Questionable Claims for Simple Versions of the Bootstrap

Abstract

Recent years have seen increasing interest in incorporating resampling methods into introductory statistics courses and the high school mathematics curriculum. While the use of permutation tests for data from experiments is a step forward, the use of simple bootstrap methods for sampling situations is more problematical. This article demonstrates via counterexamples that many of the claims made for these simple methods are simply wrong. Their use with beginners can only be justified after their true properties have been fully researched, and their many limitations explained to students. Supplementary materials for this article are available online.

Keywords

• bootstrap
• common core
• confidence interval
• resampling
• robust
• simulation