The Role of Shape and Weight in Self-Concept: The Shape and Weight Based Self-Esteem Inventory 1

Transcription

1 Cognitive Therapy and Research, Vol. 21, No. 1, 1997, pp The Role of Shape and Weight in Self-Concept: The Shape and Weight Based Self-Esteem Inventory 1 Josie Geller, Charlotte Johnston, 2 and Kellianne Madsen University of British Columbia The purpose of this study was to establish the psychometric properties of the Shape and Weight Based Self-Esteem (SAWBS) Inventory, a newly-developed measure of the influence of shape and weight on feelings of self-worth. The results from a nonclinical sample of young women indicated that SAWBS scores were stable over time (N = 24), and correlated moderately with one of two measures of shape and weight schemata (N = 50). In a sample of 84 women, SAWBS scores also correlated moderately with two measures of eating disorder symptomatology, and in regression analyses contributed statistically significant unique variances to both measures of symptomatology, even after the effects of body mass index, depression, and global self-esteem were controlled. Finally, SAWBS scores discriminated between women reporting few or no disturbed eating symptoms and possible/probable eating disorder cases. In sum, the SAWBS Inventory is a reliable and valid measure, and may be a useful tool in the assessment of eating disorders. KEY WORDS: eating disorders; self-esteem. 1 Support for this research was provided by a Studentship award from the Medical Research Council of Canada to Josie Geller, M.A., clinical psychology doctoral candidate at the University of Bristish Columbia (UBC). Charlotte Johnston, Ph.D., is Associate Professor in the Department of Psychology at UBC. Kellianne Madsen is a Research Assistant in the Department of Psychology at UBC. 2 Address all correspondence, including requests for reprints, to Charlotte Johnston, Department of Psychology, University of British Columbia, Vancouver, BC, Canada V6T 1Z /97/ $12.50/0 C 1997 Plenum Publishing Corporation

2 6 Geller, Johnston, and Madsen Recently, researchers have attempted to understand the eating disorders within the context of structural models of the self. Vitousek and Hollon (1990), for instance, proposed that the core psychopathology of the eating disorders may be represented in cognitive schemata that unite views of the self with beliefs about shape and weight. Such conceptual models maintain that shape and weight concerns are not merely salient to eating disordered individuals, but rather have a central influence (positive or negative) on feelings of self-worth. In support of this view, Cooper and Fairburn (1993) recently proposed that overvalued ideas about shape and weight, or the extent to which self-worth is judged in terms of shape and weight, constitute the "central cognitive substrate" of bulimia (p. 386). This linking of shape and weight to self-esteem in the eating disorders is also reflected in recent changes to the Diagnostic and Statistical Manual of Mental Disorders (DSM). In the fourth edition of DSM (DSM-IV), the new criterion "selfevaluation is unduly influenced by shape and weight" (American Psychiatric Association, 1994, pp. 545 and 550) has been accorded central status in the diagnosis of bulimia nervosa, and is one of three possible components of a criterion for anorexia nervosa [the other two are "disturbance in the way in which one's body weight or shape is experienced" and "denial of the seriousness of the current low body weight" (p. 54)]. Although a number of self-report measures have been developed to address the importance of shape and weight concerns to individuals who have eating disorders (see Ben-Tbvim & Walker, 1991, for a review), many of these lack specificity, in that they assess multiple dimensions (e.g., evaluations, perceptions, and feelings about the body) within a single scale or instrument. In addition, these measures do not assess the importance of shape and weight within the context of other dimensions upon which selfesteem is based. The Eating Disorders Examination (EDE; Cooper & Fairburn, 1987), a structured interview for diagnosing eating disorders, is a notable exception in that it contains two single-item scales that assess the extent to which self-esteem is based on shape and weight, respectively. Unfortunately, administration of the full interview requires at least 1 hour with a skilled interviewer, and the validity of the two items, taken out of the context of the remainder of the interview, has not been established. As a result, the EDE may not be practical for research or clinical purposes when time and/or resources are limited. In sum, there is a need for a specific, and easily administrable measure that assesses the extent to which self-esteem is based on shape and weight. The purpose of the present study was to develop a measure that met these needs; the Shape and Weight Based Self-Esteem (SAWBS) Inventory, and to determine the stability, as well as the concurrent, discriminant, and predictive validities of SAWBS scores.

