Structural equation modeling and its relationship to multiple regression and factor analysis

Abstract

Using a conceptual and nontechnical approach, the meaning of structural equation modeling (SEM) and the similarities to, and differences from, more commonly used procedures such as correlation, regression, path analysis, and factor analysis are explained. Application of the statistical technique is presented using data from a study of the relationships among stresses, strains, and physical health in a random sample of 492 community-dwelling elders aged 65 and older. Advantages of each statistical procedure are described. Theoretical issues related to the use of each procedure are presented with emphasis on the need for a sound theoretical model and match between the statistical procedure and the aims of the analysis.