Analysis of driver injury severity: Logit models of truck involvement/truck causation.

摘要

This research explores how various risk factors and in particular truck-causation vs. truck-involvement affect the probability of injury severity to drivers of vehicles in truck-involved accidents. The study also explores the differences in probability of injury severity categories sustained by the drivers in truck-involved accidents in rural vs. urban areas. Five logit model structures were hypothesized for estimating driver injury severity models for truck-involved accidents on California urban and rural highways. The models included two multinomial logit models (MNL) for urban and rural areas, and eight nested logit structures (i.e., 4 for urban and 4 for rural). The nested logit structures were rejected and the MNL models were accepted as valid logit models for the data. MNL probability models were estimated to predict the probability of four injury severity categories: Property Damage Only (PDO), Complaint of Pain (CP), Visible Injury (VI), and Severe/Fatal Injury (SFI) conditioned on an accident occurring. The results show significant differences in injury severities sustained by the automobile drivers compared to the truck drivers. The urban model results showed a more pronounced increase in the probabilities of VI and SFI injuries to drivers of automobiles if the automobile drivers were at fault in truck-involved accidents. Significant differences with respect to other risk factors including driver, vehicle, environmental, road geometry and traffic factors exist between urban and rural models. Despite significant differences between the rural and urban models, thirty variables out of a total of 50 variables that were significant in either the rural or the urban model entered both models, though with varying effects on driver injury probabilities.