ABSTRACT
In Chapter 3 we described some statistical tests which are applicable to the numbers that emerge from certain common experimental designs. These numbers were generally assumed to represent randomly and independently drawn samples from normally distributed populations whose variances were equal. In Chapter 4 several nonparametric tests were presented for analyzing data when the assumptions of population normality and equal variances could not be met. Moreover, the techniques of Chapter 4 do not require numbers — data initially in the form of ranks or signs ( + or —) can be subjected to analysis. However, investigators sometimes end up with even grosser behavioral measures in the form of a limited number of classes like: solved vs did not solve, survived vs died, improved vs no change vs became worse, and so on. The present chapter will be concerned with some non parametric methods for the analysis of such nominal data. The chapter wil l follow the previous organizational scheme of examining two and k independent and matched groups. After the discussions of k groups, we will be concerned with the analogue of multiple comparisons —the partitioning of contingency tables.
