Developing Distributions for Use in Probabilistic Exposure Assessment. A. C. Cullen, Departments of Environmental Health, and Health Policy and Management, Harvard School of Public Health, 718 Huntington Ave., Boston, MA 02115; and H. C. Frey, Department of Civil Engineering, Box 7908, North Carolina State University, Raleigh, NC 27695-7908
The propagation of variance through complex exposure and risk models using Monte Carlo techniques has been simplified by simulation software running on desktop computers; however, the appropriate representation of variability in, and uncertainty about, models and model inputs remains a challenge. The use of distributions based on empirical evidence is certainly desirable; however, it is not always possible to obtain measurements for all inputs under the conditions of interest. Where empirical information is scarce or unavailable, some use of judgment is necessary. To date, guidance has been scattered and incomplete. As a result, probabilistic exposure assessments are carried out somewhat inconsistently, sometimes with great attention to distribution development and its influence on results, and sometimes without. The ambitious goal of the present project is to produce specific recommendations in the form of a guidance document focusing on techniques to be used for the development of probability distributions to be published by the Society for Risk Analysis. To get input to this project, the authors have interviewed dozens of experts in the fields of exposure assessment, uncertainty analysis, statistics, and other related fields. The authors will prepare an initial draft of the guidance document during the early Fall of 1994. In October 1994 the draft document will be presented and critiqued at a small two day workshop comprised of the participating experts and peer reviewers. A second revision of the document will be completed prior to the SRA Annual Meeting. We will present the work-in-progress at the Annual Meeting and invite additional comments. A final draft will be prepared after the Annual Meeting.