The problem of selecting the population with the largest mean from several normal populations having a common unknown variance is considered in the context of the indifference-zone approach. A three-stage elimination type procedure is developed. For the choice of the design constants of the procedure a minimax criterion is proposed. A table of design constants is provided which can be implemented in practice. Monte Carlo sampling results are given to compare the performance of our procedure with the two-stage nonelimination type procedure of Bechhofer, Dunnett and Sobel (1954).
|Journal||Sankhyā: The Indian Journal of Statistics, Series B|
|State||Published - Nov 1976|