MODELING SPATIAL ACCESSIBILITY WITHIN DISCRETE CHOICE FRAMEWORK

Spatial accessibility is concerned with the opportunity that an individual at a given location possesses to participate in a particular activity or set of activities. The main objective of this paper is to highlight the shortcomings of traditional accessibility measures and provide some appropriate methodological suggestions for their improvement. Traditional measures derived from cumulative opportunities and gravity models focus on physical proximity leaving out individual and spatial attributes as potential explanatory variable. The improvement through random utility theory relies mainly on Multinomial logit models under Independently and Identically Distributed (IID) and individual response homogeneity assumptions that often do not hold in case of choices involving spatial units. In this paper we briefly present the process of relaxing MNL assumptions. Using the MNL, Nested logit (NL), and Mixed Multinomial logit (MMNL) models we derive related accessibility measures. The application of MNL, NL and MMNL on choice model of residential location underlines possible consequences of a misspecification of the distribution of the error term and that of model parameters. The results clearly suggest that a decision process can be corrupted, and therefore lead to erroneous policy measures because of model misspecification.