On some shrinkage estimators of multivariate location (Q1058238)

Let \(F_{\theta}\) be a p (\(\geq 1)\) variate continuous distribution function (d.f.) defined on the Euclidean space \(E^ p\). The d.f. \(F_{\theta}\) is assumed to be diagonally symmetric about its location \(\theta =(\theta_ 1,...,\theta_ p)^ T\). In a shrinkage estimation problem, one has, a priori, some reason to believe that \(\theta\) is likely to lie in a small region containing a specified point \(\theta_ 0\) (which may be assumed to be zero without loss of generality). Using a preliminary test for the hypothesis \(\theta =0\), a variant form of the James-Stein type estimation rule is used to formulate some shrinkage estimators of location based on rank statistics and U- statistics. A smaller critical value for the preliminary test has been recommended. In an asymptotic set up, the relative risks for these shrinkage estimators are shown to be smaller than their classical counterparts.