Travellers’ perceptions of travel time reliability in the presence of rare events

Travellers account for variability in transport system performance when they make choices about routes, modes and destinations. Modellers try to quantify travel time reliability through various dispersion measures, most commonly the standard deviation of travel time. However, standard deviation is only one attribute of the nuanced travel time distribution. This paper considers whether standard deviation is sufficient to describe the travellers’ understanding and value of travel time reliability and how we might include other aspects of variability such as the frequency of exceeding a lateness threshold or the likelihood of rare events. Car drivers in New South Wales, Australia, were asked to reconstruct the distribution of their commuting time and identify a lateness threshold. Further, we asked them about their preferences in a series of stated choice experiments using three representations of travel time reliability pivoted around their regular commute. The results show reliability ratios consistent with those in the literature for all three presentations. Moreover, the standard error of the estimated coefficient on the risk of rare events indicates that standard deviation alone may not sufficiently capture travellers’ preferences towards travel time reliability.