Bayesian analysis of nonlinear structural equation models with mixed continuous, ordered and unordered categorical, and nonignorable missing data

Abstract

Structural equation models (SEMs) have been widely ap-plied in examing inter-relationships among latent and ob-served variables in social, psychological, and medical re-search. Motivated by the fact that correlated discrete vari-ables and missing data are frequently encountered in practi-cal applications, a nonlinear SEM (NSEM) that accommo-dates covariates, mixed continuous and discrete variables, and nonignorable missing data is proposed. Bayesian meth-ods for estimation and model comparison are discussed. One real-life data set about cardiovascular disease is used to il-lustrate the methodologies.