Traditionally, efforts to increase the customer base of public transportation agencies have focused primarily on attracting first-time users. Customer retention, however, has many benefits not often realized. Loyal customers provide recommendations to others, increase and diversify their use of the service, and do not require acquisition costs associated with new customers. An earlier study identified key drivers of customer loyalty, with the Chicago Transit Authority (CTA) in Illinois as a case study. A customer loyalty model was created with service value, service quality, customer satisfaction, problem experience, and perception of CTA as constructs. The present study examined customer loyalty differences of captive and choice riders. Captive riders had no viable travel alternatives and might have continued to use transit even if unhappy with service. Choice riders chose to use transit after they compared travel options and might have switched to an alternative if service degraded. Captive riders reported experiencing more problems and were more sensitive to problems; each additional problem brought significant drops in service quality ratings. Captive riders tolerated problems and continued to use transit but showed discontent through their ratings of service quality. Service value was insignificant in captive riders' loyalty decisions because cost-benefit analysis defined service value as irrelevant to them. The relationship between perceptions of CTA and of service quality was stronger for choice riders. If they began the service with high opinions of the transit agency, they were much more likely to have high ratings of service quality than were captive riders.

Automated data collection systems are becoming increasingly common in urban public transport systems, both in the US and throughout the developed world. These systems, which include Automatic Vehicle Location (AVL), Automatic Passenger Counting (APC), and Automatic Fare Collection (AFC), are often designed to support specific and fairly narrow functions within the transport agency. However, it is clear that the data obtained from these systems can have wide-ranging applications within public transport, well beyond the design applications. Of particular interest in the planning of public transport is the opportunity to make use of these increasingly ubiquitous databases to develop a better picture of how public transport systems are performing and being used. In some cases, better estimates of certain performance measures and usage attributes may be made at lower cost than by using conventional data collection methods, even though there are important limitations on the detailed attributes typically available from these systems. In other cases it is possible for the first time to estimate important performance attributes, such as those related to reliability and its impacts, which have hitherto been virtually impossible to quantify because of paucity of data. This paper describes two applications, focusing on system usage and passenger behavior, which have been developed jointly between MIT and the Chicago Transit Authority (CTA), taking advantage of CTA's AFC and AVL systems. The specific applications are the estimation of passenger origin-destination matrices for the CTA rail system and the estimation of path choice models for CTA rail passengers. Next steps in the development of further applications for urban public transport systems are also discussed.

Bus service reliability has long been a top concern for transit agencies and their customers. Improving service reliability, however, has not been easy to accomplish. The use of appropriate recovery times, improved operator training, and better street supervision has produced limited results. Supervision deployment strategies and the use of real-time automatic vehicle location (AVL) information are investigated to improve current supervision practices and enhance bus service reliability. The Chicago (Illinois) Transit Authority's real-time AVL pilot project for Route 20 Madison is the case study for evaluation of the effectiveness of real-time AVL to improve reliability. A simulation model of the route was developed on the basis of archived AVL data and was used to predict the effects on service reliability when real-time AVL information is used in bus supervision. A week-long experiment was carried out both to verify the model and to address the feasibility and scalability of the system. The main conclusion is that real-time AVL does indeed have great potential to improve service reliability. Service restoration strategies previously impossible to execute are now feasible because of this new information stream. However, many obstacles remain to networkwide implementation, including the supervision communications structure and manpower deployment questions. The flood of information into a central control center must also be addressed. Automation techniques and exception-based reporting are strategies to deal with the problem of information overload.