نقش انگیزه های خودبهبودی و خودارزیابی در مقایسه اجتماعی با بدن ایده آل زنان در رسانه ها

Abstract This study investigates the effect of social comparisons with media models on women's body image based on either self-evaluation or self-improvement motives. Ninety-eight women, for whom appearance was a relevant comparison dimension, viewed advertisements that did, or did not, feature idealised models, after being prompted to engage in self-evaluation or self-improvement comparisons. The results indicate that, when focusing on self-evaluation, comparisons with thin models are associated with higher body-focused anxiety than viewing no model advertisements. In contrast, when focusing on self-improvement, comparisons with thin models are not associated with higher body-focused anxiety than viewing no models. Furthermore, women's general tendency to engage in social comparisons moderated the effects of self-evaluative comparisons with models, so that women who did not habitually engage in social comparisons were most strongly affected. It is suggested that motive for social comparison may explain previous inconsistencies in the experimental exposure literature and warrants more careful attention in future research.

Introduction It has been widely reported that idealised media models displayed in the media are unrealistic both in terms of extreme thinness and artificially glamorized attractiveness (Grogan, 1999). A growing body of research indicates that exposure to ultra-thin models, ubiquitous in the media and advertising, leads to increased body dissatisfaction amongst a large proportion of women (e.g., Dittmar, Stirling, & Halliwell, 2004; Grogan, Williams, & Conner, 1996; Halliwell & Dittmar, 2004a; Heinberg & Thompson, 1995, Irving, 1990 and Posavac et al., 1998). Indeed, a meta-analysis assessing the results of 25 experimental studies demonstrated that young women feel worse after exposure to thin images than other types of images (Groesz, Levine, & Murnen, 2002). However, some studies have produced inconsistent results, whereby exposure to idealised female models has either no effect (e.g., Champion & Furnham, 1999; Irving, 1990) or a positive effect on women's self-evaluations (e.g., Henderson-King & Henderson-King, 1997; Joshi, Herman, & Polivy, 2004; Mills, Polivy, Herman, & Tiggemann, 2002; Myers & Biocca, 1992). Social comparison theory presents a theoretical framework that may explain these seemingly contradictory findings and, in a recent review of the area, Levine and Harrison (2004) argue that further investigations of appearance-related social comparisons are necessary in order to advance our understanding of media effects. Social comparison theory (Festinger, 1954) proposes that individuals continually evaluate themselves in order to assess their own standing on a wide range of characteristics. Originally, Festinger proposed that individuals make subjective comparisons with other people around them only in the absence of objective information about norms and standards, and that the aim of these comparisons is accurate self-evaluation. However, further developments in social comparison theory suggest that subjective social comparisons occur even when objective information is available (Marsh & Parker, 1984; Ruble, 1983) and that individuals are motivated to engage in social comparisons not only for self-evaluation, but also for other motives, such as self-improvement (Wood, 1989).1 The self-evaluation motive drives judgements about one's ability or standing on a dimension, whereas the self-improvement motive refers to attempts to learn how to, or to be inspired to, improve a particular attribute. The self-improvement motive generally prompts upward comparisons with targets deemed to be inspiring on a particular characteristic. Therefore, it is likely that women who are invested in their appearance, or are actively attempting to improve their appearance, see models as inspirational figures and use them as comparison targets for self-improvement. Theoretically, self-improvement comparisons can lead to positive or neutral outcomes, so long as the comparison target is not seen as a competitor (Wood, 1989). Wood (1989) argues that, on occasions, individuals are forced into upward comparisons, as is the case for comparisons with media models. Indeed, young women report making spontaneous comparisons with media models and TV actors (Smith & Leach, 2004). The majority of research on the impact of the media assumes that self-evaluation motivates social comparisons with models and, therefore, posits these upward comparisons will have detrimental effects. However, the inconsistencies in findings of exposure research may be explained by differences in women's motives for social comparisons with models (Martin & Gentry, 1997; Mills et al., 2002). This proposition has received surprisingly little research attention in relation to body image. Two recent papers demonstrate that restrained and unrestrained eaters respond differently to media exposure; specifically that various self-perceptions of restrained eaters are enhanced or unaffected by viewing thin models. Mills et al. (2002) found that restrained eaters who viewed thin models reported more positive appearance self-esteem than restrained eaters who viewed plus-size models, and they also rated their ideal and actual body size as smaller than restrained eaters who viewed plus-size models or no model, control images. Similarly, Joshi et al., 2004 found that restrained eaters reported higher social self-esteem and a more positive self-image after exposure to thin models than to control images. Furthermore, restrained eaters’ appearance self-esteem did not differ after viewing thin models or control images, whereas viewing thin models had a negative impact on unrestrained eaters. Mills et al. (2002) propose that restrained eaters use media models as inspirational figures, because these models represent an ideal self on an important dimension, and that, therefore, the outcome of social comparisons with models can be self-enhancing or neutral. Mills et al. (2002) and Joshi et al. (2004) were primarily