Using Binary Logistic Regression to Explain the Impact of Accident Factors on Work Zone Crashes

For consolidated road networks, the identification, programming, and implementation of maintenance actions enables addressing the deficiencies identified in the infrastructure, ensuring the provision of an adequate service to users. The performance of such actions along the infrastructure lifetime makes it necessary to study the impact that road work zones may have on road crashes since these areas change locally and temporarily the traffic conditions offered to users (lower speeds, the presence of work equipment and workers, narrow lanes, changes in vertical and horizontal signs, etc.). This study aims to analyze the Portuguese official road work zones crash data from 2013-2015 period by using binary logistic regression models to identify the most significant factors influencing work zone crashes. Official data was processed in order to be used in a statistical analysis software and the binary logistic regressions were performed for the analysis of Portuguese work zone crashes by the type of crash (pedestrian, angle, rear-end and runoff road), driver age groups (under 25 years, 25 to 64 and over 65 years) and a predominant contributing factor as speeding, unexpected obstacle on the road and the disregard for vertical road signs and safety distance (main contributing factors identified in this study). Results obtained shows that factors as “urban environment”, “one driver involved is running straightly”, “clean and dry pavement” and “daylight” have positive impact in a large number of models. The identification of these factors allows supporting the definition of strategies aimed at the reduction of the number and severity of crashes in road work areas.