weight 27% 52% 63% Overweight 63% 35% 19%

Calculated weight Underweight (<85% ideal) 10% 6% 5%

Overweight (111-119% ideal) 7% 19% 18% Obesity (2120% ideal) 7% 26% 21%

Medium (2-3) 35 % 35% 35% Low (4-6) 27% 9% 1%

High (4MO) 45% 23% 8%

Normal weight (85110% ideal) 76% 49 % 56%

Self-esteem High (0-1) 38% 56% 64%

Anxiety

Medium (3039) 43% 52% 46% Low (2G29) 12% 25% 46%

Note. EAT = Eahng Attitudes Test.

Eating Attitudes 71

63% reported being overweight (although only 11% said they were “very overweight”). In contrast, calculated weights based on the reported weights and heights of the sub- urban females demonstrate that 10% can be classified as underweight and 76% as normal weight, with only 7% overweight and 7% obese. The urban students present a very different picture, with only 35% of urban females and 19% of urban males considering themselves to be overweight. However, calculated weights, based on the actual weights and heights of the urban students, indicate that 45% of the urban females and 39% of the urban males can be classified as overweight or obese, with over 20% of each meeting the criteria for obesity. Thus, although only 14% of suburban females are actually over- weight, a majority consider themselves to be so. On the other hand, although 45% of urban females and 39% of urban males are actually overweight, many fewer perceive themselves as such.

Self-Esteem and Anxiety

Data from the self-esteem and anxiety inventories (Table 1) demonstrate that suburban females are fairly evenly distributed in high, medium, and low categories of self-esteem (38%, 35%, and 27% respectively). The urban students, both females and males, are grouped mostly in the categories of high and medium self-esteem, with only 9% and 1% of urban females and males, respectively, showing low self-esteem on the Rosenberg scale. Similarly, although the suburban females were mostly grouped in the categories of high and medium anxiety, with only 12% scoring in the low anxiety range on the Spielberger Trait Anxiety Inventory, many more urban females (25%) and urban males (46%) scored in the low anxiety range. The suburban population scored within the expected range on both of these tests; the scores of the urban students were unexpected.

Correlation Coefficients

We performed Spearman rank correlation coefficients to determine the association of various findings in both the suburban and urban populations. In the suburban females, as published previously (Fisher et al., 1991), we found strong intercorrelations between abnormal eating attitudes, low self-esteem, high anxiety, and increased weight concerns (including believing oneself overweight and wanting to be thinner). In the urban females those who weighed more were somewhat more likely to score higher on the EAT, to have lower self-esteem, and to believe themselves to be overweight (each with correla- tion coefficients of .ll to .14). As was the case for the suburban females, the urban females with higher EAT scores had significantly lower self-esteem and higher anxiety (at r = .29 and .33, respectively) but were only somewhat more likely to believe they were overweight (r = .lo). Those with lower self-esteem had significant more anxiety ( r = -38) and were more likely to consider themselves overweight (r = .15). Fewer positive correlations were found for the urban males. In the males, higher EAT scores were correlated with lower self-esteem and higher anxiety, but not with expressed weight concerns or actual weight status.

The correlations with EAT scores found in the three study groups are summarized in Table 2. These demonstrate that EAT scores were highly correlated with lower self- esteem and higher anxiety in all three groups; that believing oneself overweight was highly correlated with EAT scores in the suburban females, but not in either the urban females or males; and that actually being overweight was mildly correlated with EAT scores in both groups of females, but not in the males.

72 Fisher et al.

Table 2. Spearman rank correlation coefficients of EAT-26 scores

Suburban Females Urban Females Urban Males (n = 268) (n = 399) (n = 281)

Self-esteem -

Anxiety Believe overweight Percent ideal weight

.46*

.38'

.29*

.l2**"

- .29* .33* .10 .13**

.2T

.16**

.09

.03

Note. EAT = Eating Attitudes Test. * p < ,001.

