Validity assessment of self-reported construction tasks.

This study assessed agreement between workers' and observers' daily estimates of exposure to construction work tasks. The ultimate aim was to develop a valid method and instrument for the collection of self-reported data on duration of exposure to a priori identified work tasks for use in characterizing exposure in settings with substantial task variability. Forty-nine shop workers and 52 construction site sheet metal workers were observed for up to 3 full workdays. Observers sampled approximately 25% of each worker's day, recording the work performed from a prespecified list of tasks. Each participant completed a daily questionnaire, indicating the tasks he or she performed that day and time spent on each task. Shop workers tended to specialize in particular tasks, while at the construction site, the workers' tasks reflected substantial day-to-day variability. Agreement between worker and observer estimates was generally better for major shop tasks (with intraclass correlation coefficients [ICCs] ranging from 0.52 to 0.85), than for major construction site tasks (with ICCs ranging from 0.36 to 0.64). Workers tended to overestimate the amount of time spent at tasks of longer duration and to underestimate time spent at short-duration tasks. Rank order analysis of time spent on task revealed fairly high agreement. Agreement was acceptable for shop-based work, which has less day-to-day variability than construction site work. Overall, however, the data suggest that, for highly variable work, the use of task as the unit of exposure does not improve recall over assessment approaches focusing on questions about posture and material handling.