Why Don’t More Commuters Consider Buses for Their Work Trip?—A Geographically Weighted Segmented Logistic Regression Modelling Approach

Empirical data from several cities worldwide show that low consideration rates of bus transit constitute an important reason for declining transit mode share. This trend raises a critical question: why do not more commuters consider buses for travel? The aim of this paper is to investigate the key deterrents to bus transit consideration among commuters. Consideration refers to the decision of an individual to include a given mode in his/her choice set. Along this line, this study focuses on two important directions. The first is to gain an understanding of how the key factors and their influence on consideration propensity differ across market segments based on personal vehicle availability and accessibility. The second is to quantify the spatial heterogeneity in the influence of those key factors across geographical locations. A new geographically weighted segmented logistic regression (GWSLR) model is proposed to address these research issues. The model is developed using household survey data from a sample of work-commuters from Chennai city, India. The findings reveal that neglecting either segmentation or spatial heterogeneity cannot only result in inaccurate model predictions and inferences, but also lead to sub-optimal policy interventions. Results show that factors influencing consideration differ significantly across segments based on captivity and accessibility. Choice users within walking distance to bus stops can benefit from direct bus service, whereas those beyond walking distance prioritise first-mile connectivity. Results also highlight significant variations in the influence of different factors across locations. Improved walkability in central business districts (CBD) could increase consideration, while non-CBD areas should focus on first-mile connectivity. Finally, this study illustrates that customised policy interventions for specific segments and locations can be more effective in enhancing bus consideration than segment-agnostic or geographically uniform policies.