Predicting Truck Crash Involvement: A Commercial Driver Behavior-Based Model

摘要

With safety as a top priority, the trucking industry has worked diligently to identify novelmethods for further reducing an already record low number of truck crashes. In an attempt torealize that goal, this research has utilized an archival research design to identify truck drivererrors that frequently accompany future crash records. Specifically, the researchers collectedtwo years’ worth of recent crash, violation and conviction data for 582,772 U.S. commercialmotor vehicle drivers and found dozens of driver infractions that were significant predictors offuture crash involvement. In addition to pinpointing which specific events were significantlylinked to commercial vehicle crashes, the researchers quantified the strength of each association,revealing future crash risk increases that ranged from 26 to 96 percent. Beyond measuringindividual driver events, principal components analysis was used to group the more than thirtydriver infractions from this study into 11 related factors. This smaller set of negative driverbehavior categories was then entered into a logistic regression model, which revealed the classesof behaviors that trucking companies and enforcement agencies should be most proactive intargeting for enhanced driver training and enforcement efforts, respectively.