Sensitivity of Light-Duty Vehicle Crash Frequency per Vehicle Mile of Travel to Additional Vehicle and Driver Variables

Publication Type


Date Published




This report examines the sensitivity of the estimated effect of mass reduction on crash frequency in the National Highway Traffic Safety Administration (NHTSA) baseline regression model to including several additional vehicle and driver characteristics. The additional variables include handling and braking capability by vehicle model, from three Consumer Reports road tests; initial vehicle purchase price and vehicle manufacturer; average household income and “bad driver” rating by vehicle model; and whether the driver was using alcohol or drugs, or was properly restrained. The three Consumer Reports road tests are associated with an unexpected increase in crash frequency, in both all crashes and single vehicle crashes with a stationary object. For this reason they were not included in the sensitivity regression models. As expected, vehicle initial purchase price, median household income, and whether the driver was wearing a seat belt are associated with statistically significant decreases in crash frequency, while whether the driver was using alcohol or drugs is associated with a statistically significant increase in crash frequency. A poor average driving record by vehicle model is associated with an expected increase in crash frequency in cars, but unexpected decreases in crash frequency in light trucks and crossover utility vehicles (CUVs)/minivans. Including these variables, either individually or including all in the same regression model, does not change the general results of the baseline NHTSA regression model: that mass reduction is associated with an increase in crash frequency in all three types of vehicles, while footprint reduction is associated with an increase in crash frequency in cars and light trucks, but with a decrease in crash frequency in CUVs/minivans. The variable with the biggest effect is initial vehicle purchase price, which dramatically reduces the estimated increase in crash frequency in heavier-than-average cars and light trucks, and all CUVs/minivans. These results suggest that other, more subtle, differences in vehicles and their drivers account for the unexpected finding that lighter vehicles have higher crash frequencies than heavier vehicles, for all three types of vehicles.

Year of Publication



Research Areas

Related Files