ABSTRACT
In the preceding three chapters we have tried to set the scene for a detailed account of how randomization tests work. In chapter 1, we considered when it would be appropriate to use a randomization test, assuming that a statistical test of some kind was wanted. In chapter 2, we considered the case for using statistical tests at all in clinical research, and the related issue of using randomization procedures instead of response-guided procedures in clinical research. In chapter 3, we considered why randomization tests have been so little used and why we believe that their time has come, especially in relation to single-case experiments. In this chapter we use several examples to illustrate how randomization tests work. We begin with a hypothetical experiment described by Fisher (1935).
