Abstract
Adequate statistical methods should be used to assess the compliance with the respirable dust standard accurately. The objectives of this article are to study the statistical methods that can be used in compliance hypothesis testing under the assumption that the time-weighted-average dust concentrations follow a lognormal distribution, and to analyze the testing power under different situations. Through simulation, the statistical methods are compared in terms of testing power, at different given type I errors for samples of different sizes and different geometric standard deviations. Based on the results, suggestions are given concerning the possible use of these methods. The power of a test depends on the chosen type I error, geometric standard deviation, sample size, and the ratio of the mean dust level upon which compliance is based to the respirable dust standard. The impact of each factor is studied. With samples of small size (e.g., 5), the testing powers can be high only under very specific conditions. In such case, it is very important to balance type I and type II errors to avoid either the power of the test or the confidence level being too low, especially when the geometric standard deviation is large, and/or the ratio of the mean dust level to the standard is close to 1.