Statistical models for occupational exposure measurements and decision making.
Advances in air sampling. ACGIH Air Sampling Procedures Committee, eds. Chelsea, MI: Lewis Publishers, 1988 Jun; :319-336
The usefulness of distributional models for occupational exposure measurements was reviewed. Topics discussed included full period single samples; full period consecutive samples and partial periods consecutive samples; one worker, multiple time weighted average (TWA) data set; grab samples during a single shift for a given employee; multiworker, single TWA data set; proportion of worker shifts with high TWA exposures; and autocorrelation between successive samples. These statistical distributional models underlie data analysis procedures which in turn are used in several areas of occupational exposure monitoring by employers as well as by regulatory enforcement personnel. These areas have included point estimates of exposure distribution parameters; interval estimates of exposure distribution parameters; and statistical significance tests of null hypotheses concerning true values of statistical parameters. Data models were presented which were more detailed than the models used in practice. Separate distributional models were given for the different components of the total variability in exposure measurements: air sampling device error, chemical analytical error, and changes in true exposure level variations over time and space. The separate models were then combined into a distributional model for the total errors in exposure measurements. This more detailed model was then simplified to a conventional lognormal model such as usually employed to analyze actual exposure measurements or exposure monitoring data for repeated work shifts.
Chemical-analysis; Air-quality-monitoring; Analytical-methods; Trace-analysis; Air-sampling; Sampling-equipment; Air-sampling-techniques
ACGIH Air Sampling Procedures Committee
Advances in air sampling