Incorporating random constraints in discrete models of choice set generation

This paper proposes a behavioral interpretation of the choice set generation process that is useful for structuring and specifying discrete models that incorporate this stage of choice. The key concept is the notion of random constraints, which is operationalized in the form of probabilistic choice set formation models; we thus explicitly recognize our imperfect understanding and possible lack of data about the process. Several such models are presented and discussed with respect to their behavioral plausibility and estimability. A related paper reports on the calibration of one such specification with work mode choice data from São Paulo, Brazil.