Rank-based statistical procedures in analytic hierarchy process

Rank-based procedures which do not require any assumption on statistical distributions of pairwise comparisons entries are proposed for a few relevant questions arising in the context of Analytic Hierarchy Process. One of them relates to the issues of finding out whether various groups of individuals are alike in judgments or not. In case not all groups are alike, ways are suggested to form cluster of homogeneous groups. Another concern is that of establishing a particular order of preferences for alternatives in a homogeneous group of individuals.