Bus use behavior in multi-route corridors

Abstract

Multi-route corridors are a common feature of bus networks. In these corridors, passengers select between a set of parallel routes. Understanding how passengers make these decisions can help better measure passenger experience and inform network and service planning. This thesis develops three methods for characterizing passenger behavior based on automatically collected data. The first is an empirical analysis relating bus arrivals to bus ridership on each route. The second is a probabilistic model that infers passengers' route choice strategies based on the headways that preceded their bus boarding. The third method is a panel analysis of individuals' stop and route choices over time. These methods are applied to two corridors in London, one that has only local service and another that has both local and limited stop service. On the local-only corridor, the analysis infers that the majority of passengers board the first bus that serves their destination. On the corridor with limited stop service, many passengers opt to wait specifically for the limited stop service route. This boarding strategy is increasingly prevalent as the length of the bus trip increases. Passenger behavior was also found to be affected by crowding, passenger experience on the corridor, and access to real-time information. In addition to the analysis of automated data, this research includes a web-based surveys of users of the limited stop corridor. This survey demonstrates the viability of web-based surveys for collecting detailed information about passenger behavior on a large scale. The survey data shows that passengers' route choice strategies are influenced by factors including trip length, trip purpose, respondent income, use of countdown information, attitudes towards crowding, waiting, and walking, and levels of risk aversion. The thesis relates the analysis of passenger behavior to a set of recommendations for multi-route corridor planning. The advantages and disadvantages of corridor-level scheduling and operation are discussed, and service configuration changes for the limited stop corridor are proposed. Given the prevalence of multi-route corridors and the variety of passengers' route choice behavior within them, the incorporation of an understanding of passenger behavior into network and service planning has the potential to improve passenger experiences on bus networks.