Hybrid resampling methods for confidence intervals

This paper considers the problem of constructing confidence intervals for a single parameter theta in a multiparameter or nonparametric family. HS bri d resampling methods, which "hybridize" the essential features of bootstrap and exact methods, are proposed and developed for both parametric and nonp arametric situations. In particular, we apply such methods to construct con fidence regions, whose coverage probabilities are nearly equal to the nomin al ones, for the treatment effects associated with the primary and secondar y endpoints of a clinical trial whose stopping rule, specified by a group s equential test, makes the approximate pivots in the nonsequential bootstrap method highly "non-pivotal", We also apply hybrid resampling methods to co nstruct second-order correct confidence intervals in possibly non-ergodic a utoregressive models and branching processes.