** p < .01. *** p < .05.

DISCUSSION

The results of these two studies demonstrate that abnormal eating attitudes are highly prevalent in suburban females, less prevalent in urban females, and found in few urban males. That abnormal eating attitudes are highly prevalent among suburban females is in keeping with several other studies published in the pediatric and adolescent medicine literature showing that large numbers of adolescents are unhappy with their weight and involved in chronic dieting (Moore, 1988; Moses, Banilivy, & Lifshitz, 1989; Maloney, McGuire, Daniels, & Specker, 1989; Casper & Offer, 1990; Story et al., 1991; Paxton et al., 1991; Koff & Rierdan, 1991). That these abnormal attitudes are less prevalent among urban adolescents adds to the findings of previous studies that have shown fewer abnormalities in eating attitudes and behaviors among black female college students compared with their white counterparts (Gray, Ford, & Kelly, 1987; Abrams, Allen, & Gray, 1992). However, the finding that 15% of minority high school females achieved a high score on the EAT, compared with 17.5% of suburban females, implies that concerns with weight and dieting may be increasing in the black and Hispanic adolescent popu- lations. This may be the start of a trend that will bear watching in future years. The finding that there were some, although a small number, of urban males who achieved high scores on the EAT is also of interest, although it is not yet clear what distinguishes these males. Initial data from the study of Casper and Offer (1990) also demonstrated similar findings in suburban males.

Abnormal eating attitudes were highly correlated with both self-esteem and anxiety. In fact, when we performed multiple linear regression analysis of our suburban data, we found that self-esteem explained 44% of the variance in EAT scores (Fisher et al., 1991). Low self-esteem and high anxiety have each been well reported in patients with both anorexia nervosa and bulimia (Swift, Bushnell, Hanson, & Logemann, 1986; Weiss & Ebert, 1983; Mitchell, Hatsukam, Eckert & Pyle, 1985; Baird & Sights, 1986), and have recently been described in both black and white college students with disordered eating attitudes and behaviors (Abrams et al., 1992).

Several of our findings among the minority adolescents are both interesting and puzzling. We were struck by how many of the minority youth are overweight by actual measurements, although this is in keeping with national data (Gortmaker, Dietz, Sobol, & Wehler, 1987; Dawson, 1988). That fewer minority youth are concerned about their weight is an important finding, implying that even if weight concerns are increasing in the minority adolescent population this may be more appropriate for many of these youth. In fact, it seems that our attention in this population should still be aimed at determining what interventions are most appropriate for those who are overweight and

Eating Attitudes 73

not concerned. This is contrary to our growing attention in the suburban population to those who are not overweight but are overly concerned. We were especially surprised to find that minority youth had higher self-esteem and lower anxiety than their suburban counterparts. We surmised that this finding may be a result of the many self-esteem messages and classes being delivered to minority youth on a regular basis, that the urban youth attending school in some communities may be a self-selected and psycho- logically healthier group, or that the Rosenberg scale is too global to provide meaningful population comparisons. We were also surprised to find that abnormal eating attitudes in the urban students were not correlated with perceived weight, as opposed to the strong correlation found in the suburban students.

The EAT proved a useful measure in both settings but we believe that our studies point out some of the caveats necessary when analyzing results of this test. Many of the questions have different implications when answered by those who are overweight, and actual weight did correlate with EAT scores in both of our settings. As our data show, the implications of EAT scores in various settings may be very different. These findings have been demonstrated in several previous studies employing the EAT (Williams, Hand, & Tarnopolsky, 1982; Vanderdeycken & Vanderlinden, 1983; Wells, Coope, Gabb, & Pears, 1985; Eisler & Szmuckler, 1985). More research on the varying weight and eating attitudes of adolescents in different populations is needed to corroborate these findings and further assess their implications.

REFERENCES

Abrams, K. K., Allen, L., & Gray, J. J. (1992). Disordered eating attitudes and behaviors, psychological adjustment, and ethnic identity: A comparison of black and white female college students. International Journal of Eating Disorders, 14, 49-57.

Baird, P., & Sights, J. R. L. (1986). Low self-esteem as a treatment issue in the psychotherapy of anorexia and bulimia. Journal of Counsel Development, 64, 449451.

Baldwin, B. T., & Wood, T. D. (1961). Average weight for height tables (9 and 11). In F. T. Proudfit & C. H. Robinson. (Eds.), Normal and therapeutic nutrition (pp. 782-784). New York: Macmillan.

Casper, R. C., & Offer, D. (1990). Weight and dieting concerns in adolescents, fashion and symptom? Pedi- atrics, 86, 384-390.

Dawson, D. A. (1988). Ethnic differences in female overweight: Data from the 1985 National Health Interview Survey. American Journal of Public Health, 78, 1326-1329.

