Keywords
mathematical modelling; Bayesian inference; imprecise probabilities; stochastic models; decision analysis; stakeholder involvement; elicitation; expected expected utility
Start Date
16-9-2020 11:00 AM
End Date
16-9-2020 6:20 PM
Abstract
The current practice of uncertainty quantification in environmental decision support is in-complete by not covering all sources of uncertainty. In particular, typically used parameterizations of value functions that quantify preferences of stakeholders are insufficiently flexible to describe stake-holder preferences, the uncertainty of stakeholder opinions and of aggregated “societal preferences” are usually not quantified, and the sensitivity regarding prior assumptions in modelling of the outcomes of decision alternatives as well as on elicited preferences is usually not investigated. In this presentation, the attempt will be made to demonstrate how uncertainty can be considered more comprehensively in environmental decision support. In particular, the points mentioned above will be addressed by showing how more flexible parameterizations can be fitted to elicitation data, how the uncertainty of elicited preferences can be estimated and considered in the decision support process, and how ambiguity about prior distributions can be considered. The methodological approach is based on “expected expected utility theory” that extends expected utility theory to the consideration of uncertain preferences, and on imprecise, intersubjective Bayesian probabilities. The theoretical considerations will be illustrated by examples from water management in Switzerland and recent methodological research.
Comprehensive Uncertainty Assessment in Modelling, Elicitation of Stakeholder Preferences, and Decision Support
The current practice of uncertainty quantification in environmental decision support is in-complete by not covering all sources of uncertainty. In particular, typically used parameterizations of value functions that quantify preferences of stakeholders are insufficiently flexible to describe stake-holder preferences, the uncertainty of stakeholder opinions and of aggregated “societal preferences” are usually not quantified, and the sensitivity regarding prior assumptions in modelling of the outcomes of decision alternatives as well as on elicited preferences is usually not investigated. In this presentation, the attempt will be made to demonstrate how uncertainty can be considered more comprehensively in environmental decision support. In particular, the points mentioned above will be addressed by showing how more flexible parameterizations can be fitted to elicitation data, how the uncertainty of elicited preferences can be estimated and considered in the decision support process, and how ambiguity about prior distributions can be considered. The methodological approach is based on “expected expected utility theory” that extends expected utility theory to the consideration of uncertain preferences, and on imprecise, intersubjective Bayesian probabilities. The theoretical considerations will be illustrated by examples from water management in Switzerland and recent methodological research.
Stream and Session
false