Constructing Highly Consistent Pairwise Comparison Matrices in Analytic Hierarchy Process (AHP) 

Abstract

AHP provides a decision making framework by quantifying the decision elements in order to evaluate alternative solutions with respect to a specified objective in multiple criteria decision making problems. AHP uses pairwise comparison (PC) data for generating weight vectors of decision elements for final results, which is limited to the consistency of PC matrices. Final results (resulting weight vector) can only be considered as a reliable reflection of the evaluator's opinion, if and only if relevant data is sufficiently consistent. By determining the causes of inconsistency, we develop a new method for constructing highly consistent PC matrices. This study investigates underlying reasons of inconsistency, and explores new tools/methods to derive consistent matrices.