METHODS OF TEST FOR TRANSITIVITYOF INDIVIDUAL EXPERT PREFERENCES

Abstract

Summary. Result of expert information processing is to construct a binary preference relation on the basis of which will be the choice. One of the most important properties of this relationship is transitive. However, due to the incompleteness of the human information processing system, when evaluating the sampling results are random errors, due to which may occur intransitive and inconsistency in preferences. Largely a problem of violations of transitivity individual expert ranking stems from the fact that the expert, comparing alternatives in pairs, generally speaking, is not obliged to think about how to relate these alternatives with all other. In this case it is likely a violation of consistency preferences that make useless the further processing of examination results. In this paper we propose: new universal algorithm for estimating transitive relations expert preferences when ordering of the original sample of alternatives on ordinal scale; modified algorithm for estimating the consistency of expert judgment and construction of structural matrices of various expert preferences between all pairs of alternatives sample which the ranking on difference- classification scale. The basic idea of the algorithm is to on the basis of the support matrices of paired comparisons of individual alternatives to sample to provide not only a review of profile expert orderings on the transitivity, but at the end of the survey of experts to form the structural matrix of all possible preferences between all pairs of sampling alternatives. These algorithms make use of the original information in the form of a matrix of paired comparisons of alternatives to individual sample. The article gives examples illustrating the operation of the above algorithms.