Refining the analytic hierarchy process: A statistical and methodological exploration of expert judgments

Abstract

The research aims to bridge the gap between theoretical assumptions and the actual behavior of experts in real-world scenarios. This gap has significant practical implications, as it affects the validity of the decisions made using AHP. The adjustments proposed are based on a deep statistical and methodological analysis. The analysis of nearly 500 expert judgment matrices reveals that the upper bounds of Saaty’s 1–9 scale are rarely used, supporting the adoption of a simplified 1–7 scale to reduce experts’ cognitive load and enhance judgment reliability. The authors compared various methods for deriving priority vectors, consistency indices, and deviations. This comparative analysis exposes inconsistencies in traditional approaches and underlines the need for a more context-driven, empirically informed methodology. By decomposing the AHP methodology into eight features with multiple alternatives, our morphological analysis uncovers 4860 potential configurations. This framework not only reveals the latent complexity of AHP but also paves the way for selecting or adapting configurations that better align with the specifics of different decision-making scenarios. The researchers suggest concrete adjustments revising the simulation algorithm for the random consistency index (R.I.) to reconcile theoretical assumptions with empirical behavior.