Survey-based measurement of transit customer loyalty: Evaluation of measures and systematic biases

Transit customer attrition is an important issue for transit agencies and a key component of transit customer loyalty. In order to investigate what drives changes in transit use, they must be measured, and a common form of doing so are single-wave surveys in which respondents report their likelihood of using the service in the future. This paper evaluates the accuracy of a set of such measures by contrasting them with observed behavior a year later. Furthermore, it evaluates the accuracy of a measure where respondents report on past changes in transit use. The measures considered here are subject to considerable errors, which makes their usefulness in loyalty surveys questionable. However, there is also variability in accuracy across measures, and those that stated a time frame and asked about concrete changes performed best. Predicted attrition was most often followed through on, whereas predictions of stable or increased levels of transit use had higher error rates. All measures appear to generally underestimate attrition, which can lead to an overestimation of loyalty. Using the example of one stated intention measure, two models are constructed to determine whether any of the discrepancies are systematic, i.e., associated with certain traveler characteristics. The results suggest that in several cases, there may indeed be such systematic effects. In particular, the frequency of previous transit use, vehicle ownership, transit pass ownership, existing strategies to avoid routes or times with service quality problems, and attitudes toward the car are identified as being potentially influential.