ABSTRACT
This chapter provides an introduction in specifying and testing multilevel models for observed categorical outcomes. It also extends some of the previous models presented to consider a variety of modeling situations where categorical variables are present. First it develops several multilevel models with binary, ordinal, multinomial, and counts as the dependent variable. And it also develops a multilevel factor model with ordinal observed indicators used to define latent variables. Mplus provides considerable flexibility in specifying a wide variety of measurement and structural models with both categorical and continuous indicators. The goal of the multilevel analysis is to summarize the within-and between-group variation in this confirmatory factor analysis (CFA) model and establish whether the same individual-level model holds at the organizational level. For a two-level CFA model, the total covariance matrix can be decomposed into it's between and within-organization covariance matrices and restrictions placed on each matrix to test the proposed model against the data.
