Factor structure of the CES-D-20 scale in young adults in Singapore: An examination of six alternative structures and measurement invariance by sex

Abstract

Building on past research that has indicated cultural variations in the factor structure of the 20-item Center for Epidemiologic Studies Depression Scale (CES-D-20), as well as several limitations validating the CES-D-20 within a Singapore context, this study investigated the CES-D-20's factor structure and sex-based measurement invariance in Singapore. A large young adult sample in Singapore (N = 1008) was utilised, and the original CES-D-20 was administered to all participants. The findings confirmed the four-factor model's robustness over five other alternative models (RMSEA = .05, SRMR = .04), where each of the 20 items loaded significantly onto its respective factor (all loadings≥.42, all ps < .001). All four factors were also significantly inter-correlated to a medium to very large extent (all ps < .001). Configural and weak invariance were established across sexes; however, strong invariance required modification, where the release of a single intercept constraint resulted in strong partial invariance, implying sex differences in the expression of crying. Homogeneity tests suggested no significant sex differences in latent factor means, variances, and covariances. The importance of cultural- and sex-specific considerations when utilising the CES-D-20 are discussed.

Keywords

• CES-D-20
• Depression
• Factor structure
• Measurement invariance
• Latent variable analysis