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

Advances in Psychological Science ›› 2020, Vol. 28 ›› Issue (11): 1961-1969.doi: 10.3724/SP.J.1042.2020.01961

• Research Method • Previous Articles     Next Articles

Equivalence testing: A new perspective on structural equation model evaluation and measurement invariance analysis

WANG Yang1, WEN Zhonglin2(), FU Yuanshu3   

  1. 1School of Public Administration, Guangdong University of Finance, Guangzhou 510521, China
    2School of Psychology/Center for Studies of Psychological Application, South China Normal University, Guangzhou 510631, China
    3School of Education, Zhaoqing University, Zhaoqing 526061, China
  • Received:2020-03-06 Online:2020-11-15 Published:2020-09-23
  • Contact: WEN Zhonglin E-mail:wenzl@scnu.edu.cn

Abstract:

There are some limitations in the commonly used fit indices of structural equation model. For example, χ 2, which is set up to reject the traditional null hypothesis, cannot be used to endorse a model, while descriptive fit indices such as RMSEA and CFI do not have inferential statistical properties. Equivalence testing can effectively compensate for the limitations mentioned above. In this paper, the way to use equivalence testing in evaluating the fit of a single model and its difference from the null hypothesis testing were introduced first. Then the approach to analyze measurement invariance by equivalence testing was described. Furthermore, empirical data was used to demonstrate the effectiveness of equivalence testing in single model evaluation and measurement invariance test, and to compare equivalence testing with traditional model evaluation method.

Key words: structural equation model, fit indices, equivalence testing, null hypothesis testing, measurement invariance

CLC Number: