The consistency adjustment problem of AHP pairwise comparison matrices

Abstract

The Analytic Hierarchy Process (AHP) is an often applied and well researched method in the area of multiattribute decision making. One of the main recurring tasks in AHP situations is the creation of pairwise comparison matrices, the examination of their consistencies, and the derivation of weights for the underlying objective functions. The importance of controlling these consistencies and the consequences of subsequent adjustments of comparison matrices are sometimes unvalued issues in AHP applications. We conduct a simulation study to show that increasing inconsistencies caused by superimposed stochastic error-terms on consistent starting pairwise comparison AHP data result in decreasing correlations between the corresponding weights of the error-perturbed matrices and those of the consistent starting data and demonstrate that the weights derived from adjusted matrices do not show the improvements expected.