interested in comparing the outcomes of comparisons made by restrained and unrestrained eaters and, therefore, did not manipulate or measure the nature of the social comparison processes involved. Although one may speculate about differences in comparison motives, this methodology does not allow for direct inferences about the processes underlying these effects. To our knowledge, only one previous study explicitly manipulated motivation for appearance comparisons, focusing on pre-adolescent and adolescent girls (Martin & Gentry, 1997). It examined three types of motive, using pre-exposure instructions to manipulate girls’ motives for social comparisons with media models. There were five exposure conditions. Four presented models, but with different motives: self-evaluation, self-improvement, self-enhancement through a downward comparison, or self-enhancement through discounting the beauty of the models. The fifth condition presented an advertisement with no model and no comparison instructions. Consistent with hypotheses, self-perceptions of physical attractiveness were lower amongst girls who engaged in self-evaluative social comparisons with models than amongst girls who had not viewed models. However, there were no significant differences in self-evaluations in the self-improvement, self-enhancement, and control conditions. These findings pertain to young girls, who are likely to engage in appearance-related social comparisons particularly frequently, because they are concerned with evaluating the dramatic appearance changes they are experiencing during adolescence as they attempt to conform to adult ‘beauty’ norms (Martin & Gentry, 1997). Therefore, it is unclear whether these findings are relevant for adult women. In addition, this study contrasts the effects of comparisons with models when focused on self-evaluation, self-improvement or self-enhancement with a condition in which there is no model and no comparison instruction. It is therefore unclear whether negative outcomes from self-evaluation comparisons would occur when merely asking women to evaluate their appearance, even in the absence of a model, or whether negative effects are dependent on the presence of an upward comparison target. In order to address these issues, the present study examines the impact of priming self-evaluation and self-improvement motives on young adult women when viewing advertisements that feature either thin models or products only. We predict that, when focusing on self-evaluation, social comparisons with models will be associated with higher levels of body-focused anxiety than viewing product only advertisements. In contrast, when a self-improvement motive predominates in comparisons with media models, the goal is to learn how to improve or be inspired, therefore these comparisons should not lead to negative experiences of the body, but they would also not necessarily lead to positive self-evaluations. Therefore, we predict that social comparisons with models when focusing on self-improvement will not be associated with higher body-focused anxiety than viewing product only advertisements. Theoretically, social comparisons should only have psychological impact if the domain of comparison is relevant to an individual (Tesser, 1988); upward comparisons on a dimension of little relevance do not pose a threat to the individual's self-evaluation or present a relevant target for self-improvement. Indeed, several studies demonstrated that only women for whom thinness is personally relevant react negatively to thin-ideal exposure (Dittmar & Howard, 2004; Dittmar et al., 2004 and Halliwell and Dittmar, 2004a; Heinberg & Thompson, 1995). Specifically, only women scoring above the mid-point on a measure of internalization of sociocultural attitudes toward appearance were negatively impacted by social comparisons with ultra-thin media models (Dittmar & Howard, 2004; Halliwell & Dittmar, 2004a). Therefore, the present study specifically aimed to recruit women for whom appearance is a relevant social comparison dimension and who, therefore, are invested in their appearance. However, there is evidence that other individual difference variables moderate media exposure amongst women who are invested in their appearance. Dittmar and Howard (2004) found that social comparison tendency moderated the impact of viewing attractive average-size models. Women who were high on internalization reported a negative effect after exposure to average-size models compared to control images if they habitually engaged in appearance-related social comparisons, whereas a relief effect occurred if they did not. In contrast, the only other study to investigate this failed to find empirical support for social comparisons as a moderator of exposure to idealized female bodies (Tiggeman & McGill, 2004). However, possible interactive effects between social comparison and internalization were not addressed in this study. Therefore, social comparison tendency was examined in the present study as a moderator. The analysis was exploratory; it may be that the impact of prompted self-improvement or self-evaluative comparisons may be more or less powerful amongst women who habitually engage in appearance-related social comparisons. Theoretically, the impact of social comparisons with idealised bodies could also depend on an individual's chronic self-discrepancies. Women who internalise sociocultural attitudes towards appearance will vary in their discrepancies between the way they actually look and the way they would ideally like to look. The impact of self-evaluative upward social comparisons with models may be most strongly associated with negative outcomes in individuals with large pre-existing appearance-related self-discrepancies. Although research has demonstrated that actual-ideal appearance-related self-discrepancies are associated with body-focused negativity and eating disturbance (e.g., Dittmar et al., 2004 and Halliwell and Dittmar, 2004b; Szymanski & Cash, 1995), no research has investigated the role of adult women's chronic self-discrepancies as a moderator of exposure to idealized bodies.

