A Multiple Comparison Procedure for Classifying all Pairs Out of k Means as Close or Distant

Abstract

A procedure, using the indifference zone formulation, for making simultaneous pairwise inferences about k means is presented. This procedure has the advantage of offering protection against both Type I and Type II errors. To determine a sample size large enough to afford this protection, least favorable configurations must be determined. This is done in the two- and three-population cases. Tables for the practical application of the procedure to two, three or four means are provided and a method for extending the tables to higher numbers of means is given.