Midblock Pedestrian Crash Predictions in a Systemic,Risk-Based Pedestrian Safety Process

Although pedestrian fatalities and injuries in the United States decreased for decades at a rate similar to vehicule occupant fatalities, recent years have seen substantial increases in the pedestrian fatality counts and rate. Most concerning is that the growth in pedestrian fatalities seems to be outstripping any gains in safety. There may be many contributing factors to these increases, including changes in population dynamics, vehicular design, and travel trends, but under more traditional, crash-focused roadway safety management practices, systemic risk patterns are difficult to discern and address. Moreover, locations of risk for pedestrians may be overlooked because important, network-level data types are not collected or analyzed, and pedestrian crashes are often relatively infrequent at specific locations. This paper presents the results of efforts to develop the data profile and analysis methods for a risk-based, systemic pedestrian safety approach. Using 8 years of segment data from the entire street network of the city of Seattle, the research team developed safety performance functions for two types of collision between motor vehicles and pedestrians. These predictive models were used, in conjunction with identified risk factors and countermeasures effectiveness data, to develop a systemic screening tool to identify sites that may benefit from treatment. The end goal of this research is a framework that allows practitioners to identify and prioritize locations within a jurisdiction that are risky for pedestrians and to identify and implement effective, appropriate treatments at many such locations.