Exploring injury severity in head-on crashes using latent class clustering analysis and mixed logit model: A case study of North Carolina

Abstract

Although only 2 % of crashes are head-on crashes in the United States, they account for over 10 % of all crash-related fatalities. This study aims to investigate the contributing factors that affect the injury severity of head-on crashes and develop appropriate countermeasures. Due to the unobserved heterogeneity inherent in the crash data, a latent class clustering analysis is firstly conducted to segment the head-on crashes into relatively homogeneous clusters. Then, mixed logit models are developed to further explore the unobserved heterogeneity within each cluster. Analyses are performed based on the data collected from the Highway Safety Information System (HSIS) from 2005 to 2013 in North Carolina. The estimated parameters and associated marginal effects are combined to interpret significant variables of the developed models. The proposed method is able to uncover the heterogeneity within the whole dataset and the homogeneous clusters. Results of this research can provide more reliable and insightful information to engineers and policy makers regarding the contributing factors to head-on crashes.