Influence of a new rail transit line on travel behavior: Evidence from repeated cross-sectional surveys in Hong Kong

Document Type

Journal Article

Publication Date


Subject Area

place - asia, place - urban, mode - bus, mode - car, mode - rail, planning - methods, planning - surveys, ridership - behaviour, ridership - mode choice, land use - urban density, land use - impacts


Urban rail transit, Travel behavior, Two-dimensional propensity score matching, Repeated cross-sectional survey


It has been well established that the infrastructural development for rail transit stimulates rail transit use. However, there is little agreement about the source of the increased rail transit use. Using data from two repeated cross-sectional surveys, we examine changes in individual travel behavior resulting from the introduction of a rail transit line in Hong Kong. To address some methodological limitations inherent to repeated cross-sectional research design (e.g., selection bias and longitudinal incomparability), a two-dimensional propensity score matching method (2DPSM) is adopted to pair samples between the treatment and control groups in both cross-sectional and longitudinal dimensions. Paired t-tests are used to compare the longitudinal changes in travel behavior between the treatment and control groups of the matched samples. To get a more comprehensive understanding of the net treatment effects of the new rail line on travel behavior, we examined its impacts on both home-based trips (trips originating or terminating at home) and all trips for both treatment and control group. For home-based trips, the opening of the new rail line increased the rail mode share by 10.4%, and rail trip number by 0.126 (in terms of net effect, i.e., difference in the change between treatment and control group). It reduced the bus mode share by 17.1% and bus trip number by 0.208, showing a significant bus-to-rail modal shift. For all trips, the new rail line increased rail mode share by 9.5% and total trip number by 0.189. It also decreased bus mode share by 13.2%, and bus trip number by 0.191. Hence, the source of the increased rail transit use came from both the modal shift from bus and the increased travel demand induced by the new transit infrastructure. For both home-based trip and all trips, there was no significant influence on car use and total trip distance. Our findings provide new evidence that the development of rail transit in a high-density urban setting encourages a modal shift from bus to rail transit and stimulates flexible travel behaviors, but fails to control private vehicle use.


Permission to publish the abstract has been given by Elsevier, copyright remains with them.


Journal of Transport Geography home Page: