Introduction

In the past several chapters, we have worked extensively with regression analysis. Two common threads in all of the examples have been the use of continuous data and of linear models. In this chapter we introduce techniques to accommodate categorical data and to fit several common curvilinear patterns. Throughout this chapter, all of the earlier concepts of inference, residual analysis, and model fitting still hold true. We'll concentrate here on issues of model specification, which is to say selecting variables and functional forms (other than lines) that reasonably and realistically suit the data at hand.