Evaluation of Probability Density Functions Used for Human Exposure Factors: Tracking and Presenting Critical Information. R. L. Maddalena, T. E. McKone, and A. Bodnar; Lawrence Berkeley National Laboratory, and University of California Berkeley
The outcome from a probabilistic risk assessment (PRA) may be highly sensitive to the properties of the input distributions. Thus, when judging the quality of a PRA, one first must judge the adequacy of the distributions that are used in the assessment. Judging the adequacy of an input distribution requires intimate knowledge of how the distribution in question is constructed and applied. We introduce an evaluation framework that tracks and communicates this information in a consistent and useful manner. The range of information, both qualitative (subjective) and quantitative (objective), that emerges from the process of constructing and using a distribution is fit into one of three categories. The first category relates to the original data that is used to construct the distribution. This category includes information about the quantity, quality and relevance of the data. The second category characterizes how well the distribution represents the original data (goodness of fit) and any auxiliary data (cross validation), as well as any theoretical basis that may support the distribution. The final category relates to the importance of the distribution in the context of the overall assessment. Qualitative scores of the information in the three categories are combined and communicated using radial information plots. Developing radial information plots for each exposure factor distribution is a useful way of tracking and presenting the information needed to judge the quality of a PRA.
Go to . . .
2000 SRA Annual Meeting Table of Contents
2000 SRA Annual Meeting Author Index
Main Abstracts Menu Page
RiskWorld Home Page