Factors Affecting the Robust Weighted Least Squares Model Fit Measures in Confirmatory Factor Analysis

• In Event: Estimation Accuracy and Conditions for Complex Latent Variable Models

Abstract

This study examined the performance of five robust weighted least squares (RWLS) fit measures in evaluating confirmatory factor analysis models when different types and degrees of model misspecification presented, with various model, sample, and data conditions. Hu and Bentler’s (1999) cut-off thresholds for the fit indices were also empirically tested. Results showed that the Mplus estimator outperformed the LISREL estimator. The fit measures studied generally had good Type I error control, but the power was significantly affected by model size. Large model sizes can camouflage detrimental model misspecifications. This study will contribute to the understanding of the RWLS fit measures in detecting model misspecification, which is a big concern when applied researchers specify and test their hypothesized models.

Authors

• Yu Zhao, The Pennsylvania State University - University Park

• Pui-Wa Lei, The Pennsylvania State University