We develop large deviation results with Cramér's series and the best possible remainder term for bootstrapped U -statistics with non-degenerate bounded kernels. The method of the proof is based on the contraction technique of Keener, Robinson and Weber [R.W. Keener, J. Robinson, N.C. Weber, Tail probability approximations for U -statistics, Statist. Probab. Lett. 37 (1) (1998) 59-65], which is a natural generalization of the classical conjugate distribution technique due to Cramér [H. Cramér, Sur un nouveau théoréme-limite de la theorie des probabilites, Actual. Sci. Indust. 736 (1938) 5-23].
We adapt the techniques in Stigler [Ann. Statist. 1 (1973) 472-477] to obtain a new, general asymptotic result for trimmed U -statistics via the generalized L-statistic representation introduced by Serfling [Ann. Statist. 12 (1984) 76-86]. Unlike existing results, we do not require continuity of an associated distribution at the truncation points. Our results are quite general and are expressed in terms of the quantile function associated with the distribution of the Ustatistic summands. This approach leads to improved conditions for the asymptotic normality of these trimmed U -statistics.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.