Body image change in obese and overweight women enrolled in a weight-loss intervention: The importance of perceived versus actual physical changes

Body Image 9 (2012) 311–317

Contents lists available at SciVerse ScienceDirect

Body Image journal homepage: www.elsevier.com/locate/bodyimage

Body image change in obese and overweight women enrolled in a weight-loss intervention: The importance of perceived versus actual physical changes Kathleen A. Martin Ginis ∗ , Desmond McEwan, Andrea R. Josse, Stuart M. Phillips McMaster University, Hamilton, ON, Canada

a r t i c l e

i n f o

a b s t r a c t

Article history: Received 14 February 2012 Received in revised form 12 April 2012 Accepted 12 April 2012 Keywords: Self-efficacy Physical self-perceptions Fitness Physical activity Exercise Diet

Using the exercise and self-esteem model as a guiding framework, this study examined variables related to body image change among 88 overweight and obese women (Mage = 28.4 ± 7.8; MBMI = 31.6 ± 3.5) participating in a 16-week diet and exercise weight-loss intervention. Measures of body image and potential mechanisms of body image change (actual and perceived physical changes, self-efficacy) were administered at baseline, Weeks 8 and 16. Body image improved significantly over the study time-points (ps < .001). Perceived physical changes accounted for most explained variance (12–37%) in body image change (total R2 adj = .21–.50). Improved perceptions of body fat were a particularly important predictor in each model (ps ≤ .04). Results support the notion that to improve body image, perceived changes to the body are more important than actual changes. Such information is integral to developing theories to explain body image change and interventions to maximize it. © 2012 Elsevier Ltd. All rights reserved.

Introduction For many women, being overweight or obese has profound effects on body image (Sarwer, Thompson, & Cash, 2005). Indeed, obese and overweight women consistently score higher on measures of body dissatisfaction than women at healthy body weights (Schwartz & Brownell, 2004). Because greater body dissatisfaction has been linked to an increased risk for psychological disturbances such as depression and anxiety, and an increased risk for maladaptive eating behaviors that may actually contribute to weight gain (Sarwer et al., 2005), there is a need for research that sheds light on the processes by which body image may be improved in these at-risk populations. It is often assumed that the key to improving body image is to lose weight. Although a reduction in body weight generally leads to an improvement in body image (Cash, 1994; Foster, Wadden, & Vogt, 1997), the correlation between these variables is not particularly strong (Latner & Wilson, 2011; Sarwer et al., 2005). For example, one study showed that a weight-loss program consisting of a calorie-reduced diet plus cognitive behavior therapy yielded significant decreases in body weight and improvements in body satisfaction and appearance evaluation among obese women at both study mid-point and end-point (Foster et al., 1997).

∗ Corresponding author at: Department of Kinesiology, McMaster University, Hamilton, ON L8S 4K1, Canada. Tel.: +1 905 525 9140x23574; fax: +1 905 523 6011. E-mail address: [email protected] (K.A. Martin Ginis). 1740-1445/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.bodyim.2012.04.002

However, the correlations between amount of weight lost and body image improvements were modest at both time points (rs ranged from −.25 to −.33). The authors suggested that other therapeutic aspects of the intervention, rather than weight loss, may have elicited body image improvements. They also suggested that relatively small reductions in weight are needed for positive changes in body image to occur. Another possible explanation for the weak association between weight loss and body image change is that actual changes to the body are less important than perceived changes. Research examining correlates of body image change in exercise interventions supports this interpretation (for reviews, see Martin Ginis & Bassett, 2011; Martin Ginis, Bassett, & Conlin, 2012). Like dietary weightloss interventions, exercise interventions have significant positive effects on body image (Campbell & Hausenblas, 2009). A systematic review of the exercise and body image literature revealed that perceived changes to one’s body (i.e., perceived changes in physical fitness) and perceived improvements in one’s physical abilities (i.e., physical self-efficacy), were more consistent correlates of body image change than were actual physique-related changes (Martin Ginis et al., 2012). In fact, only half of the reviewed studies yielded a significant correlation between body image change and changes in body composition (e.g., fat mass, lean mass) or anthropometric variables (e.g., body mass index, waist-to-hip ratio). Just one-third of studies reported a significant correlation between body image change and actual changes in physical strength (e.g., maximum amount of weight lifted) and endurance (e.g., maximum heart rate, maximum volume of oxygen consumption during exercise).