3 The SAWBS Inventory 7 The SAWBS Inventory is an innovative and easy-to-administer measure of the structure of self-esteem, developed to assess the contribution of shape and weight, in the context of other dimensions, to overall feelings of self-worth. In the SAWBS Inventory, individuals select and rank order from a list of attributes those that are important to their feelings of selfworth, and then divide a circle into pieces, such that the size of each piece reflects the importance of each attribute to overall feelings of self-worth. The unique design of the SAWBS Inventory allows for the relative contribution of shape and weight to overall feelings of self-esteem to be objectively assessed, in the context of other dimensions upon which self-esteem is based. As a first step toward the validation of this new instrument, the properties of the SAWBS Inventory were investigated in a sample of undergraduate women. Female undergraduates were considered a suitable population for this study because a range of eating-disordered attitudes and behaviors has been reported in such samples, and risk in such groups has been judged to be high (e.g., Klemchuk, Hutchinson, & Frank, 1990; Klesges, Beatty, & Berry, 1985; Vanderheyden, Fekken, & Boland, 1988). With regards to stability of SAWBS scores, given that overvalued ideas about shape and weight are purported to be relatively enduring over time (Cooper & Fairburn, 1993), it was hypothesized that SAWBS scores assessed 1 week apart would be significantly and highly correlated. Although one would expect shape- and weight-based self-esteem to be stable over much longer periods of time, 1 week was considered to be a minimal testretest interval for this study, and chosen as a compromise with the practical limitations of this study. Three types of concurrent validity of SAWBS scores were investigated. First, shape- and weight-based self-esteem was predicted to be related to individuals' perceptions of their own shapes and weights. It is important to note that shape- and weight-based self-esteem is not a measure of positive or negative feelings about the body, but rather of the extent to which feelings of self-worth are based on the body. However, given that shapeand weight-based self-esteem is a diagnostic criterion for eating disorders, and individuals who have eating disorders report high body dissatisfaction (e.g., Hsu & Sobkiewicz, 1989), it was hypothesized that higher SAWBS scores would be associated with the tendency to view oneself as more overweight and one's shape as less attractive. Next, the relationship between SAWBS scores and eating disorder symptomatology was investigated. Again, because an undue influence of shape and weight on self-esteem is described as a necessary and unique feature of the eating disorders, it was hypothesized that SAWBS scores would be positively related to disturbed eating patterns. Furthermore, to examine whether shape- and weight-based self-esteem accounts for a

4 8 Geller, Johnston, and Madsen unique aspect of eating disorder symptomatology, independent from global self-esteem and depression [two constructs that are not eating disorderspecific, but that are frequently associated with eating disorder symptomatology (e.g., Hsu, 1982; Kaplan, Busner, & Pollack, 1988)], the relationship between SAWBS scores and eating disorder symptomatology was examined controlling for the effects of depression and global self-esteem. Finally, given that views of the self are closely tied to cognitive schemata (e.g., Markus, 1977), it was hypothesized that shape- and weight-based self-esteem would be related to the extent to which shape- and weight-based schemata influence individuals' information processing styles on two measures of cognitive processing. Shape, weight, and food-related schemata have been widely investigated in the eating disorders using Stroop (1935) colornaming interference paradigms. A number of studies have shown that, relative to women who do not have eating disorders, women who have eating disorders take longer to color-name eating disorder-related target words (e.g., cakes, diet, and hips) than control words (e.g., Cooper, Anastasiades & Fairburn, 1992; Cooper & Fairburn, 1992). However, a number of inconsistencies in work using the Stroop paradigm have been noted. Ben-Tbvim, Walker, Fok, and Yap (1989) found that when shape and weight words were presented separately (without food words), color-naming interference was significantly reduced, reaching statistical significance only in bulimic, and not anorexic, subjects. In addition, the occurrence of Stroop effects in subclinical populations has been inconsistent, with some authors finding such effects (e.g., Green & Rogers, 1993), and others not (e.g., Cooper & Fairburn, 1992). Finally, the Stroop task has been shown to be susceptible to priming influences, which can significantly affect the sensitivity of this task (e.g., Warren, 1974). Although not yet applied to the eating disorders, a second, perhaps more promising, methodology for the assessment of cognitive schemata involves the false recall of schema-consistent words in false alarm effect (FAE) paradigms (e.g., Rogers, Rogers, & Kuiper, 1979). In this task, participants are presented with a list containing both schema words they have previously seen, as well as new schema-consistent words. The task is to identify the words that were presented in the previous list only. The incorrect identification of schema-consistent words has been related to the extent to which the individual is schematic along that dimension (e.g., Mathews & MacLeod, 1983; Rogers et al., 1979). In sum, Stroop paradigm research suggests that women who have eating disorders may selectively process information related to eating, weight, and shape, and supports the idea that elaborated cognitive schemata representing these categories of information are characteristic of eating disorders. The false alarm effect paradigm is a possible methodology for future cognitive research with the eating disorders, and may be an attractive alternative to Stroop paradigm research,

5 The SAWBS Inventory 9 given some of the shortcomings that have been identified in this method. In the present study, it was hypothesized that shape- and weight-based selfesteem would be related to the extent to which shape- and weight-based schemata influence individuals' responding on the Stroop and FAE tasks. Finally, with regard to predictive validity, the ability of SAWBS scores to discriminate between women who have an eating disorder and women who do not have an eating disorder has implications for the utility of the SAWBS Inventory as a screening instrument. In this study, the ability of shape- and weight-based self-esteem to discriminate between probable and possible eating disorder cases, and individuals not expected to have an eating disorder were therefore examined. It was hypothesized that SAWBS scores would be higher in the group identified as probable and possible cases. With regard to discriminant validity, because shape- and weight-based self-esteem is the extent to which feelings of self-worth are based upon shape and weight, and not a measure of actual shape and weight, SAWBS scores were not expected to be correlated with body mass index (BMI). It was also hypothesized that SAWBS scores would not be associated with the tendency to respond in a socially sanctioned manner, or to family socioeconomic status. METHOD Subjects One hundred and ten female undergraduate students between the ages of 18 and 45 years (M = 21.2, SD = 6.6) participated in a study on women's attitudes and self-perceptions in exchange for course credit. Fiftytwo (48%) of the women described their ethnic background as European or Anglo Saxon, 33 (31%) as Asian, 5 (5%) as Indo-Canadian, and one woman (1%) described herself as African. Seventeen (16%) women either failed to respond to this question or gave uncodable responses (e.g., "Canadian"). Most of the women (91%) indicated that they were raised in two-parent families. Family socioeconomic status ranged from 1 to 5 on the Hollingshead Four-Factor Index (Hollingshead, 1975), with a mean of 2.3 (SD = 1.0), indicating upper middle class. Measures Shape and Weight Based Self-Esteem Inventory. In the SAWBS Inventory, individuals select from a number of personal attributes those that are

