Evaluating the Analytic Hierarchy Process and Recommended Modifications for its Use in Multi-Attribute Decision Makinkg

Abstract

The Combined Arms Analysis Directorate of the Training and Doctrine Analysis Command (TRAC) uses the Analytic Hierarchy Process (AHP) to evaluate the contribution of modernization initiatives to U.S. Army capabilities. This thesis identifies several problems with using AHP. Most significantly, AHP can cause rank reversal of alternatives if a new alternative is considered, even if the new alternative has the same attribute levels as one of the previous alternatives. This thesis proposes several modifications that would improve results when AHP is used. It contains a different method of weight fitting that appears to provide alternative weights that are more accurate than the traditional AHP eigenvalue method. This thesis has two proposals for improving the nine point integer scale by which pairwise comparisons are made. Most significantly, this thesis proposes a modification to AHP that will maintain a ratio scale and avoid rank reversals. This last improvement requires the decision maker to establish and maintain units of measurement. Additionally, the decision maker must make comparisons of attributes to establish a meaningful scale not sensitive to the abundance or lack of alternatives considered. If units are maintained and the decision maker is consistent in the pairwise comparisons, there will be no rank reversals.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1993
Accession Number
ADA275065

Entities

People

  • William H. Mcquail

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Human Systems

DTIC Thesaurus Topics

  • Algorithms
  • Computer Programming
  • Consistency
  • Decision Support Systems
  • Doctrine
  • Eigenvalues
  • Equations
  • Hierarchies
  • Mathematical Programming
  • Measurement
  • Operations Research
  • Optimization
  • Reactive Armor
  • Rolled Homogeneous Armor
  • Standards
  • Training
  • United States

Readers

  • Life Cycle Cost Analysis
  • Regression Analysis.
  • Riverine Ecology