Injury severity analysis of time-of-day fluctuations and temporal volatility in reverse sideswipe collisions: A random parameter model with heterogeneous means and heteroscedastic variances

Abstract

Problem: Sideswipe collisions in the opposite direction often result in more severe injuries than the typical same-direction crashes, especially when light trucks are involved. This study investigates the time-of-day fluctuations and temporal volatility of potential factors that affect the injury severity of reverse sideswipe collisions.

Methods: A series of random parameters logit models with heterogeneous means and heteroscedastic variances are developed and utilized to explore unobserved heterogeneity inherent in variables and preclude biased parameter estimation. The segmentation of estimated results is also examined through temporal instability tests.

Results: Based on crash data in North Carolina, a number of contributing factors are identified that have profound associations with obvious and moderate injuries. Meanwhile, significant temporal volatility is observed in the marginal effects of several factors such as driver restraint, alcohol or drugs impact, Sport Utility Vehicle (SUV) at fault, and adverse road surface across three different periods. Fluctuations in the time of day indicate that restraint with belts is more effective in mitigating the obvious injury in the nighttime, and high-class roadway sustains a higher probability of resulting in more serious injury compared to the daytime.

Practical applications: The findings of this study could help further guide the implementation of safety countermeasures related to atypical sideswipe collisions.

Keywords: Crash injury severity; Random parameter; Sideswipe collisions; Temporal instability; Time of day.