OBJECTIVETo discuss application of structural equation model (SEM) approach in epidemiological research.

METHODSA brief overview on major components of SEM, statistical assumptions underlying the use of SEM, and current software available to users and how SEM can be used were discussed through a practical epidemiological research project.

RESULTSAdvantages of SEM comparing with conventional epidemiological approach were shown. SEM, having the nature of comprehensive thinking and analytic approach, not only exploring the association between factors and diseases but also among factors. It also served a confirmatory, rather than exploratory approach on data modeling, as well as having the capability of correcting estimates by separating measurement error from the equations, to provide modeling the latent variables.

CONCLUSIONSEM approach could be used in epidemiological research as having some advantages comparing with conventional epidemiological approaches.