Body Image Treatment for a Community Sample of Obligatory and Nonobligatory Exercisers Jane Ellen Smith,* Brenda L. Wolfe, and Denise E. Laframboise Department of Psychology, University of New Mexico, Albuquerque, New Mexico Accepted 2 July 2001 Abstract: Objective: Cognitive-behavioral therapy (CBT) was used to treat body dissatisfaction in obligatory and nonobligatory exercisers within a community sample of normal weight women. Method: Ninety-four women (36% obligatory exercisers, 64% nonobligatory exercisers) were assigned randomly to CBT or the waiting-list (WL) control group. Results: The hypotheses that obligatory exercisers would show poorer pretreatment body image and greater compulsivity than nonobligatory exercisers were supported partially. The prediction that obligatory exercisers would respond less favorably to treatment was not supported. Overall, CBT participants evidenced signi®cantly better body image outcomes than the WL at posttreatment, but many effects were lost by the follow-up. Discussion: Treatment response is considered in light of the unique characteristics of this ethnically diverse, older community sample when compared with the young students in earlier body image intervention studies. The high rate of physical activity among even the nonobligatory exercisers is highlighted for its mood-regulation properties and its treatment implications. Ó 2001 by John Wiley & Sons, Inc. Int J Eat Disord 30: 375 388, 2001. Key words: body image; obligatory; exercise; compulsive INTRODUCTION Body dissatisfaction has become exceedingly common among the general female population (Rodin, Silberstein, & Striegel-Moore, 1984; Thompson, Heinberg, Altabe, & Tantleff-Dunn, 1999). Importantly, poor body image has been linked to various types of psychological distress (Adams, Katz, Beauchamp, Cohen, & Zavis, 1993; Cash & Pruzinsky, 1990; Elam & Kimbrell, 1995) and to eating pathology (Herzog, Hopkins, & Burns, 1993; Kiemble, Slade, & Dewey, 1987; Rosen, 1992). Six controlled body image treatment studies have targeted body dissatisfaction in non-eating disordered normal weight women. The efficacy of these cognitive-behavioral interventions generally has been supported; most clearly when compared with a waiting-list (WL) condition (Butters Presented at the 9th International Conference on Eating Disorders, New York City, May 7, 2000. *Correspondence to: Jane Ellen Smith, Ph.D., Department of Psychology, University of New Mexico, Logan Hall, Albuquerque, NM 87131. E-mail: janellen@unm.edu Ó 2001 by John Wiley & Sons, Inc. 376 Smith et al. & Cash, 1987) or with minimal treatment (Rosen, Saltzberg, & Srebnik, 1989). In two studies, cognitive-behavioral therapy (CBT) was superior to a control condition, but was only comparable to exercise therapy (Fisher & Thompson, 1994) and was less effective than cognitive therapy (Dworkin & Kerr, 1987). When CBT was contrasted with a modest-contact version of it (Grant & Cash, 1995) or one that added size perception training (Rosen, Cado, Silberg, Srebnik, & Wendt, 1990), clinically relevant within-group changes were found, but there were no between-group effects. Although the use of CBT to treat body dissatisfaction in nonclinical samples appears promising, there are questions still to be addressed. To begin with, it is difficult to draw firm conclusions in several cases because CBT showed no advantage over the active treatment comparison condition (Fisher & Thompson, 1994; Grant & Cash, 1995; Rosen et al., 1990). Moreover, because each study utilized primarily young, Anglo college students, the treatments were never tested with older, heavier, ethnically diverse community participants. Also, the initial incentive for participating in each case was class research credit, and so the generalizability of the results to natural treatment-seeking samples was unknown. Additionally, two of the studies did not include follow-ups (Dworkin & Kerr, 1987; Fisher & Thompson, 1994) and one did not have follow-up data on the control group (Butters & Cash, 1987). Another limitation of previous studies is that only one study monitored weight throughout treatment (Grant & Cash, 1995), a necessary step because weight change can serve as a confound in body image therapy (Cash, 1994a). Further, none of the studies reported the amount of exercise in which participants engaged. This is regrettable because individuals with body dissatisfaction often attempt to alter their size or shape through exercise (Davis, 1997; Eliot & Smith, 2001) and because changes in exercise habits are suspected of contributing to changes in body satisfaction (McAuley, Bane, Rudolph, & Lox, 1995; Tucker & Mortell, 1993). Also relevant is the literature that discovered both positive and negative associations between exercise and body image (Loland, 1998; Sonstroem, Harlow, & Josephs, 1994); equivocal results that were sometimes attributed to the type of exercise or to the reasons motivating it (Brownell, Rodin, & Wilmore, 1992; Davis, Fox, Cowles, Hastings, & Schwass, 1990; Imm & Pruitt, 1991). And yet, these distinctions have not produced consistent findings (Davis, 1997; Eliot & Smith, 2001). Arguably, the most important exercise factor linked with body image is the extent to which it is excessive or compulsive. Terms like obligatory and compulsive exerciser have been adopted to represent those individuals who exercise often and who seem unable to control their need to exercise despite physical injury or social demands (Coen & Ogles, 1993; Pasman & Thompson, 1988; Steffen & Brehm, 1999). A number of conclusions have been drawn about these compulsive exercisers that could be of consequence should they participate in a body image intervention. First, several researchers detected higher levels of disturbed eating in obligatory exercisers than in controls (Pasman & Thompson, 1988; Slay, Hayaki, Napolitano, & Brownell, 1998). When obligatory exercisers were compared with nonobligatory exercisers, they had lower weights (Slay et al., 1998), but greater weight concerns (Davis et al., 1990; Yates, Shisslak, Allender, Crago, & Leehey, 1992), and poorer body images (Davis, Shapiro, Elliott, & Dionne, 1993; Imm & Pruitt, 1991). As far as personality characteristics, obligatory exercisers had elevated levels of trait anxiety, perfectionism, and obsessive-compulsiveness (Coen & Ogles, 1993; Davis et al., 1995; Goldfarb & Plante, 1984; Thompson & Sherman, 1999; Yates, 1991). Given this profile, one might expect that obligatory exercisers who receive body image therapy are at a disadvantage compared with nonobligatory exercisers and, therefore, may require a more specialized or extended treatment. Conceivably, their presence to Obligatory Exercise and Body Image 377 varying degrees in such programs could explain some of the variability in treatment response (Grant & Cash, 1995). Nevertheless, this had not been tested. The current study addressed this issue by recruiting normal weight women for a CBT program for body dissatisfaction and by categorizing them as obligatory or nonobligatory exercisers on the basis of their Obligatory Exercise Questionnaire (OEQ) scores. It was first predicted that the obligatory exercisers would enter treatment with a significantly more disturbed body image than the nonobligatory exercisers. Second, obligatory exercisers also were expected to score significantly higher than nonobligatory exercisers on the Maudsley ObsessionalCompulsive Inventory (MOCI) at pretreatment. High-level exercisers had scored significantly higher than moderate exercisers on the Symptom Checklist-90 obsessive-compulsive scale (Davis et al., 1995), but this construct had not been tested with OEQ-defined obligatory exercisers. Third, it was hypothesized that the nonobligatory exercisers would have significantly better body image treatment outcomes than the obligatory exercisers. Finally, it was predicted that the treated (CBT) participants overall in this diverse community sample would have significantly better outcomes than the WL control group. METHOD Participants Individuals were recruited via flyers posted throughout Albuquerque, NM, announcing a free body image program for normal weight women who were dissatisfied with their bodies. A total of 308 telephone screens were conducted, which resulted in excluding individuals who either had an eating disorder within the last 5 years (n = 19), were receiving psychotherapy (n = 11), were on psychotropic medication (n = 10), were outside the normal weight range (20%) for their height (n = 23), or had unresolvable schedule con¯icts (n = 12). Another 76 individuals were no longer interested upon hearing that it was not a weight-loss program. The Multidimensional Body-Self Relations Questionnaire (MBSRQ; Brown, Cash, & Mikulka, 1990; Cash, 1994b) and the OEQ (Pasman & Thompson, 1988) were mailed to the 157 remaining potential participants. Six of them failed to return the questionnaires. The 31 women who scored above 23 on the MBSRQ Appearance Evaluation scale were excluded, as these scores are not considered indicative of negative body image (Cash, 1995a). Excluded individuals were offered treatment at a later date. Respondents who scored above 50 on the OEQ were classified as obligatory exercisers (Coen & Ogles, 1993; Pasman & Thompson, 1988) and those scoring 50 or below were designated as nonobligatory exercisers. Pretreatment assessments were scheduled with the 120 remaining individuals, but 26 women did not attend. Among the 94 who did, 34 (36%) were obligatory and 60 (64%) were nonobligatory exercisers. Random assignment to treatment condition resulted in 46 participants in the CBT group (18 obligatory exercisers, 28 nonobligatory exercisers) and 48 in the WL (16 obligatory exercisers, 32 nonobligatory exercisers). On average, participants were 36.66 years of age (SD = 12.08, range = 17 69) and had 0.85 children (SD = 1.23, range = 0 5). They had an average height of 64.02 in. (SD = 2.49, range = 58 69) and a weight of 139.89 lb (SD = 18.97, range = 95 190). Their mean body mass index (BMI: kg/m2) was 23.87 (SD = 2.74, range = 18.12 29.82). Regarding marital status, 43% of the participants were single, 40% were married, and 17% were divorced, separated, or widowed. The ethnic breakdown was 59% Anglo (n = 56), 28% Hispanic (n = 26), and 13% ``Other'' (n = 12). The most common occupations by category (Hollingshead, 1965) were business manager/lesser professional (28%), administrative per- 378 Smith et al. sonnel/minor professional (16%), and clerical/sales (13%). Another 24% were classi®ed as students or unemployed. Materials In addition to its screening function, the Appearance Evaluation scale and the four remaining scales of the MBSRQ were used as dependent measures of body satisfaction. These scales included Appearance Orientation (efforts to maintain or improve appearance), Body Areas Satisfaction (satisfaction with specific body parts), Overweight Preoccupation (fat anxiety and weight vigilance), and Self-Classified Weight (weight perception). This 69-item instrument has excellent psychometric properties (Brown et al., 1990; Fisher & Thompson, 1994; Keeton, Cash, & Brown, 1990), including high internal consistency (.75 .91) and test-retest reliability (.78 .94). The Adjustable Light Beam Apparatus (ALBA; Thompson & Spana, 1988; Thompson & Thompson, 1986) was used as a measure of body satisfaction for a specific area that is viewed notoriously as problematic by women, that is, the hip and thigh region (Bailey, Goldberg, Swap, Chomitz, & Houser, 1990). Participants adjusted the width of horizontal beams of light projected onto a wall to indicate the width of these areas. Body dissatisfaction was measured by three scores: (1) Feel-Ideal = the discrepancy between how wide participants felt they were and how wide they ideally wanted to be, (2) Feel-Think = the discrepancy between how wide they felt they were and how wide they thought they actually were, and (3) Think-Ideal = the discrepancy between how wide they thought they were and how wide they ideally wanted to be. The ALBA has a test-retest reliability of .75 and an internal consistency of .83 (Thompson & Spana, 1988). The OEQ (Pasman & Thompson, 1988) was used to determine initial exercise status and to track this attitude/behavior