Stability of Ranking-Dependent Pair-wise Comparison Patterns in AHP
A recent study posted on arXiv, paper number 2608.05958, examines the reliability of three ranking techniques in the Analytic Hierarchy Process (AHP): the Best-Worst method, the Top 2 method, and the classic maximum difference method. The authors investigate how leveraging ordinal data can enhance the quality of expert evaluations while minimizing required comparisons. While the first two methods are found to be incomplete, the maximum difference approach is deemed complete. The research also highlights the susceptibility of these methods to errors by experts and includes findings from a simulation, pinpointing the most stable incomplete ranking pattern to aid decision-making.
Key facts
- The paper is published on arXiv with identifier 2608.05958.
- It compares three ranking-dependent pair-wise comparison patterns in AHP.
- The methods compared are Best-worst, Best-Second Best (Top 2), and maximum difference.
- The first two methods are incomplete; the third can be complete.
- The study determines conditions for comparing stability to expert errors.
- A simulation-type experiment was conducted.
- The research aims to define the most stable incomplete pattern.
- The goal is to reduce the number of expert comparisons.
Entities
Institutions
- arXiv