15.1.2 Multi-criteria Evaluation Models
Broadly, decision models examined can be classified into four categories:
(1) Simple ‘non-compensatory’ methods
(2) Simple Additive Weighting Method
(3) Analytic Hierarchy Process
(4) Concordance Analysis Techniques
Such approaches as dominance, satisficing, sequential elimination and attitudeoriented techniques are called ‘simple’ because the choice of the preferred alternative does not comprise trading-off the compensations on one criterion against the
shortcomings on another. There is no decomposition process followed by an aggregation. A superior performance on one criterion cannot offset an inferiority on
another one. Although it is a simpler process, it leads, however, to less coherent
outcomes than a few of the more complex ‘compensating’ multi-criteria techniques
available.
In dominance techniques, alternatives are designated as members of an exclusive
‘non-dominated’ set if no other alternative exists that achieves more than it on any of
the criteria. Else, the alternative is considered to be ‘dominated’ by another.
Satisficing techniques do not actively select the best alternatives, but merely
exclude those that do not satisfy specified minimum performances corresponding
to each of the criterion.
With sequential elimination approaches, the decision maker decides using only
one criterion, the choice of the alternative depending only on its relative performance
solely on this criterion. After this initial assessment on the first selected criterion,
should there be more than one alternative still remaining, another criterion can help
separate them, and so on, until only one alternative is left for implementation.
Attitude-oriented methods consider the decision maker’s attitude with respect to
the environment surrounding the selection being made. This attitude can be (1) pessimistic, where an alternative will be chosen depending on how badly it performs on
its worst-scoring criterion, or (2) optimistic, where an alternative will be chosen
depending on how well it performs on its best-scoring criterion.
The above are called simple decision models. Other methods include the decomposition and aggregation processes, and entail some form of performance compensation between the different criteria. These methods include:
The Simple Additive Weighting Method is a classical ‘compensatory’ model.
To enable compensation to be considered, the performances of a given alternative on
the different criteria must be measured on one common scale. The different criterion
scores for each alternative can then be handled mathematically to allow computing
an overall performance index for each project option. These indices will allow the
alternatives to be ranked relative to each other. The Simple Additive Weighting
Method is the simplest form of a more general decision model created in Multi
Attribute Utility Theory (MAUT). MAUT, or utility theory, is widely applied in the
financial and actuarial sectors to resolve project appraisal problems, as well as in the
15.1 Multi-criteria Analysis
413
Broadly, decision models examined can be classified into four categories:
(1) Simple ‘non-compensatory’ methods
(2) Simple Additive Weighting Method
(3) Analytic Hierarchy Process
(4) Concordance Analysis Techniques
Such approaches as dominance, satisficing, sequential elimination and attitudeoriented techniques are called ‘simple’ because the choice of the preferred alternative does not comprise trading-off the compensations on one criterion against the
shortcomings on another. There is no decomposition process followed by an aggregation. A superior performance on one criterion cannot offset an inferiority on
another one. Although it is a simpler process, it leads, however, to less coherent
outcomes than a few of the more complex ‘compensating’ multi-criteria techniques
available.
In dominance techniques, alternatives are designated as members of an exclusive
‘non-dominated’ set if no other alternative exists that achieves more than it on any of
the criteria. Else, the alternative is considered to be ‘dominated’ by another.
Satisficing techniques do not actively select the best alternatives, but merely
exclude those that do not satisfy specified minimum performances corresponding
to each of the criterion.
With sequential elimination approaches, the decision maker decides using only
one criterion, the choice of the alternative depending only on its relative performance
solely on this criterion. After this initial assessment on the first selected criterion,
should there be more than one alternative still remaining, another criterion can help
separate them, and so on, until only one alternative is left for implementation.
Attitude-oriented methods consider the decision maker’s attitude with respect to
the environment surrounding the selection being made. This attitude can be (1) pessimistic, where an alternative will be chosen depending on how badly it performs on
its worst-scoring criterion, or (2) optimistic, where an alternative will be chosen
depending on how well it performs on its best-scoring criterion.
The above are called simple decision models. Other methods include the decomposition and aggregation processes, and entail some form of performance compensation between the different criteria. These methods include:
The Simple Additive Weighting Method is a classical ‘compensatory’ model.
To enable compensation to be considered, the performances of a given alternative on
the different criteria must be measured on one common scale. The different criterion
scores for each alternative can then be handled mathematically to allow computing
an overall performance index for each project option. These indices will allow the
alternatives to be ranked relative to each other. The Simple Additive Weighting
Method is the simplest form of a more general decision model created in Multi
Attribute Utility Theory (MAUT). MAUT, or utility theory, is widely applied in the
financial and actuarial sectors to resolve project appraisal problems, as well as in the
15.1 Multi-criteria Analysis
413