“…Unobserved heterogeneity is defined as the unobservable factors or data that affect crash potential or severity, and they may generate biased estimations if their correlations with observed variables are not accounted for in model design (Mannering and Bhat, 2014). The unobserved heterogeneity could be attributed from different types of factors, including roadways (Flask et al, 2014;Haleem and Gan, 2013;Malyshkina and Mannering, 2010;Morgan and Mannering, 2011), drivers' demographic and behavior characteristics (Haleem and Gan, 2013;Islam and Mannering, 2006;Kim et al, 2013Kim et al, , 2010Morgan and Mannering, 2011;Ulfarsson and Mannering, 2004), spatial and temporal variations Ukkusuri et al, 2011;Xiong et al, 2014;Xu and Huang, 2015), etc. For instance, Kim et al (2010) evaluate pedestrian injury severity patterns in pedestrian-vehicle crashes considering the unobserved pedestrian heterogeneity regarding health, strength and behavior.…”