Measures of Association for Nominal and Ordinal Variables

Abstract

Chapter 12 introduced eta (η) and the more general concept of measures of association. Eta is a descriptive statistic that allows us to define how strongly the categorical variable or sample in an analysis of variance is related to the interval-level variable or trait we examined across the samples. But there are many other useful measures of association that allow us to define relationships among variables. Over the next few chapters, we will focus on some of these that are particularly useful in studying criminal justice. We will still be concerned with statistical significance in these chapters, but we will examine not only whether a measure is statistically significant but also how strong the relationship is.