over time. This 20-item scale has an internal consistency alpha value of .96, a test-retest reliability of .96, and has been shown to successfully discriminate obligatory exercisers from controls (Coen & Ogles, 1993; Pasman & Thompson, 1988). The Exercise Involvement Questionnaire was introduced (Eliot & Smith, 2001) as a means to record the frequency and duration of all types of physical activity, such as gardening and dancing, as well as formal exercise. It was shown to have good test-retest reliability (r = .65, p < .0001), convergent validity with the OEQ (r = .51, p < .0001), and discriminant validity with the MOCI cleaning scale (p > .55). Finally, the MOCI (Hodgson & Rachman, 1977), a general measure of obsessive-compulsive behaviors, was administered as well. This instrument is a 30-item true-false scale whose total score correlates signi®cantly with other measures of obsessive-compulsiveness, and which has high testretest reliability (Hodgson & Rachman, 1977). Of particular interest was the MOCI's checking scale, because obligatory exercisers were expected to share certain characteristics with the nonclinical ``checkers'' who had been distinguished from controls on this scale (Frost, Sher, & Green, 1986; Sher, Mann, & Frost, 1984). These characteristics included exaggerated interpersonal fears and general anxiety (De Coverly Veale, 1987; Yates, 1991). Procedure Pretreatment Assessment After signing the consent form, participants were given the ALBA. The order of their estimates of how wide they judged their two body parts to be in terms of their actual Obligatory Exercise and Body Image 379 perception (Think), their affective experience (Feel), and their ideal (Ideal) was counterbalanced. Next, the Exercise Involvement Questionnaire was administered, followed by measurements of participants' height and weight, and their body part widths using a caliper. The assessment was finished at an athletic field, where participants were timed while walking a mile as quickly as possible per the Rockport Fitness Walking Test (Kline et al., 1987). Treatment Participants were randomly assigned to the CBT or the WL group. Six CBT groups were conducted at the university's clinic over the course of 2 years, with 6 10 members per group. The 1 hr sessions ran for 8 weeks. Two experienced, advanced clinical psychology graduate students served as therapists. Their work was observed and supervised by the first author. Therapists were not informed about participants' exercise status. The CBT program was modeled closely after Cash's program (Cash, 1995b), with an emphasis on clarifying cognitive distortions and modifying habitual behaviors that triggered negative body image experiences. Individuals assigned to WL had treatment delayed until after the final assessment. Posttreatment and Follow-Up Assessments The posttreatment assessment was conducted when the 8-week therapy phase was completed. The follow-up occurred 2 months later. Except for the MOCI, all instruments were repeated at each assessment. Upon completing the follow-up, individuals who had given a $20 deposit initially were returned their deposit.1 RESULTS Pretreatment Characteristics There were no significant pretreatment demographic differences between the CBT and WL groups. As far as the dependent measures, separate 2 (CBT, WL) ´ 2 (obligatory exercisers, nonobligatory exercisers) analyses of variance (ANOVAs) were used for the MBSRQ scales, as they measure unique constructs (Brown et al., 1990; Smith, Thompson, Raczynski, & Hilner, 1999). A 2 ´ 2 repeated measures ANOVA was used for the six ALBA discrepancy scores, due to the similar nature of these variables. A significant pretreatment difference was detected for the treatment factor on only one variable, the MBSRQ Overweight Preoccupation scale. The CBT group was more weight preoccupied than the WL, F (1, 90) = 4.31, p = .04. This was addressed later in the treatment outcome analyses, as analyses of covariance (ANCOVAs) were used, with pretreatment values and current BMI as covariates. 1 A deposit was not collected initially so as not to exclude participants for ®nancial reasons. A $20 deposit was introduced for the ®nal 34 participants (36% of the total sample) because it became apparent that many participants were not attending follow-up assessments. A chi-square test showed that signi®cantly more participants completed both the posttreatment and follow-up assessments if they had been required to give the deposit (85.3%) than if they had not (48.3%), v2 (df = 1) = 12.55, p < .001. Causality cannot be assumed because participants were not randomly assigned to the deposit/no deposit condition. Importantly, when those who paid the deposit were compared on demographics and pretreatment dependent variables with those who had not paid, no signi®cant differences were found. Therefore, these groups were combined for all analyses. Later tests demonstrated that there were no group differences in outcome. 380 Smith et al. Table 1. Comparison of demographics and pretreatment characteristics by exercise status Exercise Status Variable Age (years) Children (number) Height (in.) Weight (lb) Body mass index Caliper: hips (cm) Caliper: thighs (cm) Marital status (%) Married Singlea Ethnicity (%) Anglo Hispanic Other Occupation (%)b Professional Skilled Unskilled Student/unemployed OEQ Walk time (s) MOCI Checking Total Exercise hours (per week) Exercise activities (per week) Obligatory (N = 34) M 35.61 0.85 64.06 134.71 22.96 34.52 30.20 (SD) (11.70) (1.20) (2.50) (15.71) (2.22) (2.84) (3.17) Nonobligatory (N = 60) M (SD) 37.25 0.85 63.98 142.84 24.39 35.93 32.10 (12.35) (1.26) (2.50) (20.12) (2.88) (2.88) (3.14) Value Probability 0.43 0.00 0.04 3.97 6.33 4.85 7.74 0.20 ns ns ns .05 .01 .03 .007 ns 0.34 ns 1.04 ns 35.29 64.71 40.00 60.00 55.88 29.41 14.71 61.67 26.67 11.67 50.00 14.71 8.82 26.47 56.67 855.75 (5.01) (120.45) 45.00 21.67 11.67 21.67 41.47 919.51 (6.09) (100.57) 147.05 6.79 .0001 .01 2.19 10.71 9.82 13.03 (2.30) (5.72) (5.30) (6.84) 1.32 8.67 4.90 6.87 (1.58) (4.42) (4.00) (4.72) 4.70 3.53 25.15 25.52 .03 .06 .0001 .0001 Note: OEQ = Obligatory