On the identification of thresholds in travel choice modelling. (December 2015)

• Record Type: Journal Article
• Authors: Obermeyer, Andy; Treiber, Martin; Evangelinos, Christos
• Is Part Of: Journal of choice modelling. Volume 17(2015)
• Page Start: 1
• Page End: 9
• Publication Date: 2015-12
• Subjects: Discrete choice model -- Logit model -- Value of travel time savings -- Threshold; Decision making -- Periodicals; Social choice -- Periodicals; Decision making; Social choice; Periodicals; 302.13
• DOI: 10.1016/j.jocm.2015.12.001
• Languages: English
• ISSNs: 1755-5345

Abstract

Abstract: In this paper we demonstrate different approaches for estimating thresholds with respect to attribute differences in discrete choice models. The proposed new transformation functions can easily be applied in estimation software. The usefulness of these functions is tested with synthetic data. Extensive simulation shows that many observations are necessary to detect thresholds. And furthermore, the value of travel time savings of a model ignoring these thresholds is biased. The application to stated choice data reveals a significant threshold. However, the cause and the precise form of this threshold cannot be inferred from the data. The presented approach might also be useful to model thresholds in other contexts of choice behaviour. Highlights: We show approaches for modelling thresholds using synthetic/stated choice data. Transformation functions can easily be applied in estimation software. If thresholds are not considered, the value of travel time savings might be biased. Many observations are necessary to detect thresholds. A time threshold between two and three minutes has been found in the SC data.