Strict preference and sensitivity analysis in additive utility functions with interval data

Several approaches in the literature on multi-criteria decision making deal with decision problems under incomplete information to support decision makers who cannot specify the parameters required by a decision method precisely. In the context of additive weighting models, these parameters involve both decision weights and partial utility values of alternatives. Standard methods for decision making under incomplete information require the solution of optimization models for each pair of alternatives. We show that if imprecise information is available in the form of bounds on weights and attribute values, the resulting model can be solved efficiently and the relation between alternatives which it implies has desirable properties.