ABSTRACT
This chapter focuses on models designed specifically for scenarios in which the outcome of interest is either categorical or count in nature and the data have been collected in a multilevel framework. In order to introduce Multilevel Generalized Linear Models for dichotomous outcomes, the chapter considers an example. The Mplus program needed to fit the multilevel dichotomous logistic regression model represents a mixture of examples. Many of the modeling commands for fitting multilevel logistic regression are identical to those for single-level logistic regression models. As was the case for non-multilevel data, the cumulative logits link function can be used with ordinal data in the context of multilevel logistic regression. Model parameter estimation is achieved using maximum likelihood, based on the Newton–Raphson method. In the first example, the chapter examines the relationship between computation scores in a formative exam and the ordinal scores in the final mathematics test.
