On the Confidence Region Comparison of Some Solutions for the Multivariate Behrens-Fisher Problem

Autor: Kocherlakota Subrahmaniam, Kathleen Subrahmaniam
Rok vydání: 1975
Předmět:
Zdroj: Communications in Statistics - Simulation and Computation. 4:57-67
ISSN: 0361-0918
DOI: 10.1080/03610917508548497
Popis: In this paper we compare the performance of three procedures suggested for the classical Behrens-Fisher problem in the P-dimensional ease. The comparison is based on the expected volume of the confidence regions for the three procedures. it is shown that in general James' series solution is superior to both Bennett's extension of Scheffe's solution and the approximate degrees of freedom solution presented by Yao.
Databáze: OpenAIRE