Abstract
Several estimators of the probability that the observed best treatment is the true best treatment in a balanced one way ANOVA experiment are compared in terms of their respective mean square errors. For the case of two treatments minimax members of the family of estimators given by Faltin (1980) and of two Bayes families are found and compared with the maximum likelihood and the global minimax estimators. Some of the results are extended to the case of more than two treatments.

This publication has 4 references indexed in Scilit: