Measuring Safety Performance

Abstract Partial-interval (PIR), whole-interval (WIR), and momentary time sampling (MTS) estimates were compared against continuous measures of safety performance for three postural behaviors: feet, back, and shoulder position. Twenty-five samples of safety performance across five undergraduate students were scored using a second-by-second continuous recording procedure to calculate actual levels of safety performance. Then 15, 30, 45, 60, and 75 s (seconds) PIR, WIR, and MTS estimates of safety performance were calculated. The results of Analysis 1 show that across all samples, PIR overestimated and WIR underestimated safety performance. MTS both under- and overestimated safety performance, but with substantially smaller errors than both PIR and WIR. Actual levels of safety performance were then categorized into low, moderate and high levels of safety performance for further analysis in Analysis 2. The results demonstrate that as safety performance increases, the degree of WIR underestimation increases across all behaviors, and the degree of PIR overestimation varies across levels of safety and behaviors, the level of safety occurrence did not appear to differentially affect MTS estimates.