NIOSHTIC-2 Publications Search
Combining professional judgment and exposure data - Bayesian decision analysis.
American Industrial Hygiene Conference and Exposition, May 8-13, 2004, Atlanta, Georgia. Fairfax, VA: American Industrial Hygiene Association, 2004 May; :1-17
Bayesian Decision Analysis (BDA): An adjunct or alternative to the calculation and interpretation of traditional statistics. The goal of BDA is to estimate the probability that the true exposure profile falls into a particular category, or Rating Zone. How do we decide which rating zone best describes exposures in a particular work environment, and express our confidence in that decision? Summary: Conventional IH Data Analysis and Interpretation Rate the SEG. Collect n measurements per exposure group Estimate group statistics. Calculate the sample X0.95 and its 95%UCL Compare to OEL (or exposure ratings). Mentally factor in professional judgment and experience, and reach a decision: Acceptable vs. Unacceptable or HC, WC, C, PC, or UC. The Bayesian Method: Combining professional judgment (or other information) with our current data to reach a decision is a Bayesian process. Bayesian statistics: branch of statistics that acknowledges that decisions are rarely based upon the data alone. Is our IH a Bayesian statistician? The Bayesian Method - Key Concepts: Prior distribution 1.) Represents our expectation based upon professional judgment, past data, experience, modeling, etc. 2.) Two types: Non-informative prior - represents complete ignorance of the work environment; Informative prior - mathematically represents our judgment or experience with this work environment.
Analytical-processes; Industrial-hygiene; Industrial-hygiene-programs; Industrial-hygienists; Mathematical-models; Qualitative-analysis; Quality-standards; Quantitative-analysis; Safety-education; Standards; Statistical-analysis; Statistical-quality-control; Work-analysis; Work-environment; Work-operations; Work-performance; Workplace-studies; Work-practices
American Industrial Hygiene Conference and Exposition, May 8-13, 2004, Atlanta, Georgia
University of Minnesota Twin Cities