Influential Factors on Level of Injury in Pedestrian Crashes: Applications of Ordered Probit Model with Robust Standard Errors

Pedestrian-involved crashes that occurred in the city of San Francisco over a six-year period, 2002–2007, were analyzed to examine various influential factors on the injury severity of pedestrian crashes. The crash data extracted from the Statewide Integrated Traffic Records System (SWITRS) include five categorical levels of injury severity in traffic crashes also in addition to detailed information about the features of each crash. This study applied an ordered probit model for injury severity analysis to specify the ordinal nature of injury categories. To draw unbiased implications from the estimated parameters, statistical tests were performed on the parameters based on robust standard errors. Then, the marginal effects of each variable on the likelihood of each injury level were computed. The variables that significantly increased the probability of severe injury and fatality were: i) age (under age 15 and over age 65), alcohol consumption and cell phone use among pedestrian characteristics; ii) nighttime, weekends and rainy weather among environmental characteristics; and iii) influence of alcohol, larger vehicles (pickups, buses and trucks) and vehicle proceeding straight when striking a pedestrian among crash characteristics. Crash characteristics were found to influence significantly on the level of pedestrian injury. Based on the findings of this analysis, policy implications and countermeasures are also discussed.