Measuring the Determinants of Bus Dwell Time: New Insights and Potential Biases

Abstract

Dwell time is a major component of bus travel time and travel time variability. In turn, the distribution of bus travel times affects transit operators’ costs and customer satisfaction. Previous research used dwell time from bus stop–level data to understand the key factors that contribute to dwell time duration. However, bus stop–level data have significant shortcomings when bus stops are located near intersections or at time points. Regression results show that the use of only stop-level data can significantly bias estimation of boarding and alighting coefficients. This research complements bus stop data with bus GPS trajectory data around bus stops to prevent estimation bias and to measure better the key factors that determine dwell time. Regression results from individual and pooled bus stop models are compared to provide new insights into the impacts of traffic conditions, signalized intersections, bus bays, and time points on dwell times. The impacts of nearside, midblock, and farside bus stops are included in the analysis. The number of passengers boarding and alighting has a nonlinear effect with economies or efficiencies of scale.