Sensitivity Analysis of Structural Parameters to Measurement Noninvariance: A Bayesian Approach

• In Event: 61.084 - Criteria and Bias in Measuring Latent Constructs and Classes

Abstract

Previous studies have demonstrated that measurement noninvariance could yield incorrect conclusions on structural parameter differences within the multiple-group confirmatory factor analysis (MGCFA) framework (Beuckelaer & Swinnen, 2011; Kaplan & George, 1995). However, results of these previous studies regarding impact of measurement noninvariance on the correct conclusion for the structural parameter differences were confounded with model misspecification. Therefore, this study empirically investigated the extent to which measurement noninvariance affects structural parameter differences across populations taking a Bayesian approach. The simulation results showed that measurement noninvariance did not have an impact on accuracy of statistical conclusion on structural parameter differences as well as bias and precision in the posterior mean of structural parameter difference.

Authors

• Yoonjeong Kang, American Institutes for Research

• Gregory R. Hancock, University of Maryland