Measurement Invariance Assessment With a Bifactor Model

• In Event: Response Styles and Measurement Invariance in Factor Modeling

Abstract

Measurement invariance is a primary assumption when comparing performance between groups or across time, in order to help ensure comparisons are accurate. The bi-factor model is increasing in popularity in many areas. However, little evidence relating to the accuracy of measurement invariance assessment with this model exits. To address this need, the current study assessed the bi-factor model’s Type I error and power rates under non-invariance conditions with a focus on the general factor. Results suggest that missing the lack of invariance in the specific factors inflates the Type I error rate for the general factor whereas accounting for the lack of invariance in the specific factors will control Type I error rates for the general factor. Implications are discussed.

Authors

• William Holmes Finch, Ball State University

• Brian F. French, Washington State University