The structure and validity of directional measures of appearance social comparison among emerging adults in China

Body Image 11 (2014) 464–473

Contents lists available at ScienceDirect

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

The structure and validity of directional measures of appearance social comparison among emerging adults in China夽 Jun Liao, Todd Jackson ∗ , Hong Chen School of Psychology, Southwest University, Chongqing, China

a r t i c l e

i n f o

a b s t r a c t

Article history: Received 20 February 2014 Received in revised form 1 July 2014 Accepted 2 July 2014 Keywords: Upward appearance social comparisons Downward appearance social comparisons Factor structure Validity Body image Chinese

We evaluated the structure and validity of the Upward Appearance Comparison Scale (UPACS) and Downward Appearance Comparison Scale (DACS) (O’Brien et al., 2009) in Chinese samples. In Study 1, principal component analysis on an initial sample (427 women, 123 men) and confirmatory factor analysis on another sample (447 women, 121 men) found that a 15-item, two component model had the best overall fit. Derived components had moderate correlations with most conceptually related measures and low correlations with less conceptually related indices. Study 2 participants (310 women, 201 men) completed the UPACS and DACS as well as measures of disordered eating, fatness concern, and negative affect; they were re-assessed one year later. Baseline UPACS scores predicted changes in disordered eating for women and fatness concerns for men, independent of initial disturbances, but DACS responses were not related to outcomes. Findings highlighted the potential utility of derived UPACS and DACS within a Chinese context. © 2014 Elsevier Ltd. All rights reserved.

Introduction Social comparison is a possibly innate process that helps people to understand ambiguous circumstances and evaluate their attitudes, attributes and abilities with those of others (e.g., Buunk & Gibbons, 2007; Festinger, 1954; Wood, 1996). While “similar” others are often sought as targets of social comparisons, “downward” comparisons with less skilled or less fortunate others may be made to improve perceptions of one’s self or circumstances (Wills, 1981) and “upward” comparisons with more skilled or more fortunate others can reflect a desire to improve one’s status or skills (Collins, 1996). Appearance social comparisons have had a central role in recent theory and research on body image disturbances. For example, frequent physical appearance comparisons with other people and media images have been linked to body dissatisfaction, weight regulation, and disordered eating (e.g., Jones, 2001, 2004; Keery, Van den Berg, & Thompson, 2004; Shroff & Thompson, 2006; Thompson, Coovert, & Stormer, 1999; van den Berg, Thompson,

夽 This research was supported by grants from the National Natural Science Foundation of China (31371037 and 31170981), Education Ministry of China. ∗ Corresponding author at: School of Psychology, Southwest University, Chongqing 400715, China. Tel.: +86 13883224482. E-mail addresses: [email protected], [email protected] (T. Jackson). http://dx.doi.org/10.1016/j.bodyim.2014.07.001 1740-1445/© 2014 Elsevier Ltd. All rights reserved.

Obremski-Brandon, & Coovert, 2002), as well as explicit and implicit anti-fat attitudes (O’Brien, Hunter, Halberstadt, & Anderson, 2007). Recently, O’Brien et al. (2009) developed directional measures to evaluate the potentially distinct effects of upward and downward appearance comparisons on attitudes and functioning among Australian university students (60% women). Principal components analyses (PCA) resulted in a 10-item Upward Physical Appearance Comparison Scale (UPACS) that assesses comparisons of oneself with more physically attractive others and an eight-item Downward Physical Appearance Comparison Scale (DACS) that reflects comparisons of oneself with others judged to be less physically attractive. The scales were found to have satisfactory internal consistency, stability over two weeks, and construct validity. For example, relative to DACS scores, UPACS scores had stronger relations to negative appearance evaluations and eating problems. Conversely, DACS scores had stronger links with anti-fat attitudes. Subsequently, Vartanian and Dey (2013) reported that both the UPACS and DACS have high alphas, modest correlations with age and body mass index (BMI), and moderate associations with a media ideal internalization scale among university-age Australian women. Although each measure has promise, O’Brien et al. (2009) acknowledged the scales should be evaluated more fully in diverse samples. Because research on the UPACS and DACS has been limited to Australian samples, it is not clear whether they are structurally equivalent and valid in other cultures. Appearance comparisons

J. Liao et al. / Body Image 11 (2014) 464–473

are highly relevant in collectivist nations such as China where sensitivity to standards within one’s social network are important to self-definitions, useful for achieving personal ends, and crucial for gaining social approval (Jung & Lee, 2006). Notably, body dissatisfaction and disordered eating have become increasingly common in Chinese samples (e.g., Chen & Jackson, 2008; Chen, Jackson, & Huang, 2006; Jackson & Chen, 2010b), and frequent appearance comparisons correlate reliably with such concerns (Chen, Gao, & Jackson, 2007; Chen & Jackson, 2012; Jackson & Chen, 2007, 2008c, 2008d, 2010b). Hence, the utility of directional appearance comparison scales warrants consideration in this cultural milieu. Another issue of note regarding the UPACS and DACS was their high correlation with one another (r = .66) in O’Brien et al.’s (2009) sample. This correlation indicated people who made frequent upward appearance comparison also made more downward comparisons. As a result, unique correlates of elevations in upward versus downward comparisons may be obscured by substantial overlaps the scales have with appearance comparison frequency. While brief situational effects of appearance comparison direction can be illuminated through experimental manipulations involving exposure to upward or downward comparison targets (e.g., Galioto & Crowther, 2013), identifying and comparing subgroups reporting predominant upward versus predominant downward appearance comparison tendencies is a strategy that may clarify less contextually bound correlates of each tendency. On one hand, the tendency to make upward comparisons with more physically attractive others should correspond to valuing unrealistic attractiveness ideals and heightened body image/eating disturbances (e.g., Keery et al., 2004; Leahey, Crowther, & Mickelson, 2007; Myers, Ridolfi, Crowther, & Ciesla, 2012), while downward comparison tendencies might correlate with positive self-concept (Wills, 1981). On the other hand, downward comparisons with the less fortunate are sometimes made by overwhelmed people to protect self-esteem or bolster optimism about their circumstances, while upward comparisons with “better-off” peers are more likely to be made, among people whose self-esteem is secure, to improve skills or strive for growth (see Collins, 1996). Assessing differences between groups with stronger upward versus stronger downward appearance comparison tendencies may help to clarify experiences unique to each tendency. Finally, prospective studies have identified frequent appearance comparisons as a risk factor for later increases in body image and eating concerns (e.g., Chen & Jackson, 2009a, 2009b; Jackson & Chen, 2008a, 2011; Jones, 2004). However, to our knowledge, prospective effects of appearance comparison direction have been limited to intervals of several days rather than months or years (e.g., Leahey et al., 2007; Myers et al., 2012). For example, Leahey et al. (2007) found that American college women who compared themselves to more attractive reference group members experienced more negative affect, body dissatisfaction, and thoughts of exercising over one week, while downward appearance comparisons with others perceived as less attractive predicted decreases in negative affect and body dissatisfaction. Extensions are needed to assess long-term ramifications of upward and downward appearance comparison on these outcomes. Based on this overview, two studies evaluated the UPACS and DACS in Chinese samples. Study 1 assessed the structure of UPACS and DACS items via PCA and confirmatory factor analysis (CFA) on independent samples as well as the construct validity of derived components. In addition, differences between subgroups with predominant upward versus predominant downward appearance comparison tendencies were explored on demographic and psychosocial measures common to both samples. In Study 2, we evaluated the status of responses on derived UPACS and DACS as risk factors for exacerbations in eating disturbances, fatness concerns, and negative affect over 12 months in a third sample.

465

Study 1 Following from O’Brien et al.’s (2009) findings, we hypothesized that the structure of UPACS and DACS items would reflect moderately correlated, distinct upward and downward appearance comparison components. Furthermore, based on the premise that derived appearance comparison components should correlate with theoretically related constructs from the same content domains (i.e., body image influences/outcomes or social comparison processes), it was hypothesized that UPACS, in particular, and DACS, to a lesser extent, would have significant positive correlations with other measures of appearance comparisons, pressure, and investment, disturbances in eating and body image, and selfesteem contingent on friendship quality (e.g., Patrick, Neighbors, & Knee, 2004). Given that preferences for attractiveness ideals reflect upward comparisons, UPACS scores were expected to have positive relations with thin feminine ideal preferences while DACS scores were not. Conversely, based on O’Brien et al. (2009), UPACS scores were expected to have a weaker positive correlation with anti-fat attitudes than DACS scores would. Appearance comparison components were also expected to have weaker correlations with sociodemographics (age within emerging adulthood, household income) and general psychological constructs (e.g., individual selfworth, coping strategies used for non-specific stressors, positive affect) that lie outside the domains of body image and social comparison. The exception to this premise was negative affect which was expected to have moderate correlations with both UPACS and DACS scores due to its inclusion in models of disordered eating (Stice, 2001) and moderate relations to other sociocultural measures and eating problems in Chinese adolescents (e.g. Jackson & Chen, 2011). Method Participants. PCA sample. Participants were first year students (427 women, 123 men) attending Southwest University (SWU), Chongqing. They ranged in age between 16 and 24 years (M = 19.35 years, SD = 1.04) and were predominantly of Han majority ethnicity (84.73%), or members of Zhuang (3.09%), Miao (2.90%), Tujia (2.72%), or one of 12 other ethnic minorities. The sample had a mean body mass index (BMI) of 19.97 (SD = 2.15; M = 19.68, SD = 2.01 for women; M = 20.97, SD = 2.31 for men); 8.5% had a BMI over 23, the level at which risk for obesity-related diseases is elevated in Asia-Pacific samples (WHO, 2000). CFA sample. An independent sample of first year SWU students (447 women, 121 men) between 15 and 23 years of age (M = 19.30 years, SD = 1.02) participated. For ethnicity, 85.7% were Han, 4.05% were Tujia, 1.76% were Miao, and 8.49% were from 16 other minorities. In this sample, 11.44% of participants had a BMI over 23 (M = 20.12, SD = 2.24 overall; M = 19.77, SD = 2.00 for women; M = 21.39, SD = 2.60 for men). The sample did not differ from the PCA sample on age, F(1, 1116) = 1.78, p = .221,  = .04, ethnicity, 2 (18) = 0.15, p = .735 or average BMI, F(1, 1116) = 0.46, p = .265,  = .03. Measures. Demographics. Age, sex, ethnicity, and estimated monthly household income (below 1000 Yuan, 1000–2999 Yuan, 3000–4999 Yuan, 5000 Yuan or above) were assessed. BMI was calculated from self-reported height and weight which correlates highly with objective BMI measurement (e.g., Gorber, Tremblay, Moher, & Gorber, 2007). Upward Physical Appearance Comparison Scale (UPACS) and Downward Physical Appearance Comparison Scale (DACS). The

466

J. Liao et al. / Body Image 11 (2014) 464–473

10-item UPACS and eight-item DACS (O’Brien et al., 2009) are directional appearance comparison scales. Items are rated from 1 = never to 5 = always and summed to derive total scale scores reflecting more frequent upward and downward comparisons. As summarized above, the scales had satisfactory psychometrics in Australian samples (O’Brien et al., 2009; Vartanian & Dey, 2013). Physical Appearance Comparison Scale – Revised (PACS). The four-item PACS (Shroff & Thompson, 2006) assesses frequency of physical appearance comparison with others. Items are rated from 1 = never to 5 = always and summed; higher total scores reflect more frequent comparisons. PCA in Chinese samples replicated the initial structure in both sexes (Jackson & Chen, 2008a). Significant correlations with other eating disorder risk factors, body image concerns, and eating disturbances support the scale’s construct validity in Chinese samples (e.g., Chen et al., 2007; Jackson & Chen, 2008a, 2008b, 2011). The PACS had alphas of ˛ = .80 and ˛ = .80 in PCA and CFA samples, respectively. Eating Disorder Diagnostic Scale (EDDS). The 18-item EDDS symptom composite (Stice, Telch, & Rizvi, 2000) assessed overall eating disturbances by summing responses to each standardized item; higher scores reflected more eating pathology. The composite is reliable, stable, and has excellent concordance with related indices among American women (Stice et al., 2000). In Chinese samples of both sexes, PCA found all 18 items load on one component (Jackson & Chen, 2008a). EDDS composite scores are also internally consistent and have convergent validity in Chinese samples (e.g., Jackson & Chen, 2008a, 2008b, 2011, 2014). Alphas were ˛ = .84 and ˛ = .82 in PCA and CFA samples, respectively. Negative Physical Self Scale – Fat Concern (NPS-Fat). This 11item scale (Chen et al., 2006) was developed in Chinese samples and taps thoughts, feelings, behaviors and projections (e.g., “I think I am fat in others’ eyes”) related to having a large body size. Items are rated from 0 (not at all like me) to 4 (very much like me) and summed; higher total scores indicating more fatness concerns. The scale is reliable, stable, and valid (e.g., Chen et al., 2006, 2007; Chen & Jackson, 2007; Jackson & Chen, 2008c, 2008d). In the PCA sample, its alphas was ˛ = .91. Ideal Body Stereotype Scale Revised (IBSS-R). The 10-item IBSSR (Stice, 2001) assesses preferences for physical features (e.g., thin, long legs, shapely) that characterize a thin feminine ideal. Items were rated between 1 = Definitely disagree and 5 = Definitely agree and summed; higher total scores indicate stronger thin ideal preferences. The scale has satisfactory psychometrics in American samples (Stice, 2001), a univariate structure in Chinese samples (Jackson & Chen, 2008b), and utility in predicting later increases in disordered eating among young women in each culture (Jackson & Chen, 2008b; Stice, 2001). Alphas were ˛ = .81 (PCA sample) and ˛ = .80 (CFA sample). Concern about Appearance Scale (CAAS). Carver et al. (1998) developed the four-item CAAS to assess investment in appearance as a source of self-worth. Items (e.g., “It’s important to me to look my best all the time”) are rated from 1 (strongly disagree) to 5 (strongly agree). Higher summed scores reflect more investment in appearance. Carver et al. noted the CAAS is reliable (˛ = .78) and stable (r = .75) over four weeks. A PCA in the initial sample found a univariate structure comprising all items (KMO = .90, p < .001) and explained 53.89% of the scale variance. Alphas were ˛ = .71 (PCA sample) and ˛ = .69 (CFA sample). Sociocultural Attitudes Towards Appearance Questionnaire-3 Pressure subscale (SATAQ-3-Pressure). Thompson, van den Berg, Roehrig, Guarda, and Heinberg (2004) developed these seven items to tap reported pressure to conform to media portrayals of physical appearance. Items are rated from 1 = Definitely disagree to 5 = Definitely agree; higher summed scores reflect more perceived pressure. The scale is reliable and valid in American samples (e.g., Thompson et al., 2004). A single factor structure, high reliability,

