Linear Rank Statistics for Truncated Data
受Prentice对删失数据线性秩统计量推广的启发，本文为截断数据引入了类似的核来推导线性秩统计量，并证明其可等价表示为加权对数秩形式，从而提供自然的方差估计，还给出了Wilcoxon和对数秩检验的显式形式，并扩展到左截断右删失数据。
Motivated by Prentice's (1978) generalization of linear rank statistics for censored data, we introduce a similar kernel for deriving linear rank statistics for truncated data. They take very similar forms to those of Prentice's statistics. Moreover, we also show that they can be expressed equivalently in weighted log rank-type forms, thus providing natural variance estimators. Explicit forms corresponding to Wilcoxon and log rank tests are obtained. Extensions to data subject to both left truncation and right censoring are provided.