A task-based statistical model of a worker's exposure distribution: Part II--Application to sampling strategy

A task-based statistical model of a worker's exposure distribution for an airborne chemical toxicant is applied to estimating the long-term average exposure level, mu. The precision in estimation is represented by the variance of the sample estimator, denoted by Var[mu]. A traditional sampling strategy consists of integratively measuring the 8-hr time-weighted average exposure level on randomly selected workdays, and computing the sample mean; this strategy is termed "simple one-stage cluster sampling," where each 8-hr workday is a cluster of thirty-two 15-min periods. Three alternative strategies involving measurements of 15-min TWAs are examined: simple random sampling of 15-min periods, and stratified random sampling of 15-min periods with proportional allocation by task, and with optimum allocation by task. All four survey designs provide unbiased estimates of mu. However, for a fixed cost, the stratified sampling designs may provide a lower Var[mu] than simple one-stage cluster sampling for less work time monitored.