and convergent validity were evident in Chinese samples that used the SATAQ-3 Pressure subscale (Chen & Jackson, 2012; Jackson & Chen, 2010b). In PCA and CFA samples, its alphas were ˛ = .88. Perceived Sociocultural Pressure Scale – Interpersonal (PSPSInt). Six PSPS items (Stice & Agras, 1998) assessed perceived pressure to alter one’s appearance from friends, family, and dating partner (or best friend) from 1 (Not at all) to 5 (Very much). Higher summed scores reflect more perceived interpersonal pressure. PCA on Chinese students found six interpersonal PSPS items load together on one component while two PSPS mass media items load together on a separate component (Jackson & Chen, 2010b). In this sample, the scale had alphas of ˛ = .82 (PCA sample) and ˛ = .81 (CFA sample). Rosenberg Self-Esteem Scale – Chinese (RSES-C). The nineitem RSES-C (Song, Cai, Brown, & Grimm, 2011) includes five-item Self-Appreciation and four-item Self-Derogation components. One other item (“I wish I could have more respect for myself”) loads on neither component (Song et al., 2011). Each item is rated from 1 (strongly disagree) to 4 (strongly agree). Higher summed scores reflect higher levels on each component. Alphas were satisfactory for Self-Appreciation (˛ = .81 and ˛ = .77 for PCA and CFA samples, respectively) and Self-Derogation (˛ = .76 (PCA sample) and ˛ = .75 (CFA sample)). Except for NPS-Fatness Concerns, the above measures were used in both samples. Conversely, the self-report scales described below were assessed only in the CFA sample. Dissatisfaction with Body Parts Scale (DBPS). DBPS (Bearman, Presnell, Martinez, & Stice, 2006) scores were calculated by summing ratings of dissatisfaction with nine body parts from 1 = extremely satisfied to 5 = extremely dissatisfied; higher scores represented more overall dissatisfaction. The DBPS is reliable and valid in American samples (e.g., Bearman et al., 2006). In Chinese samples, the scale had univariate structures in four different age-groups (Jackson & Chen, 2011) and satisfactory psychometrics (Chen & Jackson, 2012; Jackson & Chen, 2011, 2014). Its alpha was ˛ = .86 in this research. Anti-fat Attitudes Questionnaire – Dislike subscale (AFAQ-D). Crandall’s (1994) AFAQ-D examines explicit prejudice toward fat people (e.g., “I don’t like fat people much”). Items are scored from 0 (very strongly disagree) to 9 (very strongly agree), with higher summed scores indicating more prejudice. A PCA run on the CFA sample found six items loaded on one component and accounted for 46.43% of the variance (KMO = .82, p < .001). The item that failed to load (“I don’t have many friends who are fat”) may reflect prejudice less in China where obesity base rates are lower than rates in some developed nations. The PCA-derived scale had an alpha of ˛ = .81 in the CFA sample. Friendship Contingent Self-Esteem Scale (FCSE). The eightitem FCSES (Cambron, Acitelli, & Steinberg, 2010) was added to examine self-worth contingent upon the quality of one’s friendships. Items were rated between 1 (very unlike me) and 5 (very much like me), with higher summed scores representing more friendshipcontingent self-worth. In the CFA sample, a PCA replicated the original structure, explaining 45.80% of total variance (KMO = .88, p < .001). Its alpha was ˛ = .83 in the CFA sample. Positive and Negative Affect Scale (PANAS). The 20-item PANAS (Watson, Clark, & Tellegen, 1988) assessed positive affect and negative affect during the last month. Item are rated between 1 (none or little of the time) and 4 (most of the time), and summed; higher total scores reflect more affect. The original PANAS structure has been replicated in Chinese samples, with the exception that “alert” loads and is included on negative affect rather than positive affect (Jackson & Chen, 2008a). Alphas were ˛ = .81 for positive affect and ˛ = .82 for negative affect in this sample. Coping. Based on evidence of low correlations between upward/downward comparisons and coping (e.g., Jackson & Phillips, 2011), brief general coping scales were added to assess

J. Liao et al. / Body Image 11 (2014) 464–473

discriminant validity, specifically, the Measure of Current Status (MOCS; Antoni et al., 2006) Awareness of Tension and Coping Confidence subscales, and COPE-Denial subscale (Carver, Scheier, & Weintraub, 1989). On each scale, items were rated from 0 (not at all) to 4 (extremely well), with higher summed scores representing more frequent use of the relevant strategy. In the CFA sample, PCA replicated MOCS subscale structures and explained 58.82% of the total variance (KMO = .83, p < .001). A PCA of Denial items also yielded a one-factor solution, explaining 55.83% of total variance (KMO = .71, p < .001). Alphas were ˛ = .71 for Awareness of Tension, ˛ = .77 for Coping Confidence, and ˛ = .73 for Denial. Procedure. The human research ethics committee at SWU approved the study. Initially, class advisors of first year students were contacted about a new study on body image among university students. In turn, advisors informed their advisees about the research. Interested students under age 18 needed parental or guardian permission to participate. Volunteers completed the study in a classroom setting during December, 2012 (PCA sample) or April, 2013 (CFA sample). They were given a survey packet consisting of a cover page describing the main research focus, an informed consent form, and the measures above. All English- language scales had undergone Mandarin to English back-translations previously by two bilingual Mandarin-English-speakers from Foreign Languages at SWU. Discrepancies from original versions were discussed between translators and the corresponding author to better approximate intended meanings. Students were told to carefully read and honestly respond to each item. The research took 25–40 min to complete. Eight Yuan was provided as remuneration. After analyses, a summary of the study purposes and results was forwarded to the pool of potential participants via their class advisors. Results Principal Components Analysis (PCA). PCA were performed using SPSS FACTOR to identify components underlying upward and downward appearance comparison items. Extracted components were judged to be adequate when their eigenvalues exceeded 1.0 and the values for corresponding components (e.g., 1, 2, 3) derived from parallel analysis (O’Connor, 2000). Items loading on each surviving component were retained when (1) they loaded >.40 on one component, (2) they loaded <.30 on other components, and (3) their content was relevant to that of all other items loading on the same component (Jackson & Chen, 2010a). Following the approach taken by O’Brien et al. (2009), a PCA with direct oblimin rotation was performed, although UPACS and DACS items were included together in one analysis rather than two separate PCAs to determine whether they loaded on unique components. Values of the Kaiser–Meyer–Olkin Measure of Sampling Adequacy (KMO = .92, Bartlett’s Test, p < .0001) revealed an interpretable solution. The analysis yielded components with eigenvalues of 7.33, 2.42, and 1.10. The first component had seven UPACS items and accounted for 40.72% of the variance. The second component had all eight DACS items and explained 13.43% of the variance. The third component comprised three other UPACS items and explained 6.12% of the variance. Eigenvalues for the first and second components exceeded respective parallel analysis values of 1.33 and 1.27 but the third component was not retained because its eigenvalue was smaller than the corresponding parallel analysis value (1.22). Item loadings and communalities are presented in Table 1. Alphas were ˛ = .89 for the seven-item UPACS and ˛ = .88 for the eight-item DACS. Within each sex, PCA performed on the 15 surviving items completely replicated the two component solution. Confirmatory Factor Analyses (CFA). AMOS 20.0 tested fits of the 15-item PCA-derived model of appearance comparisons, O’Brien et al.’s (2009) 18-item model, and single component

