对齐(Alignment)——一种新的多群组分析法

Alignment: A new method for multiple-group analysis

进行跨群组的因素均值比较需要检验测量工具的可比较性,常见的做法是使用多群组CFA检验各群组的截距恒定性,但截距恒定假设因其过于苛刻而往往无法满足。对齐法在这一背景下应运而生,通过检验构置恒定模型中参数的近似恒定性而进行跨群组因素均值比较。文章介绍了多群组CFA和对齐法的原理,运用某省4个学校的本科生职业价值观研究实例来演示如何在Mplus软件中使用对齐法进行研究,并总结了对齐法的优势与局限,回顾了对齐法近年的研究进展。

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.