Eisler, I., & Szmukler, G. I. (1985). Social class as a confounding variable in the Eating Attitudes Test. Journal of Psychiatric Research, 19, 171-176.

Fisher, M., Schneider, M., Pegler, C., & Napolitano, B. (1991). Eating attitudes, health-risk behaviors, self- esteem, and anxiety among adolescent females in a suburban high school. Journal of Adolescent Health, 1 2 , 377-384.

Gamer, D. M., & Garfinkel, P. E. (1979). The Eating Attitudes Test: An index of the symptoms of anorexia nervosa. Psychological Medicine, 9, 273-279.

Garner, D. M., & Garfinkel, P. E. (1980). Socio-cultural factors in the development of anorexia nervosa. Psychological Medicine, 10, 647456.

Gamer, D. M., Olmstead, M. P., Bohr, Y., Garfinkel, P. E. (1982). The Eating Attitudes Test: Psychometric features and clinical correlates. Psychological Medicine, 12, 871478.

Gortmaker, S. L., Dietz, W. H., Sobol, A. M., & Wehler, C. A. (1987). Increasing pediatric obesity in the United States. American Journal of Diseases of Children, 141, 535-540.

Gray, J. J., Ford, K., &Kelly, L. M. (1987). The prevalence of bulimia in a black college population. International Journal of Eating Disorders, 6 , 733-740.

Hamill, P. V. V., Drizd, T. A., Johnson, C. L., et al. (1979). Physical growth National Center for Health Statistics percentiles. American Journal of Clinical Nutrition, 32, 607429.

Hollander, M., & Wolfe, D. A. (1973). Nonparametriclstatisfical methods. New York: John Wiley & Sons, Inc. Koff, E., & Rierdan, J. (1991). Perceptions of weight and attitudes toward eating in early adolescent girls.

Journal of Adolescent Health, 12, 307412.

74 Fisher et al.

Maloney, M. J., McGuire, J., Daniels, S. R., & Specker, 8. (1989). Dieting behavior and eating attitudes in

Mitchell, J. E., Hatsukami, D., Eckert, E. D., & Pyle, R. L. (1985). Characteristics of 275 patients with bulimia.

Moore, D. C. (1988). Bodv image and eating behavior in adolescent girls. American rournal of Diseases in

children. Pediatrics, 84, 482489.

American Journal of Psychiatry, 142, 482-485.

Children, 242, 1114-1118. - 393-398.

I - Moses, N., Banilivy, M. M., & Lifshitz, F. (1989). Fear of obesity among adolescent girls. Pediatrics, 83,

Pastore, D., Fisher, M., & Friedman, S. B. (1991). Abnormalities in weight, eating attitudes, and eating behaviors among urban adolescents. Paper presented at the annual research meeting of the Society for Adolescent Medicine. Denver, Colorado, March 1991.

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.

Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Rosenberg, M. (1979). Conceiving the self. New York: Basic Books. Spielberger, C. (1973). Manual for the State-Trait Anxiety Inventory for Children. Palo Alto, CA: Consulting

Psychologists Press, Inc. Story, M., Rosenwinkel, K., Himes, J. H., Resnick, M., Harris, L. I., & Bium, R. W. (1991). Demographic and

risk factors associated with chronic dieting in adolescents. American Iournal of Diseases in Children, 145, 994-998.

Swift, W. J., Bushnell, N. J., Hanson, P., & Logemann, T. (1986). Self-concept in adolescent anorexics. Journal of the American Academy of Child Psychiatry, 25, 82-35,

Vanderdeycken, W., & Vanderlinden, J. (1983). Denial of illness and the use of self-reporting measures in anorexia nervosa patients. International journai of Eating Disorders, 2 , 101-107.

Weiss, S. R., & Ebert, M. H. (1983). Psychological and behavioral characteristics of normal-weight bulimics and normal-weight controls. Psychosomatic Medicine, 45, 293-303.

Wells, J. E., Coope, P. A,, Gabb, D. C., &Pears, R. K. (1985). The factor structure of the Eating Attitudes Test with adolescent schoolgirls. Psychological Medicine, 15, 141-146.

Williams, P., Hand, D., & Tarnopolsky, A. (1982). The problems of screening for uncommon disorders-A comment on the Eating Attitudes Test. Psychological Medicine, 12, 431434.