Nonparametric Tests for Independence in the Presence of Cured Data

Abstract

We address the problem of testing independence between two randomly right-censored variables under the assumption that the data follow an additive mixture model with a cured fraction. Four classes of test statistics, inspired by Kendall’s τ and its extensions, are proposed. We investigate their asymptotic behavior as well as their performance in finite samples. A comprehensive empirical study confirms the strong competitiveness of the proposed methods.