ABSTRACT
One view of the development of the Empirical Bayes (EB) approach is that it is an attractive compromise between the classical non-Bayes and the full Bayes approaches to statistical inference. The EB approach uses previous data to get an estimate of the prior distribution. The EB method is actually only one of several methods of more effective utilization of data from such a two-stage sampling scheme. Established competitors of the EB method are: compound decision theory, the full Bayesian multiparameter approach and a modified likelihood approach. Compound estimation and decision theory is another alternative to EB techniques, which could be regarded as a stimulus for the creation of the EB approach. In general EB estimates can only be better than conventional ones if the prior variance is smaller than the data variance. The risk convergence criterion was introduced for compound decisions as an analogue of asymptotic optimality in EB theory.
