“…Many earlier injury severity studies have used such a specification to appropriately recognize the ordinal nature of the injury severity levels (see, for example, Xie et al, 2009, Christoforou et al, 2010, Haleem and Abdel-Aty, 2010, Quddus et al, 2010, Jung et al, 2010, Zhu and Srinivasan, 2011. In recent years, unordered-response specifications have also been considered (see, for example, Rifaat et al, 2011, Yan et al, 2011, Malyshkina and Mannering, 2009, Huang et al, 2008, Milton et al, 2008, Kim et al, 2010, and Moore et al, 2011. These unordered-response specifications, while not recognizing the ordinal nature of injury severity levels, provide additional flexibility in capturing variable effects and are also less vulnerable to parameter inconsistency problems caused by varying under-reporting rates (across injury severity levels) in the data (see Ye and Lord, 2011).…”