Exploring Variation in Built Environment Predictors of Ridership by Transit Mode (Paper No: 20-01322)
Supplementary presentation to support poster presented at Transportation Research Board Annual Meeting, 2020.
1695, Current Topics in Public Transportation
Session Location: Hall A / Convention Center
Session Time: Wednesday, Jan 15, 2020 8:00AM 9:45AM
Paper: 20-01322 - Exploring Variation in Built Environment Predictors of Ridership by Transit Mode
Many studies have identified links between the built environment and transit use. However, little is known about whether the built environment predictors of bus, train, tram and other transit services are different. Studies to date typically analyze modes in aggregate, by combining bus, train, and tram; or analyze each mode separately. Findings from these studies demonstrate the built environment attributes that are relevant for apportioning trips to transit. However, they do not differentiate demand for competing modes. This study aims to investigate if built environment impacts on transit ridership vary according to mode, by analyzing two types of co-located (matched) transit modes (train-bus and tram-bus) in Melbourne. Multivariate multiple linear regression models were estimated to identify the relationships between different indicators of the built environment with patronage of each mode.
This research indicates built environment impacts on ridership vary in type and relative importance according to mode. The strongest land use predictor of train ridership was proximity to the CBD, while bus ridership was most closely associated with the presence of Activity Centres in the catchment. Tram and bus ridership shared no built environment predictors, with land use diversity the strongest predictor for tram use, compared to intersection density for bus. These differences provide evidence that built environment impacts on transit cannot be generalized for all modes. The study’s findings suggest that strategies to encourage transit use could be made more effective if they are differentiated by mode.