312

K.A. Martin Ginis et al. / Body Image 9 (2012) 311–317

When significant, these correlations accounted for <20% of the variance in body image change. In contrast, all of the studies that measured perceived changes in physical fitness (i.e., body composition, strength, aerobic endurance) and self-efficacy yielded significant correlations between these variables and body image change. Taken together, these findings suggest that how an exerciser perceives changes to her body (e.g., decreased fat, increased muscle mass) is a stronger determinant of body image change than the actual magnitude of those changes (e.g., percentage of body fat lost, kg of muscle mass gained). This perspective is consistent with theoretical approaches used to examine the effects of exercise on body image. In particular, the exercise and self-esteem model (Sonstroem & Morgan, 1989) posits that actual physical changes, perceived physical changes and selfefficacy changes all lead to changes in feelings about one’s body which, in turn, affect self-esteem. However, perceived changes and self-efficacy changes are more proximal determinants of body image change, than are actual physical changes. Although the exercise and self-esteem model was developed to explain the effects of exercise on self-esteem (not body image), it does capture variables that have been identified as potentially key mechanisms of body image change (Martin Ginis et al., 2012). As such, it is a viable framework for examining the processes by which a weight-loss intervention affects body image. Accordingly, the primary purpose of the present study was to examine correlates of body image change among overweight and obese women who participated in a 16-week diet and exercise intervention. To date, no published study has tested the extent to which a comprehensive model of actual physical changes, perceived physical changes, and self-efficacy changes can account for body image improvements following a weight-loss intervention. The present study was designed to test the relative contributions of these variables. Based on previous research syntheses (Martin Ginis et al., 2012; Martin Ginis & Bassett, 2011) and the exercise and self-esteem model (Sonstroem & Morgan, 1989), it was predicted that in a hierarchical regression model, perceived physical and selfefficacy changes would explain significant variance in body image change beyond that explained by actual physical changes. Furthermore, once all variables were entered into the model, perceived physical change and self-efficacy change variables were predicted to be more strongly associated with body image change than actual physical change measures. A secondary purpose was to examine variables associated with body image change early in an intervention versus those associated with change at the end of the intervention. Within the context of an exercise-training program, improvements in some of the proposed mechanism variables – such as strength and endurance (perceived and actual), and physical self-efficacy – can emerge relatively quickly and be quite drastic for out-ofshape beginner exercisers (Martin Ginis & Bassett, 2011). These sudden improvements could serve to challenge long-held, negative body image attitudes and beliefs, leading to enhanced body image. But as the intervention progresses, participants may become accustomed to their improved physicality; by the end of the intervention, such improvements may no longer be playing a role in body image change. Within the context of a dietary intervention, it is not clear when fat loss plays the strongest role in body image change. On the one hand, the association may be strongest early in an intervention if relatively small amounts of weight loss do indeed trigger body image improvements (Foster et al., 1997; Schwartz & Brownell, 2004). On the other hand, the association could be strongest later in the intervention when participants achieve their maximum weight loss. These possibilities were explored by examining variables that explained change in body image at the intervention mid-point versus the intervention end-point.

Method Participants Participants were 88 women (Mage = 28.4 ± 7.8 years, range = 19–45) recruited from a larger study of the effects of exercise and a low, moderate, or high dairy hypoenergetic diet, on weight loss (for further description, see Josse, Atkinson, Tarnopolsky, & Phillips, 2011, 2012). The larger study was targeted to young women for two key reasons. First, to help promote dairy food and calcium intake as part of a healthy weight loss diet in a group for whom habitual intakes of these nutrients are consistently sub-optimal (Briefel & Johnson, 2004; Health Canada, 2004); second, to develop a healthy exercise routine that incorporates resistance training, given that young women tend to avoid this type of training (cf. Prichard & Tiggemann, 2008). Participants in the larger trial (N = 90) were recruited from a university campus and surrounding community through posters and newspaper advertisements seeking women interested in participating in a 16-week diet and exercise weight-loss study. Participants met the following study inclusion criteria: body mass index (BMI) between 27 and 40 kg/m2 (M = 31.6 ± 3.5) sedentary, non-smoker, regular menstrual cycle, not menopausal, otherwise healthy, and no contraindications to exercise. The study coordinator verbally invited all participants in the larger study to participate in the body image study. Design This ancillary study utilized a single group, pre-test/post-test study design with repeated measures taken at baseline, Week 8 and Week 16 of the intervention. The larger experiment utilized a three-group, randomized experimental design. There were no differences between the three conditions on the measures of body image change (ps > .05). Given the purpose of the present study, it was considered appropriate to collapse the three conditions into a single group to examine correlates of change in body image. Measures Body image. The cognitive and subjective satisfaction dimensions of body image were measured with the 9-item Body Areas Satisfaction Scale (BASS) and the 7-item Appearance Evaluation (AE) subscale, respectively. The BASS and AE are subscales of the Multidimensional Body-Self Relations Questionnaire (MBSRQ; Brown, Cash, & Mikulka, 1990). The BASS assesses the degree of satisfaction with nine aspects of one’s body (e.g., face, upper torso, muscle tone). Participants rated their satisfaction with each aspect using a scale ranging from 1 (very dissatisfied) to 5 (very satisfied). A total score was calculated with higher scores indicating greater satisfaction. The AE subscale assesses overall satisfaction with physical appearance. Participants indicated the degree to which they agreed with each of 7 statements (e.g., “I like my looks just the way they are”) using a scale ranging from 1 (definitely disagree) to 5 (definitely agree). After reverse scoring two items, a total score was calculated with higher values indicating greater appearance satisfaction. The affective dimension of body image was assessed using the 9item version (Martin, Rejeski, Leary, McAuley, & Bane, 1997) of the Social Physique Anxiety Scale (SPAS; Hart, Leary, & Rejeski, 1989). The SPAS measures the anxiety experienced in response to others’ evaluations of one’s physique. Using a 5-point Likert scale (1 = “not at all characteristic of me” and 5 = “extremely characteristic of me”), participants indicated the extent to which they agreed with each statement (e.g., “I wish I wasn’t so uptight about my physique”). After reverse scoring two items, a total score was calculated, with a higher score indicating greater social physique anxiety. The AE,

K.A. Martin Ginis et al. / Body Image 9 (2012) 311–317

BASS, and SPAS have all been used to assess body image change in overweight/obese women participating in diet and/or exercise interventions (e.g., Foster et al., 1997; Kerksick et al., 2009). All three scales showed acceptable internal consistency at all three study measurement points (BASS ˛ ≥ .77; AE ˛ ≥ .81; SPAS ˛ ≥ .85).

