Saddlepoint approximations to studentized bootstrap distributions based onM-estimates |
| |
Authors: | Diego Kuonen |
| |
Institution: | (1) Statoo Consulting, PO Box 107, 1015 Lausanne, Switzerland |
| |
Abstract: | Summary Saddlepoint methods can provide extremely accurate approximations to resampling distributions. This article applies them to
distributions of studentized bootstrap statistics based on robustM-estimates. As examples we consider the studentized versions of Huber’sM-estimate of location, of its initially MAD scaled version, and of Huber’s proposal 2. The studentized version of Huber’s
proposal 2 seems to be a preferable measure of location. Remarks on implementation and related problems are given. |
| |
Keywords: | Bootstrap Huber’ s proposal 2 Measure of location M-estimation Robust estimation Saddlepoint Studentized bootstrap |
本文献已被 SpringerLink 等数据库收录! |