Individual judgments are sought in order to elicit values of the entries of pairwise comparison matrices in analytic hierarchy process. Some of these matrices are more consistent than others. But throwing out inconsistent matrices reduces the number of matrices. In this article, we propose a resampling technique to generate sets of pairwise comparison matrices, which pass the consistency check. The advantage of the resampling technique is that one can generate as many sets of pairwise comparison matrices as needed to select the ones that satisfy the consistency requirement. Based on these selected matrices, the priority weights of the alternatives are then estimated. We propose rank-based statistical procedures to check the significance in the difference between estimated priorities of the alternatives to establish their most significant rank order.
All Science Journal Classification (ASJC) codes
- General Decision Sciences
- Strategy and Management