In earlier chapters we considered the problem of how to compare two different methods or treatments and how to decide which, if either, was better. It often happens that more than two treatments are available for study. In this chapter, we generalize our two-sample test procedures to cover the comparison of more than two treatments. As in the two-sample case, we consider two different experimental setups: (i) the completely randomized design when, as in Chapter 6, available test subjects are randomly assigned to the various treatments; and (ii), the randomized block design when, as in Chapter 12, test subjects are, or can be, divided into homogeneous groups, each group containing as many test subjects as there are treatments. For (i), we develop a test procedure that generalizes the Wilcoxon two-sample test of Chapter 7. For (ii), we generalize the sign test procedure of Chapter 12 for paired observations.
KeywordsPlacebo Insurance Coverage Gasoline
Unable to display preview. Download preview PDF.