The University of Massachusetts Amherst
University of Massachusetts Amherst

Search Google Appliance

Links

A rank-dependent expected utility model for strategic route choice with stated preference data

TitleA rank-dependent expected utility model for strategic route choice with stated preference data
Publication TypeJournal Article
Year of Publication2013
AuthorsRazo M, Gao S
JournalTransportation Research Part C: Emerging Technologies
Volume27
Start Page117
Pagination117-130
Date Published02/2013
KeywordsAdvanced traveler information systems, Choice models, Real time information, Risk taking, Route choice, Stated preferences, Travel behavior
Abstract

Route choice behavior under real-time traffic information needs to be adequately modeled for the proper analysis of a transportation system in the presence of Advanced Traveler Information Systems (ATIS). This paper focuses on strategic route choice, where a traveler is able to plan ahead for traffic information that s/he will receive in the future. A Stated Preference (SP) survey was conducted with interactive maps showing two types of networks with risky travel times, one type eliciting risk attitude and the other allowing for strategic route choice with a detour to an incident-prone road segment and real-time traffic information. The preliminary analysis suggests that a traveler’s risk attitude is probability-dependent. A rank-dependent expected utility (RDEU) model is adopted to account for such a phenomenon, where the decision weight of a probabilistic outcome depends on its ranking among all outcomes and a non-linear transformation of the cumulative probability. A latent-class mixed Logit model for panel data is specified with a RDEU component and two latent classes, strategic and non-strategic route choice. The estimated strategic class probability is significantly different from 0 and 1 respectively, suggesting that a route choice model under real-time information should consider both types of behavior. The estimated RDEU parameters show significant diminishing sensitivities to both outcome and probability and explain the probability-dependent risk attitude.

DOI10.1016/j.trc.2011.08.009