Alternatives to incorporate uncertainty and risk attitude in multicriteria evaluation of forest plans

This article studies alternative possibilities to incorporate uncertainty and risk attitude into multicriteria forest planning calculations. The interest is in studying the uncertainties involved in forest owner's subjective preferences, but the presented approach could be applied also more generally with measurement and modeling errors concerning forest inventory and forecasting, for example. The models applied are based on ratio scale pairwise comparisons of decision elements and their statistical regression analysis, which enables versatile possibilities to incorporate the uncertainty into decisionmaking. In traditional statistical inference, uncertainty is not used directly as a decision criterion that would be required to include risk attitude. However, the statistical approach can be used, for example, to derive pairwise winning probabilities for forest plans beating each other and rank probabilities for forest plans for attaining a given rank. Several types of indices can be derived from these probabilities enabling the incorporation of risk attitude, but a more promising approach was to reduce the decisionmaking problem to contain mean and SD of the utility distributions for each forest plan and provide them as decision support to the forest owner. The approach is based on classical mean-variance utility in portfolio theory. Also, weighted average of certain percentiles of the utility distributions or weighted average of rank probabilities have their advantages.