313

amounts of weight. An average self-efficacy score was calculated for each exercise and then an average composite self-efficacy score was calculated across all four exercises (˛ ≥ .92 at all measurement points). Procedure

Actual physical changes. Physical change was conceptualized as change in three aspects of physical fitness: body composition, aerobic fitness, and muscular strength. Body composition was operationalized as body fat percentage (%) and lean (muscle) mass (kg), measured using Dual-Energy X-ray Absorptiometry scans (DXA; QDR-4500A, Hologic Inc., Waltham, MA). Whole body scans using the fan-beam mode were performed from head to toe. Participants were scanned with their hands lying flat on the bed at their sides and their feet internally rotated. They wore hospital gowns or the same light exercise clothing for every scan. All scans took place in the afternoon and were performed by the same researcher. Aerobic fitness was measured using a modified Astrand Submaximal Fitness Test protocol (Astrand & Rodahl, 1986; Legge & Banister, 1986). Participants cycled on a stationary bicycle (Monark Ergomedic 828 E, Monark Exercise AB, Vansbro, Sweden) at two consecutive workloads (ranging from 75 W to 125 W) for 6-min each, following a 2-min warm up at a workload of 50 W (1.0 kp). Heart rate (HR) was measured continuously throughout the test with average HR in the last 2 min/workload (of a 6-min workload) serving as the dependent measure of aerobic fitness. A lower HR indicates better aerobic fitness. Muscular strength was measured by assessing participants’ 1 repetition maximum (1RM) for the following exercises: bench press, seated row, leg press, and hamstring curl. The 1RM is the maximum amount of weight (kg) a participant can move through the entire range of motion, just once, for a particular exercise. For the purpose of this study, an overall composite strength score was calculated by summing the amount of weight lifted on the 1RM test, across all four exercises. Perceived physical changes. Measures of perceived physical changes paralleled the measures of actual physical changes. Specifically, three 6-item subscales of the Physical Self-Description Questionnaire (PSDQ; Marsh, Richards, Johnson, Roche, & Tremayne, 1994) were administered: the Body Fat subscale measured perceived fatness, the Body Strength subscale measured perceived muscular strength, and the Body Endurance subscale measured perceived aerobic endurance. Participants rated how well each item described them (e.g., “I have too much fat on my body,” “I am a physically strong person,” and “I could jog 5 kilometers without stopping”) using a scale ranging from 1 (false) to 6 (true). Items for each subscale were summed (to simplify interpretation, Body Fat subscale items were not reverse-scored) with higher values indicating greater perceived body fat, strength and endurance. Each subscale has shown validity and reliability amongst university-aged females (e.g., Marsh et al., 1994) and demonstrated satisfactory internal consistency at all three measurement points (Body Fat ˛ ≥ .84, Body Strength ˛ ≥ .85, Body Endurance ˛ ≥ .90). Self-efficacy. Self-efficacy was conceptualized as confidence in one’s abilities to perform activities requiring (a) aerobic endurance and (b) muscular strength. To assess aerobic self-efficacy, participants indicated their confidence, on a scale ranging from 0 (not at all) to 100 (completely) to run on a treadmill at a moderate intensity for 5, 10, 15, 20, 30, 40, 50, and 60 min without stopping. A mean self-efficacy value was calculated across these eight levels, with higher scores indicating greater aerobic self-efficacy (˛ ≥ .94 at all measurement points). Using the same 0–100 response scale, strength self-efficacy was measured by asking participants to indicate their confidence to perform the four 1RM exercises (seated row, bench press, leg extension and hamstring curl) with increasing

The protocol was approved by university and hospital research ethics boards and the study was conducted from June 2008 until May 2010. The exercise intervention consisted of 45–60 min of aerobic exercise 7 days/week and progressive resistance exercise 2 days/week. On weekdays, exercise was performed at a supervised training center on campus. On weekends, women exercised unsupervised. At each aerobic workout (e.g., jogging, cycling, fitness classes), participants were required to expend 250 kcal as determined by a SenseWear ProTM energy expenditure device (BodyMedia Inc. Pittsburgh, PA) worn on the exerciser’s arm. The resistance training regimen consisted of an upper and lower body workout supervised by a personal trainer or kinesiologist. Once participants could complete 3 sets of 10 repetitions or more at the prescribed weight, the prescription (i.e., weight) was increased. Exercise logs were completed after each session and checked frequently by study personnel to ensure compliance. The dietary intervention consisted of restricting daily caloric intake to 500 kcal below each participant’s weight maintenance energy requirements. Depending on experimental group assignment, participants’ diets differed in the proportion of kcal consumed as protein (15% or 30%) and the amount of protein derived from dairy sources (high, medium, or low). Participants submitted 3-day food records bi-weekly and met individually with study dietitians to monitor nutritional intake and reinforce hypocaloric healthy eating patterns. Study questionnaires and DXA scans were completed at baseline, Weeks 8 and 16. Because aerobic fitness and strength were not primary endpoints in the larger trial, these variables were measured at baseline and Week 16 only. Conducting the aerobic fitness and 1RM strength tests at mid-point would have required an extra day of testing and disrupted the prescribed training regimen. Hence, these tests were not part of the mid-point assessment. At baseline and Week 16, questionnaires were administered after the fitness tests to ensure that all participants had feedback regarding their fitness prior to reporting on their self-perceptions. Data Analysis Analyses were conducted using SPSS v. 19 (IBM SPSS Predictive Analytics, Chicago IL). For descriptive purposes, General Linear Model repeated measures ANOVAs were conducted to determine changes in the study variables over time. Significant ANOVAs were followed-up with post hoc t-tests to determine when significant changes occurred. To test the study hypotheses, hierarchical regression models were computed in which residualized body image change scores were regressed on simple change scores for actual fitness, and residualized change scores for perceived fitness, and self-efficacy. Separate regression analyses were computed to predict change in each body image measure (i.e., Appearance Evaluation, Body Areas Satisfaction, and Social Physique Anxiety) from baseline to Week 8, and baseline to Week 16. Because preliminary analyses revealed age to be significantly negatively correlated with body image change, it was controlled for in each model. To test our hypotheses, the predictor variables were entered in blocks in the following order: Block 1: age, actual fitness changes (body fat, lean mass, aerobic fitness, strength [note that aerobic fitness and strength were included as predictors only in the Week 16 model]); Block 2: self-efficacy changes (aerobic and strength); Block 3:

314

K.A. Martin Ginis et al. / Body Image 9 (2012) 311–317

Table 1 Means, standard deviations, significance tests, and effect sizes (partial 2 ) for measures of body image and perceived and actual body fat, muscularity, and strength at baseline, Week 8, and Week 16. Measure

Body image Appearance Evaluation Body Areas Satisfaction Social Physique Anxiety Body composition and fitness Lean mass (kg) Fat percentage Aerobic fitness (HR) Strength (kg) Self-perceptions Perceived strength Perceived body fat Perceived endurance Self-efficacy Aerobic Strength

Baseline

Week 8

2.21 ± 0.66 2.46 ± 0.47 33.51 ± 6.55 48.73 40.14 165.69 104.08

± ± ± ±

Week 16

2.76 ± 0.77 2.91 ± 0.58 31.65 ± 7.21

3.14 ± 0.84 3.36 ± 0.57 28.33 ± 7.64

48.64 ± 5.88 37.96 ± 4.27 – –

5.91 4.02 15.51 16.80

F

48.54 36.59 148.85 117.67

± ± ± ±

5.71 4.54 16.83 16.32

3.96 ± 0.89 5.33 ± 0.70 2.20 ± 1.02

4.59 ± 0.69 4.74 ± 0.93 3.14 ± 1.11

5.02 ± 0.68 3.99 ± 1.24 3.87 ± 1.15

35.49 ± 28.13 42.58 ± 16.95

48.09 ± 28.75 55.09 ± 16.59

62.44 ± 26.03 62.92 ± 14.46

2

p

p-Values

n

W0–16

W0–8

W8–16

66.90 114.59 33.36

<.001 <.001 <.001

0.50 0.64 0.33

<.001 <.001 <.001

<.001 <.001 .002

<.001 <.001 <.001

67 67 68

0.79 161.28 157.58 95.69

0.45 <.001 <.001 <.001

0.01 0.67 0.66 0.54

.25 <.001 <.001 <.001

.50 <.001 – –

.52 <.001 – –

82 82 81 82

70.20 77.65 101.15

<.001 <.001 <.001

0.50 0.52 0.59

<.001 <.001 <.001

<.001 <.001 <.001

<.001 <.001 <.001

72 72 72

39.99 53.63

<.001 <.001

0.36 0.43

<.001 <.001

<.001 <.001

<.001 <.001

72 72

Note: W0–16 = significance of change from baseline to Week 16; W0–8 = significance of change from baseline to Week 8; W8–16 = significance of change from Week 8 to Week 16.

perceived fitness changes (strength, fat, and endurance). Variance explained (R2 ) by each block was examined, along with the regression coefficients, once all predictors had been entered into the model. Given the potential for collinearity among the predictors, the variance inflation factor was examined for each block, but was <2.5 in every model (well below the tolerance level; Tabachnick & Fidell, 2007). Scatterplots of the residuals and model diagnostic statistics were examined to ensure regression assumptions were met (Tabachnick & Fidell, 2007). Results Preliminary Analyses Six participants dropped out before Week 8 for reasons unrelated to the study. These women did not differ from the remaining participants on any baseline measure. Their data were excluded from all subsequent analyses. In addition, four women did not return the questionnaire packet at Week 8 and six did not return it at Week 16. Consequently, sample sizes vary across measurement points (n = 78 at Week 8; n = 76 at Week 16). SPSS missing values analysis revealed <5% missing data for all variables except the BASS and SPAS (7% at Week 16) and that the data were missing at random. Given the relatively small amount of missing data and

its randomness, missing values were replaced with the group mean (Tabachnick & Fidell, 2007). Means and standard deviations for each study variable at each measurement point are in Table 1, along with significance tests and effect sizes. Measures of the actual physical changes revealed significant improvements in aerobic fitness and strength, and significant decreases in % body fat over the 16-week intervention (ps < .001). Percent body fat also decreased significantly from baseline to Week 8 and from Week 8 to Week 16 (ps < .001). Lean body mass did not change significantly over the intervention period. All of the body image, perceived physical change, and self-efficacy measures improved significantly over the 16-week intervention (ps < .001). There were significant changes in these variables from baseline to Week 8 and from Week 8 to Week 16 (ps < .002). Predictors of Body Image Change: Baseline to Week 8 For descriptive purposes only, bivariate correlations between the 8-week change in body image and the 8-week change in the potential mechanisms, are shown in Table 2. Results for the regression analyses are presented in Table 3. Appearance Evaluation. The full set of predictors accounted for 49% of the variance in change in AE over the first 8 weeks of the