6 10 Geller, Johnston, and Madsen important to how they have felt about themselves in the past 4 weeks. The list of personal attributes was generated from previous measures that have identified specific dimensions of self-esteem (Harter, 1993; Marsh, 1986, 1993; O'Brien, 1980). Several possible personal attributes were initially pilot-tested on female graduate students, who provided suggestions regarding the relevance of the attributes for young women. The final list of attributes included intimate or romantic relationships, body shape and weight, competence at school/work, personality, friendships, face, personal development, competence at activities other than school/work, and other (individuals are asked to list an attribute if it is not covered in the preceding list). After selecting the personal attributes that are important to how they feel about themselves, individuals rank-order the attributes in terms of how much their opinion of themselves is based on each. Then, they divide a circle into pieces, such that the size of each piece reflects how much their self-opinion is based on each of the ranked attributes. Thus constructed, the SAWBS Inventory has the advantage of measuring the relative contribution of shape and weight attributes to overall self-esteem. The SAWBS score is the angle of the shape and weight piece of the circle 3. Eating Disorders Inventory (EDI). The EDI (Garner & Olmstead, 1984; Garner, Olmstead, & Polivy, 1983) is a 64-item self-report scale designed to measure attitudes, personality features, and eating disorder symptoms thought to be relevant to anorexia nervosa and bulimia nervosa. Subjects are asked to rate each item on a 6-point scale ranging from never to always. The EDI was constructed to have eight independent subscales (Garner et al., 1983). Some of these assess personality factors thought to be associated with the eating disorders (i.e., Perfectionism, Interpersonal Distrust), and some assess behaviors that are more directly tied to eating disorder symptomatology (i.e., Drive for Thinness, Bulimia). A factor analysis in a nonclinical sample revealed that three of the scales (Drive for Thinness, Bulimia, and Body Dissatisfaction) loaded on the same factor tapping a general concern with shape, weight, and eating, and this factor was recommended as a screening measure of eating disorder symptoms (Welch, Hall, & Walkey, 1988). Accordingly, in this study, the sum of the Drive for Thinness, Body Dissatisfaction, and Bulimia subscales was used to measure severity of eating disorder symptomatology 4. Scores on this composite scale can range from 0 to 69, with higher scores corresponding to higher levels of eating disorder symptomatology. 3 Copies of the SAWBS Inventory can be obtained from the second author at University of British Columbia, Department of Psychology, 2136 West Mall, Vancouver BC, V6T 1Z4, Canada. 4 All analyses were completed with the three EDI scales separately, and an identical pattern of results was obtained.

7 The SAWBS Inventory 11 Beck Depression Inventory (BDI), The BDI (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) is a 21-item self-report questionnaire assessing depressive symptoms. This measure has demonstrated good internal consistency, test-retest reliability, and concurrent validity (Beck, Steer, & Garbin, 1988). Scores can range from 0 to 63, with higher scores corresponding to more depressed mood. General Information Sheet. Designed for this study, this form asked each woman to provide her date of birth, year of study, faculty, major, family situation (two-parent family, one-parent family, step-family, or "other"), parents' occupations and educations, ethnic background, and height and weight. Parents' occupations and educations were used to determine family socioeconomic status (Hollingshead, 1975), and height and weight were used to calculate BMI. Women also rated the extent to which they felt they had an attractive shape (on a 7-point scale ranging from not at all attractive to extremely attractive) and rated how they felt about their weight (on a 7-point scale ranging from extremely overweight to extremely underweight). These variables were used to describe the sample and were explored as possible correlates of SAWBS scores. Stroop Task. Two word cards (18 x 25 cm) and one practice card (9 x 10 cm) were used. The word cards consisted of 12 stimulus words repeated five times. The words were printed in block capitals (4 mm in height) in one of six colors of ink; red, yellow, green, blue, black, or grey. The words were presented in random order in 12 rows of 6. On the shape and weight card, the words were stomach, shape, buttocks, weight, thighs, hips, fat, obese, flabby, thin, firm, and slim. Each word on the shape and weight card was matched to a control word on the basis of first letter, word frequency, and, when possible, word length. The control words were soldier, seven, bribe, welcome, theatre, helm, folk, orchard, fired, thorn, farms, and shoe. The control card was constructed in the same way as the shape and weight card. A practice card was designed for subjects to familiarize themselves with the ink colors and the color-naming task. This was a smaller card in which the capital letter o was printed in the six colors, in random order, in three columns of six. Instructions for the Stroop informed participants that the task involved naming the colors of words each of which was printed in a different colored ink. Subjects were shown the practice card, the experimenter identified the six ink colors, and subjects were instructed to name the colors by placing their finger below each word prior to naming its color. Subjects were then asked to name the colors of the words printed on the word cards as quickly as possible, and to correct any mistakes as they occurred. The experimenter measured the tune taken to color-name the words on each of the word cards using a hand-held stop watch. The order of presentation

