ABSTRACT
In this chapter we examine structural relations between latent variables within and between groups. The models can include combinations of observed predictors and latent variables, mean structures, and random slopes, as well as direct, indirect, and reciprocal effects. The examination of factor structures at the individual and group levels emphasizes the usefulness of the SEM approach to account for sources of error in the measurement of constructs at multiple levels of a data hierarchy, which improves the accuracy of the model’s estimated structural parameters. We develop a series of models that illustrate some of the possible relationships that can be investigated where latent variables are the major focus of the analyses.
