Ordinal Discrete Choice Analyses of Wisconsin Cross-Median Crash Severities

A cross-median crash (CMC), in which a vehicle crosses the median, is one of the most severe crashes because of the risk of colliding with an opposing vehicle. Ordinal discrete choice modeling efforts for investigating the nexus between the underlying severity propensity and miscellaneous roadway-safety-related factors for single- and multivehicle CMCs that occurred from 2001 to 2007 in Wisconsin are described. Ordinal logit (ORL) and probit (ORP) models were employed for the severity analyses. For multivehicle CMCs, the final ORP model found that road condition has a significant effect on severity. Adverse road conditions enhance the likelihood of a more severe consequence if a CMC occurs. Winter precipitation negatively affects CMC severity, and logically Wisconsin's geographical location plays a significant role. The final ORL model found that alcohol and drug use incurs more severe consequences when a CMC occurs. Both models found that more severe injuries occur on roadways posted with higher speed limits. The similarity and dissimilarity in findings by both models imply that it is necessary for safety researchers to apply distinct statistical methods when pursuing a comprehensive understanding of a study topic. The final ORP model for single-vehicle CMCs shows that alcohol and drug use, lane curvature, and unfriendly lighting conditions exacerbate the severity tendency if a CMC happens. A dry road surface is found to incur more severe consequences; this result implies that more severe single-vehicle CMCs are closely related to maintaining overly high speeds. All ORL regression models for single-vehicle CMCs were found statistically invalid. Median width and average daily traffic were found insignificant for both multivehicle and single-vehicle CMCs.