Results Preliminary analyses Checking for possible differences between women assigned to each of the four exposure conditions, we found no significant differences in BMI, F(3, 93) = 1.36, ns, but there was an effect for participant age, F(3, 94) = 5.19, p < .01. The mean age of women in the no-model evaluation condition was 24.20 years (SD = 8.03), in the no-model improvement condition 24.10 years (SD = 6.89), in the model evaluation condition 19.62 years (SD = 3.58), and in the no-model improvement condition 19.82 years (SD = 2.44). We decided to control for both age and BMI in all further statistical analyses. Multivariate analysis of variance confirmed that were no significant differences between women assigned to each condition in terms of internalisation and the proposed moderators, λ = .85, F(9, 194) = 1.55, ns. Moreover, none of the univariate analyses were significant: internalisation, F(2, 82) = 1.88, ns; social comparison tendencies, F(2, 82) = 1.77, ns; and appearance-related self-discrepancies, F(2, 82) = 2.20, ns. Appearance-related discrepancies and appearance-related social comparisons were positively related to each other, r (88) = .43, p < .001. Social comparison tendency was significantly related to body-focused anxiety amongst women who viewed the no model advertisements, r (41) = .40, p = .01, but not amongst women who viewed models, r (52) = .08, ns. In contrast, self-discrepancies were significantly related to body-focused anxiety regardless of whether women viewed advertisements with models, r (52) = .37, p < .05, or without, r (41) = .39, p < .05. Motive for comparison We conducted an ANCOVA with body-focused anxiety as the dependent variable, controlling for age and BMI. The exposure conditions factor was partitioned a priori into a set of orthogonal contrasts that represent three theoretically derived contrasts,6 which allow us to test our hypotheses directly. The first contrast compared the effect of priming a self-evaluation motive with priming a self-improvement motive. The second contrast compared viewing thin models with viewing no models after priming a self-evaluative motive. The third and final contrast compared viewing thin models with viewing no models after priming a self-improvement motive. As well as examining the main effect for comparison motive (represented by the first contrast), this ANCOVA in fact constitutes a simple effects analysis (Keppel et al., 1992), where the impact of each self-comparison motive was examined separately, comparing the effect of exposure to a model against no-model (represented by the second and third contrasts, which test our two hypotheses directly). The difference in levels of body-focused anxiety reported by women in the self-evaluation and self-improvement conditions approached significance, F(1, 96) = 3.45, p = .067, partial η2 = .04, with self-evaluative comparisons (M = 2.76, SD = .83) associated with higher anxiety than self-improving comparisons (M = 2.44, SD = .81). Testing our first hypothesis directly, the findings for the second comparison showed a significant effect F(1, 96) = 6.40, p < .05, partial η2 = .07, supporting our prediction that, amongst participants in the self-evaluation condition, body-focused anxiety was significantly higher after viewing models (M = 2.98, SD = .69) than after viewing no model advertisements (M = 2.54, SD = .91). The final, third comparison—testing our second hypothesis directly—revealed a non-significant effect F(1, 96) = .49, ns, partial η2 = .01. As expected, amongst participants in the self-improvement condition, there was no difference in levels of body focused anxiety after viewing models (M = 2.46, SD = .86) compared to viewing no models (M = 2.49, SD = .74). Tests of potential moderators A stepwise hierarchical multiple regression analysis was conducted to examine whether self-discrepancies or social comparison tendency moderated the effects of comparison motive and exposure on women's body-focused anxiety. Following the procedure outlined by Baron and Kenny (1986), a moderation effect is identified if the interaction between the proposed moderator, self-discrepancies or social comparison, and the predictor variable, in this case the two central condition contrasts, explains a significant proportion of the variance in the outcome variable: body-focused anxiety. As correlations revealed a differential relationship according to model exposure between body-focused anxiety and social comparison tendency, but not between body-focused anxiety and self-discrepancies, interactions with social comparison were entered first.7 Predictors were entered in eight steps, and the increase in variance explained assessed at each step. Age and BMI were entered first in order to control for them (Step 1), and then the three condition contrasts (effect-coded, Step 2). Next, we entered social comparison tendencies (mean-centred, Step 