Random parameters logit models applied to public transport demand

The multinomial logit model has been used widely as a fundamental tool for the analysis of discrete choices and has found large application in transport studies. However, its restrictive assumptions, such as independence from irrelevant alternatives (IIA) and preference homogeneity across respondents, have motivated the development of more flexible model structures that allow for an increasingly realistic representation of travel behaviour. Among these, a primary role is played by random parameter models. This paper proposes a comparison between two different specifications of random parameter models, namely the mixed logit and the discrete mixture model. An application to public transport demand is illustrated.