Analyzing Pedestrian Injury Severity Using Uncorrelated and Correlated Random Parameters Logit Model with Heterogeneity in Means and Variances

Abstract

Pedestrians are an integral part of traffic stream. However, in spite of their undeniable presence in the traffic stream, they are most susceptible to road traffic crashes due to least protection. Pedestrian safety is an issue of great concern in transportation sector especially in low and middle income countries as compared to the developed world where pedestrians are involved in nearly 50% of the road traffic fatalities. For the development and effective enforcement of pedestrian safety strategies, a thorough understanding of the ground realities and crash risk factors is very essential. The present study investigates the impact of various key factors on pedestrian injury severity using pedestrian crash data from Rescue 1122 Peshawar, a National Emergency Response Unit. Using pedestrian crash data for a period of four years (Jan 2016 to Dec 2019) for Peshawar city, present study estimated uncorrelated and correlated random parameters logit models with heterogeneity in means and variances, as well as random parameter logit model with no heterogeneity in means and variances, to investigate numerous risk factors associated with pedestrian injury severities. From a statistical performance standpoint, the correlated random parameters logit model with heterogeneity in means was statistically superior and was selected as the final model. Minor injury, major injury and fatal injury are used as three levels of pedestrian injury severity. Major risk factors that were considered for analysis include pedestrian and driver-specific attributes, roadway geometric details, climatic and visibility conditions and other accident-specific characteristics. Factors found with an increased likelihood of fatal injury to pedestrians were: old age pedestrians (above 50 years old), collision with heavy vehicles such as truck and bus, foggy weather and occurring in the month of March (Spring Season).