ABSTRACT
Introduction Estimating the power of randomization tests is not always straightforward. Also, especially for phase designs, low power can be a problem. In this chapter, we consider ways to estimate power and also we review some of the ways in which power can be maximized. These are generally good experimental practice and apply to designs other than the randomization designs we are concerned with here. Phase designs and their special problems are discussed in the nal section of the chapter. For these designs, effort spent on maximizing power is particularly important.
