Alternative Non- Systematic Heterogenity in Destination, Mode and Simultaneous Destination- Mode Choice (Case Study: Qazvin Shopping Trips)

Destination and mode choice play an important role in travel demand analysis and policy assessment as a consequence. To study these two choices, most researchers have used multinomial logit which suffers from the assumption of all error terms being independently and identically Gumbel distributed. This leads to the inability to consider the non-systematic heterogeneity in the alternatives. Heteroscedastic extreme value model assuming independent and non-identical distribution can analyze the alternatives with non-systematic heterogeneous utility. The purpose of this paper is to analyze this heterogeneity and compare results with those of multinomial logit in destination, mode and simultaneous destination- mode choice for shopping trips in Qazvin city. Three choices for destination based on distance; and three choices for modal split including walking, private car and bus (totally 9 choices) are considered. The results for 1570 person- trip according to different criteria (such as log-likelihood function and likelihood ratio index) show the superiority of heteroscedastic extreme value model in destination and mode choice and superiority of multinomial logit in simultaneous destination- mode choice. It is also observed that by increasing the number of alternatives, the non- identical and independent assumption of error term does not increase the explanatory power of the model: travelers consider the non- observed factors as independent and with the same distribution. This paper is a step towards relaxing choice constraints, allowing researchers to consider heterogeneity of the alternatives.