8 12 Geller, Johnston, and Madsen of the two word cards was counterbalanced across subjects. The Stroop interference score is the difference between the time taken to color-name the shape and weight words and the time taken to color-name the control words. Word Recognition Task. For this task, the 24 words used in the Stroop task (12 shape and weight, and 12 control) were presented on a sheet of paper, randomly interspersed with 24 distractor words (also 12 shape and weight, 12 control). Each distractor word was matched to a target word on the basis of word length, word meaning and structure, and, when possible, first letter and word frequency. The 12 shape and weight distractor words were bottom, rotund, pelvis, figure, fat, hard, waist, curves, round, tummy, chubby, and trim. The 12 control distractor words were orange, eleven, held, ranch, salute, sailor, clan, sole, discharged, sliver, bait, and movie. Participants were instructed to read through the word list and indicate which words they had seen on the previous two cards (target words). The false alarm effect score was the number of shape and weight distractor words that were incorrectly identified as target words. FAE scores can range from 0 to 12. Short Marlowe-Crowne Social Desirability Scale [M-C 1(10)]. The M-C 1(10) (Strahan & Gerbasi, 1972) is a 10-item, true-false measure of the tendency to respond in a socially sanctioned manner. The scale has shown moderate internal consistency and correlates highly with the longer version of this measure (Strahan & Gerbasi, 1972). Scores on this measure can range from 0 to 10, with higher scores corresponding to a greater tendency to respond in a socially sanctioned manner. Rosenberg Self-Esteem Scale (RSES). The RSES (Rosenberg, 1979) is a self-report questionnaire composed of 10 items that assess global attitudes toward the self. Individuals rate the extent to which statements are descriptive of them on a 5-point Likert scale ranging from not at all descriptive of me to very descriptive of me. The RSES has been shown to have strong construct, convergent, and discriminant validity (Rosenberg, 1979; Wylie, 1989). It correlates with other measures of self-esteem, as well as with peer ratings (Demo, 1985). Scores on the RSES can range from 10 to 50, with higher scores corresponding to higher levels of self-esteem. Health Information Questionnaire (HIQ). This measure was designed for this study to assess the presence and severity of disturbed eating practices. The HIQ is based on DSM-IV eating disorders criteria (APA, 1994) and other self-report surveys developed for assessing eating disorder symptomatology (Greenfeld, Quinlan, Harding, Glass, & Bliss, 1987; Kagan & Squires, 1983; Whitaker et al., 1989). It assesses recent weight history, menstrual history, fear of weight gain, worries about eating habits, and presence and severity of binge eating, perceived lack of control over eating, and six compensatory behaviors (excessive exercise, fasting, use of diet pills, use

9 The SAWBS Inventoiy 13 of diuretics or water pills, vomiting, and use of laxatives). The response scales were designed to assess DSM-IV threshold criteria, as well as subthreshold levels of severity for binge eating and all relevant compensatory behaviors. The HIQ yields a total disturbed eating score ranging from 0 to 69, with higher scores corresponding to more eating disorder symptomatology. Procedure Fifty of the women completed, in the following order, the SAWBS Inventory, the Stroop color-naming task followed by the word-recognition task, the EDI, the BDI, and a General Information Sheet. Although it was anticipated that completing the SAWBS Inventory prior to performing the Stroop task might prime participants for shape and weight words, this order was nevertheless considered more desirable than having SAWBS Inventory responses biased by prior presentation of the Stroop, or indeed of any other measure. Therefore, because it was our goal to minimize bias in responses to the SAWBS Inventory, possible compromise to the sensitivity of the Stroop task was tolerated. Based on an a priori decision that a sample of 20 would be sufficient to test the stability of the SAWBS Inventory, women were invited to return for retest 1 week later until this number was reached. Forty women were asked back in order to get a sample of 24 (60%) to agree to return. These 24 women completed the SAWBS Inventory a second time, as well as the additional questionnaire packet described below. Sixty other women participated on only one occasion and did not complete the Stroop or word recognition task. These women completed a questionnaire packet that included, in the following order, a General Information Sheet, the SAWBS Inventory, the BDI, the EDI, the M-C 1(10), the RSES, and the HIQ. As part of a larger study, these subjects also completed five other measures assessing health-related perceptions, beliefs, and behaviors. Two of these measures were placed at the end of the questionnaire packet and therefore did not influence responses to the other questionnaires. The other three measures were completed after the SAWBS Inventory. In summary, a sample of 50 women completed the SAWBS Inventory, the cognitive tasks, BDI, and EDI. Twenty-four provided test-retest data for the SAWBS. Eighty-four women completed the SAWBS, EDI, BDI, social desirability measure, and the HIQ (24 of these women completed these measures as part of the SAWBS retest).

