Evaluating Latent Mediated Effect and Regression Coefficients With Ordinal Data Using Alternative Methods

• In Event: Structural Equation Modeling With Ordinal/Categorical Data

Abstract

Variables featured with an ordinal level of measurement are common in many empirical situations within the social and behavioral sciences. Even though simple models (one- or two-factors) of latent variables with ordinal data had been empirical evaluated, applied researchers are likely to encounter more complex situations in practice. This study proposes a MCMC-based method to focus on latent mediated effect analysis with ordinal data. Beyond that, this study attempts to demonstrate the potential of using MCMC as compared with conventional methods when analyzing complex models of ordinal data through empirical evaluation. The results could be useful to applied researchers when they need to model ordinal data in complicated settings.

Authors

• Jinsong Chen, Rutgers University

• Jaehwa Choi, The George Washington University