Confirmatory Factor Analysis, Exploratory Structural Equation Modeling, and Bayesian Structural Equation Modeling: Systematic Comparisons

• In Event: Growth Curve Modeling Extensions and Factor Analysis Choices

Abstract

We contrast competing approaches balancing CFA/SEM rigor with flexibility to fit realistically complex data. Exploratory SEM (ESEM) provide an optimal compromise between EFA and CFA/SEM, whilst three Bayesian SEMs (BSEMs) replace fixed-zero estimates with informative, small-variance priors for cross-loadings (CL), residual covariances (RC) or both (CLRC). For two simulated data designs (Study 1), BSEM-RC and BSEM-CLRC fit the data better than ESEM and BSEM-CL, but ESEM parameter estimates were more accurate in most conditions. For longitudinal Big-Five personality data (60-item NEO-FFI, N = 3390; Study 2), ESEM performed similarly to BSEM-CLRC and BSEM-CL, but BSEM-RC performed aberrantly. Comparison of ESEMs and BSEMs provided better understanding of BSEM specification, choice of priors, cross/external validation, and guidance for applied researchers.

Authors

• Jiesi Guo, Australian Catholic University

• Herbert W. Marsh, Australian Catholic University

• Philip D Parker, Australian Catholic University

• Theresa Dicke, Australian Catholic University

• Oliver Lüdtke, Leibniz Institute for Science and Mathematics Education, Kiel, Germany