10 14 Geller, Johnston, and Madsen RESULTS Description of Sample In the 84 women who completed the larger questionnaire package, EDI Drive for Thinness, Bulimia, and Body Dissatisfation mean subscale scores were within the nonclinical range, and were similar to values reported in other university samples (e.g., Klemchuk et al., 1990; Vanderheyden et al., 1988). However, also similar to previous research (e.g., Klemchuk et al., 1990), a wide range of eating disorder symptoms was reported, and a clinical-level pathology was evident in some women. For instance, on the HIQ, 20% of the women stated that they felt afraid to eat for fear of weight gain "often" or "all the time," and 12% indicated that they binge-ate two or more times a week. The percentage of women who reported that they had fasted, used diuretics, vomited, or used laxatives at least once per month in order to get rid of unwanted calories was 7%, 1%, 6%, and 5%, respectively. Using the HIQ items that assessed specific DSM-IV (APA, 1994) eating disorder criteria, three individuals from the sample met the behavioral criteria for bulimia nervosa (i.e., the occurrence of both binge-eating episodes and compensatory behaviors two or more times per week), and were identified as probable cases. An additional seven participants met criteria for one behavioral category (e.g., binge eating), but were subthreshold on one or more other categories (e.g., used self-induced vomiting, but only once per week). These individuals were identified as possible cases. BMI scores in this sample ranged from 17.0 to 40.5, with a mean of 21.7 (SD = 3.1), which is within the range associated with the lowest risk of illness for most people (Health and Welfare Canada, 1988). Table I provides means and standard deviations on the HIQ, as well as of participants' Perceived Weight and Shape, BMI, EDI composite, BDI, and RSES scores. Properties of the SAWBS Inventory In dividing the self-esteem circle, the 110 women in this study used between two and eight pieces (attributes), with a mean of 5.5 pieces. The SAWBS score, or the angle of the shape and weight piece, ranged from 0 to 230, with a mean angle of 58. Looking at the rankings of shape and weight compared with other attributes, 9 (11.9%) of the 84 women ranked shape and weight most important to their feelings of self-worth, 16 (19.0%) ranked shape and weight second, 19 (22.6%) third, 8 (9.5%) fourth, 8 (9.5%) fifth, 6 (7.1%) sixth, 4 (4.8%) seventh, and 14 (16.7%) women did

11 The SAWBS Inventory 15 Table L Means and Standard Deviations for Aspects of Shape and Weight Based Self-Esteem (SAWBS) Scores, the Health Information Questionnaire (HIQ), the Eating Disorders Inventory (EDI), the Beck Depression Inventory (BDI), and the Rosenberg Self-Esteem Scale (RSES) SAWBS score (angle) Total number of pieces used HIQ Total a Weight Perception score* Shape Perception score c Body Mass Index EDI Composite score BDI RSES Mean Standard deviation a Score can range from 0 to 69, with higher scores indicating more disturbed eating practices. b Score can range from 1 to 7, with higher scores indicating seeing oneself as more underweight, and lower scores indicating seeing oneself as being more overweight. A score of 4 indicates seeing one's weight as just right. c Score can range from 1 to 7, with higher scores indicating seeing oneself as having a more attractive shape. not asign any ranking to shape and weight. Means and standard deviations of SAWBS Inventory characteristics are presented in Table I. Stability The 1-week test-retest correlation of the SAWBS score as provided by the sample of 24 women, was.81, p <.001. Concurrent Validity Relationship Between SAWBS, Perceptions of Shape and Weight, and Eating Disorder Symptoms. In the sample of 84 women who completed the questionnaire packet, SAWBS scores were negatively related to the extent to which women perceived their shapes to be attractive, and to the extent to which they perceived themselves to be underweight. SAWBS scores were correlated with the two measures of eating symptomatology, the EDI composite score and the HIQ symptom score. As predicted, both correlations were statistically significant. Table II illustrates

12 16 Geller, Johnston, and Madsen

13 The SAWBS Inventory 17 correlations among SAWBS and Perceived Shape, Perceived Weight, HIQ, and EDI Composite scores. To determine whether SAWBS scores contributed significant unique variance to eating symptom scores after the effects of BMI, depression, and global self-esteem were controlled, two regression analyses were performed in which the dependent variables were the EDI composite score and the HIQ symptom score, respectively. Table III displays the standardized regression coefficients (b) for the full prediction models, and the R 2 change for each step when entered into the equations. The full models were both statistically significant for EDI scores, F(4, 79) = 25.61, p <.001, and for the HIQ scores, F(4, 79) = 17.49, p <.001. In both cases, even after the variance accounted for by BMI, depression, and self-esteem was controlled, SAWBS scores accounted for additional statistically significant variance in EDI (10%) and HIQ (12%) symptom scores. Stroop Task. Most (90%) of the 50 women who completed the Stroop task took longer to color-name the words on the shape and weight card than they did to color-name the words on the control card. The average interference score (shape and weight card time minus control card time) was 5.5s (SD = 4.7). A Mest examining whether Stroop color-naming interference was related to the order of presentation of the Stroop cards (shape and weight first vs. control first) was not statistically significant (p =.40). Table III. Regression Analysis with Body Mass Index (BMI), Beck Depression Inventory (BDI), and Rosenberg Self-Esteem Scale (RSES) Scores as Independent Predictors of Eating Disorder Inventory (EDI) Composite and Health Information Questionnaire (HIQ) Total Symptom Scores Predictor variable EDI Composite score b R 2 incremental HIQ Total Symptom score b R 2 incremental Step 1 Body Mass Index BDI score RSES score Step 2 SAWBS score a *.10 b a SAWBS = Shape and Weight Based Self-Esteem Inventory. b p < b.12 b

