This article deals with the problem of selecting the population most equivalent to a control from among k independent normal populations using the parametric empirical Bayes approach. By combining useful information from the past data, an empirical Bayes selection procedure is studied. It is proved that the regret of converges to zero at a rate , where n is the number of past observations at hand. A simulation study is carried out to investigate the performance of for small to moderate values of n.
Communications in Statistics: Simulation and Computation 38(8), pp.1690-1713