Nonparametric Statistical Methods, 3rd Edition

8.5 A Distribution-Free Test for Independence Based on Ranks (Spearman)

Hypothesis

Let be a random sample from a continuous bivariate population (i.e., Assumption A is satisfied) with joint distribution function and marginal distribution functions and . In this section, we return to the problem of testing for independence between the and variables corresponding to the null hypothesis (8.1). Here, however, alternatives to will no longer be stated in terms of the correlation coefficient (8.2). Instead, the alternatives of interest in this section are less specifically interpretable, corresponding ...