Mitigating the impact of errors in travel time reporting on mode choice modelling. (June 2017)

• Record Type: Journal Article
• Authors: Varotto, Silvia F.; Glerum, Aurélie; Stathopoulos, Amanda; Bierlaire, Michel; Longo, Giovanni
• Is Part Of: Journal of transport geography. Volume 62(2017)
• Page Start: 236
• Page End: 246
• Publication Date: 2017-06
• Subjects: Reported travel time -- Measurement error -- Latent variables -- Hybrid choice -- Mode choice -- Discrete choice; Transportation -- Periodicals; Telecommunication -- Periodicals; Transport -- Périodiques; Télécommunications -- Périodiques; Telecommunication; Transportation; Periodicals; 388
• DOI: 10.1016/j.jtrangeo.2017.05.016
• Languages: English
• ISSNs: 0966-6923

Abstract

Abstract: Travel time is a major component in understanding travel demand. However, the quantification of demand and forecasting hinges on understanding how travel time is perceived and reported. Travel time reporting is typically subject to errors and this paper focuses on the mitigation of their impact on choice models. The aim is to explain the origin of these errors by including elements of travel behaviour (e.g., activities during the trip), which have been shown to significantly affect mode choices and commuting satisfaction. Based on responses from a revealed preferences survey, we estimate a mode choice model that treats travel time as a latent variable and incorporates different sources of data along with information on travel activities. Employing these multiple – sometimes incongruent – sources of information in the choice model appears to be beneficial. Results from comparing a logit model assuming error-free inputs and the integrated hybrid model revealed significant impacts on the generated policy scenarios. The model results also contributed to identifying the main travel activity features that affect travel time reporting, providing indications that can assist in understanding and mitigating the impact of imprecise measurements. Highlights: The impact of errors in reported travel time on mode choice was captured explicitly. Elements of travel behaviour significantly influenced the travel time perception. This framework led to more plausible estimates of value of time. This framework improved the forecast accuracy in zones where the errors were large.