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 ›› 2011, Vol. 19 ›› Issue (2): 274-283.

• 研究方法 • 上一篇    下一篇

因子分析的元分析技术及其应用

范会勇   

  1. 渤海大学 教育与体育学院; 课程与教学论研究中心教育心理研究所, 辽宁锦州 121013
  • 收稿日期:2010-08-23 修回日期:1900-01-01 出版日期:2011-02-15 发布日期:2011-02-15
  • 通讯作者: 范会勇

Meta-analysis of Factor Analyses and Its Applications

School of Education and Physical Education; Institute of Educational Psychology of Research Center for Curriculum and Teaching, Bohai University, Jinzhou 121013, China   

  1. School of Education and Physical Education; Institute of Educational Psychology of Research Center for Curriculum and Teaching, Bohai University, Jinzhou 121013, China

  • Received:2010-08-23 Revised:1900-01-01 Online:2011-02-15 Published:2011-02-15
  • Contact: FAN Hui-Yong

PDF (PC)

2671

被引次数

18

摘要： 因子分析的元分析指对采用因子分析的原始研究进行分析, 是知识生产和更新的重要一环, 但尚未引起研究者的注意。主要有5种主要技术, 即因子配对旋转法、多组验证性因子分析、基于汇总相关矩阵的因子分析、基于估计的总体相关矩阵的验证性因子分析、基于显著负荷共生矩阵的探索性因子分析等。每种技术的介绍都包括其基本思想、适用范围、优缺点以及典型应用等。因子分析的元分析可分成7个基本步骤; 资料整理、数据合成和数据分析三步与其它类型的元分析有所不同。未来研究应注意因子分析的元分析在方法发展和应用方面的一系列问题。

关键词: 因子分析, 元分析, 共生矩阵, 因子负荷, KHB方法

Abstract:

Meta Analysis of Factor Analyses (MAFA) is defined as utilizing meta-analysis techniques to synthesize numerous primary studies with factor analysis paradigms. Although MAFA is a key stage in the process of producing and updating knowledge, it is not well known to the community of researchers. The five main MAFA techniques were summarized, including their basic premises, conditions for application, advantages and disadvantages. Typical examples corresponding to 1) pair-wise rotation of results to congruence (KHB), 2) multiple group confirmatory factor analysis (MGCFA), 3) factor analysis based on aggregate correlation matrix (ACMFA), 4) confirmatory factor analysis based on pooled correlation matrix (PCMCFA), and 5) exploratory factor analysis based on co-occurrence matrix of salient factor loadings (COEFA) are also presented The MAFA process can be divided into seven stages of which three, i.e., data extraction, data transformation, and data analysis, differ from other types of meta-analyses. Finally, several potential issues concerning the method itself and its application were also discussed.

Key words: factor analysis, meta analysis, co-occurrence matrix, factor loadings, KHB