Answers are grounded in this paper and cite their sources.
The American National Standards Institute (ANSI) standard on fit testing, Z88.10 includes a provision for the acceptance of new fit test methods. The new test method is compared to one of the ANSI Z88.10 accepted quantitative fit tests (QNFT). To determine if the underlying distribution of fit factors affects the Z88.10 sensitivity calculation, four different possible fit factor distributions were examined. Pass/fail rates were estimated based on a logistic regression analysis of actual fit test performance. When a simple sensitivity test is used as recommended by the ANSI standard, the apparent performance of a new fit test can be affected by the distribution of data. It is proposed that a binary logistic regression analysis be used to determine the probability of passing with a new fit test method. An acceptable fit test method should have a low probability of passing the test with low reference method fit factors. Passing criteria with probabilities of less than 0.2 at a fit factor of 50 and less than 0.5 at a fit factor of 100 are recommended by the authors. These criteria are consistent with the current Z88.10 standard’s accepted fit test methods, in that the irritant smoke test does not meet the criteria while the other tests are found to be acceptable.
No takes yet. Share an insight, caveat, or question.
Nelson et al. (2004) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: