Pedestrian behavior and safety at signalized intersections

Abstract

Signalized intersections are among the probable location for pedestrian crashes. To formulate effective pedestrian-safety strategies, it will be necessary to enhance understanding of pedestrian behavior and develop predictive models to identify the influence of possible contributory factors on pedestrian decision making. This research investigates pedestrian crossing behavior at signalized intersections, and develops predictive models to identify factors that contribute to pedestrian crash frequency and injury severity.

To address pedestrian crossing behavior, a two-tier model is proposed to represent pedestrian route choice decisions when crossing signalized crosswalks. A mixed logit model is used for the direction choice and an exponential model for determination of step size. Pedestrian jaywalking behavior is also investigated. A binary logit model is used to identify factors that may influence a pedestrian’s decision to jaywalk. In both models of pedestrian crossing behavior, random-parameter models are used to accommodate inter-pedestrian heterogeneity.

To address pedestrian safety, a count-data model is used to explore the relationships between pedestrian crash frequency and various site-specific characteristics. A binary logit model is used to indicate the effects of various individual-specific and site-specific factors on injury severity. A random parameter model is used to account for the heterogeneity of pedestrians and intersections.

With the use of predictive models, the factors that influence pedestrian crossing behavior are identified. In addition, the relationships between pedestrian crash frequency, injury severity and other intersection characteristics are explored. These results will aid in the formulation of effective pedestrian-safety strategies.