Exercise Questionnaire; MOCI = Maudsley Obsessional-Compulsive Inventory. a Single includes never married, divorced, separated, and widowed. b Occupation is collapsed across Hollingshead (1965) categories: professional (categories 1 3), skilled (4 6), unskilled (7 8), and student/unemployed (9). As expected for the second factor, significant pretreatment differences between obligatory and nonobligatory exercisers were found using 2 ´ 2 ANOVAs, beginning with the obligatory group scoring higher than the nonobligatory group on the OEQ screening instrument, p < .0001 (see Table 1 for all Ms, SDs). Additionally, obligatory exercisers reported signi®cantly greater time devoted to physical activity weekly (M = 9.82 hr, SD = 5.30) than nonobligatory exercisers (M = 4.90 hr, SD = 4.00), as well as more activities (ps < .0001). Obligatory exercisers also completed the 1-mile walk faster than did the nonobligatory group (p = .01). For demographics, obligatory exercisers had lower BMIs than the nonobligatory exercisers (p = .01) and smaller caliper measurements for hips (p = .03) and thighs (p = .007). As predicted, obligatory exercisers also scored signi®cantly higher than the nonobligatory group on the MOCI checking scale (p = .03). There was a trend for MOCI total score differences as well (p = .06). Several predicted differences were found on the pretreatment body image variables (Table 2). Obligatory exercisers showed more fat anxiety than nonobligatory exercisers on the MBSRQ Overweight Preoccupation scale (p = .05) and more overall hip and thigh dissatisfaction than the nonobligatory group on the repeated measures ANOVA for ALBA discrepancy scores, F (1, 89) = 6.15, p < .02. Individual 2 ´ 2 ALBA ANOVAs Obligatory Exercise and Body Image 381 Table 2. Pretreatment MBSRQ and ALBA means (SDs) by exercise status Exercise Status Variable MBSRQ Appearance Evaluationa Appearance Orientation Body Areas Satisfactiona Overweight Preoccupation Self-Classi®ed Weight ALBA Feel-Ideal: hips Feel-Ideal: thighs Feel-Think: hips Feel-Think: thighs Think-Ideal: hips Think-Ideal: thighs Obligatory (N = 34) M (SD) Nonobligatory (N = 60) M (SD) F Value Probability 16.41 47.53 20.88 14.24 7.21 (4.24) (8.50) (4.77) (3.21) (0.73) 16.57 46.33 20.73 12.97 7.42 (3.72) (7.41) (4.30) (2.58) (1.08) 0.03 0.41 0.01 3.95 1.02 ns ns ns .05 ns 14.33 15.36 2.90 3.21 12.14 12.15 (9.67) (10.97) (5.87) (5.50) (9.38) (9.85) 10.51 10.44 1.71 0.75 8.81 9.69 (9.75) (8.11) (8.10) (5.37) (6.65) (8.26) 3.53 6.14 0.62 4.49 4.04 1.64 .06 .02 ns .04 .05 ns Note: MBSRQ = Multidimensional Body-Self Relations Questionnaire; ALBA = Adjustable Light Beam Apparatus. The df for the MBSRQ and individual ALBA exercise status effects are 1,90. ALBA: Feel-Think = the discrepancy between how large a participant feels and how large she thinks she is. Feel-Ideal = the discrepancy between how large a participant feels and how large she ideally would like to be. Think-Ideal = the discrepancy between how large a participant thinks she is and how large she ideally would like to be. Larger discrepancies indicate greater body dissatisfaction. a For these two MBSRQ scales, higher values signify better body image. For the remaining three scales, lower values represent better body image. showed signi®cant differences on two measures for thighs (Feel-Ideal, p = .02, and FeelThink, p = .04) and on one for hips (Think-Ideal, p = .05). Treatment Outcome Between Groups Drop-outs and Assessment Rates Treatment (CBT) drop-outs were defined as women who attended fewer than one half of the eight sessions. Among these 46 treated participants, 9 dropped out (19.6%; 2 obligatory and 7 nonobligatory exercisers). This difference was not significant. Nor were significant differences found when demographics and pretreatment values were contrasted between the drop-outs and the completers. Overall, an average of 5.67 (SD = 2.28) sessions were attended. There was no difference in attendance between the exercise conditions. Furthermore, there were no differences in drop-out rates, attendance, or treatment outcome when the two therapists' participants were compared or when ethnicity was examined. For the total sample, posttreatment and follow-up rates varied slightly by instrument, due to some incomplete MBSRQ data. Posttreatment MBSRQ data were collected for 68% of the original 94 participants (n = 64) and follow-ups were obtained for 61% (n = 57). For the ALBA, 74% of the total sample completed the posttreatment (n = 70) and 66% ®nished the follow-up (n = 62). There were no differences in assessment completion rates across treatment conditions or exercise status. In comparing pretreatment values on all variables for the participants who had completed assessments with those who had not, a signi®cant difference was found on one instrument, the MOCI. The women lost to attrition scored higher on both the MOCI total score, t(86) = 2.63, p < .02, and on the checking scale, t(18.3) = 2.54, p < .03. 382 Smith et al. Table 3. MBSRQ outcome means and SDs by treatment condition Treatment Variable M Control (SD) M (SD) F Value Probability a Appearance Evaluation Pre Post Follow-Up Appearance Orientation Pre Post Follow-Up Body Areas Satisfactiona Pre Post Follow-Up Overweight Preoccupation Pre Post Follow-Up Self-Classi®ed Weight Pre Post Follow-up 16.46 22.83 21.71 (3.64) (4.05) (4.42) 16.56 18.32 18.77 (4.15) (5.05) (4.59) 24.74 13.52 47.74 43.77 43.94 (7.27) (7.45) (7.63) 45.83 45.26 42.35 (8.24) (8.34) (8.23) 4.42 0.00 21.18 25.13 25.35 (4.27) (4.32) (3.67) 20.42 21.74 22.81 (4.62) (5.10) (3.89) 16.50 15.03 14.04 12.43 12.23 (3.03) (3.06) (3.00) 12.83 12.41 11.77 (2.62) (3.21) (2.80) 0.68 0.47 ns ns 7.30 7.03 7.06 (0.92) (1.13) (0.93) 7.38 7.44 7.12 (1.02) (0.99) (0.91) 4.98 0.42 .03 ns .0001 .001 .04 ns .0001 .001 Note: MBSRQ = Multidimensional Body-Self Relations Questionnaire. Only the unadjusted means are presented. Samples sizes differ for pretreatment (n = 94), posttreatment (n = 64), and follow-up (n = 57). Pretreatment means (SDs) are included for comparison purposes. The df for the MBSRQ treatment effects are 1,58 at posttreatment and 1,51 at