Inadmissibility of the Best Equivariant Estimators of the Variance-Covariance Matrix and the Generalized Variance under Entropy Loss.
Abstract
Based on a data matrix and an independent Wishart matarix testmators dominating the best equivariant estimators of sigma and /sigma/ are obtained under two types of entropy loss. For simultaneous estimation of the mean vector and the variance covariance matrix of a multinormal population, a suitable entropy loss is developed and testimators dominating the pair consisting of the sample mean vector and the best multiple of the sample Wishart matrix are derived. A technique of SINHA is heavily exploited. Keywords: MANOVA test; Roy's maximum root test; Wishart distribution.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1985
- Accession Number
- ADA159059
Entities
People
- B. K. Sinha
- M. Ghosh
Organizations
- University of Pittsburgh