The Multivariate Point Null Testing Problem: A Bayesian Discussion

Abstract

In this paper the problem of testing a multivariate point hypothesis is considered. Of interest is the relationship between the p-value and the posterior probability. A Bayesian test for simple H0 V � D �0 versus bilateral H0 V 6D 0, with a mixed prior distribution for the parameter , is developed. The methodology consists of fixing a sphere of radius around 0 and assigning a prior mass,0, to H0 by integrating the density ./ over this sphere and spreading the remainder, 1 0, over H1 according to ./. A theorem that shows when the frequentist and Bayesian procedures can give rise to the same decision is proved. Then, some examples are revisited.