follow-up. a For these two MBSRQ scales, higher values signify better body image. For the remaining three scales, lower values represent better body image. MBSRQ Using separate 2 ´ 2 ANCOVAs, a main effect for treatment condition that favored CBT over WL was found at posttreatment for four of the five scales (Table 3). The strongest effects were for Appearance Evaluation and Body Areas Satisfaction (ps < .0001), which measure the affective component of body image. These same scales were the only ones for which signi®cant group differences were detected at follow-up (ps < .001). Posttreatment differences also were found for Appearance Orientation (p < .04), which measures cognitive and behavioral investment in appearance, and for Self-Classi®ed Weight (p < .03), which measures weight perception.2 There was neither a main effect of exercise status nor any significant interaction at either posttreatment or follow-up. Planned comparisons tested specifically the prediction that treated (CBT) nonobligatory exercisers would respond better than the treated obligatory exercisers. The prediction was not supported.3 2 Intention-to-treat analyses were run using pretreatment values for missing data. Only the weakest signi®cant ®nding from the treated analyses was lost (Appearance Orientation). 3 Planned comparisons were then conducted between the obligatory exercisers in the CBT and WL conditions to see if treatment appeared bene®cial for them as a subgroup. The obligatory CBT group showed signi®cantly better body satisfaction than the obligatory WL at both posttreatment and follow-up on the two MBSRQ scales that had shown the most robust effects overall: Appearance Evaluation (post), F (1, 58) = 8.26, p = .006; (followup), F (1, 51) = 5.76, p = .02; Body Areas Satisfaction (post), F (1, 58) = 6.06, p = .02; (follow-up), F (1,51) = 4.13, p = .05. When similar planned comparisons were run for the ALBA measures, only trends were detected that favored the obligatory CBT participants over the obligatory WL group. Obligatory Exercise and Body Image 383 Table 4. ALBA outcome means and SDs by treatment condition Treatment Variable Feel-Ideal: hips Pre Post Follow-Up Feel-Ideal: thighs Pre Post Follow-Up Feel-Think: hips Pre Post Follow-Up Feel-Think: thighs Pre Post Follow-Up Think-Ideal: hips Pre Post Follow-UP Think-Ideal: thighs Pre Post Follow-Up Control M (SD) M (SD) F Value Probability 10.40 6.55 5.51 (8.53) (6.70) (7.29) 13.24 11.88 10.61 (10.83) (8.33) (9.21) 6.79 .01 12.37 8.28 7.64 (9.77) (7.85) (7.66) 12.08 12.00 11.09 (9.32) (10.04) (12.16) 3.93 .05 1.27 0.84 )1.64 (7.30) (5.10) (7.48) 2.93 3.44 2.37 (7.44) (5.52) (8.37) 4.16 .05 1.58 1.15 0.25 (5.76) (5.53) (6.32) 1.70 2.23 2.50 (5.34) (5.18) (8.59) 0.91 ns 9.70 5.71 7.16 (7.92) (5.77) (10.33) 10.31 8.44 8.23 (7.89) (5.72) (5.82) 3.89 .05 10.79 7.13 7.39 (9.43) (6.82) (10.74) 10.38 9.77 8.59 (8.45) (7.38) (8.58) 2.56 ns Note: ALBA = Adjustable Light Beam Apparatus. Only the unadjusted means are presented. Sample sizes differ for pretreatment (n = 94), posttreatment (n = 70), and follow-up (n = 62). Pretreatment means (SDs) are included for comparison purposes, as are follow-up means (SDs) despite there being no repeated measures effect. The df for the individual ALBA treatment effects are 1, 64 at posttreatment. ALBA: Feel-Think = the discrepancy between how large a participant feels and how large she thinks she is. Feel-Ideal = the discrepancy between how large a participant feels and how large she ideally would like to be. Think-Ideal = the discrepancy between how large a participant thinks she is and how large she ideally would like to be. Larger discrepancies indicate greater body dissatisfaction. ALBA A 2 ´ 2 repeated measures ANCOVA detected a significant overall treatment factor effect at posttreatment, F (1,64) = 5.83, p < .02, with the CBT group being more satis®ed with their hips and thighs than the WL. The overall effect was not signi®cant at followup. Individual 2 ´ 2 ANCOVAs found a main effect for treatment condition on four of the six ALBA variables at posttreatment (p values .01 .05). Three of the four differences were for hip measures (Table 4). Speci®cally, the CBT group showed more satisfaction than the WL for both hips and thighs on Feel-Ideal. This was indicated by less of a difference for the CBT group than for the WL between how they felt they looked and how they ideally wanted to look. Relatedly, the CBT condition showed less of a discrepancy than the WL between how large their hips felt and how large they believed them to be (Feel-Think). Also, the CBT group showed a smaller discrepancy than the WL for the Think-Ideal hip measure, suggesting that the CBT participants' ideal hip size was closer to how they thought their hips actually looked than it was for the WL.4 There was no overall exercise status effect or any interaction at either posttreatment or follow-up on the 2 ´ 2 repeated measures ANCOVA. Planned comparisons found one significant disadvantage for the obligatory exerciser group, which occurred at follow-up 4 The repeated measures ANCOVA remained signi®cant in the intention-to-treat analysis. 384 Smith et al. for hips (Think-Ideal), F (1,56) = 6.46, p = .01. Obligatory exercisers receiving CBT had more of a discrepancy between their ideal and perceived hip size (M = 12.50, SD = 14.64) than did nonobligatory exercisers in CBT (M = 4.02, SD = 4.95). Within-Group Changes Given that few significant between-group differences were found for treatment condition at follow-up, matched pair t tests were used to investigate pretreatment to followup within-group improvement. For the MBSRQ, the CBT group showed signi®cant improvement on four of the ®ve scales: Appearance Evaluation, t(30) = 6.61, p < .0001, Appearance Orientation, t(30) = )3.77, p = .001, Body Areas Satisfaction, t(30) = 6.28, p < .0001, and Overweight Preoccupation, t(30) = )3.58, p = .001. The WL showed some improvement, but it occurred for only two scales and was less robust: Appearance Evaluation, t(25) = 2.66, p = .05, and Appearance Orientation, t(25) = )3.12, p = .01. In terms of pretreatment to follow-up ALBA change, the