对齐(Alignment) —— 一种新的多群组分析法 postprint

Abstract: Comparing the latent factor means across groups need to test the comparability of the instrument. Researchers usually test the scalar invariance to achieve this goal but often get unsatisfactory results. The alignment is proposed under this background. It allows the comparison of the latent factor means by testing the approximate invariance of the configural model parameters and has become a significant complement to multiple-group CFA. This article gives a detailed description of the multiple-group CFA and alignment, summarizes the research procedures and points to which researchers need to pay attention when applying alignment, uses an undergraduates’ work value data to illustrate how to use alignment to do a research with Mplus. In the end, this article summarizes the advantages and limitations of alignment and reviews its research developments and empirical applications.