Sankhya: The Indian Journal of Statistics

1993, Volume 55, Series A, Pt. 3, 481--493

PREDICTION IN A MULTIVARIATE NORMAL SETTING: COHERENCE AND INCOHERENCE

By

MORRIS L. EATON

And

WILLIAM D. SUDDERTH, University of Minnesota, U. S. A.

 

SUMMARY. For sampling from a multivariate normal population with mean zero and unknown covariance matrix, we consider the problem of obtaining a predictive distribution for a future observation. It is shown that using the common "flat" invariant prior distribution on the covariance matrix yields an incoherent predictive distribution. However, an alternative improper prior yields a coherent predictive distribution.

Subject classification. 62A15,62C10,62F15

FULL PAPER.

This article in Mathematical Reviews.