Understanding mode choice decisions for shopping mall trips in metro cities of developing countries

Abstract

The traditional travel demand model for an urban region generally considers work trips as the major constituent which subsequently attracts the maximum focus on mode choice analysis. However, with the increase in traffic-related problems coupled with the evolving lifestyle of people in general, it has become imperative to focus on trips intended for other purposes as well. In most of the urban scenarios, shopping trips constitute the second most frequented trips after work trips. The present study aims to develop a mode choice model exclusively for shopping malls in Mumbai city and determine the factors which influence a trip maker’s behavior of mode choice in the context of a developing country. Choice of mode for shopping mall trips is important in relation to the impact of the vehicular traffic generated from the shopping malls on the adjacent road networks as well as in the planning of the parking space allocations at the malls along with providing critical insights on the influential factors to the transport planners for policy recommendation. Revealed Preference (RP) survey technique is used to collect 650 samples of data regarding trip makers’ choice of mode to shopping malls by interviewing individuals at various shopping malls across Mumbai. A Multinomial Logit (MNL) model is initially developed, followed by a Nested Logit (NL) model by grouping the private modes in a single nest. The significance level of the variables used in both the models are mostly consistent. Both the models reveal that travel time has a significant role in the mode choice behavior. While walking time and access time notably affect the utility of public transport mode, factors like number of accompanying persons and driving license possession also influence the private modes significantly. Socio-demographic characteristics like age, gender and occupation are also found to be critical in the mode choice behavior for shopping mall trips. A comparative analysis of the MNL and NL models reveals that the NL model outperforms the MNL model both in terms of goodness of fit and prediction success rate. However, both the models (MNL with a predictive success rate of over 72% and NL with a predictive success rate of about 77%) fairly captures the factors affecting the mode choice behavior for trips to shopping malls. The study can help policy makers to understand the factors affecting the choice of mode for such trips and can aid in planning strategies to minimize traffic congestion resulting from the increasing number of shopping malls.