NONPARAMETRIC STATISTICS IN MULTICRITERIA ANALYSIS

Abstract

The paper deals with a method of hierarchization of alternatives in a multicriteria environment by means of statistical nonparametric procedures. For each criterion, alternatives are disposed on an ordinal scale. After that a procedure similar to ANOVA is activated on the data. The differences relating to the average ranks of each action are used to build the hierarchical algorithm. The concept of outranking, in a probabilistic meaning, is reached in this way. Thereafter, we arrive at the concept of dominance using pairwise comparisons. A case study is presented which implements the methodological suggestion.