Observed and unobserved correlation between crash avoidance manoeuvers and crash severity.

Understanding drivers' responses to critical events, analyzing drivers' abilities to perform corrective manoeuvers, and investigating the correlation between these manoeuvers and crash severity provide the opportunity of increasing the knowledge about how to avoid crash occurrence or at least mitigate crash severity. We extend existing research on the determinants of engaging in crash avoidance manoeuvers by considering that observable and unobservable factors relate to both the selection of corrective manoeuvers and the severity outcome. Accordingly, we propose a joint multinomial-logit ordered-probit model of single-vehicle crashes extracted from the NASS GES database for the years 2005-2009. Results show (1) the existence of unobserved correlation between crash avoidance manoeuvers and crash severity, and (2) the link between drivers' attributes, risky driving behaviour, road characteristics, and environmental conditions, with the propensity to engage in crash avoidance manoeuvers and experience severe crash outcomes.