A common framework for deriving preference values from pairwise comparison matrices
Introduction
In evaluating n competing alternatives A1,…,An under a given criterion, it is natural to use the framework of pairwise comparisons represented by an n×n square matrix from which a set of preference values for the alternatives is derived. Many methods for estimating the preference values from the pairwise comparison matrix have been proposed and their effectiveness comparatively evaluated. Some of the proposed estimating methods presume interval-scaled preference values [1], [2]. But most of the estimating methods proposed and studied are within the paradigm of the analytic hierarchy process that presumes ratio-scaled preference values [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16]. The main challenge is how to reconcile the inevitable inconsistency of the pairwise comparison matrix elicited from the decision makers in real-world applications.
When the decision maker is unable to rank the alternatives holistically and directly with respect to a criterion, pairwise comparisons are often used as intermediate decision support. The approach taken herein is to specify a common way of measuring effectiveness of the solutions based upon what the estimating procedures are attempting to achieve. The stronger form of effectiveness is how close the ratio of derived preferences (wi/wj) come to the pairwise ratios (tij) in the comparison matrix. Those estimating techniques that rely on minimizing the aggregated distance between wi/wj and tij are called the “distance minimization” methods.
A weaker form of effectiveness relates to different ways of calculating preference values from the column of an error free comparison matrix. Since these techniques are known to give the correct preference values in error free cases, they may possess some efficacy when used under imperfect conditions with some errors. We call them “correctness in error free cases” methods. Although the distance minimization methods achieve the correctness in every error free case, we consider them under their stronger features of distance minimization.
In the paper, we are interested in the ability of the methods to estimate ratio-scaled preference values from pairwise comparison matrices. As mentioned above, our approach is to use the two effectiveness concepts as a common framework for the many methods of deriving preference values. Some comparison results of these 18 methods on two sets of matrices with small and large errors are presented. We also give insight regarding the underlying mathematical structure of some of the methods.
Problem structure is given in Section 2 and notations used are given in the appendix. A common framework based on two general concepts of effectiveness is introduced and discussed in Section 3. Some estimating methods based on distance minimization are introduced in Section 4. In Section 5, we describe several estimating methods that satisfy “correctness in error free cases.” Some numerical comparisons are given in Section 6 to illustrate the strength of these 18 estimating methods on two sets of judgment matrices with small and large errors. Insight from the underlying mathematical structure of some of the methods is discussed. Finally we highlight the importance of commensurate scales and provide some concluding remarks.
Section snippets
Problem structure
In evaluating n competing alternatives A1,…,An under a given criterion, it is common to use the framework of pairwise comparisons. The basic assumptions are that there exist preference values v1,…,vn such that vi represents the preference intensity of and the decision maker (DM) is able to provide as answers to simple questions on pairwise comparison of vi with vj for all i,j=1,…,n. Usually, the pairwise question is in two parts: (1) first an ordinal question that asks which
Common framework of effectiveness for estimating preference vector
The pairwise comparison judgment matrix with can be regarded as the n approximations of , one approximation for each column. Thus, an estimating method with is effective whenever the distance between [wi/wj] and is very small or minimized. There are many distance functions from and B are n×n matrices} to nonnegative real numbers [14]. Definition For each distance function D on n×n matrices, we define an estimating method associated to
Distance minimization estimation methods
For every distance function D, there is an associated estimating method . We consider only the commonly used distance functions between n×n matrices in this paper. Estimating methods associated with some selected distance functions are given below. The distance functions used are basically defined in terms of the squares of deviations or absolute deviations. These deviations can be modified by preference weights, logarithmic transformation and mini–max operator.
(1)
“Correctness in error free cases” estimating methods
Instead of aggregating by minimizing deviations between tij and wi/wj, we can use an estimation method that is known to produce correct results in error free cases. Any computational procedure that can compute correctly from an error free may be regarded as an effective estimating method. Let be the columns of T. When is error free, all the columns of are positive multiples of since , j=1,…,n. Thus, we can regard each column of as an approximation of
Numerical tests
All the above methods can compute the preference vector v correctly when the judgment matrix is error free with . When is not error free, different solution vectors would be generated by different methods. Two sets of judgment matrices with some errors are used to compare the effectiveness of the 18 methods. Let be the true preference vector and be the reciprocal judgment matrix obtained from [vi/vj] by introducing error εij to
Discussion and conclusion
It is shown that a common framework to most methods of deriving preference values from pairwise judgment matrices can be established from the common concept of effectiveness except the statistical procedures by Basak [4]. Distance-based estimating methods are more appealing because it is undisputable that the matrix [wi/wj] associated to the solution vector w must be “close” to the judgment matrix . Weighted by preference values seems to be a desirable property but does not seem to enhance the
Acknowledgements
We gratefully acknowledge the support of the Natural Sciences and Engineering Research Council of Canada (RGPIN8432). We also wish to thank the editors and anonymous referees for their constructive comments and suggestions.
Eng-Ung Choo is a full professor of Management Science in the Faculty of Business Administration, Simon Fraser University, Burnaby, BC, Canada. His primary research interest is multi-criteria decision analysis. He has published articles in Computers & Operations Research, Journal of operational Research Society, European Journal of Operational Research, Computers & Industrial Engineering, Annals of Operations Research and Journal of Multi-Criteria Decision Analysis.
References (27)
Comparison of statistical procedures in analytic hierarchy process using a ranking test
Mathematical and Computer Modeling
(1998)- et al.
A Monte Carlo investigation of incomplete pairwise comparisons matrices in AHP
European Journal of Operational Research
(1997) - et al.
Statistical theory for the ratio model of pairwise comparisons
Journal of Mathematical Psychology
(1992) - et al.
Interpretations of criteria weights in multicriteria decision making
Computers and Industrial Engineering
(1999) - et al.
Deriving weights from pairwise comparison ratio matricesan axiomatic approach
European Journal of Operational Research
(1988) On deriving priority vectors from matrices of pairwise comparisons
Socio-Economic Planning Sciences
(1986)- et al.
A Statistical look at Saaty's method of estimating pairwise preferences expressed on a ratios Scale
Journal of Mathematical Psychology
(1994) Ranking alternatives—comparison of different methods based on binary comparison matrices
European Journal of Operational Research
(1987)A simulation study of estimation methods in the analytic hierarchy process
Socio-Economic Planning Sciences
(1986)- et al.
A multicriteria evaluation of methods for obtaining weights from ratio-scale matrices
European Journal of Operational Research
(1993)
An alternative scaling method for priorities in hierarchical structures
Journal of Mathematical Psychology
Comparison of eigenvalue, logarithmic least squares and least squares methods in estimating ratios
Mathematical and Computer Modeling
A note on the analysis of subjective judgment matrices
Journal of Mathematical Psychology
Cited by (0)
Eng-Ung Choo is a full professor of Management Science in the Faculty of Business Administration, Simon Fraser University, Burnaby, BC, Canada. His primary research interest is multi-criteria decision analysis. He has published articles in Computers & Operations Research, Journal of operational Research Society, European Journal of Operational Research, Computers & Industrial Engineering, Annals of Operations Research and Journal of Multi-Criteria Decision Analysis.
William Wedley is a full professor of Management Science in the Faculty of Business Administration, Simon Fraser University, Burnaby, BC, Canada. He obtained his Ph.D. from Columbia University, specializing in international business and production management. He has published articles in Decision Sciences, Social-Economic Planning Sciences, Canadian Journal of Administrative Sciences, Academy of Management Review, Journal of operational Research Society, European Journal of Operational Research and Journal of Multi-Criteria Decision Analysis.