14 18 Geller, Johnston, and Madsen Contrary to prediction, the correlation between SAWBS and Stroop color-naming interference scores was not statistically significant, r(50) =.13, p >.05. However, Stroop scores were related to the number of shape and weight "hits" in the word recognition task (the correct identification of shape and weight target words), r(50) =.38, p <.01. Therefore, not unexpectedly, individuals whose color-naming times were most delayed by the shape and weight card, and who consequently spent more time looking at the shape and weight words, were able to recall more of those words than did individuals whose color-naming was not as delayed by the shape and weight target words. Word Recognition Task. As predicted, the correlation between SAWBS scores and the FAE was positive, and statistically significant, r(50) =.44, p <.001. As expected, SAWBS scores were not associated with the false recognition of control words, r(50) =.05, p =.36. Because previous literature has shown measures of cognitive schemata to be associated with depression (e.g., Cooper et al., 1992), and because there was a statistically significant relationship between BDI scores and the FAE in this study [r(50) =.40, p <.01], a multiple-regression analysis was performed to determine whether SAWBS scores contributed unique variance to FAE scores after the effect of depression was controlled. In this analysis, the dependent variable was the FAE score, BDI scores were entered as a first predictor, and SAWBS scores were entered second. The overall model was statistically significant, F(2, 47) = 9.37, p <.001. BDI scores accounted for 16% of the variance in FAE scores, F(1, 47) = 9.02, p <.01, and SAWBS scores accounted for a further 13% of the variance in FAE scores, F(1, 47) = 9.37, p <.001. Therefore, the relationship between SAWBS scores and shape and weight false alarms was statistically significant, even after the effect of depression was controlled. Looking at the relationship between the two measures of cognitive schemata, Stroop interference scores were unrelated to participants' incorrect identification of shape and weight target words in the word recognition task (FAE), r(50) = -.06, p >.05. Predictive Validity To further test the relationship between SAWBS scores and eating disorder symptomatology in the group of 84 women, the three probable and seven possible eating disorder cases (identified on the basis of the HIQ) were combined into one group. First, the validity of this classification based on the HIQ was verified by comparing the cases with the remainder of the sample on the EDI composite score. This t-test was statistically significant, t(83) = 6.40, p <.001, confirming that individuals identified as

15 The SAWBS Inventory 19 probable and possible cases (M = 49.50) scored higher on eating disorders symptomatology than the remainder of the sample (M = 18.68) as assessed by the EDI Composite score. A t-test comparing SAWBS scores in the probable and possible eating disorder case group with women with no suspected eating disorder was also statistically significant [t(83) = 3.10, p <.01], indicating that SAWBS scores were higher in the probable and possible eating disorder group (M = 99.70) than in the remainder of the sample (M = 56.19). Discriminant Validity Non-Eating Disorder Correlates of SAWBS. In the sample of 84 women, SAWBS scores were not significantly related to the tendency to respond in a socially desirable manner, to socioeconomic status, or to BMI (see Table II). SAWBS scores were negatively correlated with RSES scores, indicating that the more self-esteem was based on shape and weight, the lower were global feelings of self-worth. As noted earlier, the relationship between SAWBS and BDI scores was positive, and statistically significant. That is, the tendency to base one's self-esteem on shape and weight was related to higher depressed mood scores. DISCUSSION Growing interest in shape- and weight-based self-esteem has created the need for a specific measure of this construct. The purpose of this study was to establish the psychometric properties of the SAWBS Inventory, a newly developed measure of shape- and weight-based self-esteem that is easily administered and scored, and requires less than 10 min to complete. Findings from this study, using a nonclinical sample of young women, indicated that SAWBS scores were stable over 1 week and correlated moderately with a measure of shape and weight schemata, the false alarm effect. SAWBS scores did not correlate significantly with Stroop interference times using shape and weight words. Consistent with the idea that a unification of views of the self with beliefs about shape and weight characterizes the eating disorders, SAWBS scores were moderately correlated with two measures of eating disorder symptomatology. In further support of the view that individuals who have eating disorders can be distinguished by the degree of influence (positive or negative) that shape and weight have on their feelings of self-worth, SAWBS scores discriminated between individuals who reported few or no disturbed eating symptoms and possible and probable

16 20 Geller, Johnston, and Madsen eating disorder cases. In addition, SAWBS scores were not related to family socioeconomic status or to the tendency to respond in a socially sanctioned manner. Given these characteristics, the SAWBS Inventory shows promise as a screening instrument to identify individuals at high-risk to develop an eating disorder, and as a tool to measure changes in shape- and weightbased self-esteem in the course of treatment. The correlation between SAWBS scores and the FAE was consistent with the hypothesis that shape- and weight-based self-esteem may be associated with a shape and weight schema. In this study, individuals who based a greater portion of their self-worth on shape and weight were more apt to confuse novel shape and weight words with ones they had already seen, whereas no such effect occured with neutral control words. As previous research has shown false alarms to increase with the degree to which the words refer to important aspects of the self-concept (Rogers et al., 1987), at the very least, these findings suggest that SAWBS scores tapped a salient dimension of beliefs about the self. The fact that SAWBS scores were associated with the FAE even after the effects due to depressed mood were controlled points to the specificity of the SAWBS Inventory in measuring a central cognitive dimension of eating disorders (shape- and weightbased self-esteem), independent of depressed mood. This finding is in contrast to previous research (Cooper et al., 1992) in which the association between general eating disorder attitudes and behaviors and a shape, weight, and eating schema assessed using a Stroop paradigm was reduced to nonsignificance when the effect of depression was controlled. A number of differences between the methodology used in the present study and that used in the Cooper et al. study may account for this difference in results, including, most importantly, use of the SAWBS Inventory, a specific measure of shape- and weight-based self-esteem instead of the more general Eating Attitudes Test (Garner & Garfinkel, 1979), and use of the FAE to assess cognitive schemata, instead of the Stroop task. Future investigation is clearly warranted to clarify the possible greater specificity of the measures used in the present study. The failure to detect a relationship between SAWBS and Stroop scores, the second measure of shape and weight schemata used in this study, was contrary to prediction. Follow-up inspection of scores revealed that this failure was not due to a problem in the score distribution, as Stroop interference scores were approximately normally distributed and had an adequate range (20.5) and variance (SD = 4.7). In addition, as noted earlier, the significant positive relationship between Stroop scores and shape and weight hits in the word recognition task suggests that the problem was not one of inaccuracies in the recording of participants' response time, as individuals who took longer to color-name shape and weight

