T. Kubokawa [J. Multivariate Anal. 36, No. 1, 121-126 (1991; Zbl 0733.62059)] constructed a shrinkage estimator of a p-variate (p≥3) normal distribution which dominates the famous James-Stein estimator uniformly. Kubokawa showed that the estimator is generalized Bayes and also claimed the admissibility of the improved estimator using L.D. Brown and J.T. Hwang’s [Statistical Decision Theory and Related Topic III, Vol. 1, 205-230 (1982; Zbl 0585.62016)] admissibility proof. Unfortunately the admissibility proof is incorrect and here we present a correct proof using L.D. Brown’s result in Ann. Math. Stat. 42, 855-903 (1971; Zbl 0246.62016).
關聯:
Tamsui Oxford Journal of Mathematical Sciences 17(1), pp.23-33
