Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of GraphSearch.
An increasing number of studies are concerned with the use of alternatives to random utility maximisation as a decision rule in choice models, with a particular emphasis on regret minimisation over the last few years. The initial focus was on revealing which paradigm fits best for a given dataset, while later studies have looked at variation in decision rules across respondents within a dataset. However, only limited effort has gone towards understanding the potential drivers of decision rules, i.e. what makes it more or less likely that the choices of a given respondent can be explained by a particular paradigm. The present paper puts forward the notion that unobserved character traits can be a key source of this type of heterogeneity and proposes to characterise these traits through a latent variable within a hybrid framework. In an empirical application on stated choice data, we make use of a mixed random utility-random regret structure, where the allocation to a given class is driven in part by a latent variable which at the same time explains respondents' stated satisfaction with their real world commute journey. Results reveal a linkage between the likely decision rule and the stated satisfaction with the real world commute conditions. Notably, the most regret-prone respondents in our sample are more likely to have aligned their real-life commute performance more closely with their aspirational values.
Michel Bierlaire, Shadi Sharif Azadeh, Meritxell Pacheco Paneque