ISSN 1671-3710
CN 11-4766/R
主办：中国科学院心理研究所
出版：科学出版社

心理科学进展 ›› 2003, Vol. 11 ›› Issue (5): 593-599.

• • 上一篇    

潜变量交互效应分析方法

温忠麟,侯杰泰，马什赫伯特西悉尼大学教育学院，悉尼 澳大利亚   

  1. 华南师范大学教科院，广州 510631；香港中文大学教育学院，香港；
  • 收稿日期:2002-10-10 修回日期:1900-01-01 出版日期:2003-09-15 发布日期:2003-09-15
  • 通讯作者: 温忠麟

Methods and Recent Research Development in Analysis of Interaction Effects between Latent Variables

Wen Zhonglin， Hau Kit-Tai， Herbert W. Marsh   

  1. Faculty of Education Science, South China Normal University, 510631, China，Faculty of Education, The Chinese University of Hong Kong, Hong Kong，Faculty of Education, The University of Western Sydney, NSW2560, Australia
  • Received:2002-10-10 Revised:1900-01-01 Online:2003-09-15 Published:2003-09-15
  • Contact: Wen Zhonglin

PDF (PC)

7835

被引次数

222

摘要： 简要回顾了分析显变量交互效应的常用方法。详细讨论了目前分析潜变量交互效应的主要方法，包括用潜变量的因子得分做回归分析、分组线性结构方程模型分析、加入乘积项的结构方程模型分析和两步最小二乘回归分析，并比较和评价了这些方法的优缺点。最后归纳了潜变量交互效应分析方法的研究趋势，并介绍了新近进展（包括LMS方法和GAPI方法）

关键词: 潜变量, 交互效应, 回归, 结构方程

Abstract: Analysis of interaction, the phenomenon that the effect (including the size and direction) of a certain independent variable (or predictor) depends on the state (size, value) of another independent variable, has always been important in psychological or social research. Methods for the analysis of interaction effects between observed variables were briefly reviewed. The main concern of the article was the detailed comparison and discussion on the analysis of latent variable interactions. Four basic approaches, including regression on factor scores, multiple-group structural equation modeling, structural equation modeling with product terms, and two-stage least square regression, were illustrated and contrasted. Advances in these and other analytical methods, including recently developed latent moderated structural equations (LMS) approach and generalized appended product indicator (GAPI) procedure, were also described and evaluated

Key words: latent variable, interaction effect, regression, structural equation modeling (SEM)

中图分类号: