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 ›› 2015, Vol. 23 ›› Issue (3): 529-538.doi: 10.3724/SP.J.1042.2015.00529

• 研究方法 • 上一篇    

因子混合模型:潜在类别分析与因子分析的整合

陈宇帅;温忠麟;顾红磊   

  1. (华南师范大学心理应用研究中心/心理学院, 广州 510631)
  • 收稿日期:2014-05-26 出版日期:2015-03-15 发布日期:2015-03-15
  • 通讯作者: 温忠麟, E-mail: wenzl@scnu.edu.cn
  • 基金资助:

    国家自然科学基金项目(31271116, 31400909)资助。

Factor Mixture Model: An Integration of Latent Class Analysis and Factor Analysis

CHEN Yushuai; WEN Zhonglin; GU Honglei   

  1. (Center for Studies of Psychological Application/School of Psychology, South China Normal University, Guangzhou 510631, China)
  • Received:2014-05-26 Online:2015-03-15 Published:2015-03-15
  • Contact: WEN Zhonglin, E-mail: wenzl@scnu.edu.cn

摘要:

因子混合模型(FMM)是考虑了群体潜在异质性后的因子分析模型, 它将潜在类别分析(LCA)与传统的因子分析(FA)整合在同一框架内, 既保留了两种分析技术的优点, 同时又展现出独特优势。FMM的应用主要包括描述变量的潜在结构、对被试进行分组以及探测社会称许偏差等。我们建议分别采用FA、LCA与FMM三种模型拟合数据, 参考拟合指数和模型可解释性选择最优模型。总结了FMM的分析步骤以及软件使用, 并用于探讨大学生社会面子意识的测量模型。未来研究应关注FMM分析过程的简化, 继续深化对拟合指数等方面的探讨。

关键词: 因子混合模型, 潜在类别分析, 因子分析

Abstract:

Factor Mixture Model (FMM) is a factor analysis model in which the latent population heterogeneity is considered. Combined with latent class analysis (LCA) and traditional factor analysis (FA), the FMM model consistently preserves the advantages of these two statistical methods, and has some unique features as well. Present empirical applications of FMM include the description of latent structure of variables, classification of subjects, and detection of social desirability bias. We suggest to fit data with FA, LCA and FMM respectively, and to choose an optimal model according to the fit indexes and practical implications. By applying FMM to build the measurement model of consciousness of social face, we illustrate the analysis steps and software operation procedures. Future research efforts are needed for some issues on FMM, such as the simplification of analytical process and the selection of fit index.

Key words: factor mixture model, latent class analysis, factor analysis