This project used data from the North Carolina crash database system and driver license history files to determine driver risk factors that predict fault in fatal and serious injury crashes involving passenger vehicles. After analyzing univariate trends, logistic regression models were developed using driver, vehicle, and environmental factors from the crash data only, from the driver history data only, and using both (combined models) to predict fault in multi-vehicle crashes and single-vehicle crash involvement. The comparison group for both types of analyses was comprised of the non-culpable (no contributing circumstances cited) drivers involved in multi-vehicle collisions.
Results were similar to earlier studies indicating that driver alcohol/drug impairments, impairments due to sleep/fatigue, or other causes, lack of safety belt use, young driver age at the time of the study crash; and having prior at-fault crashes, driving on a suspended license or unlicensed, and having a graduated driver license restriction in the prior five years were reliable predictors of fault in both multi-vehicle and single-vehicle crashes. Older drivers were associated with fault in multi-vehicle, but not single-vehicle collisions. Few other driver history factors were reliable predictors. Driver risk factors were more strongly associated with involvement in single-vehicle collisions than fault in multi-vehicle crashes and model prediction efficiency reflected this result. Additional environmental factors were also associated with single-vehicle crashes including night-time occurrence, rural, and higher-speed locations. These latter factors may also suggest in part driver behavioral patterns. Discussion of potential countermeasures and further research are also provided.