467

solutions based on 15 and 18 items, respectively. CFA models were assessed using the Comparative Fit Index (CFI) and Tucker-Lewis coefficient (TLI) as incremental fit indices, the Root Mean Square Error of Approximation (RMSEA) and Standardized Root Mean Squared Residual (SRMR) as residuals-based indices, and the chisquare, associated degrees of freedom (CMIN/df ratio), and p-value. No modification indices were used to improve hypothesized model fits. Table 2 indicates the PCA-derived model approached or exceeded thresholds of acceptability of .90–.95 for the CFI and TLI, .05–.08 for the RMSEA and SRMR, and 2–5 for CMIN/df (Hu & Bentler, 1999; Jackson & Chen, 2010a; Kline, 2005), and had better fits than those obtained for O’Brien et al.’s original model, which were adequate except for the TLI. In contrast, fits for both one factor solutions were poor. CFA performed within each sex yielded the same pattern of results. In sum, the PCA-derived two component model was retained for subsequent analyses. Validity Analyses. Relations of derived components with demographics and self-report scales were evaluated with analyses of variance (ANOVA), chi-square tests, and bivariate correlation coefficients. A p < .01 significance level was set to reduce the risk of Type I errors. Within each sample, standard scores on each component were calculated by dividing the sum of responses to all relevant items by the total number of items comprising that component. Univariate ANOVAs indicated PCA and CFA samples did not differ on mean UPACS, F(1, 1117) = 0.76, p = .382,  = .001, or DACS, F(1, 1117) = 0.02, p = .890,  < .01, scores. Repeated-measures ANOVA indicated average UPACS scores (PCA sample: M = 2.85, SD = 0.78, CFA sample: M = 2.77, SD = 0.77) exceeded mean DACS scores (PCA sample: M = 2.39, SD = 0.69, CFA sample: M = 2.25, SD = 0.69) in both the PCA sample, F(1, 548) = 257.26, p < .001,  = .32, and the CFA sample, F(1, 567) = 243.59, p < .001,  = .30, thus replicating O’Brien et al.’s (2009) findings. Regarding sex differences, UPACS scores were higher for women (M = 2.78, SD = 0.78) than for men (M = 2.57, SD = 0.79) in the PCA sample, F(1, 548) = 7.08, p = .008,  = .11; however, average DACS scores did not differ between women (M = 2.27, SD = 0.66) and men (M = 2.12, SD = 0.74), F(1, 548) = 5.10, p = .024,  = .10. In the CFA sample, mean UPACS scores were elevated among women (M = 2.81, SD = 0.74) relative to men (M = 2.61, SD = 0.85), F(1, 566) = 6.52, p = .011,  = .11; conversely, average DACS scores did not differ between CFA sample women (M = 2.27, SD = 0.68) and men (M = 2.17, SD = 0.73), F(1, 566) = 2.01, p = .156,  = .06. The convergent validity of derived components was supported by moderate positive associations with all conceptually related scales (appearance comparison frequencies, appearance pressure, investment in appearance, eating disturbances, fatness concerns, contingent self-esteem), except body dissatisfaction and negative affect. Anti-fat attitudes were linked more strongly to DACS than UPACS scores as hypothesized, but the difference between these correlations was not significant (ZH = −1.56, p = .118). Regarding discriminant validity, derived components had low correlations with measures falling outside the specific domain of body image (age, BMI, income, self esteem, positive affect, coping) (Table 3). Based on formulas from Hoerger (2013), relations between UPACS and DACS scores and convergent validity measures were stronger than relations between UPACS and DACS scores and discriminant validity indices, except for body dissatisfaction and negative affect. When analyses were rerun in each sex, strengths of relation were replicated, except for UPACS-disordered eating correlations among men in PCA (r = .196, p = .030) and CFA (r = .18, p = .056) samples. Sub-Group Analyses. To assess how tendencies toward making upward versus downward appearance comparisons differed from one another on measures completed by all participants, PCA and CFA samples were pooled together and subgroups more strongly oriented toward each tendency were identified. Due to the lack

468

J. Liao et al. / Body Image 11 (2014) 464–473

Table 1 Principal components analysis results for upward and downward appearance comparison components among Chinese university students (N = 550). Item

Component 1

8. At parties or other social events, I compare my physical appearance to the physical appearance of very attractive people. 7. When I see good-looking people I wonder how I compare to them. 9. I find myself comparing my appearance with people who are better looking than me. 6. When I see a person with a great body, I tend to wonder how I ‘match up’ with them. 5. I tend to compare myself to people who I think look better than me. 10. I compare my body to people who have a better body than me. 4. At the beach or athletic events (sports, gym, etc.) I wonder if my body is as attractive as people I see there who have very attractive bodies. 17. I often compare myself to those who are less physically attractive. 14. I compare myself to people less good looking than me. 18. I tend to compare my physical appearance with people whose bodies are not as physically appealing. 16. At parties I often compare my looks to the looks of unattractive people. 13. At the beach, gym, or sporting events I compare my body to those with less athletic bodies. 15. I think about how attractive my body is compared to overweight people. 12. I tend to compare my body to those who have below average bodies. 11. When I see a person who is physically unattractive I think about how my body compares to theirs. 1. I tend to compare my own physical attractiveness to that of magazine models. 2. I find myself thinking about whether my own appearance compares well with models and movie stars. 3. I compare myself to those who are better looking than me rather than those who are not.

Communality 2

3

.81

.01

.03

.61

.79 .79

.04 .09

.07 .13

.67 .65

.78

−.05

−.04

.68

.72 .69 .68

.04 .13 −.11

.39 .15 .02

.42 .57 .57

−.07 −.04 −.07

.89 .87 .86

.09 .01 .05

.75 .73 .69

.01 .07

.82 .69

.14 −.05

.70 .53

−.02 .07 .29

.65 .64 .52

.01 −.17 −.08

.40 .48 .53

−.01

−.01

.85

.10

.05

.76

.72

.09

−.09

.64

.63

51

Note. Factor loadings of items on each component are in boldface.

of precedents to guide subgroup selection, participants oriented toward upward comparisons (n = 98) were selected based on having z scores .333 above and below UPACS and DACS medians, respectively. Conversely, those oriented toward downward comparisons (n = 78) had z scores .333 above and below, respective DACS and UPACS medians. Use of more extreme z score cut-offs (.500) resulted in an insufficient number of participants with predominant downward comparison tendencies. Derived subgroups did not differ on age, t(174) = 0.45, p = .651,  = .02, sex, 2 (1) = 0.07, p = .794,  = .20, ethnicity, 2 (1) = 1.60, p = .206,  = .10, income, 2 (4) = 9.91, p = .078,  = .21, or BMI, t(174) = 1.92, p = .057,  = .03. However, a multivariate ANOVA revealed an overall difference between appearance comparison tendency subgroups, F(8, 167) = 8.10, p < .001,  = .28. Table 4 indicates the upward comparisons subgroup reported more frequent appearance comparisons with others, stronger investment in physical appearance as a source of self-worth, and stronger thin ideal preferences. The upward comparison subgroup also had marginally higher and lower RSES Self-Appreciation (p = .017) and Self-Derogation (p = .018) scores, respectively. No other differences emerged. Supplementary Analyses. Based on reviewer comments highlighting the need for concise measures that reduce response burdens, particularly in clinical contexts, PCA were re-run using the highest loading five items from each scale (Table 1). The analysis yielded a five-item UPACS that had an eigenvalue of 4.99

and explained 49.92% of the scale variance and a five-item DACS having an eigenvalue of 1.82 and accounting for 18.20% of the variance. In the CFA sample, all model fits were acceptable (CFI = 0.960, TLI = 0.947, RMSEA = 0.078, SRMR = 0.045, CMIN/df = 4.45). In each sample, relations with more and less conceptually related measures replicated the pattern shown in Table 3. Discussion PCA and CFA yielded interpretable solutions largely in line with DACS and UPACS derived by O’Brien et al. (2009). Specifically, the 15-item two component model from PCA had the best fit from a priori alternatives in the CFA sample and may have utility in Chinese samples. The original DACS structure was replicated but three UPACS items failed to load with related items in PCA. These three items also had the lowest loadings on the UPACS in O’Brien et al.’s sample. One excluded item differs from all other UPACS/DACS items in that it involves the choice of comparing oneself to better-looking people rather than those who are not. The other two excluded items reflected comparisons with fashion models and movie stars instead of people in general as all other items do. Based on evidence that body satisfaction is reduced from exposure to pictures of attractive women perceived to be peers but not those of professional models (Cash, Cash, & Butters, 1983), O’Brien et al. (2009) posited that conscious distinctions are made between realistic and unrealistic comparison targets. PCA findings may support this view, as content

Table 2 Goodness-of-fit indices for hypothesized appearance comparison models in confirmatory factor analyses sample (N = 568). Model

Description

p

CMIN/df

TLI

CFI

RMSEA

SRMR

PCA-derived O’Brien et al. One-component (1) One-component (2)

15 items, two components 18 items, two components 15 items, PCA solution, one component 18 items, O’Brien et al., one component

.001 .001 .001 .001

3.74 4.41 17.71 15.44

.933 .895 .592 .555

.943 .908 .650 .608

.070 .078 .172 .160

.053 .058 .145 .136

Note. PCA = principal components analysis.

