ABSTRACT
The models described in the previous chapters are all multilevel variants of the conventional multiple regression model. There are models that cannot be analyzed with multiple regression, notably factor analysis and path analysis models. A general approach that includes both factor and path analysis is structural equation modeling, or SEM. This chapter describes the multilevel confirmatory factor analysis model, and the two major estimation methods currently in use: weighted least squares (WLS) and maximum likelihood (ML). In addition, it reviews methods to obtain standardized coefficients, and goodness of fit in multilevel SEM models.
