Modeling Injury Severity of Unconventional Vehicle Occupants: Hybrid of Latent Segments and Random Parameters Logit Models

Abstract

Road safety research in developing countries has evolved in two categories: (a) crash frequency prediction modeling and (b) injury severity analysis. In injury severity analysis, the focus is to identify the influential factors for different injury severity categories. However, limited research has been undertaken in this domain, especially to assess the injury severity of the occupants of unconventional vehicles (UVOs) (including both human-pulled and engine-operated vehicles). This study investigates the injury severity of UVOs in Dhaka, Bangladesh adopting a hybrid of latent segments and random parameters logit (LSRPL) models. The model is developed utilizing police-reported collision records for the years 2011–2015. The LSRPL model captures multi-dimensional heterogeneity by allocating victims into discrete latent segments (i.e., inter-segment heterogeneity) and allowing a continuous distribution of parameters within the segments (i.e., intra-segment heterogeneity). The model is estimated for two segments using victim and crash attributes, where segment one is lower risk and segment two is higher risk. The model results suggest that victim and driver profiles, crash attributes, environmental factors, road network attributes, transportation infrastructure, and land use attributes influence the injury severity of UVOs. The model confirms the existence of significant inter-segment heterogeneity. For example, mid-block crashes are more likely to result in severe injury in higher-risk segments, and less likely to result in severe injury in lower-risk segments. The model further confirms intra-segment heterogeneity for areas with higher levels of mixed land use. For example, for mid-block crashes, higher mixed land use shows a significantly lower mean in high-risk segments, revealing lower likelihood of sustaining severe injury.