J. Liao et al. / Body Image 11 (2014) 464–473

469

Table 3 Convergent and discriminant validity of derived upward (UPACS) and downward (DACS) appearance comparison scales in Study 1 samples. Measure

PCA sample

Upward Appearance Comparisons Downward Appearance Comparisons Appearance Comparison Frequency Perceived Media Appearance Pressure Perceived Interpersonal Appearance Pressure Investment in Physical Appearance Thin Ideal Preferences Eating Disturbances Fatness Concerns Body Dissatisfaction Negative Affect Friendship Contingent Self-Esteem Anti-Fat Attitudes-Dislike Self-Appreciation Self-Derogation Positive Affect Denial Awareness of Tension Coping Confidence Age Body mass index Income

CFA sample

UPACS

DACS

UPACS

DACS

– .53** .62** .50** .47** .61** .28** .26** .26** – – – – .04 .05 – – – – −.10 −.11 −.11

.53** – .46** .41** .43** .47** .09 .26** .29** – – – – −.04 .12* – – – – −.11 .05 −.05

– .40** .56** .38** .45** .54** .22** .27** – .08 .17** .40** .17** .05 .02 .00 .04 −.01 .01 −.11 −.01 −.11

.40** – .35** .37** .33** .41** .09 .23** – .12* .12** .32** .24** −.04 −.10 −.06 .12* −.07 −.14* −.08 .07 −.05

Note. * p < .01. ** p < .001. Table 4 Differences between subgroups having predominant upward (UPACS) and predominant downward (DACS) appearance comparison tendencies (N = 176). Measure

Orientation subgroup High UPACS

Appearance Comparison Frequencies Perceived Media Appearance Pressure Perceived Interpersonal Appearance Pressure Investment in Physical Appearance Thin Ideal Preferences Eating Disturbances Self-Appreciation Self-Derogation

High DACS

M

SD

M

SD

11.66 22.33 14.62 17.66 13.21 1.88 16.36 7.72

2.92 6.45 3.88 2.38 2.50 8.65 2.53 2.62

9.91 20.96 15.45 16.00 11.79 0.83 15.42 8.76

2.47 5.86 4.02 3.94 2.79 7.91 2.58 2.98

F



17.89* 2.11 1.91 11.98* 12.63* 0.69 5.84 5.98

.30 .11 .10 .25 .26 .06 .18 .18

Note. * p < .001.

reflecting distal media targets did not load with UPACS content reflecting potentially proximal targets in one’s environment. Mean UPACS and DACS scores were comparable to those reported in O’Brien et al.’s (2009) sample, suggesting that appearance social comparisons are as salient to young adults in China as they are for cohorts in Australia. Extending evidence of sex differences in appearance comparison frequencies in Chinese samples (e.g., Chen & Jackson, 2012; Jackson & Chen, 2008c, 2011), women had higher UPACS scores than men did in line with O’Brien et al.’s (2009) results. This sex difference may explain, in part, why eating and body image disturbances are more common among young women (e.g., Chen & Jackson, 2008, 2012). The discriminant validity of derived UPACS and DACS was supported by low correlations with demographics and general constructs (i.e., positive affect, self-esteem, coping) having little conceptual relevance to the body image domain. Convergent validity support was evidenced by moderate, positive associations between UPACS/DACS scores and responses on 10 of 12 measures related to the body image domain (Table 3). However, low correlations of UPACS/DACS to body dissatisfaction were unexpected given that body dissatisfaction has been linked previously to frequent appearance comparisons (e.g., Jones,

2001, 2004; Thompson, Coovert, et al., 1999). Notably, these studies focused on comparisons about specific attributes such as weight, shape, hips, stomach, and thighs. By contrast, general appearance comparisons, as reflected in Study 1, can have low, non-significant correlations with overall body dissatisfaction derived from rating specific body parts (e.g., van den Berg et al., 2007). Negative affect was also expected to have stronger relations with UPACS/DACS scores in light of its role in approaches such as the dual pathway model (Stice, 2001). Regardless, because negative affect scale content is not specific to the body image domain relative to scales that tap distressing emotional experiences such as body shame and fear of negative appearance evaluation, weaker relations social comparison measures reflecting physical appearance are at least understandable. Also contrary to results from O’Brien et al. (2009), anti-fat attitudes did not have significantly higher correlations with DACS than UPACS scores. In part, this discrepancy may reflect lower base rates of obesity in Chinese samples (Chen & Jackson, 2012) and more emphasis on features such as facial appearance in defining attractiveness in Chinese contexts (Chen et al., 2006). However, given dramatic rises in obesity among Chinese youth during the last 30 years (Yu et al., 2012), links between anti-fat attitudes and

470

J. Liao et al. / Body Image 11 (2014) 464–473

appearance comparison direction should continue to be tracked in future work. Subgroup analyses indicated students with stronger upward than downward comparison tendencies were more likely to make appearance comparisons with others, embrace a thin feminine ideal, and invest in their appearance. While it was not clear whether such comparison tendencies reflected stable dispositions, prospective effects of frequent appearance comparisons (e.g., Jackson & Chen, 2008b, 2011; Jones, 2004) and thin ideal preferences (e.g., Jackson & Chen, 2008b; Stice, 2001) on exacerbations in body image and eating problems suggest enduring upward comparison tendencies, in particular, increase risk for disturbances over time. However, the upward comparison subgroup also had higher and lower self-appreciation and self-derogation levels, respectively. These effects were only marginally significant and cannot be overstated without replication but are in line with conclusions of Collins’ (1996) review indicating upward social comparisons are sometimes self-enhancing and correspond to higher self-esteem and more positive self-evaluations. Finally, supplementary analyses indicated that five-item UPACS and DACS were psychometrically sound in both samples. Of note, these may not be the “best” five-item scales for use in non-Chinese samples and the optimal number of items such scales should have (e.g., four, five, or six) is not clear. Nonetheless, these brief versions may have utility in Chinese samples and provide foundations for extensions in other countries. In sum, the structure and construct validity of derived UPACS and DACS were largely supported among urban young adults in China. However, their status as risk factors for later increases in disturbances could not assessed due to the use of cross-sectional designs in each sample. Study 2 addressed this limitation.

Study 2 Past research has found directional appearance comparisons predict changes in body dissatisfaction, eating disturbances and negative affect (e.g., Leahey et al., 2007; Myers et al., 2012), but prospective studies have not been conducted beyond one week. Several lines of evidence suggest elevated UPACS scores increase longer-term risk for poor outcomes while DACS scores do not. First, frequent appearance comparisons predict later increases in weight and eating concerns in Chinese samples nine-18 months later (e.g., Chen & Jackson, 2009a; Jackson & Chen, 2008a, 2008b, 2011). Given that PACS scores, fatness concerns, and disordered eating all had stronger relations with UPACS scores than DACS scores in Study 1, frequent baseline upward comparisons, in particular, should predict more fatness and eating concerns at follow-up. Conversely, weak correlations of UPACS and DACS scores with negative affect in Study 1 suggested their status as risk factors may not extend to general affective distress. Second, effects of upward appearance comparisons on short-term outcomes (e.g., Leahey et al., 2007; Myers et al., 2012) suggest such comparisons could have negative long-term implications. Third, while Leahey et al. (2007) reported downward comparisons predicted decreases in negative affect and body dissatisfaction for college women, DACS scores were positively related to these responses in Study 1 and O’Brien et al. (2009). Hence, it seemed unlikely that high initial DACS scores would predict decreases in distressing outcomes. Following this rationale, we hypothesized that high baseline (T1) UPACS scores would predict increases in disordered eating and fatness concerns but not negative affect at 12 month followup (T2), independent of significant demographics and T1 levels on each criterion measure. Conversely, baseline DACS scores were not expected to make unique contributions to related prediction models.

Method Participants and Procedure. Of 613 first year SWU students (277 men, 336 women) who participated at T1, 83.4% (201 men, 310 women) completed the T2 assessment. Participants ranged from 17 to 22 years old at T1 (M = 18.80 years, SD = 0.91) and were typically Han (84.50%), Zhuang (4.10%), Miao (2.70%), Tujia (2.00%), or members of 17 other minorities. The mean T1 BMI was 20.06 (SD = 2.14, range: 15.43–31.74) and 7.6% had a BMI over 23. T1 BMIs were similar for women (M = 19.71, SD = 2.61 and men (M = 20.60, SD = 2.00). More women than men completed both study phases, 2 (1) = 42.48, p = .001,  = .26. Within each sex, T2 completers and non-completers did not differ on most indices. However, among men, completers were younger, F(1, 272) = 21.97, p < .001,  = .08, and had fewer T1 fatness concerns, F(1, 272) = 15.15, p < .001,  = .05, than did non-completers. Among women, completers were younger, F(1, 334) = 8.02, p < .005,  = .02, and had higher mean T1 UPACS, F(1, 334) = 11.47, p < .001,  = .03, and DACS, F(1, 334) = 6.28, p = .013,  = .02, scores than did non-completers. Procedures were identical to those of Study 1, except that smaller batteries of previously used scales were completed in a new sample during December, 2012 and December, 2013. Typically assessments took 10–20 min each to complete and students received eight Yuan as compensation for each assessment. Measures. Age, sex, ethnicity, income, height, and weight were demographics assessed in Study 2. Date of birth and mother’s maiden name were also solicited to permit matching of otherwise de-identified baseline and follow-up data from each student. The seven-item UPACS and eight-item DACS derived from Study 1 were completed at T1 and had alphas of ˛ = .82 and ˛ = .85, respectively. Alphas were also acceptable for the EDDS composite (T1: ˛ = .84, T2: ˛ = .86), NPS-Fat (T1: ˛ = .88, T2: ˛ = .89), and negative affect (T1: ˛ = .80, T2: ˛ = .84). Results Within each sex, bivariate correlations were first run between T1 age, income, BMI, UPACS, DACS and criterion measures (T2 eating disturbances, fatness concerns, negative affect). Neither age (all ps > .27) nor income (all ps > .13) was related any T2 outcome for either sex. However, BMI correlated with T2 eating disturbances (women: r = .24, p < .001, men: r = .29, p < .001) and fatness concerns (women: r = .50, p < .001, men: r = .55, p < .001). UPACS scores were related to T2 fatness concerns (women: r = .28, p < .001, men: r = .32, p < .001) and negative affect (women: r = .13, p < .023, men: r = .16, p = .021) as well as eating disturbances for women (r = .32, p < .001) but not men: (r = .10, p = .168). DACS scores correlated with T2 fatness concerns within each sex (women: r = .22, p < .001, men: r = .20, p = .003), negative affect among men (r = .15, p = .039) but not women (r = .10, p = .080) and eating disturbances for women (r = .20, p < .001) but not men (r = .07, p = .335). Next, the impact of T1 UPACS and DACS scores having correlations (p < .05) with each criterion (i.e., T2 levels of eating disturbances, fatness concerns, negative affect) were assessed after controlling for baseline responses on the criterion (Block 1) and BMI (Block 2), where relevant. This strategy generated estimates of relations between initial UPACS or DACS scores and each T2 criterion, independent of T1 criterion responses or overlaps between UPACS and DACS scores (Jones, 2004). Among women, substantial variance in T2 eating disturbances was explained by T1 eating disturbances (ˇ = .64, Adj. R2 = .408, p < .001). The effect for BMI was not significant but was included in the model to be conservative because it approached significance (ˇ = .08, R2 Ch. = .006, p = .067). Controlling for these factors, T1 UPACS scores added to the model (ˇ = .11, R2 Ch. = .010, p = .024)

J. Liao et al. / Body Image 11 (2014) 464–473

but DACS scores did not (ˇ = .02, p = .744). For men, T2 eating disturbances were predicted by T1 disturbances (ˇ = .42, Adj. R2 = .173, p < .001) and BMI (ˇ = .15, R2 Ch. = .019, p = .033) but neither UPACS nor DACS gained entry within the model due to non-significant relations with the criterion. Fatness concerns among women at T2 were predicted by T1 fatness concerns (ˇ = .77, Adj. R2 = .592, p < .001) and BMIs (ˇ = .12, R2 Ch. = .010, p = .006). Beyond these factors, effects of UPACS (ˇ = .06, p = .126) and DACS (ˇ = .01, p = .850) were negligible. For men, T1 fatness concerns (ˇ = .73, Adj. R2 = .528, p < .001) and BMIs (ˇ = .14, R2 Ch. = .013, p = .020) contributed to the model. Controlling for these factors, UPACS scores had an impact (ˇ = .12, R2 Ch. = .014, p = .015) but DACS scores did not (ˇ = .06, p = .221). Finally, T2 negative affect among women was predicted by T1 negative affect (ˇ = .77, Adj. R2 = .291, p < .001) but UPACS scores failed to add to the model (ˇ < .01, p = .969). For men, after controlling for T1 negative affect (ˇ = .73, Adj. R2 = .248, p < .001), T1 scores on the UPACS (ˇ < .01, p = .970) and DACS (ˇ < .01, p = .967) did not add to the model. Discussion In sum, high T1 UPACS scores predicted exacerbations in eating disturbances among women and fatness concerns among men one year later, independent of baseline concerns and BMIs. Past work has reported frequent appearance comparisons are a risk factor for losses of body satisfaction (Jones, 2004) and body esteem (Chen & Jackson, 2009b) as well as increases in disordered eating (Jackson & Chen, 2008a, 2011). Frequent upward comparisons also predicted increases in emotional distress, body dissatisfaction, and thoughts of exercising over one week (Leahey et al., 2007). Building upon these findings, Study 2 implicated upward appearance comparison as a risk factor for increases in eating and body image concerns over an extended interval. As predicted, contributions of baseline UPACS scores were specific to body image-related outcomes and did not extend to negative affect. Also as hypothesized, T1 DACS scores did not predict decreased T2 disturbances at one year follow-up in line with Study 1 findings and O’Brien et al. (2009). While downward appearance comparisons predicted temporary decreases in negative affect and body dissatisfaction in Leahey et al.’s (2007) sample, this study suggested any such effects are time-limited or situation-bound. Study 2 findings have possible implications for preventing fatness concerns and disordered eating within a Chinese cultural milieu. First, high positive correlations between baseline and 12-month follow-up disturbances suggested these problems are highly entrenched among those most affected. As a result, singlesession treatments or brief interventions may not produce lasting change (Jones, 2004). Unique effects of UPACS scores on later concerns suggested such comparisons warrant attention within treatment protocols. For example, extending the logic of supported approaches (e.g., Stice, Marti, Rohde, & Shaw, 2011), having program enrollees critically evaluate advantages and disadvantages of making upward appearance comparisons with peers may produce dissonance that weakens the pursuit of unrealistic attractiveness ideals as a way of judging one’s appearance or value. General Discussion In sum, Study 1 found unique upward and downward appearance comparison dimensions in applying O’Brien et al.’s (2009) appearance social comparison scales among emerging adults in China. Study 1 findings also supported the convergent and discriminant validity of each component in this context and identified unique experiences that differentiated subgroups having

471

predominant upward versus predominant downward appearance comparison tendencies. The overall pattern of results underscored the salience of directional appearance comparisons for young urban Chinese adults. In Study 2, high initial UPACS scores predicted increases in disordered eating and fatness concerns one year later for young Chinese women and men, respectively. However, effects of baseline UPACS scores did not extend to negative affect and initial DACS scores failed to bolster prediction models for all outcomes. Notwithstanding its implications, the main limitations of this research should be noted. First, even though emerging adults in China, particularly women, are at-risk for body image and eating disturbances (Chen & Jackson, 2008; Chen et al., 2006), extensions to younger and older age groups as well as clinical samples are needed to elucidate generalizability. Second, given that Study 2 completers were less likely than non-completers to be men or, among women, to have lower baseline UPACS and DACS scores, findings may be less applicable to young Chinese men or young Chinese women who do not report making frequent appearance comparisons in either direction. Third, because findings were based on self-reports, the use of informant perceptions in assessment protocols may clarify whether self-reported appearance comparisons affect functioning beyond self-perceptions. These limitations aside, this research indicated that UPACS and DACS factor structures are conceptually similar among mainland Chinese young adults relative to Australian young adult samples evaluated in earlier work. To establish whether upward and downward appearance comparison tendencies are culturally universal, further evaluations are needed in samples from other countries. Upward appearance comparisons were reported to be relatively more common within each of our samples and predicted negative outcomes related to body image over a one year interval in Study 2. Hence, future longitudinal research should assess whether upward appearance social comparison scales, in particular, are more useful than frequency measures of appearance social comparison in predicting disturbances in body image and eating over time. Finally, examining interventions targeting upward and downward comparison may clarify whether treatment-induced changes in appearance comparisons are markers for treatment effectiveness.