17 The SAWBS Inventory 21 words correctly identified more of those words in the subsequent word recognition task. Instead, one possible explanation for the lack of relationship between SAWBS and Stroop scores may have been our failure to make control words content consistent. That is, although the control words in this study were carefully controlled with respect to word frequency and length, they did not belong to a common content area. Green and Rogers (1993) reported that, when they controlled for this oft-seen methodological shortcoming in eating disorder Stroop research, Stroop sensitivity increased. Performance on the Stroop task may also have been affected by participants' awareness of the purpose of the study. As noted earlier, the Stroop has been shown to be susceptible to extraneous influences (e.g., Warren, 1974), and it is possible that some individuals guessed that the focus of the research pertained to shape and weight. Finally, the inconsistencies reported earlier in eating disorder Stroop research (e.g., Cooper & Fairburn, 1992; Green & Rogers, 1993) suggest that the Stroop may simply not be a good measure of the construct of interest. Possibly, the FAE task is a more promising measure of eating disorder schemata for future research. The relationship between SAWBS and eating disorder symptom scores supports the view that shape- and weight-based self-esteem is a central cognitive feature of the eating disorders. The independent relationship between shape- and weight-based self-esteem and eating disorder symptomatology was further demonstrated in regression analyses, in which SAWBS scores accounted for unique variance in eating symptom scores, even after the effects of depression, global self-esteem, and BMI were controlled. It is noteworthy that SAWBS scores were also negatively correlated with selfesteem and positively correlated with depression. These findings suggest that, in addition to being associated with eating disorder symptomatology, basing one's self-worth largely on shape and weight is associated with poorer global feelings about the self, and a greater likelihood of depression. It was also noteworthy that, while SAWBS scores were not associated with actual weight and shape (as assessed by the BMI), they were associated with individuals' perceptions of their shapes and weights. A limitation of the present research is the self-report nature of the SAWBS Inventory, and the implicit assumption that individuals are able to accurately report on the extent to which shape and weight affect their selfesteem. Although this limitation cannot be discounted, the fact that SAWBS scores correlated with the FAE, a measure of low face validity, and were unaffected by social desirability scores, suggests that individuals may indeed have such awareness, and are able to accurately report on a structural aspect of their self-esteem, independent of their need to present themselves in a socially sanctioned manner. A second possible limitation of the SAWBS Inventory is that it is based on a single score. As the SAWBS

18 22 Geller, Johnston, and Madsen Inventory requires the relative importance of a number of selected attributes to feelings of self-worth to be assessed, this is simply an intrinsic property of the measure, and unfortunately prevents the calculation of its internal consistency. However, it might be argued that participants' prior selection and rank-ordering of attributes increase the validity of the SAWBS score, in that the final SAWBS score is the culmination of a number of preliminary steps. In addition, the ability of the SAWBS Inventory to distill two complex constructs (self-esteem and shape and weight concerns) into a single score is as much a strength of this measure as it is a limitation. In interpreting our findings regarding the relationship between the SAWBS Inventory and one of our measures of shape and weight schemata (the FAE), we also wish to note that the assessment of schemata is a difficult and controversial area (e.g., Higgins & Bargh, 1987), and alternative interpretations are possible. Nevertheless, our data are consistent with our hypotheses, and we find the schemata to be a useful heuristic in making sense of our findings. Perhaps a more serious shortcoming of the present research was the short test-retest interval of only 1 week. Future research is clearly warranted to determine whether SAWBS scores are stable over longer periods of time. In addition, although use of a female undergraduate sample was considered appropriate as a first step toward the validation of this new instrument, future research should examine SAWBS scores in women who have eating disorders, and compare their scores with those from psychiatric and nonpsychiatric control groups. Research in our lab is currently under way to examine SAWBS scores in such clinical samples. Finally, although it is tempting to consider the possible causal role of shape- and weightbased self-esteem in the development of eating disorders, the correlational design of this study prevented such causality questions from being addressed. Longitudinal designs offer a necessary and valuable avenue for future research. REFERENCES American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed). Washington DC: Author. Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8, Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, Ben-Tovim, D. I., & Walker, M. K. (1991). Women's body attitudes: A review of measurement techniques. International Journal of Eating Disorders, 10,

