Abstract
The use of appropriately designed computer aids for policy could improve the standards of risk analysis and other quantitative policy analysis in several important ways. They could make it easier to treat uncertainties more thoroughly and systematically than is now typical. To do this, they should provide a broad variety of techniques for representing uncertain quantities as ranges of alternative values or as probability distributions, for propagating uncertainties through a model, for analyzing and comparing the impacts of different sources of uncertainty, and for displaying results in various numerical and graphic formats. A nonprocedural modeling language allowing interactive editing of input values and model structure could encourage exploration and progressive refinement of models, and comparison of alternative formulations. The integration of model documentation and explanatory text within the computer representation could encourage maintenance of consistency between different versions of the mathematical structure and their descriptions. It could also allow interactive scrutiny of the model assumptions and sensitivities by outside reviewers. We describe a particular system, Demos, designed to provide these facilities and test their usefulness. The use of Demos is illustrated by an analysis of the risks and optimal control level for a hypothetical air pollutant, with uncertainty about the population exposure, health effects, and control costs. This example demonstrates progressive refinement of a model, and various kinds of parametric and probabilistic uncertainty analysis. Demos is now being used by a growing number of risk analysts, students, and policy researchers in a wide variety of tasks.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call