Bayesian Structural Equation Models: A Health Application

Concepts of health are often multivariate or multidimensional. Structural equation modelling (SEM) is a multivariate method that incorporates ideas from regression, path-analysis and factor analysis. A Bayesian approach to SEM may enable models that reflect hypotheses based on complex theory. The development and application of Bayesian approaches to SEM has, however, been relatively slow but with modern technology and the Gibbs sampler, is now possible. This paper contributes to the knowledge of Bayesian methods in the SEM framework by illustrating how different sources of uncertainty in data can be incorporated into the modelling process. The particular aim is to develop a preliminary Bayesian approach to SEM for the longitudinal relationship between life events and health, which is extended to account for the suspected effect of telescoping. Telescoping is a tendency to recall events from the past as having happened more recently than when they actually occurred, and is suspected to have occurred due to the time recall component of some questions. It is expected that this basic model be extended to include other issues as required.