19 The SAWBS Inventory 23 Ben-Tovim, D. I., Walker, M. K., Fok, D., & Yap, E. (1989). An adaptation of the Stroop test for measuring shape and food concerns in eating disorders: A quantitative measure of psychopathology? International Journal of Eating Disorders, 8, Cooper, M. J., Anastasiades, P., & Fairburn, C. G. (1992). Selective processing of eating-, shape-, and weight-related words in persons with bulimia nervosa. Journal of Abnormal Psychology, 101, Cooper, P. J., & Fairburn, C. G. (1987). The eating disorder examination: A semi-structured interview for the assessment of the specific psychopathology of the eating disorders. International Journal of Eating Disorders, 6, 1-8. Cooper, M. J., & Fairburn, C. G. (1992). Selective processing of eating, weight and shape related words in patients with eating disorders and dieters. British Journal of Clinical Psychology, 31, Cooper, P. J., & Fairburn, C. G. (1993). Confusion over the core psychopathology of bulimia nervosa. International Journal of Eating Disorders, 13, Demo, D. H. (1985). The measurement of self-esteem: Refining our methods. Journal of Personality and Social Psychology, 48, Fairburn, C. G., & Gamer, D. M. (1988). Diagnostic criteria for anorexia nervosa and bulimia nervosa: The importance of attitudes to shape and weight. In D. M. Garner & P. E. Garfinkel (Eds.). Diagnostic issues in anorexia nervosa and bulimia nervosa (pp ). New York: Brunner/Mazel, pp Garner, D. M., & Garfinkel, P. E. (1979). The Eating Attitudes Test: An index of the symptoms of anorexia nervosa. Psychological Medicine, 9, Garner, D. M., & Olmstead, M. P. (1984). The eating disorder inventory manual. Odessa, FL: Psychological Assessment Resources. Gamer, D. M., Olmstead, M. P., & Polivy, J. (1983). Development and validation of a multidimensional eating disorder inventory for anorexia nervosa and bulimia. International Journal of Eating Disorders, 2, Green, M. W., & Rogers, P. J. (1993). Selective attention to food and body shape words in dieters and restrained nondieters. International Journal of Eating Disorders, 14, Greenfeld, D., Quinlan, D. M., Harding, P., Glass, E., & Bliss, A. (1987). Eating behavior in an adolescent population. International Journal of Eating Disorders, 6, Harter, S. (1993). Visions of self: Beyond the me in the mirror. In J. Jacobs (Ed.). Nebraska Symposium on Motivation: Vol. 40. Perspectives on Motivation (pp ). Lincoln: University of Nebraska Press. Health and Welfare Canada. (1988). Promoting healthy weights: A discussion paper. Ottawa, Ontario: Author (Health Sciences and Promotion Branch). Higgins, T. E., & Bargh, J. A. (1987). Social cognition and social perception. Annual Review of Psychology, 38, Hollingshead, A. B. (1975). Four-Factor Index of Social Status. New Haven, CT: Author. Hsu, L. K. (1982). Is there a disturbance of body image in anorexia nervosa? Journal of Nervous and Mental Disease, 170, Hsu, L. K., & Sobkiewicz, T. A. (1989). Body image disturbance: Time to abandon the concept for eating disorders? International Journal of Eating Disorders, 10, Kagan, D. M., & Squires, R. L. S. (1983). Dieting, compulsive eating, and feelings of failure among adolescents. International Journal of Eating Disorders, 3, Kaplan, S. L., Busner, J., & Pollack, S. (1988). Perceived weight, actual weight, and depressive symptoms in a general adolescent sample. International Journal of Eating Disorders, 7, Kelly, G. A. (1955). The psychology of personal constructs. New York: Norton. Klemchuk, H. P., Hutchinson, C. B., & Frank, R. I. (1990). Body dissatisfaction and eating-related problems on the college campus: Usefulness of the the eating disorders inventory with a nonclinical population. Journal of Counseling Psychology, 37, Klesges, R. C., Beatty, W. W., & Berry, S. L. (1985). Some behavioral, attitudinal, and perceptual correlates of obesity in a university population. International Journal of Eating Disorders, 4,

20 24 Geller, Johnston, and Madsen Markus, H. (1977). Self-scemata and processing information about the self. Journal of Personality and Social Psychology, 35, Marsh, H. W. (1986). Global self-esteem: Its relation to specific facets of self-concept and their importance. Journal of Personality and Social Psychology, 51, Marsh, H. W. (1993). Relations between global and specific domains of the self: The importance of individual importance, certainty, and ideals. Journal of Personality and Social Psychology, 65, Mathews, A., & MacLeod, C. (1985). Selective processing of threat cues in anxiety states. Behavioral Research Therapy, 23, O'Brien, E. J. (1980). The Self-Report Inventory: Development and validation of a multidimensional measure of the self-concept and sources of self-esteem. Unpublished doctoral dissertation, University of Massachusetts at Amherst. Rogers, T. B., Rogers, P. J., & Kuiper, N, A. (1979). Evidence for the self as a cognitive prototype: The "false alarms effect." Personality and Social Psychology Bulletin, 5, Rosenberg, M. (1979). Conceiving the self. New York: Basic Books. Strahan, R., & Gerbasi, K. C. (1972). Short, homogeneous versions of the Marlow-Crowne Social Desirability Scale. Journal of Clinical Psychology, 28, Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18, Vanderheyden, D. A., Fekken, G. C, & Boland, F. J. (1988). Critical variables associated with bingeing and bulimia in a university population: A factor analytic study. International Journal of Eating Disorders, 3, Vitousek, K. B., & Hollon, S. D. (1990). The investigation of schematic content and processing in eating disorders. Cognitive Therapy and Research, 14, Warren, R. E. (1974). Association, directionality, and stimulus encoding. Journal of Experimental Psychology, 102, Welch, G., Hall, A., & Walkey, F. (1988). The factor structure of the Eating Disorders Inventory. Journal of Clinical Psychology, 44, Whitaker, A., Davies, M., Shaffer, D., Johnson, J., Abrams, S., Walsh, B. T., & Kalikow, K. (1989). The struggle to be thin: A survey of anorexic and bulimic symptoms in a non-referred adolescent population. Psychological Medicine, 19, Wylie, R. C. (1989). Measures of self-concept. Lincoln: University of Nebraska Press.