Estimating possible fraud in coal mine operators' samples of respirable dust.

Coal mine operators are required to sample miners' exposure to respirable dust on a regular basis. A method has been developed for screening all operator samples for detecting possible instances of fraud. This method relies on the occurrence of very low concentration (VLC = 0.1 mg/m3) samples more frequently than expected. The number of underground mine sections from which 50% or more of all samples for FY 1990 were VLC (VLC-50 sections) was compared to the expected number, which was estimated by assuming that the occurrence of VLC samples could be described with a binomial probability distribution. Out of 1983 continuous mining sections, there were 147 VLC-50 sections vs. 7.3 expected (O/E = 20.1). Moreover, these 147 sections occurred more frequently than expected among small as opposed to large mines. For sections using conventional mining methods, 19 of 175 were VLC-50 vs. 0.9 expected (O/E = 20.8). There was one VLC-50 longwall section vs. 0.003 expected. For 99 bituminous and anthracite hand-loading sections the observed number of VLC-50 sections was 65 vs. 78 expected (O/E = 0.8). It is suggested that, except for hand-loading and longwall sections, sampling programs should be investigated at mines with VLC-50 sections.