ABSTRACT
All of the varying effects, whether they were intercepts or slopes, have been defined over discrete, unordered categories. This chapter shows how to specify varying slopes in combination with the varying intercepts. It analyses how to model covariation among continuous categories, as well as how to generalize the strategy to seemingly unrelated types of models such as phylogenetic and network regressions. The chapter explores the sort of data that is well-suited to a varying slopes model. There are multiple clusters in the data and each cluster is observed under different conditions. To see the consequence of the adaptive regularization, shrinkage, the chapter plots the posterior mean varying effects. The angled shrinkage lines reflect the negative correlation between intercepts and slopes. The chapter also considers the variation in gender bias among departments.
