Distribution Free Tests for Stochastic Ordering in the Competing Risks Model
针对个体面临两个独立竞争风险的情形，基于观测到的失效时间和原因，提出了检验两个失效分布是否相等的无分布方法，涵盖位置、尺度及一般随机排序备择假设，并推导了局部最优秩检验及Wilcoxon检验的推广。
The competing risks set-up is considered where an individual is subject to failure due to two independent competing risks. The available data consist of observed times to failure and the causes of failure. On the basis of this information, distribution-free tests are proposed for testing the equality of the two failure distributions against location, scale and general stochastic ordering alternatives. Locally most powerful rank tests are derived and a generalization of the Wilcoxon test has been proposed. Exact critical points are provided for the newly proposed tests, and Pitman efficiency comparisons made.