ABSTRACT
This notion plays an important role since it selects the best de cision, according to an effectiveness criterion. Loosely speaking, a decision is admissible only if there is no decision strictly more effec tive than itself. Bayes decisions can be proved to be admissible when the prior is proper, although admissibility is often not accepted as part of the Bayesian approach since it violates the Likelihood prin ciple. However, the converse, that every admissible rule is the Bayes rule for some prior distribution, is generally not true.
