Number and severity of BI victims, assuming dependence between vehicles involved in the crash

Abstract

The number of victims in vehicles in Spanish motor crashes is analyzed by bodily injury (BI) severity level. Generalized linear mixed models (GLMMs) are applied to model the number of non-serious victims, serious victims and fatalities. Dependence between vehicles involved in the same crash is captured including random effects. After comparing between error distributions, the binomial GLMM is selected. The effect of the driver, vehicle and crash characteristics on the number of BI victims by severity level is analyzed, paying special attention to the influence of the age of the driver and the age of the vehicle. We found a nonlinear relationship between driver’s age and severity, with young and older drivers being the riskiest groups. On the other hand, the expected severity of the crash linearly increased with the vehicle age until the vehicle was 18 years old and then remained constant at the highest severity level from that age. These results are relevant in countries such as Spain with increasing longevity of drivers and aging of the car fleet.