Isotonic Procedures for Selecting Populations Better than a Control Under Ordering Prior

Abstract : The problem of selecting a subset containing all populations better than a control under an ordering prior is considered. Three new selections procedures which satisfy a desirable basic requirement on the probability of a correct selection are proposed and studied. Two of the three procedures use the isotonic regression over the sample means of the k-treatments with respect to the given ordering prior. Tables of constants which are necessary to carry out the selection procedures with isotonic approach for the selection of unknown means of normal populations are given. The results including Monte Carlo studies indicate that, in general, the stepwise procedure delta sub 1, using isotonic estimators, is the best. (Author)