Almost sure asymptotic representation for a class of functionals of the Kaplan-Meier estimator (Q1178948)

Let \(X_ 1,\ldots ,X_ n\) be nonnegative i.i.d. random variables with distribution function \(F\), called the survival time distribution, and consider the right censoring scheme. If no censoring is present, the substitution of \(F\) by the empirical d.f. \(F_ n\) in a functional like \[ \theta_ t (F) = \int \cdots \int h_ t (x_ 1, \dots , x_ m) \prod_{i=1}^ m dF(x_ i) \] leads to V-statistics and U-statistics. In the case of censored data the Kaplan-Meyer estimator \(\hat F_ n\) is used instead of \(F_ n\). Using an asymptotic representation for the Kaplan-Meyer estimator in which \(\hat F_ n\) is decomposed into a leading term, which is an average of i.i.d. r.v., and a lower order term, \(\theta_ t (\hat F_ n) - \theta_ t (F)\) is presented as an ordinary V-statistic plus a remainder term of the order \(O(n^{-1} \log n)\) uniformly in \(t\).