Unusual properties of bootstrap confidence intervals in regression problems (Q1102055)

We show that the percentile-t method, and one of the two percentile methods, have unusually good performance when employed to construct bootstrap confidence intervals in a regression setting. In the case of slope parameters, percentile-t produces two-sided intervals with coverage error \(n^{-2}\), and one-sided intervals with coverage error \(n^{- 3/2}\), where n is sample size. The errors are only \(n^{-1}\) in most other problems. One of the percentile methods produces critical points which are third-order correct for \textit{B. Efron}'s [J. Am. Stat. Assoc. 82, 171-200 (1987; Zbl 0622.62039)] relatively complex accelerated bias- corrected points.