Skip directly to search Skip directly to A to Z list Skip directly to page options Skip directly to site content

NIOSHTIC-2 Publications Search

Search Results

Statistical protocol for the NIOSH validation tests.

Bush-KA; Taylor-DG
Chemical Hazards in the Workplace, Measurement and Control. ACS Symp Ser 1981; :503-517
A statistical protocol was developed and employed for laboratory validation of 311 sampling and analytical methods for monitoring employee exposure to toxic substances. Methods were evaluated to determine if they met an accuracy criterion defined relative to a single measurement from the test method rather than for a mean of several replicate test results. Accuracy was the difference between the test result and the true value, and thus combined the two sources of measurement error: the random errors of sampling and analysis represented by the total coefficient of variation of replicate measurements around their own mean, and the error due to a real bias (systematic error) represented by the difference between average results by the subject collection and measurement method and average results from an independent method. For each method, both types of errors were estimated and results were referred to a decision chart to see if the test method did or did not meet the accuracy criterion. All validation tests were carried out in a single laboratory. Initially the analytical method was tested to assure that it was acceptable for analyte recovery as well as for precision. The net precision due to both sampling and analytical procedures was then tested using known airborne concentrations at each concentration were collected using calibrated critical orifices. The data from these samples and the spiked sample results from the analytical validation were the basic statistical set of data used for the overall method validation. Of the 310 methods validated, only 31 had coefficients of variation above 9 percent. Most of the methods had precisions clustering around 6 to 7 percent. Methods were generally distributed in the minus 10 percent to 10 percent bias region. The authors conclude that the great majority of methods for personal sampling and analysis of toxic substances are of high quality.
Workplace-studies; Industrial-exposures; Safety-research; Analytical-models; Workers; Exposure-levels; Industrial-hazards; Safety-equipment; Toxic-effects; Hazards; Quantitative-analysis
Publication Date
Document Type
Conference/Symposia Proceedings
Fiscal Year
NTIS Accession No.
NTIS Price
Source Name
Chemical Hazards in the Workplace, Measurement and Control. ACS Symposium Series No. 149