Commentary on Biasing Effects of Pilot Samples and Gillett's Observations on the Stein Confidence Interval

Abstract

A previous note demonstrated that correctable bias is present when pilot sample information is used to establish sample size. Though Gillett claims that the Stein approach is a superior method for establishing sample size, we argue that it pursues an artificial goal, is impractical, and is subject to the underachievement effect that prompted our original note.