ROBUST BAYESIAN INFERENCE FOR THE HYPER GEOMETRIC PARAMETER

摘要

Following the interesting article by Steck and Zimmer [l1] which appears to make Bayesian inference from a hypergeometric sample rather arbitrary, this paper adapts the methods which Good [4, 5, 6,] proposed for the multinomial, in order to give more 'robust' Bayesian inference procedures for both the hypergeometric and the multivariate hypergeometric distribution, under assumptions of prior ignorance.