A Pre-Evaluation Step towards a Guaranteed Consistent AHP-Based Pairwise Comparison

Abstract

Reports by various researchers show that ensuring pairwise comparison consistency is one of the main issues when applying analytic hierarchy process (AHP) in a decision making activity. Normally, in the case where inconsistencies are found, the evaluators might have to redo the pairwise comparisons, hence, disturbing the original preference. In this paper, we proposed an evaluation step prior to conducting the pairwise comparison in order to eliminate inconsistent comparisons. The evaluation step involves the rating of importance of criteria using a scale of 1- 9 whereby 1 represents least important while 9 represents extremely important. The resulted ratings are then used as a guide to construct the AHP-pairwise comparison matrix by using certain simple algorithm that is also proposed in this paper. Once the pairwise comparison matrix is obtained, the weights of criteria can be calculated using the existing AHP-technique. Seven different experiments involving 1000 simulated pairwise comparisons for cases with three, four, five, six, seven, eight and nine criteria, respectively, were conducted. All results yielded consistent pairwise comparisons with consistency ratio values of less than the benchmarked value which is CR = 0.1. This technique can really be an added value to the standard AHP-technique since the users of AHP can now skip the consistency test and stop worrying about producing inconsistent pairwise comparison matrices as reported by many previous researchers.