Sensitivity of Goodness of Fit Indexes to Lack of Measurement Invariance in Exploratory Structural Equation Modeling

• In Event: Measurement Invariance and Longitudinal Modeling

Abstract

This study examined the appropriateness of using the fit indices cutoff criteria of measurement invariance testing derived in the context of structural equation modeling (SEM) (e.g., CFI change less than .01) for measurement invariance testing in exploratory SEM (ESEM). The design factors of this simulation study include model complexity, magnitude of cross-loadings, location of measurement noninvariance, proportion of items with measurement noninvariance, magnitude of measurement noninvariance, and sample size. The results indicated that when sample size was large (over 800 per group), the CFI criteria failed to detect a measurement noninvariance of .15, but successfully detected the measurement noninvariance of .30. AIC and BIC performed very well when sample size was relatively large.

Authors

• Chunhua Cao, The University of Alabama

• Xinya Liang, University of Arkansas