Testing Configural, Metric, Scalar, and Latent Mean Invariance Across Genders in Sociotropy and Autonomy Using a Non-Western Sample

Abstract

The Personal Style Inventory (PSI) was developed to assess individuals' levels of sociotropy and autonomy, two personality characteristics considered to be associated with increased vulnerability to depression. This study used the approach of latent means analysis (LMA) within the framework of structural equation modeling (SEM) to explore the factor structure and gender differences associated with the PSI in a sample of Korean undergraduates (N = 508). Results of the confirmatory factor analysis are consistent with previous work and support the cross-cultural stability of the PSI factor structure. However, in contrast to previous research, results of the LMA showed women to have higher scores than men on sociotropy and autonomy dimensions, raising the possibility of cross-cultural differences in the interaction between these personality styles and vulnerability to depression.

Gender Differences in Stress Levels and Coping Strategies in South Korea While Using Mobile Phones

“…Finally, self-efficacy was estimated with two variables derived from previous studies (Duhachek & Iacobbuci, 2005;Rosen & Weil, 1995), which included asking respondents about their levels of knowledge and ability in using mobile phones. As shown in Table 2, skewness and kurtosis statistics of the constructs met the requirements of normal distribution (Hong et al, 2003). The results of the confirmatory factor analysis and Cronbach's α coefficients, for women and men, suggested that all parameters were loaded significantly on their respective latent variables.…”

“…SEM has become one of the most popular methods in multivariate analysis; it involves greater theoretical meaningfulness and cross-population stability, by controlling measurement errors and testing for a more complex set of relationships than regression or ANOVA methods do (Kim, Kim, & Hong, 2009;Motl, Dishman, Saunders, Dowda, Felton, Ward, & Pate, 2002;Yuan & Bentler, 2000). And it is important to satisfy the invariance constraints, because multigroup comparisons may be meaningless without measurement invariance, (Hancock, 1997;Hong et al, 2003). As shown in Table 3, we conducted measurement invariance tests sequentially across gender groups, which are the pre-requisites for LMC (Steenkamp & Baumgartner, 1998): (a) configural invariance, (b) metric invariance, and (c) scalar invariance.…”

“…In the first step, configural invariance can be met if the unconstrained model structure is the same across groups. Once configural invariance is satisfied, more restrictive models can be tested and the configural model provides the basis for comparison with all subsequent models (Bollen, 1989;Hong et. al., 2003).…”

“…Because the chi-square difference between the metric invariance model (Model 2) and the unconstrained model (Model 1) was not statistically significant (∆χ 2 = 9.15), full metric invariance was supported with the better fits of NNFI and RMSEA (∆NNFI = 0.005, ∆RMSEA = -.001). In the last step, we tested scalar invariance by fixing the intercepts of the indicators to be equal across gender groups (Hong et al, 2003;Meredith, 1993). In Model 3, all seventeen intercepts of the seven indicators were fixed to be equal across gender groups.…”

A study of Australian and Chinese accountants’ attitudes towards independence issues and the impact on ethical judgements

“…It examines whether there is equality between the measure variable intercepts of the construct (Hair et al, 2010). This implies that the scalar test establishes whether two different groups use a response scale in a similar fashion (Hong et al 2003;Campbell et al 2008). Thus configural invariance, metric invariance, and scalar invariance were conducted to validate the instrument before compare the mean differences between Australian and Chinese public accountants.…”