Table 2 Correlations between change in body image and change in fitness, self-perceptions and self-efficacy at Week 8 and Week 16. Change variable 1. AE 2. BASS 3. SPAS 4. Aerobic fitness 5. Strength 6. Lean mass 7. Fat percentage 8. Perceived strength 9. Perceived body fat 10. Perceived endurance 11. Aerobic efficacy 12. Strength efficacy

1 – .62** −.42** – – −.08 −.27* .13 −.60** .33** .35** −.06

2

3 **

.80 – −.37** – – −.15 −.20 .25* −.49** .33** .20 −.08

−.66 −.67** – – – .15 .14 −.20 .36** −.22 −.37** −.02 **

4

5

6

7

−.07 −.15 .29* – – – – – – – – –

−.12 −.20 −.08 −.07 – – – – – – – –

−.13 −.23 −.00 .20 .38** – −.22 .14 .23* −.10 −.15 −.02

−.37 −.36** .32** .28* .07 .09 – −.27* .22 −.14 −.14 −.14

Note. r values below the diagonal are the correlations for the 8-week change scores. r values above the diagonal are the correlations for the 16-week change scores. AE = Appearance Evaluation; BASS = Body Areas Satisfaction Scale; SPA = Social Physique Anxiety Scale. –, aerobic and strength measurements were not taken at 8 weeks. * p < .05. ** p < .01.

8 **

9 **

.30 .41** −.34** −.17 .08 −.03 −.10 – −.05 .12 .02 .26*

10

−.61 −.67** .51** .25* .13 .29** .48** −.12 – −.21 −.20 −.01 **

11 **

.32 .42** −.18 −.10 −.07 .01 −.37** .33** −.28* – .63** .27*

12 *

.28 .32** −.16 −.03 −.08 .01 −.32** .19 −.28* .70** – .18

.04 .13 −.12 −.25* .16 .02 −.10 .41** −.09 .21 .16 –

K.A. Martin Ginis et al. / Body Image 9 (2012) 311–317

315

Table 3 Hierarchical regression models predicting change in the body image measures from Weeks 1 to 8. Appearance Evaluation

Model 1 Age Fat percentage Lean mass Model 2 Age Fat percentage Lean mass Aerobic self-efficacy Strength self-efficacy Model 3 Age Fat percentage Lean mass Aerobic self-efficacy Strength self-efficacy Strength perception Fat perception Endurance perception * ** ***

R2 change

ˇ

.12*

−.15 −.37 −.15

.11*

.27***

Body Areas Satisfaction

Social Physique Anxiety

R2 change

ˇ

p

R2 change

ˇ

p

.24 .005 .23

.07

.08 −.21 −.15

.52 .12 .22

.05

−.01 .18 .19

.97 .18 .15

−.13 −.32 −.08 .33 −.11

.26 .01 .53 .006 .31

.06

.09 −.17 −.11 .23 −.10

.45 .20 .41 .06 .40

.14**

−.03 .11 .10 −.38 .08

.81 .38 .40 .002 .52

−.13 −.15 .06 .18 −.14 .08 −.51 .15

.18 .18 .56 .16 .17 .46 <.001 .24

.27***

.07 .00 −.02 .01 −.19 .24 −.40 .27

.55 .99 .88 .92 .09 .04 .001 .06

.12*

−.01 −.02 .05 −.36 .13 −.25 .26 .01

.97 .89 .67 .02 .29 .05 .04 .94

p

p < .05. p < .01. p < .001.

intervention (R2 adj = .43, F(8, 6) = 7.31, p < .001). Inspection of the R2 values indicated that significant variance was accounted for by actual physical changes (R2 = .12, p = .04), self-efficacy (R2 = .11, p = .02), and perceived physical changes (R2 = .27, p < .001). Once all variables were entered into the model, only changes in perceived body fat (ˇ = −.51, p < .001) were a significant unique predictor, indicating that women perceiving greater decreases in body fatness had bigger improvements in appearance evaluation. Body Areas Satisfaction. The full set of predictors explained 39% of the variance in change in BASS scores (R2 adj = .31, F(3, 60) = 8.66, p < .001). However, only perceived physical changes accounted for significant explained variance (R2 = .27, p < .001). Once all variables were entered into the model, changes in perceived strength (ˇ = .24, p = .04) and perceived body fat (ˇ = −.40, p = .001) were the only significant predictors. Women who reported the greatest improvements in their body fatness and strength had the biggest increases in body satisfaction. Social Physique Anxiety. The full set of predictors accounted for 30% of the variance in SPAS change scores from baseline to Week 8 (R2 adj = .21, F(8, 60) = 3.25, p = .004). Self-efficacy changes (R2 = .14, p = .008) and perceived physical changes (R2 = .12, p = .02) accounted for significant variance. Once all variables were entered into the model, change in aerobic self-efficacy (ˇ = −.36, p = .02), perceived strength (ˇ = −.25, p = .049), and perceived body fat (ˇ = .26, p = .04) were significant predictors of change in SPAS scores. Women with the greatest increases in aerobic self-efficacy and perceived strength, and the biggest decreases in perceived fatness experienced the largest decreases in social physique anxiety. Predictors of Body Image Change: Baseline to Week 16 Bivariate correlations between the 16-week body image and predictor change scores, are shown in Table 2. Results for the regression analyses are presented in Table 4. Appearance Evaluation. The full set of predictors accounted for 49% of the variance in AE change from baseline to Week 16 (R2 adj = .39, F(10, 49) = 4.76, p < .001). Significant variance was accounted for by actual physical changes (R2 = .20, p = .03) and

