This chapter describes three paradigmatic changes in quantitative methods for transport planning: big data, causality outside of laboratory contexts, and machine learning approaches. In contrast with other scholars, I argue that big data is not all data—digital transport traces track technological objects rather than people and their transport decisions. I show this through a case of understanding micromobility, where e-scooter trips may be fully tracked under a corporate subscription model. However, many owned vehicles, including bicycles, provide no digital trace whatsoever. Therefore, reliance on data from shared mobility services may mis-represent the broader population of users. Planners relying on these forms of potentially biased ‘smart city’ information could inadvertently prioritize users of specific mobility services, creating a negative feedback loop. As an example, a planner identifying unsafe intersections by calculating the number of crashes per bike share trip—rather than performing bicycle traffic counts throughout a city—could prioritize areas in the central city and along recreational corridors frequented by tourists over the city's everyday bicyclists. Tracking rental trips is not the same as all trips for any travel mode. Understanding the prospects and limitations of quantitative analysis in transport is critical whether working on an individual project or an entire forest of transport issues.

This understanding argues in favor of new methods of data fusion—combinations of representative surveys and other methods with big data to mitigate bias while maximizing spatial and temporal accuracy.