Acknowledgements This research was supported by the National Natural Science Foundation of China (r#31371037 and #31170981), Education Ministry of China, a Chongqing 100 Persons Fellowship, and the BaYu Scholar Program of the Chongqing Government.

References Antoni, M., Lechner, S., Kazi, A., Wimberly, S., Sifre, T., Urcuyo, K., Phillips, K., Glück, S., & Carver, C. S. (2006). How stress management improves quality of life after treatment for breast cancer. Journal of Consulting and Clinical Psychology, 74, 1143–1152. http://dx.doi.org/10.1037/0022-006X.74 Bearman, S. K., Presnell, K., Martinez, E., & Stice, E. (2006). The skinny on body dissatisfaction: A longitudinal study of adolescent girls and boys. Journal of Youth and Adolescence, 35, 217–229. http://dx.doi.org/10.1007/s10964-005-9010-9 Buunk, A. P., & Gibbons, F. X. (2007). Social comparison: The end of a theory and the emergence of a field. Organizational Behavior and Human Decision Processes, 102, 3–21. http://dx.doi.org/10.1016/j.obhdp.2006.09.007 Cambron, M. J., Acitelli, L. K., & Steinberg, L. (2010). When friends make you blue: The role of friendship contingent self-esteem in predicting self-esteem and depressive symptoms. Personality and Social Psychology Bulletin, 36, 384–397. http://dx.doi.org/10.1177/0146167209351593 Carver, C. S., Pozo-Kaderman, C., Price, A. A., Noriega, V., Harris, S. D., Derhagopian, R. P., Robinson, D. S., & Moffat, F. L. (1998). Concern about aspects of body image and adjustment to early stage breast cancer. Psychosomatic Medicine, 60, 168–174. http://dx.doi.org/10.1037/0278-6133.20.1.20

472

J. Liao et al. / Body Image 11 (2014) 464–473

Carver, C. S., Scheier, M. F., & Weintraub, J. K. (1989). Assessing coping strategies: A theoretically based approach. Journal of Personality and Social Psychology, 56, 267–283. http://dx.doi.org/10.1037/0022-3514.56.2.267 Cash, T. F., Cash, D. W., & Butters, J. W. (1983). “Mirror, mirror on the wall. . .?”: Contrast effects and self-evaluations of physical attractiveness. Personality and Social Psychology Bullletin, 9, 359–364. http://dx.doi.org/10.1177/0146167283093004 Chen, H., Gao, X., & Jackson, T. (2007). Predictive models for understanding body dissatisfaction among young males and females in China. Behaviour Research and Therapy, 45, 1345–1356. http://dx.doi.org/10.1016/j.brat.2006.09.015 Chen, H., & Jackson, T. (2007). Stability of body image concerns among Chinese adolescents: Nine-month test–retest reliabilities of the Negative Physical Self Scale. Perceptual and Motor Skills, 105, 677–680. http://dx.doi.org/10.2466/pms. 105.2.677-680 Chen, H., & Jackson, T. (2008). Prevalence and sociodemographic correlates of eating disorder endorsements among adolescents and young adults from China. European Eating Disorders Review, 16, 375–385. http://dx.doi.org/10.1002/erv.837 Chen, H., & Jackson, T. (2009a). Predictors of changes in weight esteem among mainland Chinese adolescents: A longitudinal analysis. Developmental Psychology, 45, 1618–1629. http://dx.doi.org/10.1037/a0016820 Chen, H., & Jackson, T. (2009b). Predictors of changes in body image concerns of Chinese adolescents. Journal of Adolescence, 32, 977–994. http://dx.doi.org/10.1016/ j.adolescence.2008.08.002 Chen, H., & Jackson, T. (2012). Gender and age group differences in mass media and interpersonal influences on body dissatisfaction among Chinese adolescents. Sex Roles, 66, 3–20. http://dx.doi.org/10.1007/s11199-011-0056-8 Chen, H., Jackson, T., & Huang, X. (2006). The Negative Physical Self Scale: Initial development and validation in samples of Chinese adolescents and young adults. Body Image, 3, 401–412. http://dx.doi.org/10.1016/j.bodyim.2006.07.005 Collins, R. L. (1996). For better or worse: The impact of upward social comparison on self-evaluations. Psychological Bulletin, 119, 51–69. http://dx.doi.org/10.1037/ 0033-2909.119.1.51 Crandall, C. S. (1994). Prejudice against fat people: Ideology and self-interest. Journal of Personality and Social Psychology, 66, 882–894. http://dx.doi.org/10.1037/ 0022-3514.66.5.882 Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117–140. http://dx.doi.org/10.1177/001872675400700202 Galioto, R., & Crowther, J. (2013). The effects of exposure to slender and muscular images on male body dissatisfaction. Body Image, 10, 566–573. http://dx.doi.org/ 10.1016/j.bodyim.2013.07.009 Gorber, S. C., Tremblay, M., Moher, D., & Gorber, B. (2007). A comparison of direct vs. self-report measures for assessing height, weight and body mass index: A systematic review. Obesity Reviews, 8, 307–326. http://dx.doi.org/10.1111/ j.1467-789X.2007.00347.x Hoerger, M. (2013). ZH : An updated version of Steiger’s Z and web-based calculator for testing the statistical significance of the difference between dependent correlations. Retrieved from http://www.psychmike.com/dependent correlations.php Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. http://dx.doi.org/10.1080/10705519909540118 Jackson, T., & Chen, H. (2007). Identifying the eating disorder symptomatic in China: The role of sociocultural factors and culturally defined appearance concerns. Journal of Psychosomatic Research, 62, 241–249. http://dx.doi.org/10.1016/ j.Jpsychores.2006.09.010 Jackson, T., & Chen, H. (2008a). Predicting changes in eating disorder symptoms among Chinese adolescents: A 9-month prospective study. Journal of Psychosomatic Research, 64, 87–95. http://dx.doi.org/10.1016/j.jpsychores. 2007.08.015 Jackson, T., & Chen, H. (2008b). Predicting changes in eating disorder symptoms among adolescents in China: An 18 month prospective study. Journal of Clinical Child and Adolescent Psychology, 37, 874–885. http://dx.doi.org/10.1080/ 15374410802359841 Jackson, T., & Chen, H. (2008c). Sociocultural predictors of physical appearance concerns among adolescent girls and young women from China. Sex Roles, 58, 402–411. http://dx.doi.org/10.1007/s11199-007-9342-x Jackson, T., & Chen, H. (2008d). Sociocultural influences on body image concerns of young Chinese males. Journal of Adolescent Research, 23, 154–171. http://dx.doi.org/10.1177/0743558407310729 Jackson, T., & Chen, H. (2010a). Factor structure of the Sociocultural Attitudes Towards Appearance Questionnaire-3 (SATAQ-3) among adolescent boys in China. Body Image, 7, 349–355. http://dx.doi.org/10.1016/j.bodyim. 2010.07.003 Jackson, T., & Chen, H. (2010b). Sociocultural experiences of bulimic and non-bulimic adolescents in a school-based Chinese sample. Journal of Abnormal Child Psychology, 38, 69–76. http://dx.doi.org/10.1007/s10802-009-9350-0 Jackson, T., & Chen, H. (2011). Risk factors for disordered eating during early and middle adolescence: Prospective evidence from mainland Chinese boys and girls. Journal of Abnormal Psychology, 120, 454–464. http://dx.doi.org/10.1037/a0022122 Jackson, T., & Chen, H. (2014). Risk factors for disordered eating during early and middle adolescence: A two year longitudinal study of Mainland Chinese girls and boys. Journal of Abnormal Child Psychology., http://dx.doi.org/10.1007/s10802-013-9823-z Jackson, T., & Phillips, H. (2011). Social comparison performance standards, threat, and tolerance for experimentally induced pain. European Journal of Pain, 15, 1081–1086. http://dx.doi.org/10.1016/j.ejpain.2011.04.008

