Investigation of factors affecting crash severity of rear-end crashes with high collision speeds in work zones: A South Carolina case study

Abstract

The aim of this study is to identify factors that affect injury severity levels of work zone rear-end crashes with high collision speeds ($⩾$35 miles per hour (mph, 1 mph equals about 1.609 344 km/h)). Using statewide crash data provided by the South Carolina Department of Transportation from 2014 to 2020, a mixed binary logit model with heterogeneity in mean and variance is estimated. The model’s outcome variable is injury or non-injury (i.e., property damage only), and the explanatory variables include information related to vehicle, collision, time, occupant, roadway, and environmental characteristics. The estimation results show that the interstate variable is best modeled as a random parameter at a 90% confidence level. Late-night and dawn/dusk conditions influence the mean effect, while driving under the influence affects the variance of the random parameter. Factors positively influencing injury severity include multi-vehicle involvement, airbag deployment, dark conditions, and truck-involved crashes. Conversely, advanced warning area, activity area, lane shift/crossover, young and middle-aged drivers, and dawn/dusk conditions have negative effects on injury severity.