Most SEM models have assumed that the relations in the models were linear (i.e., the relations among all variables, observed and latent, are represented by linear equations). Several studies have been published where nonlinear and interaction effects are used in multiple regression models; however, these effects have seldom been tested in path models, and you will infrequently nd nonlinear factor models. It should not be surprising to nd that for several decades structural equation modeling has been based on Linear Structural Relations (LISREL).