3), self-discrepancies (mean-centred, Step 4), and then—to test for moderation—two-way interactions between social comparison tendencies and the condition contrasts (Step 5), and two-way interactions between self-discrepancies and the condition contrasts (Step 6). In the final two steps, the interaction between self-discrepancies and social comparison tendency was entered (Step 7), and three-way interactions between the proposed moderators and the condition contrasts (Step 8). The main focus of interest concerns interactions between the exposure contrasts and the proposed moderators, i.e., Steps 5 and 6. Age and BMI increased the variance explained in women's body-focused anxiety by 4%, ΔF(2, 90) = 2.07, ns, where higher BMI was associated with higher anxiety (β = .21; p < .05). The condition contrasts added a further 8%, ΔF(2, 87) = 2.78, p < .05, and, as we know from the ANOVA reported, the self-evaluation comparison instruction led to marginally higher anxiety (β = .18, p = .09), significantly so when women viewed advertisements with models compared to no models (β = −.23, p < .05). The impact of social comparison tendency (SC) approached significance and added 3%, ΔF(1, 86) = 3.54, p = .06, β = .20; p = .06. Chronic appearance-related self-discrepancies (SD) made a significant contribution to the model, 9%, ΔF(1, 85) = 9.77, p < . 01, β = .34; p < .01. The next two steps are of central importance to the present research. The addition of the interactions between SC and the condition contrasts proved marginally significant, 7%, ΔF(3, 82) = 2.61, p = .057. In contrast, the addition of interactions involving SDs proved non-significant, 2%, ΔF(3, 79) = .88, ns. 8 Thus, Step 5 produced the final significant model, which offered the best fit and a significant regression model, F(10, 82) = 3.75, p < .001, accounting for 31% of the variance in women's body-focused anxiety. Table 2 reports the regression coefficients and statistics for Step 5. With respect to the moderating effect of SC, the interaction between social comparison tendency and the self-evaluation model/no model comparison contrast was significant, β = .22, p < .05. Table 2. Summary of hierarchical regression analysis testing potential moderators (Step 5) Variable B SE B β Age −.02 .02 −.15 BMI .08 .03 .26* C1: self-evaluation (SE) vs. self-improvement (SI) .70 .32 .21* C2: models vs. no models for SE instruction −.24 .24 −.10 C3: models vs. no models for SI instruction .20 .24 .08 Social comparison (SC) .13 .14 .10 Self-discrepancies (SD) .01 .00 .39** C1 × SC −.10 .53 −.09 C2 × SC .86 .38 .22* C3 × SC .68 .38 .18 * p < .05. ** p < .01. Table options The nature of the moderator effect for the contrast of self-evaluative comparisons between viewing a model compared to a no model image was examined in a subsequent analysis, which is a regression analogue to a simple effects analysis for interpreting interactions in an analysis of variance (Jaccard, Turrisi, & Wan, 1990). The difference in anxiety after viewing models or no models in the self-evaluation condition was calculated at different levels of social comparison tendency. The mean level of social comparison was 3.17, just above 3 the mid-point of the scale. At this level of social comparison tendency, there was a significant difference between levels of body-focused anxiety reported after viewing models and no models, t(41) = 1.95, p = .05, d = .6. The mean anxiety level is, on average, .48 of a scale point higher after exposure to models. This constitutes a moderate effect, given that it corresponds to a mean difference of just over half a standard deviation ( Cohen, 1988). Amongst women who reported above average social comparison tendencies, no significant differences between anxiety after exposure to models compared to no models were found. This was true at half a scale point above the mean for social comparison, t(41) = .28, ns, d = .10, at one point above the mean, t(41) = −.67, ns, d = −.31, and .5 points above the mean, t(41) = −1.09, ns, d = −.88. In fact, amongst women who habitually engage in social comparisons, body-focused anxiety is somewhat lower after exposure to models than to no models. Amongst women who do not habitually engage in appearance-related social comparisons, the difference between anxiety reported in the model and no model conditions is significant. Differences in body-focused anxiety were significant at half a scale point below the mean for social comparison tendency, t(41) = 2.63, p < .05, d = 1.09, at one scale point below the mean, t(41) = 2.58, p < .05, d = 1.59, and at 1.5 scale points below the mean, t(41) = 2.46, p < .05, d = 2.08. The effect sizes were moderate to large. In summary, women who do not habitually engage in appearance-related social comparison appear to show particularly negative effects from self-evaluative social comparison with advertisements that depict ultra-thin models.