Integrated value function construction with application to impact assessments

In this article, an integrated structure is provided for processing various forms of imprecise preference information in the context of multicriteria impact assessments. Linear programing formulations generate best-fit value function models and associated ranking of alternatives, both when preferences are overdetermined (leading to potential inconsistencies) or when preference information is incomplete. In the latter context, the algorithm identifies a range of possible rank orders for the decision alternatives under consideration, consistent with the information provided. The approach is primarily aimed at structuring opinions of experts concerning the desirability of different actions in terms of technical aspects, intended as input into the final political decision-making process. It is demonstrated that the approach described here can be implemented with modest levels of effort by the experts. Experience is reported with the approach in the context of soil sanitation problem in the Netherlands, in which experts expressed satisfaction with the resulting rank ordering of alternatives. © 2013 International Federation of Operational Research Societies.