Some Probabilistic Revealed-Preference Models of the Choice Process in Recurrent Urban Travel

Abstract

Several probabilistic models are introduced for individual destination choice in intraurban travel. They are designed for revealed-preference estimation using travel diary data. The models are adaptations of Thurstonian' random utility models in psychology. They contain two logically distinct components that are termed here the choice model and the response model. This paper focuses on alternative forms for the choice model. Two broad classes of choice models are given. One, containing nonbinary' models, results from direct application of Thurstone's structure. Because this structure is mathematically complex for large choice sets, an alternative class ofbinary models is proposed, involving the representation of set-choice probabilities in terms of underlying (non-revealed) binary preferences. A variety of operational models results, which are being tested on data of the Uppsala Household Travel Study. Some comments on the response model and estimation are given, though these topics are covered in detail elsewhere.