Analyzing Spatial Clusters of Public Transit Unreliability Using Scan Statistics with A Social Equity Lens

Keywords: public transit, social equity, transit reliability, cluster analysis, scan statistic
Abstract Type: Virtual Paper Abstract

Abstract

This paper identifies spatial clusters of public transit unreliability (e.g., delays, early arrivals, pass-ups) and examines whether the clusters of unreliable transit operations are socioeconomically deprived areas in Winnipeg, MB, Canada. We use the spatial scan statistics method with a discrete Poisson model to detect the high-risk vs. low-risk clusters of unreliable transit operations. For the social equity analysis, we first see if there are statistically significant differences in socioeconomic status (SES) variables (e.g., proportions of transit commuter, visible minority, low-income family, refugee and unemployment rates) between the high-risk vs. low-risk clusters of public transit unreliability using the bootstrap t-test technique. Further, we also perform binary logistic regression models to estimate the odds of being a high-risk cluster of unreliable transit operation, thereby understanding the demographic and socioeconomic characteristics of neighbourhoods with high-risks of public transit unreliability. Results demonstrate inequities in the spatial distribution of public transit reliability, which can exacerbate mobility and accessibility issues that transit-dependent populations are facing within urban areas. This study provides analytical tools for urban/transit planners and city officials to analyze areas with a high-risk of transit unreliability via a social equity lens. Our approach can provide practical implications for effectively allocating resources to enhance the reliability of public transit service in vulnerable areas dominated by transit-dependent populations.