ABSTRACT
As we have seen, the linear multilevel models discussed previously work under the assumption of normality of errors. As such, they will not be appropriate for situations in which these, or other types of variables that cannot be appropriately modeled with a linear model are to be used. However, alternative models for such variables are available, thereby forcing us to consider alternative models. Taken together, these alternatives for categorical outcome variables are often referred to as generalized linear models (GLiMs). Prior to discussing the multilevel versions of these models in Chapter 8, it will behoove us to first explore some common GLiMs and their applications in the single-level context.
