This chapter introduces conditional methods for the analysis of discrete data. To provide a comparative framework, relevant unconditional large sample methods are also described. Models for the 2× 2 table of counts, and two and four group Poisson data serve as the raw material for constructing these methods. The specific aims here are:

• To define and contrast the unconditional and conditional methods of data analysis. • To relate the concepts of ancillarity and sufficiency to the use of conditional methods of

inference on discrete data.