perceived physical changes (R2 = .28, p < .001). However, when all of the variables were entered into the model, perceived body fat (ˇ = −.57, p < .001) was the only unique predictor. Women who perceived the biggest changes in body fatness had the greatest increases in AE scores. Body Areas Satisfaction. The full predictor set explained 59% of the variance in BASS change scores from baseline to Week 16 (R2 adj = .50, F(10, 48) = 6.79, p < .001). Only perceived physical changes accounted for significant variance (R2 = .37, p < .001). In the final model, change in perceived body fat (ˇ = −.63, p < .001), and perceived strength (ˇ = .27, p = .02) were significant predictors of BASS change scores. Thus, greater decreases in perceived body fatness and perceived increases in strength were associated with bigger increases in body satisfaction. Social Physique Anxiety. The full set of predictors accounted for 47% of the variance in changes in SPAS scores from baseline to Week 16 (R2 adj = .37, F(10, 50) = 4.48, p < .001). Significant variance was explained by actual physical changes (R2 = .22, p = .02) and perceived physical changes (R2 = .25, p < .001). When all predictors were entered into the model, changes in perceived strength (ˇ = −.31, p = .01) and perceived body fat (ˇ = .50, p < .001) were the only significant predictors of SPAS scores. Thus, women who perceived the biggest increases in strength and decreases in body fatness had the largest reductions in social physique anxiety. Discussion This study is the first to examine the extent to which a comprehensive, theory-based set of variables can account for body image change among overweight and obese women participating in a weight-loss intervention. All three measures of body image yielded significant improvements over the measurement time points. Although there were some differences in the pattern of predictors across the body image measures and measurement time points, overall, the results supported our hypothesis that perceived physical changes and self-efficacy changes predict variance in body image change beyond that explained by actual physical changes. Furthermore, once all three blocks of predictors were entered into the regression models, perceived change variables

316

K.A. Martin Ginis et al. / Body Image 9 (2012) 311–317

Table 4 Hierarchical regression models predicting change in the body image measures from Weeks 1 to 16. Appearance Evaluation

Model 1 Age Strength Aerobic fitness Lean mass Fat percentage Model 2 Age Strength Aerobic fitness Lean mass Fat percentage Aerobic self-efficacy Strength self-efficacy Model 3 Age Strength Aerobic fitness Lean mass Fat percentage Aerobic self-efficacy Strength self-efficacy Strength perception Fat perception Endurance perception * ***

R2 change

ˇ

.20*

−.19 .03 .11 −.06 −.47

.02

.28***

Body Areas Satisfaction R2 change

ˇ

.16 .82 .38 .63 .001

.15

.01 .06 .06 −.15 −.38

−.17 .05 .10 −.07 −.42 .14 −.04

.21 .70 .44 .62 .01 .33 .79

.07

−.13 −.03 .17 .08 −.10 −.06 −.14 .22 −.57 .17

.27 .83 .14 .52 .48 .70 .24 .07 <.001 .30

.37***

p

Social Physique Anxiety R2 change

ˇ

.97 .68 .63 .28 .01

.22*

−.15 −.24 .20 −.02 .30

.26 .08 .11 .86 .02

.04 .08 .07 −.15 −.27 .27 .02

.78 .55 .62 .28 .07 .06 .88

.01

−.16 −.25 .19 −.02 .26 −.10 −.01

.23 .08 .14 .86 .07 .46 .96

.09 −.01 .14 .00 .10 −.01 −.09 .27 −.63 .27

.37 .94 .16 .99 .44 .96 .42 .02 <.001 .09

.25***

−.21 −.16 .16 −.18 .02 .01 .12 −.31 .50 .04

.08 .17 .17 .13 .87 .96 .33 .01 <.001 .83

p

p

p < .05. p < .001.

emerged as the most consistent unique predictors of body image change. These findings have implications for understanding potential mechanisms underlying diet- and exercise-induced body image change, and for developing interventions to improve body image in overweight and obese women. At both time points, and for all three measures of body image change, perceived changes and self-efficacy changes together accounted for more variance than actual physical change. These findings are consistent with the notion that changes in beliefs about one’s body, and not actual changes to the body, are the driving force behind body image change in exercise (Martin Ginis & Bassett, 2011; Martin Ginis et al., 2012) and weight-loss interventions. As a significant unique predictor in all six regression models, decreases in perceived body fatness were particularly important to body image change in the early weeks of the intervention as well as at the intervention end-point. Thus, the importance of perceived body fat to body image change does not seem to wane over the course of a weight-loss intervention. Given the centrality of reductions in body fat to achieving the female cultural body ideal (Schwartz & Brownell, 2004), this finding makes sense. That is, our overweight and obese study participants felt increasingly better about their bodies over the course of the trial, so long as they continued to perceive reductions in their body fatness. We also note that actual fat loss was a more consistent predictor of body image change at Week 16 than at Week 8, emerging as a significant predictor for all three body image change measures and playing its strongest role when fat loss was at its maximum. Some investigators have suggested that there is a weight-loss threshold that triggers body image improvements (Foster et al., 1997; Schwartz & Brownell, 2004). The present study was not designed to address this issue. However, based on our findings, we urge researchers to explore the weight-loss threshold concept as it relates to body image change. Changes in perceived strength also played a significant role in body image change, predicting 8 and 16-week BASS and SPAS change scores. These results are consistent with previous research

