Some Combinatorial Problems Related To Rank Sums In A Two Classification

Abstract

Statistical problems may sometimes reguire the use of ranks perhaps dne to the reason that the response can only be ranked but not mesasured. Yet in some other problems, the utilization of ranked. data may be a desirable device to avoid in tlıe Analysis of Variance. an unjustified assumption of normality, implicit However, the analysis of ranked data may in turn lead to some complicated coınbinatioral problems. The author discusses some of these problems and presents theorems and formulations.