CBT group showed one trend for improvement: Feel-Ideal hips (p = .07). The WL demonstrated none. In addition to evaluating within-group body image progress, OEQ change was also tracked. The obligatory CBT group showed a significant pre to posttreatment reduction in obligatory exercise characteristics, t(5.68) = )2.40, p = .04, whereas the obligatory WL controls did not. It was necessary to track changes in physical activity and BMI to determine whether they contributed to the body image outcomes. None of the conditions showed any change in the time devoted to physical activities. One group, the nonobligatory exercisers as a whole, showed a significant increase in BMI from pretreatment to follow-up, t(40) = 2.03, p = 0.5, that was driven primarily by CBT participants. The change was nonsigni®cant from pretest to posttest. Exploratory Analysis with New Exercise Status Variable The high reported level of physical activity of the nonobligatory exercisers (M = 4.9 hr per week) could have contributed to the lack of exercise status differences. This was investigated by creating an exercise status independent variable that considered both the pretreatment OEQ score and the amount of physical activity. Obligatory/high exercisers were de®ned as scoring above 50 on the OEQ and above the median (>5.58 hr per week) for physical activity (M = 11.38, SD = 4.75). Nonobligatory/low exercisers scored less than or equal to 50 on the OEQ and fell below the median for physical activity (M = 2.58 hr, SD = 1.88). Neither a signi®cant exercise effect nor an interaction was found for any of the MBSRQ scales, or for the repeated measures ANCOVA for the ALBA discrepancy scores, at posttreatment or follow-up. DISCUSSION The prediction that obligatory exercisers would show more disturbed pretreatment body image than nonobligatory exercisers was supported partially. This was most pronounced on the ALBA, where obligatory exercisers showed greater dissatisfaction for their hips and thighs than the nonobligatory exercisers, despite these body areas being Obligatory Exercise and Body Image 385 objectively smaller for them. The significant MBSRQ pretreatment difference (Overweight Preoccupation) was noteworthy because obligatory exercisers had lower BMIs than nonobligatory exercisers, and because the "fat anxiety" represented by this scale was in line with a general theme of more anxiety symptoms for the obligatory group. As predicted, obligatory exercisers, scored significantly higher than the nonobligatory exercisers on the MOCI checking scale. Although they did not score at a level associated with obsessive-compulsive disorder (Hodgson & Rachman, 1977), research has linked elevated checking scores with social fears and symptoms of anxiety and depression (Frost et al., 1986). This interpretation is compatible with the belief that obligatory exercisers have high trait anxiety and avoid intimate relationships (Coen & Ogles, 1993; De Coverly Veale, 1987, Yates, 1991). The suggestion of underlying mood issues also is reasonable from the standpoint of exercise being empirically supported as an effective short-term strategy for regulating negative mood (Yeung, 1996). Future studies should attempt to replicate these tentative findings with other instruments and should test stringently for symptoms of comorbid conditions. Even though obligatory exercisers entered treatment at a disadvantage in terms of relevant symptoms, this did not interfere with outcome. Although this implies that obligatory exercisers respond to body image treatment just as well as nonobligatory exercisers, other explanations should be considered. One might argue that there was too much overlap between the exercise conditions in the study, given the high level of activity in the nonobligatory group. But in the Coen and Ogles (1993) study that provided the OEQ psychometrics, the nonobligatory exercisers averaged even more physical activity than our nonobligatory exercisers. Still, researchers now often require distinct groups in terms of both obligatory status and amount of exercise when examining compulsive exercisers (Davis, Kaptein, Kaplan, Olmsted, & Woodside, 1998). However, when this practice was followed in the current study's exploratory analyses, our findings did not change. Another possible explanation for the lack of exercise effects is that the nonobligatory exercisers gained a significant amount of weight from pretreatment to follow-up. This increase of approximately 2.6 lb over 4 months might have attenuated their treatment response (Cash, 1994a), thereby minimizing group differences. Yet, the weight gain was not substantial at posttreatment, and exercise status differences still were not found. In terms of overall treatment response, most predictions were supported at posttreatment, but many of the effects were lost at follow-up. The diminution in effect over time did not appear to be related to treatment exposure or content because 71% of the sessions were attended, and experienced therapists delivered an empirically supported program. Instead, the issue probably was one of sample characteristics. These women averaged 37 years of age; a minimum of 15 years older than most participants in the earlier body image studies (Butters & Cash, 1987; Rosen et al., 1989, 1990). Age alone would not explain the less robust findings, but associated features meriting attention include the fact that many of these women had families and jobs, as opposed to being primarily single students. It is also likely that the women in the current study were heavier than those in other studies, despite all of them being normal weight. Unfortunately, this cannot be tested directly, given that BMIs were not reported for any of the body image treatment studies. Still, if one uses the National Health and Nutrition Examination Survey's findings that the weight of women ages 25 44 increases by 5.2% over 10-year intervals (Williamson, Kahn, Remington, & Anda, 1990), then the current study's 5 ft. 4 in. woman was nearly 12 lb heavier than the 19 24-year-olds in other studies. 