which showed that after a strength-training program, perceived increases in strength positively correlated with improvements in women’s body image (Martin Ginis, Eng, Arbour, Hartman, & Phillips, 2005). Taken together, these studies attest to the value of including strength training in body image interventions. Strength training may help shift a woman’s focus away from her appearance, and foster greater awareness of functional aspects of her body (Martin Ginis & Bassett, 2011; Martin Ginis et al., 2005). A shift in awareness, coupled with the positive feedback provided by a progressive strength-training regimen (i.e., the exerciser knows when she becomes stronger because she is lifting more weight during her workouts), may challenge, and ultimately alleviate, negative body image thoughts and feelings. Importantly, these benefits do not seem to diminish over the course of a strength-training intervention, suggesting they play a role in body image change even after the initial period of acute strength increases. Interestingly, self-efficacy explained significant variance in body image change (AE and SPAS scores) at the study mid-point, but not the end-point. These results could indicate that self-efficacy is a more potent source of body image information in the early stages of an exercise intervention, when beginner exercisers are still developing a sense of confidence in their physical capabilities. Once participants become assured of their capabilities, other types of information – such as perceptions of body fat and strength – become more relevant to the formulation of body image. The finding that aerobic self-efficacy was a stronger predictor of body image than strength self-efficacy likely reflects women placing greater value on aerobic exercise than strength training as a means for losing weight and managing body image concerns (cf. Prichard & Tiggemann, 2008). Increased confidence to do aerobic exercise could create the sense that the exerciser is getting closer to the body ideal, thus enhancing body image. The results of this study have important theoretical and clinical implications. From a theoretical standpoint, the lack of welldeveloped frameworks has been a barrier to studying the mechanisms of body image change within the context of weight-loss

K.A. Martin Ginis et al. / Body Image 9 (2012) 311–317

interventions (Martin Ginis et al., 2012). Given that the regression models explained 21–50% (adjusted R2 ) of variance in body image change, we now have evidence that the exercise and self-esteem model (Sonstroem & Morgan, 1989) variables provide a viable starting point for examining mechanisms underlying the effects of diet and exercise interventions. Because there were some differences in the pattern of significant predictors across the time points and measures, we recommend that investigators measure all of the proposed mechanisms when designing their studies. Determining the conditions when significant predictors emerge is an important step towards developing theory in this area. From a clinical perspective, our results underscore previous observations, that large changes in body composition are not necessary to produce significant improvements in body image (Foster et al., 1997; Martin Ginis & Bassett, 2011; Martin Ginis et al., 2012; Schwartz & Brownell, 2004). Rather, maximizing perceived changes is more important. This could be facilitated by helping women to develop realistic weight/fat loss goals that will make them feel good about even relatively small losses. Another strategy is to have women log their workouts, so that they have a record of their physical achievements. Drawing attention to improvements in strength and aerobic endurance will help exercisers become aware of changes in their capabilities, which may be especially important for improving body image in those who are just beginning an exercise regimen. Despite the many contributions of this study, a few limitations must be acknowledged. First, due to logistical restrictions in the larger trial, we were unable to measure actual aerobic fitness and strength at Week 8. Given the overall pattern of results, we doubt inclusion of these variables would have altered the results substantially. Nevertheless, we acknowledge that feedback from the aerobic and strength tests could have affected the magnitude of change in other variables in the model (i.e., self-efficacy, perceptions of strength and endurance). Second, the regression analyses consisted of predictor variables that were measured concurrent with the body image outcome measures. To determine whether the proposed mechanisms drive body image change, additional research is needed using prospective designs. Because the time course of change in the proposed mechanism variables is not yet clear, we elected to measure the variables cross-sectionally to maximize the likelihood of detecting an association and identifying potential mechanisms to be addressed in future studies. A third limitation is that we cannot tease apart the effects of the dietary versus the exercise component of the intervention on body image. However, given the importance of exercise to weight loss and its maintenance (Stiegler & Cunliffe, 2006), exercise should be considered a key component of any weight loss strategy particularly when the goal is to reduce body fat. Thus, the effects of a combined exercise and diet weight-loss intervention on body image, and the underlying mechanisms of change, are particularly relevant research issues. And finally, our 16-week intervention was relatively short. One consequence is that the intervention may have been too brief to elicit significant improvements in lean body mass which, in turn, limits conclusions that can be made regarding the role of lean mass change in body image change. Furthermore, interventions of longer duration may find a different pattern of body image change predictors at study mid-point and end-point. Investigators should continue to examine body image change predictors across time-points to identify mechanisms that should be targeted for intervention during the earlier versus later stages of a weightloss program. Despite these limitations, our study yields new, important information regarding the relative roles of potential mechanisms underlying change in body image following a weight-loss intervention. Overall, our results suggest that among overweight and obese women, perceived changes to the body play a more

