TASK-BASED EXPOSURE ASSESSMENT - ANALYTICAL STRATEGIES FOR SUMMARIZING DATA BY OCCUPATIONAL GROUPS

This article presents task-based exposure data measured on multiple wo rkers within occupational categories and discusses (1) the importance of defining the basic unit of analysis and choosing an analytical plan consistent with it; (2) analylical approaches for estimating the task means and variances, including procedures for weighting the observati ons by measurement times; and (3) methods for estimating the mean and the variance of the mean; or an overall occupational category, includi ng methods for incorporating variability due to varying task proportio ns and dependence of worker exposures over tasks. The goal is to provi de analytical techniques for summarizing task-level exposure data to t he occupational group level. A simulation is used to illustrate import ant principles and show how the proposed methods perform better than o ther standard approaches. When the current survey data are used to est imate task proportions and worker dependency over tasks, a proposed ja ckknife method for computing the variance of the overall occupational category mean performs well. A proposed method for incorporating the v ariability in task proportions when their source is external to the cu rrent survey is also evaluated. When the variability is properly incor porated in the computations, stratification by task does not lead io a ny reduction in estimated variance of the overall occupational categor y mean. Weighting the observations by the measurement times provided n o reduction in the variance estimates of the means.