386 Smith et al. The main limitation of the current study was the relatively low follow-up rate, which raises questions about how representative the data are. In response to this, one must note that there were no differences in assessment completion rates across treatment condition or exercise status, and there was only one pretreatment difference (on the MOCI) for those who finished the assessments and those who did not. Further, an intention-to-treat analysis, which assumed that noncompleters did not improve, mirrored the major outcome results. Perspective also should be maintained because several earlier studies had incomplete follow-up data or none (Butters & Cash, 1987; Dworkin & Kerr, 1987; Fisher & Thompson, 1994) and others may have had higher rates because "captive" audiences (i.e., students) were involved. More generally, attrition is routinely a problem in eating disorder treatment research, with drop-out rates frequently being quite similar to the current study's (Allison, Kreibich, Heshka, & Heymsfield, 1995; Steel et al., 2000). The problem seems to be most pronounced in community samples (Cachelin et al., 1999). This appears to be the first body image therapy study for older community women who had families and jobs and who fell at the upper end of the normal weight range. It is unique in that it included a large percentage of ethnic minority participants (41%) and demonstrated no ethnic differences in treatment response. The results suggest that CBT is a reasonable body image treatment for a community sample, but that refinement may be needed to improve long-term outcomes for everyone, not just for obligatory exercisers. There were two intriguing findings that require replication. First, obligatory exercisers had somewhat elevated (albeit not pathological) compulsivity scores that may be linked with anxiety and mood problems. Second, high levels of physical activity were the norm for even the nonobligatory participants. Because exercise has mood regulation properties (Yeung, 1996), both of these findings suggest indirectly that general anxiety and mood symptoms may need to be addressed in standard body image programs. Also, because high levels of exercise serve to further increase a negative body focus (Davis et al., 1990), this study intimates that the majority of women in body image programs may benefit from discussions of their physical activity. The authors express their appreciation to the women of Albuquerque who graciously participated in this study and to Thomas Cash, Ph.D., for his permission to adapt the treatment to ®t our format. REFERENCES Adams, P.J., Katz, R.C., Beauchamp, K., Cohen, E., & Zavis, D. (1993). Body satisfaction, eating disorders, and depression: A developmental perspective. Journal of Child and Family Studies, 2, 37 46. Allison, D.B., Kreibich, K., Heshka, S., & Heyms®eld, S.B. (1995). A randomised placebo-controlled clinical trial of an acupressure device for weight loss. International Journal of Obesity, 19, 653 658. Bailey, S.M., Goldberg, J.P., Swap, W.C., Chomitz, V.R., & Houser, Jr., R.F. (1990). Relationships between body dissatisfaction and physical measurements. International Journal of Eating Disorders, 9, 457 461. Brown, T., Cash, T., & Mikulka, P. (1990). Attitudinal body-image assessment: Factor analysis of the Body Self Relations Questionnaire. Journal of Personality Assessment, 55, 135 144. Brownell, K.D., Rodin, J., & Wilmore, J. (1992). Eating, body weight and performance in athletes: Disorders of modern society. Philadelphia: Lea & Febiger. Butters, J.W., & Cash, T.F. (1987). Cognitive-behavioral treatment of women's body-image dissatisfaction. Journal of Consulting and Clinical Psychology, 55, 889 897. Cachelin, F.M., Striegel-Moore, R.H., Elder, K.A., Pike, K.M., Wil¯ey, D.E., & Fairburn, C.G. (1999). Natural course of a community sample of women with binge eating disorder. International Journal of Eating Disorders, 25, 45 54. Cash, T.F. (1994a). Body image and weight changes in a multisite comprehensive very-low-calorie diet program. Behavior Therapy, 25, 239 254. Obligatory Exercise and Body Image 387 Cash, T.F. (1994b). The Multidimensional Body-Self Relations Questionnaire users' manual. Norfolk, VA: Author. Cash, T.F. (1995a). Developmental teasing about physical appearance: Retrospective descriptions and relationships with body image. Social Behavior and Personality, 23, 123 129. Cash, T.F. (1995b). What do you see when you look in the mirror? Helping yourself to a positive body image. New York: Bantam Books. Cash, T.F., & Pruzinsky, T. (Ed.) (1990). Body images: Development, deviance, and change. New York: Guilford Press. Coen, S.P., & Ogles, B.M. (1993). Psychological characteristics of the obligatory runner: A critical examination of the anorexia analogue hypothesis. Journal of Sport and Exercise Psychology, 15, 338 354. Davis, C. (1997). Body image, exercise, and eating behaviors. In K.R. Fox (Ed.), The physical self: From motivation to well-being. Champaign, IL: Human Kinetics. pp. 143 174. Davis, C., Fox, J., Cowles, M., Hastings, P., & Schwass, K. (1990). The functional role of exercise in the development of weight and diet concerns in women. Journal of Psychosomatic Research, 34, 563 574. Davis, C., Kaptein, S., Kaplan, A.S., Olmsted, M.P., & Woodside, D.B. (1998). Obsessionality in anorexia nervosa: The moderating in¯uence of exercise. Psychosomatic Medicine, 60, 192 197. Davis, C., Kennedy, S.H., Ralevski, E., Dionne, M., Brewer, H., Neitzert, C., & Ratusny, D. (1995). Obsessivecompulsiveness and physical activity in anorexia nervosa and high-level exercising. Journal of Psychosomatic Research, 39, 967 976. Davis, C., Shapiro, C., Elliott, S., & Dionne, M. (1993). Personality and other correlates of dietary restraint: An age by sex comparison. Personality and Individual Differences, 14, 297 305. De Coverly Veale, D. (1987). Exercise dependence. British Journal of Addiction, 82, 735 740. Dworkin, S.H., & Kerr, B.A. (1987). Comparison of interventions for women experiencing body image problems. Journal of Counseling Psychology, 34, 136 140. Elam, P., & Kimbrell, D. (1995). Size, lies, and measuring tape. Cognitive and Behavioral Practice, 2, 