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): 293-300.

• 研究方法 • 上一篇    

形成性测量模型:结构方程模型的新视角

王晓丽;李西营;邵景进   

  1. (1北京师范大学发展心理研究所, 北京 100875) (2河南师范大学教科院, 河南新乡 453007)
  • 收稿日期:2010-10-10 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: 邵景进

Formative Model: A New Perspective of Structural Equation Model

WANG Xiao-Li;LI Xi-Ying;SHAO Jing-Jin   

  1. (1 Institute of Developmental Psychology, Beijing Normal University, Beijing 100875, China)
    (2 College of Education Science of Henan Normal University, Xinxiang 45300, China)
  • Received:2010-10-10 Revised:1900-01-01 Online:2011-02-15 Published:2011-02-15
  • Contact: SHAO Jing-Jin

摘要: 形成性测量模型(Formative Model, FM)是指标变异导致潜变量变异的模型, 反映性测量模型(Reflective Model, RM)是潜变量变异导致指标变异的模型。FM在模型界定、识别和估计、信效度评价以及模型应用等方面均与RM存在极大的不同。模型界定错误会使参数估计发生偏差, 影响统计结论的有效性, 应当审慎考虑指标和潜变量之间的关系, 选择恰当的测量模型。进一步揭示两者的区别和误用带来的偏差, 完善FM的识别和估计、信效度评价方法、对变量含义的解释以及高阶FM的理论解释和模型估计是未来的研究方向。

关键词: 形成性测量模型(FM), 反映性测量模型(RM), 结构方程模型

Abstract:

Formative Model (FM) refers to the measurement model in which the variation of latent variable is caused by the index variation, while Reflective Model (RM) refers to the measurement model in which the index variation is caused by the variation of latent variable. There are distinctions between FM and RM in definition, identification, estimation, the evaluation of reliability and validity and application. Wrong model definition may cause the deviation of parameters estimation, so as to lower the efficiency of statistical result. Therefore prudent consideration should be taken to the relationship between index and latent variable in order to choose the appropriate measurement model. At last, the future research should focus on the clarification of the differences between FM and RM, the effect of misuse, the improvement of the identification and estimation, the reliability and validity evaluation, the explanation of the meaning of variables and the theoretical interpretation and model estimaton of high-level FM.

Key words: Formative Model, Reflective Model, Structural Equation Model