Data aggregation impacts on built environment-mode share models around public transit stations

This study examines how data aggregation influences the relationship between the built environment (BE) and mode share around 2,794 rail and BRT stations in the United States, using both inferential and machine learning methods. The results indicate that data aggregation impacts the outcomes of BE-mode share models, regardless of the data analysis approach. Models using network buffers are less affected by data aggregation compared to those using circular buffers, Thiessen polygons, or administrative boundaries (block groups). In addition, the optimal buffer sizes for capturing BE effects and minimizing sensitivity to data aggregation for active and public transit modes are 800 meters for BRT stations and 1000 meters for rail stations, while 1200 meters is effective for private vehicle mode share at both rail and BRT stations. Furthermore, key BE features in commuting mode share models—such as employment density, jobs per household, intersection density, residential density, distance from the central business district, job accessibility (active), and regional population density—remain robust against data aggregation. We recommend that urban and transportation planners account for aggregation biases and apply multiple methods when evaluating BE's impact on mode share around public transit stations to inform more effective policy recommendations.