317

important role than actual changes. Such information is integral to the ongoing development of theories to explain body image change, and the design of interventions to maximize body image improvements. Acknowledgments We gratefully acknowledge support from Dairy Farmers of Canada, the U.S. Dairy Research Institute, the Canadian Institutes of Health Research, and the Social Sciences and Humanities Research Council of Canada. We extend our appreciation to Adrienne Sinden for her assistance with manuscript preparation. References Astrand, P., & Rodahl, K. (1986). Textbook of work physiology (3rd ed.). New York: McGraw-Hill. Briefel, R. R., & Johnson, C. L. (2004). Secular trends in dietary intake in the United States. Annual Review of Nutrition, 24, 401–431. Brown, T. A., Cash, T. F., & Mikulka, P. J. (1990). Attitudinal body image assessment: Factor analysis of the body-self relations questionnaire. Journal of Personality Assessment, 55, 135–144. Campbell, A., & Hausenblas, H. A. (2009). Effects of exercise interventions on body image: A meta-analysis. Journal of Health Psychology, 14, 1–14. Cash, T. F. (1994). Body image and weight changes in a multisite comprehensive very-low-calorie diet program. Behavior Therapy, 25, 239–254. Foster, G. D., Wadden, T. A., & Vogt, R. A. (1997). Body image in obese women before, during, and after weight loss treatment. Health Psychology, 16, 226–229. Hart, E. H., Leary, M. R., & Rejeski, W. J. (1989). The measurement of social physique anxiety. Journal of Sport & Exercise Psychology, 11, 94–104. Health Canada. (2004). Canadian Community Health Survey, cycle 2.2, nutrition. Ottawa, ON: Health Canada. Josse, A. R., Atkinson, S. A., Tarnopolsky, M. A., & Phillips, S. M. (2011). Increased consumption of dairy foods and protein during diet- and exercise-induced weight loss promotes fat mass loss and lean mass gain in overweight and obese premenopausal women. The Journal of Nutrition, 141, 1626– 1634. Josse, A. R., Atkinson, S. A., Tarnopolsky, M. A., & Phillips, S. M. (2012). Diets higher in dairy foods and dietary protein support bone health during diet- and exerciseinduced weight loss in overweight and obese premenopausal women. Journal of Clinical Endocrinology and Metabolism, 97, 251–260. Kerksick, C., Thomas, A., Campbell, B., Taylor, L., Wilborn, C., Marcello, B., et al. (2009). Effects of a popular exercise and weight loss program on weight loss, body composition, energy expenditure and health in obese women. Nutrition and Metabolism, 6, 1–17. http://dx.doi.org/10.1186/1743-7075-6-23 Latner, J. D., & Wilson, R. E. (2011). Obesity and body image in adulthood. In T. F. Cash & L. Smolak (Eds.), Body image: A handbook of science, practice, and prevention (2nd ed., pp. 189–197). New York: Guilford Press. Legge, B. J., & Banister, E. W. (1986). The Astrand-Ryhming nomogram revisited. Journal of Applied Physiology, 61, 1203–1209. Marsh, H. W., Richards, G. E., Johnson, S., Roche, L., & Tremayne, P. (1994). Physical self-description questionnaire: Psychometric properties and a multitrait–multimethod analysis of relations to existing instruments. Journal of Sport and Exercise Psychology, 16, 270–305. Martin, K. A., Rejeski, W. J., Leary, M. R., McAuley, E., & Bane, S. M. (1997). Is the Social Physique Anxiety scale really multidimensional? Conceptual and statistical arguments for a unidimensional model. Journal of Sport and Exercise Psychology, 19, 360–368. Martin Ginis, K. A., & Bassett, R. L. (2011). Exercise and changes in body image. In T. F. Cash & L. Smolak (Eds.), Body image: A handbook of science, practice, and prevention (2nd ed., pp. 378–386). New York: Guilford Press. Martin Ginis, K. A., Bassett, R. L., & Conlin, C. (2012). Body image and exercise. In E. Acevedo (Ed.), Oxford handbook of exercise psychology (pp. 55–75). London: Oxford University Press. Martin Ginis, K. A., Eng, J. J., Arbour, K. P., Hartman, J. W., & Phillips, S. M. (2005). Mind over muscle? Sex differences in the relationship between body image change and subjective and objective physical changes following a 12-week strengthtraining program. Body Image, 2, 363–372. Prichard, I., & Tiggemann, M. (2008). Relations among exercise type, selfobjectification, and body image in the fitness centre environment: The role of reasons for exercise. Psychology of Sport and Exercise, 9, 855–866. Sarwer, D. B., Thompson, J. K., & Cash, T. F. (2005). Body image and obesity in adulthood. Psychiatric Clinics of North America, 28, 69–87. Schwartz, M. B., & Brownell, K. D. (2004). Obesity and body image. Body Image, 1, 43–56. Sonstroem, R. J., & Morgan, W. P. (1989). Exercise and self-esteem: Rationale and model. Medicine and Science in Sports and Exercise, 21, 329–337. Stiegler, P., & Cunliffe, A. (2006). The role of diet and exercise for the maintenance of fat-free mass and resting metabolic rate during weight loss. Sports Medicine, 36, 239–262. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston: Allyn and Bacon.