The Effectiveness of Alternative Preference Elicitation Methods in the Analytic Hierarchy Process

Abstract

This study compares alternative preference elicitation methods that are currently available in software implementations of the analytic hierarchy process. For the simple problem used in this study, the elicitation methods may be sorted from least to most accurate as follows: (1) direct estimation, (2) graphical multiple bars, (3) graphical pairwise, (4) numeric pairwise and (5) verbal pairwise. However, the most accurate methods are not the easiest to use. Furthermore, actual accuracy differed substantially from users' perceptions, indicating a need to assist the users in the selection of an appropriate elicitation method. & 1997 by John Wiley & Sons, Ltd.

Analyzing Consensus Measures in Group Decision Making

“…, e m }. Different preference elicitation methods were compared in [5], where it was concluded that pairwise comparison methods are more accurate than non-pairwise methods.…”

Multi Criteria Decision Analysis, Hesitant Fuzzy Methodology towards to Expert Systems for analysing financial markets dynamics

“…The pairwise comparison of alternatives and criteria is at the heart of the AHP methodology. This evaluation yields a more accurate conclusion than a direct evaluation, such as that found in the more conventional weighted sum methodology [51,52,[52][53][54].…”

“…The original AHP has been transformed into AHPSort [24] and AHPSor-tII [55] to keep all the advantages of the original approach . It allows the evaluation of only two items at a time, which results in information that is more precise [51]. Moreover, given that the problem is organized in a hierarchy, it is much simpler to comprehend, explain, and find a solution to it.…”

A Hybrid FMEA‐AHP Framework for Risk Prioritization in Nontransparent Artificial Intelligence Systems

“…For the present problem we retained AHP on three operational grounds rather than on theoretical superiority. First, eliciting full single-attribute utility functions across heterogeneous criteria, clinical harm, financial loss, regulatory exposure, and reputational consequence, from the multidisciplinary panels involved in AI risk assessment imposes a substantial cognitive load and frequently produces incommensurable scales; pairwise comparison on a bounded ratio scale is closer to the way such panels actually deliberate (Millet and Harker 1992 ;Ishizaka and Labib 2011 ). Second, AHP's consistency ratio supplies a quantitative integrity check on each panel's pairwise judgments before any priority is computed, providing an audit trail that maps directly onto the documentation requirements of EU AI Act Article 9 and ISO/IEC 42001 Clause 6.1.…”