Several mode choice studies have investigated the effects of socio-demographics, vehicle ownership, LOS variables etc., but only a few studies deal with mode choice at the household-level of aggregation. Household-level mode choice analysis lends valuable insights into multiple mode choice, group decision making in a household, activity delegation and other inter-personal dependencies. Also, relatively few studies adequately address the unique features (socio-demographics, vehicle-type etc.) in developing countries, primarily due to data constraints. This paper investigates two mode-related dimensions at the household level, namely, mode choice and modal use intensity (as reflected by modal expenditures). These dimensions are analyzed jointly in the context of Chennai city in India, using a large disaggregate database consisting of over 2000 households. Specifically, the objectives of this study are to: (1) analyze mode choice decisions at the household level; (2) integrate the dimensions of mode choice and mode usage using a suitable model; and (3) analyze the effect of contextual factors (e.g. state-dependence, attitudinal indicators, quality of service variables) relevant to developing countries on the mode choice propensity and mode use intensity. Note that at the household level, the mode choice problem is a multiple discrete choice problem (multiple alternatives may be selected) in contrast to the singly discrete nature of the individual mode choice problem. Therefore, a Multiple Discrete-Continuous Extreme Value (MDCEV) model is formulated to integrate the two related dimensions based on a utility maximization framework. The proposed model provides a reasonably good fit with observed data and a coherent and computationally efficient framework to link the two related dimensions. The results reveal that several unique and context-specific features in developing countries affect household level mode choice significantly including mixed land-use, extended-family system, presence of non-worker, inter-personal interactions, use of two-wheelers and IPT, peer pressure etc. Further, the mode use intensity of alternative modes is influenced by prior mode choice decisions, inertia, and user’s perception of safety and congestion. The results have important planning and policy implications for transit improvement, and congestion mitigation strategies.
Samenvatting