The distribution of student's t-statistic for small samples from lognormal exposure distributions.
Am Ind Hyg Assoc J 1988 Oct; 49(10):512-515
Student's t-statistics were studied as methods of evaluating average exposure when based on small samples from lognormal distributions. In industrial hygiene, the compliance officer's test (for noncompliance) and the factory industrial hygienist's hypothesis test (for compliance) were based on the Student's t-statistic. Actual samplings of small samples were compared with theoretical values. When the geometric standard deviation (GSD) was 1.01 the theoretical and emperical results were nearly the same. When the GSD was greater than 1.01 there was a divergence between the actual and the empirical. The greater the variability the greater the divergence that was observed. The degree of Type-I error was shown to be greater for plant industrial hygienists than predicted but less than predicted for compliance officers. The power of the t- statistic was also studied in conjunction with some recently proposed alternatives. The author concludes that the ordinary t- statistic has some shortcomings especially if the sample is skewed but that some of the problems could be eliminated by the use of t- statistic modifications.
NIOSH-Publication; NIOSH-Grant; Statistical-quality-control; Long-term-exposure; Biostatistics; Risk-analysis; Industrial-hygiene
Physiology Department, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115
American Industrial Hygiene Association Journal
Harvard University, Boston, Massachusetts