Submitted to: International Symposium on the Analytic Hierarcy Process
Publication Type: Proceedings
Publication Acceptance Date: 7/10/1996
Publication Date: N/A
Citation: Interpretive Summary: A decision making situation may involve evaluating several competing alternatives with respect to several criteria that can be categorized in a hierarchical order consisting of groupings and subgroupings of the decision criteria. A method to quickly compute an index range of values that indicates how well an alternative does with respect to the criteria considering the priorities implied by the relationships between the criteria is developed. The method is an analysis tool to be applied after each alternative is evaluated based on each of the decision criteria. The decision maker can change the priority order of the criteria and quickly recompute the range from best to worst of the overall value index of the alternative under consideration. The method has application possibility to a broad range of decision making problems including environmental issues in which there may be several points of view to consider. The simple calculations are illustrated.
Technical Abstract: A method to quickly compute the range of values from the most optimistic to the most pessimistic viewpoint (best to worst) for a hierarchically arranged multiattribute problem under the assumption of an additive value function is presented. The method is an analysis tool to be applied after commensurate attribute values have been determined for each alternative but twithout the need to specify or determine weights on the attributes explicitly. The decision maker can change the priority order of the criteria or attributes at any tier in the hierarchy and quickly recompute the range from best to worst of the overall value measurement. A simple algorithmic method is presented that requires no Linear Program solver. This solution method makes it easy to determine the result of modifying priorities in portions of the hierarchical architecture without recalculating the contributions of unaffected branches. The simple calculations are illustrated and generalization to any hierarchical structure is readily apparent.