Accuracy of the bootstrap approximation (Q1179291)

Let \(F_ n\) be the empirical distribution generated by i.i.d. r.v.'s \(X_ 1,X_ 2,\dots,X_ n\) taking values in \(R^ m\), \(m\geq 1\). Let further \(X_{n1},X_{n2},\dots,X_{nn}\) be i.i.d. with common cdf \(F_ n\). The author proves a number of statements concerning the accuracy of approximation of \(P(X_ 1+\dots+X_ n\in A)\) by the bootstrap measure \(P(X_{n1}+\dots+X_{nn}\in A)\). Besides the sample mean, \(t\)-statistics, quantiles, etc. are considered. The proofs are based on the Edgeworth expansion techniques developped by \textit{R. N. Bhattacharya} and \textit{R. Ranga Rao} [see: Normal approximation and asymptotic expansions. (1976; Zbl 0331.41023)].