Uncertainty assessment in probabilistic risk assessment

This paper focuses on our proposal for the different roles that data and expert opinion play in uncertainty analysis. Parameters for which reliable data exist are estimated by classical statistical techniques. Their uncertainty bounds are statistical confidence limits. Uncertainty about data-free parameters is expressed as a range, or set, of plausible values, with no probabilistic connotations. For parameters with both data and opinion sources, conditional confidence limits can be used to assess both total uncertainty, and the separate contributions of data-based and data-free uncertainties