This paper presents the findings of the vehicle occupant injury severity model, particularly focusing on the collisions involving distracted driving. The study develops a latent segmentation-based logit model for analyzing crash injury severity utilizing police-reported collision data from 2007 to 2011 in Nova Scotia, Canada. A segment allocation model is estimated to capture latent heterogeneity based on individual victims’ and drivers’ profiles, and collision attributes including vehicle type, vehicle trajectory, collision object, and collision type. The segment allocation model results suggest the existence of high-risk and low-risk injury severity segments. This study extensively tests the effects of built environment characteristics. The model results suggest that rain, curved road, freeway, and mid-block collisions aggravate vehicle occupant injury severity; whereas, higher land use mix, longer length of sidewalk, and higher population density mitigate injury severity. Significant heterogeneity is found across the high- and low-risk segments. For instance, straight road alignment is found to yield higher injury severity in the high-risk segment and lower severity in the low-risk segment. Moreover, the model unveils the interplay between built environment and distraction type. Driver distraction by communication device increases injury severity at a curved road intersection. Additionally, distraction because of inattentiveness increases injury severity. The findings of this study assist road safety engineers and planners to identify effective countermeasures and awareness programs for reducing the crash injury severity or consequences for vehicle occupants.
The influence of urbanicity and built environment on the frequency of distracted driving-related crashes: a multi-state comparison