An Integrative Look at the Use of Additive and Multiplicative Covariance Structure Models in the Analysis of MTMM Data
比较了传统一阶验证性因子分析模型与两种新提出的协方差结构模型（Wothke的协方差成分分析模型和Browne的直接乘积模型）在分析多特质多方法数据时的优劣，并探讨了方法因子与特质因子的关系及模型选择问题。
First-order confirmatory factor analytic models have had widespread use in the analysis of multitrait-multimethod (MTMM) data. In contrast to the usual first-order confirmatory factor analytic model for the analysis of MTMM data, other covariance structure models have recently been proposed and advocated. Two such models are Wothke's covariance component analysis model and Browne's direct product model. The authors provide a conceptual and analytic discussion of those alternative procedures and compare them with the conventional first-order confirmatory factor analytic model. They consider the relationship between method factors and trait factors assumed under each model specification. General remarks about the nature of method factors and the likely reasons for lack of fit and ill-defined solutions frequently encountered with use of first-order factor models are presented. The authors also attempt to integrate the various approaches to modeling MTMM data and in so doing provide some perspective on selection of a particular covariance structure model for use in applied research.