Inferences from Indifference-zone Selection Procedures

Two-stage indifference-zone selection procedures have b widely studied and applied. It is known that mos indifference-zone selection procedures also guarantee m tiple comparisons with the best confidence intervals w half-width corresponding to the indifference amount. W provide the statistical analysis of multiple comparisons wi a control confidence interval that bounds the difference b tween each design and the unknown best and multiple co parisons with the best confidence intervals. The efficien of selection procedures can be improved by taking in consideration the differences of sample means, using variance reduction technique of common random numbe and using sequentialized selection procedures. An exp imental performance evaluation demonstrates the valid of the confidence intervals and efficiency of sequentializ selection procedures.