Crash Avoidance Technology Evaluation Using Real-World Crash Data

This study used data on primarily optional safety content from 1.2 million General Motors (GM) vehicles linked to State police-reported crash data by Vehicle Identification Number to estimate field performance of a variety of new safety technologies. After linkage, there were 35,401 vehicles in our analysis dataset. This data included both an indication (presence/absence) of certain types of safety equipment on each vehicle, as well as a variety of crash descriptors at the crash, vehicle, and driver levels. Available covariates were also used to attempt to control for variables that might influence system-relevant crash involvement for the systems examined, including driver age and gender, speed limit, alcohol/drug presence, fatigue, weather, road surface condition, vehicle type, and vehicle model.