In this chapter, the authors’ consider models for the analysis of categorical independent variables when observations are nonindependent because they are grouped in some way, and the independent variables vary either between groups of linked observations or within them. Independence of observations is formally defined using conditional probabilities. Two observations are nonindependent if the conditional probability of one of them having a particular value, given knowledge of the other one, is different from the unconditional probability. Nonindependence in data is likely to be found in two classic situations: grouped data and sequential data. Since nonindependence is likely to be present in both grouped and sequential data, the usual regression model that the author has employed needs to be modified in some way to deal with the fact that one of its underlying assumptions has been violated.