Optimal Allocation of Observations When Comparing Several Treatments with a Control. III: Globally Best One-Sided Intervals for Unequal Variances.

摘要

In the paper the authors continue their earlier studies of optimal allocation of observations when comparing several treatments with a control. Previously,the authors considered one-sided and two-sided comparisons and described in detail a general allocation procedure which is globally optimal for the case in which the known variances of the 'test'populations are equal,but possibly unequal to the known variance of the 'Control'population;this same procedure is suboptimal for the case in which the known variances of the 'test'populations are unequal (although it is optimal for this latter case within a more restricted class of procedures). In the present paper the authors generalize the results to obtain the globally optimal procedure for one-sided comparisons for the case in which the known variances of the 'test'populations are unequal. (Author)