Modelling travel time perception in transport mode choices

Modelling travel time perception in transport mode choices

Varotto, S.F.
Glerum, A.
Stathopoulos, A.
Bierlaire, M.
Longo, G.

Civil Engineering and Geosciences

Transport & Planning

2015-12-31

Travel behaviour models typically rely on data afflicted by errors, in perception (e.g., over/under-estimation by traveller) and measurement (e.g., software or researcher imputation error). Such errors are shown to have a relevant impact on model outputs. So far a comprehensive framework to deal with different types of biases related to travel model inputs is missing. In this paper, focusing on travel time, we include two types of measures (i.e., "calculated" and "reported")as indicators of an unobservable true travel time. The aim of including these travel time indicators is to investigate how the underlying travel time perception on behalf of travellers influences the modal choice, compared to the role of externally obtained measurements. The model framework is a latent variable structure where the different travel time indicators serve as manifestations, characterized by different types of biases, of the true travel time. The model is applied to a mode choice case study from Trieste (Italy). Notably, for this data-set, it is established that the calculated travel time distributions (i.e., measured by devices such as an assignment model developed with the software Visum and Google Maps) do not match the reported travel time distributions (i.e., reported by respondents in the survey). Therefore, a discrete choice model that employs available data and accounts for data limitations is developed. Results fromcomparing the base model assuming error-free inputs and the integrated models show more consistent and plausible model outputs such as value of time for the latter.

travel time perception
measurement error
mode choice
discrete choice

http://resolver.tudelft.nl/uuid:34c3bf54-ce8b-47e1-af8d-78d2db6957e6

TRB

94th Annual Meeting Transportation Research Board, Washington, USA, 11-15 January 2015; Authors version

Institutional Repository

conference paper

(c) 2015 The Author(s)