Analysis of duration between crashes using a hazard-based duration approach with heterogeneity in means and variances: Some new evidence

Abstract

This paper provides new evidence for the factors underlying crash involvement by modeling the time duration between crashes for drivers involved in one or more crashes between 2016 and 2020. Several random parameter hazard-based duration models with heterogeneous means and variances are presented. Among this study’s other findings, the results show that male drivers had a higher risk of being involved in one crash than female drivers (among drivers involved in only one crash). Female drivers were more likely to be involved in higher-order crashes however. Among female drivers involved in only one crash, millennials had the highest crash risk. However, baby boomers and Gen Z drivers had a greater risk of being involved in a crash than millennials or Gen X drivers, whether male or female. The analysis presents evidence for distinct crash risk patterns in men and women and among different age groups. The lagged duration dependence indicates that the longer the time from a previous crash, the sooner the driver will be involved in their next crash. In addition, the lagged duration dependence suggests two types of dependencies. The first is profound dependency. Drivers with this type of dependency tended to be tier-three male millennials, tier-three Gen X drivers, tier-three Gen Z drivers, or tier-four male millennials. The second is shallow dependency. Drivers with this type of dependency tended to be tier-three female millennials, tier-four male Gen X drivers, and tier-five male millennials. The likelihood of a crash was almost independent of the time that had transpired without a crash for those involved in more than one crash. Estimation results also revealed that crash survivors showed different subsequent behavior. Surviving a severe crash and experiencing crashes involving multiple vehicles may lead to hazardous habituation among male millennials. Moreover, many drivers seemed to alter their behavior after the first crash, particularly male and female drivers involved in one crash only. Other drivers did not show any behavioral changes, including tier-three female millennials, tier-four male Gen X, and tier-five male millennials, who had a shallow lagged dependency, and their likelihood of a crash was almost independent of the time that transpired without a crash.