Use of Empirical Bayesian Methods to Estimate Crash Modification Factors for Daytime versus Nighttime Work Zones

Abstract

Many studies have been conducted to estimate the traffic safety impacts of roadway construction. Overall, the results of the analyses have varied widely, a variance traditionally attributed to site-to-site and project-to-project differences. In this paper, researchers describe an effort to use empirical Bayesian (EB) techniques to develop crash modification factors (CMFs for construction zones on the basis of temporal factors). Specifically, separate CMFs were estimated for various conditions, including time-of-day (e.g., daytime, nighttime), work status (e.g., activity, inactivity), and temporary traffic control (e.g., lane closure, no lane closure). Daily project inspector diaries from 64 freeway construction projects in four states were analyzed to determine hours of work, hours and locations of temporary lane closures, and the number of travel lanes closed during each work period. Researchers used EB methods in a before–during study design to investigate the safety impacts of the work zone conditions. Work activities that required the temporary closure of one or more travel lanes resulted in the largest CMFs (i.e., the largest increases in crashes), followed by periods of work activity that did not require a lane closure (i.e., work was occurring in the median or beyond the edge of the travel lanes). The lowest increase in crashes occurred during periods when work was not occurring at a project. The CMFs during work activity did not vary significantly between daytime and nighttime conditions when there was a lane closure. The results suggest that EB methods can be useful in assessing temporal factors that influence road safety.