Pragmatic Modeling of Pedestrian Jay walking Behaviour at Signalised Intersections in Urban Area

Abstract

In developing countries, people often jaywalk and indulge in irregular/illicit crossing at signalised crossings which leads to an exponential increase in odds of fatal accidents. This in turn reduces the service quality of signalised crosswalks. Hence, an observational and field study have been undertaken to analyse and model the pedestrian jaywalking behaviour at prominent signalised intersections in an urban Indian city. The authors have collected pedestrian, flow, geometric and crosswalk characteristics using the video-graphic technique followed by the statistical techniques (Multi-Correlation and Exploratory Factor Analysis). The results specify 7 principal factors (socio-demographics, crossing pattern, arrival attributes, road features, dimensions, physical attributes and flow physiognomies) of the pedestrian jaywalking index. Further, binary logit model has identified 7 significant variables namely gender, crossing pattern, type of signal at arrival, number of lanes, width of crosswalk, presence of guard rails and average pedestrian delay in determining the probability of pedestrian jaywalking with 90.39% success rate. Moreover, the area under Receiver Operating Characteristic (ROC) curve (0.891) directs an excellent level of discrimination. The authors suggest to use the developed model by the transport professionals in the evaluation of pedestrian jaywalking behaviour and dealing with safety issues at signalised crosswalks; thus, improving the LOS.