233 248. Eliot, A., & Smith, J.E. (2001). Enhancing women's body image: A comparison of treatment interventions. Manuscript in preparation. Fisher, E., & Thompson, J.K. (1994). A comparative evaluation of cognitive-behavioral therapy (CBT) versus exercise therapy (ET) for the treatment of body image disturbance: Preliminary ®ndings. Behavior Modi®cation, 18, 171 185. Frost, R., Sher, K., & Green, T. (1986). Psychopathology and personality characteristics of nonclinical compulsive checkers. Behaviour Research and Therapy, 24, 133 143. Goldfarb, L.A., & Plante, T.G. (1984). Fear of fat in runners: An examination of the connection between anorexia nervosa and distance running. Psychological Reports, 55, 296. Grant, J.R., & Cash, T.F. (1995). Cognitive-behavioral body image therapy: Comparative ef®cacy of group and modest-contact treatments. Behavior Therapy, 26, 69 84. Herzog, D.B., Hopkins, J.D., & Burns, C.D. (1993). A follow-up study of 33 subdiagnostic eating disordered women. International Journal of Eating Disorders, 14, 261 267. Hodgson, R.J., & Rachman, S. (1977). Obsessional-compulsive complaints. Behaviour Research and Therapy, 15, 389 395. Hollingshead, A.B. (1965). Two factor index of social position. New Haven: Yale Station. . Imm, P.S., & Pruitt, J. (1991). Body shape satisfaction in female exercisers and nonexercisers. Women and Health, 17, 87 96. Keeton, W.P., Cash, T.F., & Brown, T.A. (1990). Body image or body images? Comparative multidimensional assessment among college students. Journal of Personality Assessment, 54, 213 230. Kiembel, G., Slade, P.D., & Dewey, M.E. (1987). Factors associated with abnormal eating attitudes and behaviors: Screening individuals at risk of developing an eating disorder. International Journal of Eating Disorders, 6, 713 724. Kline, G.M., Porcari, J.P., Hintermeister, R., Freedson, P.S., Ward, A., McCarron, R.F., Ross, J., & Rippe, J.M. (1987). Estimation of VO2max from a one-mile track walk, gender, age, and body weight. Medicine and Science in Sports and Exercise, 19, 253 259. Loland, N.W. (1998). Body image and physical activity: A survey among Norwegian men and women. International Journal of Sport Psychology, 29, 339 365. McAuley, E., Bane, S.M., Rudolph, D.L., & Lox, C.L. (1995). Physique anxiety and exercise in middle-aged adults. Journal of Gerontology Series B Psychological Sciences and Social Sciences, 50B, 229 235. Pasman, L., & Thompson, J.K. (1988). Body image and eating disturbance in obligatory runners, obligatory weightlifters, and sedentary individuals. International Journal of Eating Disorders, 7, 759 769. Rodin, J., Silberstein, L.R., & Striegel-Moore, R.H. (1984). Women and weight: A normative discontent. In T.B. Sonderegger (Ed.), Psychology and gender: Nebraska symposium on motivation (pp. 267 307). Lincoln, NE: University of Nebraska Press. Rosen, J.C. (1992). Body-image disorder: De®nition, development, and contribution to eating disorders. In J. Crowther, D. Tennenbaum, S. Hobfoll, & M. Stephens (Eds.), The etiology of bulimia nervosa: The individual and familial context (pp 157 177). Washington DC: Hemisphere. Rosen, J.C., Cado, S., Silberg, N.T., Srebnik, D., & Wendt, S. (1990). Cognitive behavior therapy with and without size perception training for women with body image disturbance. Behavior Therapy, 21, 481 498. 388 Smith et al. Rosen, J.C., Saltzberg, E., & Srebnik, D. (1989). Cognitive behavior therapy for negative body image. Behavior Therapy, 20, 393 404. Sher, K., Mann, B., & Frost, R. (1984). Cognitive dysfunction in compulsive checkers: Further explorations. Behaviour Research and Therapy, 22, 493 502. Slay, H.A., Hayaki, J., Napolitano, M.A., & Brownell, K.D. (1998). Motivations for running and eating attitudes in obligatory versus nonobligatory runners. International Journal of Eating Disorders, 23, 267 275. Smith, D.E., Thompson, J.K., Raczynski, J.M., & Hilner, J.E. (1999). Body image among men and women in a biracial cohort: The CARDIA study. International Journal of Eating Disorders, 25, 71 82. Sonstroem, R.J., Harlow, L.L., & Josephs, L. (1994). Exercise and self-esteem: Validity of model expansion and exercise associations. Journal of Sport and Exercise Psychology, 16, 29 42. Steel, Z., Jones, J., Adcock, S., Clancy, R., Bridgford-West, L., & Austin, J. (2000). Why the high rate of dropout from individualized cognitive-behavior therapy for bulimia nervosa? International Journal of Eating Disorders, 28, 209 214. Steffen, J.J., & Brehm, B.J. (1999). The dimensions of obligatory exercise. Eating Disorders: The Journal of Treatment and Prevention, 7, 219 226. Thompson, J.K., Heinberg, L.J., Altabe, M., & Tantleff-Dunn, S. (1999). Exacting beauty: Theory, assessment, and treatment of body image disturbance. Washington, DC: American Psychological Association. Thompson, J.K., & Spana, R.E. (1988). The adjustable light beam method for assessment of size estimation accuracy: Description, psychometrics, and normative data. International Journal of Eating Disorders, 7, 521 526. Thompson, J.K., & Thompson, C.M. (1986). Body size distortion and self-esteem in asymptomatic, normal weight males and females. International Journal of Eating Disorders, 5, 1061 1068. Thompson, R.A., & Sherman, R.T. (1999). "Good athlete" traits and characteristics of anorexia nervosa: Are they similar? Eating Disorders: The Journal of Treatment and Prevention, 7, 181 190. Tucker, L.A., & Mortell, R. (1993). Comparison of the effects of walking and weight training programs on body image in middle-aged women: An experimental study. American Journal of Health Promotion, 8, 35 42. Williamson, D.E., Kahn, H.S., Remington, P.L., & Anda, R.E. (1990). The 10-year incidence of overweight and major weight gain in US adults. Archives of Internal Medicine, 150, 665 672. Yates, A. (1991). Compulsive exercise and the eating disorders. New York: Brunner/Mazel. Yates, A., Shisslak, C.M., Allender, J., Crago, M., & Leehey, K. (1992). Comparing obligatory and nonobligatory runners. Psychosomatics, 33, 180 189. Yeung, R.R. (1996). The acute effects of exercise on mood state. Journal of Psychosomatic Research, 40, 123 141.