Two sample adaptive rank tests under type II censoring (Q1069598)

The two sample testing problem of randomness against general 'stochastically larger alternatives' for censored data is considered. The optimal nonparametric score function is identified and consistently estimated by some generalized kernel estimator based on the rank data. Asymptotic normality of the resulting adaptive rank statistic is proved under the null hypothesis and under fixed alternatives. The good power properties of the corresponding adaptive rank test for small sample sizes are demonstrated by Monte Carlo simulation under various types of non- parametric alternatives.