In the last chapter we introduced four coding systems-dummy codes, unweighted effect codes, weighted effect codes, and contrast codes-that may be used to represent nominal (categorical) IVs. We also saw that regression equations could easily be specified that contain a mixture of nominal and quantitative IVs. However, the presentation in Chapter 8 was limited in two important ways. First, we did not consider any models that contained interactions between IVs. And second, we did not consider models in which the effect of the quantitative IV was nonlinear.