"Stochastic Reweighting Schemes: an Empirical Investigation"

The bootstrap has become a fundamental tool for statisticians over the past 30+ years. Randomly resampling from data has shown to be very effective at estimating the sampling distribution of our data. In an attempt to improve on the bootstrap, a method called stochastic reweighting was introduced. While there have been many theoretical results of various stochastic reweighting schemes, these results have been inconclusive and unhelpful to practitioners. We study the effectiveness of various methods, including t-tests, the standard bootstrap, and various stochastic reweighting schemes, for estimating the mean with small samples of data.