We discuss how to extend the technique of homogeneity analysis (also known as multiple correspondence analysis) to deal with hierarchical data structures. We introduce various models and focus in particular to one that borrows some ideas from the hierarchical linear model literature, and differentially weighs the various groups of observations in the data. Such models allow us to make better visual representations of large datasets and uncover common patterns in the data. The new techniques are illustrated with an example from a large educational study.