Decision making under partial probability information

This paper proposes an approach to decision making under partial probability information that involves the use of linear programming (LP). The exact values of state probabilities are not required to construct and solve the LP model, therefore, the approach can be used in situations where exact state probability values cannot be specified but value ranges can be obtained. The LP solution of a decision problem identifies a set of strategies, called efficient strategies, for the given state probability value ranges. Strategies that are not efficient need not be considered further since they are dominated by efficient strategies. The LP model's ease of construction and solution makes it a viable alternative to the techniques currently used in decision analysis.