(PS8-C62) Cross-cultural Differences in the Impact of Positive and Negative Affect on Social Anxiety: A Moderated Nonlinear Factor Analytic Study

Purpose: Core dimensions of both positive affect (PA) and negative affect (NA) predict social anxiety in Western culture (Brown et al., 1998). However, cross-cultural research has shown that non-Western cultures may value PA and NA differently, which may alter the relationship between these constructs and other psychological constructs (Mesquita & Walker, 2003). While traditional assessment tools of psychological constructs rely on the sum of mean values to examine relationships among psychological constructs, this can be an imprecise approach. In contrast, moderated nonlinear factor analysis (MNLFA; Bauer & Hussong, 2009) allows cultural context to alter the measurement models of psychological constructs and also simultaneously models the impact of culture on means and variances of indicators and factors. Finally, factors accounting for the role of culture can be used in prediction, and structural relationships can themselves be moderated. Therefore, we used MNLFA to examine the impact of NA and PA on social anxiety depending on Western or non-Western culture context.

Procedure: Western cultural context was represented by a US sample and the Non-Western context was represented by a Japanese sample. The US data were collected as part of the MIDUS II Biomarker Study from 2004-2009 and the Japanese data were collected as part of a corresponding study (i.e, MIDJA) in 2008. NA and PA were measured using the Positive and Negative Affect Schedule (PANAS) and social anxiety was assessed using Liebowitz Social Anxiety Scale (LSAS). Participants in both studies completed 9-item versions of the LSAS and the PANAS. The Japanese sample (N=1027) had a mean age of 54.36 (SD=14.15), while the US sample (N=1255) had a mean age of 54.52 (SD=11.71). The Japanese sample had a balanced representation of males (49.2%) and females (50.8%), while there were slightly more females (56.8%) than males (43.2%) in the US sample. 

Results: 

Three item loadings in the LSAS differed significantly across cultures (λs > |.07|, ps < .05).  Social anxiety did not significantly differ (α=.18, p=.40) in terms of intercepts across cultures.  The factor variance for the LSAS differed significantly across cultures (ψ=.38, p=.001).

In the PANAS, three NA item loadings differed significantly across cultures (λs > |.11|, ps < .05), and no PA item loadings differed (λs < |.06| , p > .05). However, NA (α=3.39, p< .001) and PA (α=-0.64, p< .001) intercepts differed significantly across cultures.  For the PANAS, cultures differed in factor variance of NA (ψ=-.37, p< .001), while factor variance for PA did not differ significantly (ψ=.02, p=.82). 

NA (γ=-.53, p< .001) and PA  (γ=.19, p< .001) impacted social anxiety differently depending on culture.

Conclusion: Psychological constructs examined in assessment tools and the relationship among psychological constructs may differ across cultures. We found that NA is a stronger predictor of social anxiety for the US sample and PA is a weaker predictor of social anxiety for the US sample. It is important to consider how different cultural contexts may impact underlying relationships between psychological processes and further examine how traditionally Western CBT treatments can be tailored to different cultures.