Jones, D. C. (2001). Social comparison and body image: Attractiveness comparisons to models and peers among adolescent girls and boys. Sex Roles, 45, 645–664. http://dx.doi.org/10.1023/A:1014815725852 Jones, D. C. (2004). Body image among adolescent girls and boys: A longitudinal study. Developmental Psychology, 40, 823–834. http://dx.doi.org/10.1037/ 0012-1649.40.5.823 Jung, J., & Lee, S. H. (2006). Cross-cultural comparisons of appearance selfschema, body image, self-esteem, and dieting behavior between Korean and US women. Family and Consumer Sciences Research Journal, 34, 350–365. http://dx.doi.org/10.1177/1077727X06286419 Keery, H., Van den Berg, P., & Thompson, J. K. (2004). An evaluation of the tripartite influence model of body dissatisfaction and eating disturbance with adolescent girls. Body Image, 1, 237–251. http://dx.doi.org/10.1016/j.bodyim. 2004.03.001 Kline, R. B. (2005). Principles and practice of structural equation modeling. New York: Guilford. Leahey, T. M., Crowther, J. H., & Mickelson, K. D. (2007). The frequency, nature, and effects of naturally occurring appearance-focused social comparisons. Behavior Therapy, 38, 132–143. http://dx.doi.org/10.1016/j.beth.2006.06.004 Myers, T., Ridolfi, D., Crowther, J., & Ciesla, J. (2012). The impact of appearance-focused social comparisons on body image disturbance in the naturalistic environment: The roles of thin-ideal internalization and feminist beliefs. Body Image, 9, 342–351. http://dx.doi.org/10.1016/j.bodyim. 2012.03.005 O’Brien, K. S., Caputi, P., Minto, R., Peoples, G., Hooper, C., Kell, S., & Sawley, E. (2009). Upward and downward physical appearance comparisons: Development of scales and examination of predictive qualities. Body Image, 6, 201–206. http://dx.doi.org/10.1016/j.bodyim.2009.03.003 O’Brien, K. S., Hunter, J. A., Halberstadt, J., & Anderson, J. (2007). Body image and explicit and implicit anti-fat attitudes: The mediating role of physical appearance comparisons. Body Image, 4, 249–256. http://dx.doi.org/10.1016/j.bodyim. 2007.06.001 O’Connor, B. P. (2000). SPSS and SAS programs for determining the number of components using parallel analysis and Velicer’s MAP test. Behavior Research Methods, Instruments, & Computers, 32, 396–402. http://dx.doi.org/ 10.3758/BF03200807 Patrick, H., Neighbors, C., & Knee, C. R. (2004). Appearance-related social comparisons: The role of contingent self-esteem and self-perceptions of attractiveness. Personality and Social Psychology Bulletin, 30, 501–514. http://dx.doi.org/10.1177/0146167203261891 Shroff, H., & Thompson, J. K. (2006). Peer influences, body-image dissatisfaction, eating dysfunction and self-esteem in adolescent girls. Journal of Health Psychology, 11, 533–551. http://dx.doi.org/10.1177/1359105306065015 Song, H., Cai, H., Brown, J., & Grimm, K. J. (2011). Differential item functioning of the Rosenberg Self-Esteem Scale in the U.S. and China: Measurement bias matters. Asian Journal of Social Psychology, 14, 176–188. http://dx.doi.org/10.1111/j.1467-839X.2011.01347.x Stice, E. (2001). A prospective test of the dual pathway model of bulimic pathology: Mediating effects of dieting and negative affect. Journal of Abnormal Psychology, 110, 124–135. http://dx.doi.org/10.1037/0021-843X.110.1.124 Stice, E., & Agras, W. S. (1998). Predicting onset and cessation of bulimic behaviors during adolescence: A longitudinal grouping analysis. Behavior Therapy, 29, 257–276. http://dx.doi.org/10.1016/S0005-7894(98)80006-3 Stice, E., Marti, C. N., Rohde, P., & Shaw, H. (2011). Testing mediators hypothesized to account for the effects of a dissonance-based eating disorder prevention program over longer term follow-up. Journal of Consulting and Clinical Psychology, 79, 398–405. http://dx.doi.org/10.1037/a0023321 Stice, E., Telch, C. F., & Rizvi, S. L. (2000). Development and validation of the Eating Disorder Diagnostic Scale: A brief self-report measure of anorexia, bulimia, and binge-eating disorder. Psychological Assessment, 12, 123–131. http://dx.doi.org/10.1037/1040-3590.12.2.123 Thompson, J. K., Coovert, M., & Stormer, S. (1999). Body image, social comparison, and eating disturbance: A covariance structure modeling investigation. International Journal of Eating Disorders, 26, 43–57. http://dx.doi.org/10.1002/ (SICI)1098-108X(199907) Thompson, J. K., van den Berg, P., Roehrig, M., Guarda, A. S., & Heinberg, L. J. (2004). The Sociocultural Attitudes Towards Appearance Scale-3 (SATAQ-3): Development and validation. International Journal of Eating Disorders, 35, 293–304. http://dx.doi.org/10.1002/eat.1057 van den Berg, P., Paxton, S. J., Keery, H., Wall, M., Guo, J., & Neumark-Sztainer, D. (2007). Body dissatisfaction and body comparison with media images in males and females. Body Image, 4, 257–268. http://dx.doi.org/10.1016/j.bodyim. 2007.04.003 van den Berg, P., Thompson, J. K., Obremski-Brandon, K., & Coovert, M. (2002). The Tripartite Influence Model of body image and eating disturbance: A covariance structure modeling investigation testing the meditational role of appearance comparison. Journal of Psychosomatic Research, 53, 1007–1020. http://dx.doi.org/10.1016/S0022-3999(02)00499-3 Vartanian, L. R., & Dey, S. (2013). Self-concept clarity, thin-ideal internalization, and appearance-related social comparison as predictors of body dissatisfaction. Body Image, 10, 495–500. http://dx.doi.org/10.1016/j.bodyim.2013.05.004 Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. http://dx.doi.org/10.1037/ 0022-3514.54.6.1063

J. Liao et al. / Body Image 11 (2014) 464–473 Wills, T. A. (1981). Downward comparison principles in social psychology. Psychological Bulletin, 90, 245–271. http://dx.doi.org/10.1037/0033-2909.90.2.245 Wood, J. V. (1996). What is social comparison and how should we study it? Personality and Social Psychology Bulletin, 22, 520–537. http://dx.doi.org/10.1177/ 0146167296225009

473

World Health Organization. (2000). The Asia-Pacific perspective: Redefining obesity and its treatment. Sydney: Health Communications Australia. Yu, Z., Han, S., Chu, J., Xu, Z., Zhu, C., & Guo, X. (2012). Trends in overweight and obesity among children and adolescents in China from 1981 to 2010: A metaanalysis. PLOS ONE, 7, e51949. http://dx.doi.org/10.1371/journal.pone.0051949