Detecting Cross-Loadings in CFA Using Bayesian Estimation: A Monte Carlo Simulation

• In Event: AERA Roundtable Session Thursday 7:45 am - Gold 4
  In Roundtable Session: Measurement Complexity in SEM: Cross-Loading, Invariance, and Heterogeneity

Abstract

Bayesian estimation offers greater flexibility in structural equation modeling by enabling the estimation of parameters typically fixed under frequentist methods (Muthén & Asparouhov, 2012). This simulation study evaluates its effectiveness in detecting meaningful cross-loadings in confirmatory factor analysis (CFA) across varying conditions, including sample size, data skewness, number of true cross-loadings, and loading magnitude. Preliminary results show that Bayesian estimation with informative priors achieves high power when cross-loading magnitudes reach 0.4, while the number of cross-loadings and indicator distributions have minimal impact. These findings highlight Bayesian estimation as a powerful and flexible tool for improving model specification in CFA.

Authors

• Tiejun Zhang, University of South Carolina

• Ruiqin Gao, University of South Carolina

• Xiaobo Wei, University of South Carolina

• Jingtong Dou