An Estimator for a Population Mean Which Reduces the Effect of Large True Observations

Abstract

An estimator, , of μ is investigated. This estimator is formed by replacing all sample values larger than a predetermined cutoff point t by the value of t itself. A proof is given that a region for t exists for certain common distributions such that MSE is smaller than An example is presented which demonstrates the wide range of cutoff points that can be chosen which still result in a gain.