Systematic and random errors in posture percentiles assessed from limited ezposure samples
Upper arm postures are a frequent target for exposure assessment in ergonomics studies. The 10th and 90th percentiles of the angular elevation distribution have been used to assess the occurrence of low and high exposures, and variation has been quantified by the inter-percentile range. These parameters are often estimated using short samples of observed or directly measured postures. In the present study, the performance of sampling from 1 through 60 minutes of data was explored on the basis of 115 whole-day recordings of right upper arm elevation among housepainters. For each day, sample duration and parameter, the mean and standard deviation (SD) of all possible sample estimates were calculated and compared to the true whole-day result. The estimate of the 10th percentile was upward biased, and that of the 90th percentile and the percentile range strongly downward biased with limited sampling, and more severely so with shorter sample durations. As expected, shorter samples even led to decreased precision of the estimate. This risk of pronounced bias and low precision of estimated percentiles should be noted, in particular in ergonomics practice where short sampling durations may be dictated by resource limitations.
Svend Erik Mathiassen Susanne Wulff Svendsen
Centre for Musculoskeletal Research, University of G(a)ve, Sweden Danish Ramazzini Center, Department of Occupational Medicine, Aarhus University Hospital, Denmark
国际会议
17th World Congress on Ergonomics(第十七届国际人类工效学大会)
北京
英文
1-8
2009-08-09（万方平台首次上网日期，不代表论文的发表时间）