Second-order properties of a two-stage fixed-size confidence region for the mean vector of a multivariate normal distribution (Q1283925)

We consider the classical fixed-size confidence region estimation problem for the mean vector \(\boldsymbol \mu\) in the \(N_p(\boldsymbol \mu\Sigma)\) population, where \(\Sigma\) is unknown but positive definite. We write \(\lambda_1\) for the largest characteristic root of \(\Sigma\) and assume that \(\lambda_1\) is simple. Moreover, we suppose that, in many practical applications, we will often have available a number \(\lambda_* (>0)\) and that we can assume \(\lambda_1>\lambda_*\). Given this additional, and yet very minimal, knowledge regarding \(\lambda_1\), the two-stage procedure of \textit{S. K. Chatterjee} [Calcutta Stat. Assoc. Bull. 8, 121--148 (1959; Zbl 0107.13905); 9, 20--28 (1959); 11, 144--149 (1962; Zbl 0114.10401)] is revised appropriately. The highlight in this paper involves the verification of second-order properties associated with such revised two-stage estimation techniques, along with the maintenance of the nominal confidence coefficient.