Resampling: Transforming Statistics – Basic Tools Appearing in Winter 1994 Newsletter

Abstract: Recent advancements in the power, speed and availability of computers have facilitated the resurgence of resampling as an inference and estimation tool. This article proposes resampling as the treatment of first choice in all statistical problems requiring probabilistic analysis. Furthermore, the paper demonstrates the applications of resampling with a hypothesis test and a measurement reliability example. An important consideration here is that in the absence of precise knowledge about the universe of interest, a hypothetical universe is constructed based on the samples observed. Of course, just like for conventional methods, some limitations in the sample affect the suitability of resampling methods, but resampling provides answers in a number of situations in which conventional methods are inappropriate or difficult to implement.

Keywords: Resampling - Hypothesis Testing - Normal distribution