Temporal Analysis of Predictors of Pedestrian Crashes

Abstract

This study investigates the relationship between pedestrian crashes and various socio-demographic, built environment, traffic exposure, and roadway characteristics across different times of day for both weekdays and weekends. Using the street segment as the unit of analysis, multilevel generalized linear mixed models with negative binomial estimators are applied to examine predictors of pedestrian crashes, including those resulting in severe injuries and fatalities, that occurred in Philadelphia, U.S., between 2010 and 2017. It is found that most of the relationships between the predictor variables and pedestrian crashes are consistent throughout the day for both weekdays and weekends. Although traffic volumes and pedestrian trips fluctuate throughout the day, average daily measures of traffic and pedestrian exposure have consistent relationships with pedestrian crashes throughout the day for both weekdays and weekends. Certain roadway characteristics, such as the amount of secondary highways and major arterials, have stronger relationships with pedestrian crashes than others at certain times of day. Results indicate that authorities should pay particular attention to pedestrian safety at night, as well as in lower-income neighborhoods throughout the day when designing interventions to improve the walking environment. Modeling pedestrian crashes by time of day provides additional information that might not be captured by temporally aggregate analyses. Scholars should consider incorporating time of day into future traffic crash analyses.