On the Use of ESEM for Evaluating Measurement Invariance

• In Event: Applications in Systematic and Meta-Analysis

Abstract

Measurement invariance is an essential part of validity evidence that is concerned with ensuring that tests function similarly across groups and time. Evaluations of construct invariance are typically evaluated within the framework of confirmatory factor analysis (CFA) that impose strict restrictions on non-target indicators that they be constrained to zero. Exploratory structural equation modeling (ESEM) allows for a more nuanced hypothesis about factor structures by estimating both target and non-target loadings. In so doing, ESEM results in less biased factor-factor correlation estimates. Using Monte Carlo simulation, we examine the behavior of fit statistics within an ESEM framework for evaluating measurement invariance when the primary source of invariance is the presence of non-zero cross-loadings in one group, but not in another.

Authors

• Kelvin Tosin Afolabi, University of Texas at Arlington

• Timothy R. Konold, University of Virginia

• Elizabeth A. Sanders, University of Washington