Opportunities and potential bias in new transportation data

By Brian Lutenegger
A new issue brief from the Center for American Progress examines congestion on roadways in the United States and considers the potential and pitfalls of new data sources, such as those provided by private ride hailing companies including Uber and others. Although cities are eager to access these private sources of data, the report warns that planners should be careful of relying too heavily on these sources.
The issue brief first reviews congestion’s environmental impacts. Idling in traffic wastes time and fuel—$1000 or more per driver each year. It can also lead to human health impacts, and nonwhite communities tend to experience greater impacts from environmental pollution and other contamination. Traffic congestion also contributes to air pollution and climate change; the U.S. transportation sector is the largest contributor of greenhouse gases that lead to climate change.
Transportation planners have traditionally relied on publicly available datasets to forecast where and when people will travel in the coming decades as well as how they will get there. These data sources include traffic data, transit ridership, census data, and household travel surveys, the last of which can be expensive if done independently from the American Community Survey or if more detailed data is needed. However, recently private companies—including the major ride hailing companies like Uber and Lyft as well as several mapping companies such as TomTom and Waze—have offered the possibility of lower-cost data that is widespread and immediately accessible. These new data sources allow planners to reduce the lag time to update traffic models or other tools.
There are two major issues with this private data that warrants its more careful and deliberate management:
1. The data is vehicle-based, and by planning based on this data, agencies could help encourage driving to remain the primary mode of transportation in the United States into the future. In its current form, most of this existing private data fails to take into account other trips taken by walking, biking, or public transit.
In fact, a University of California Davis study released last fall showed ride hailing services are contributing to congestion overall. The study found that while ride hailing services may reduce the miles that riders drive themselves, they increase total vehicle miles traveled in cities. About half of trips taken using ride hailing service would either have not occurred or would have been taken by walking, biking, or public transit instead.
2. This leads to a focus on single occupancy vehicle congestion rather than on building and improving a multimodal transportation system that better serves not only drivers, but also pedestrians, bicyclists, and transit riders. The ready availability of single occupancy vehicle data should not serve as the only guide for transportation planning and infrastructure decision-making.
The data may have equity and demographic biases. Ride hailing drivers may intentionally choose not to drive in certain neighborhoods they perceive as dangerous or to discriminate against certain passengers. Further, demand for ride hailing services may be lower in lower-income areas and fewer residents may own the smartphones required to access them.
The impacts of ride hailing services on transit are also a concern. The same University of California Davis study found that ride hailing service users are less likely to take public transit other than commuter rail (which actually showed a small 3 percent increase among ride hailing service users). Ride hailing service users reported a 6 percent reduction in bus trips and 3 percent reduction in light rail trips.
In San Francisco, one study found that the majority of ride hailing trips are in a part of the city that is the most congested but also the part of the city best served by transit, bike lanes, and walkable streets. In order to counter the adverse effects of ride hailing on other modes, some cities, such as Chicago, have responded by instituting per-ride fees on ride hailing trips to raise funding to improve public transit.
Ride hailing companies, for their part, continue to be protective of their data, and negotiations with public agencies will be required to create data sharing agreements. Cities are a long way from the widespread usage of private data, and the data itself is not without its own biases.
Brian Lutenegger is a Program Associate at Smart Growth America.

Opportunities and potential bias in new transportation data

By Brian Lutenegger
A new issue brief from the Center for American Progress examines congestion on roadways in the United States and considers the potential and pitfalls of new data sources, such as those provided by private ride hailing companies including Uber and others. Although cities are eager to access these private sources of data, the report warns that planners should be careful of relying too heavily on these sources.
The issue brief first reviews congestion’s environmental impacts. Idling in traffic wastes time and fuel—$1000 or more per driver each year. It can also lead to human health impacts, and nonwhite communities tend to experience greater impacts from environmental pollution and other contamination. Traffic congestion also contributes to air pollution and climate change; the U.S. transportation sector is the largest contributor of greenhouse gases that lead to climate change.
Transportation planners have traditionally relied on publicly available datasets to forecast where and when people will travel in the coming decades as well as how they will get there. These data sources include traffic data, transit ridership, census data, and household travel surveys, the last of which can be expensive if done independently from the American Community Survey or if more detailed data is needed. However, recently private companies—including the major ride hailing companies like Uber and Lyft as well as several mapping companies such as TomTom and Waze—have offered the possibility of lower-cost data that is widespread and immediately accessible. These new data sources allow planners to reduce the lag time to update traffic models or other tools.
There are two major issues with this private data that warrants its more careful and deliberate management:
1. The data is vehicle-based, and by planning based on this data, agencies could help encourage driving to remain the primary mode of transportation in the United States into the future. In its current form, most of this existing private data fails to take into account other trips taken by walking, biking, or public transit.
In fact, a University of California Davis study released last fall showed ride hailing services are contributing to congestion overall. The study found that while ride hailing services may reduce the miles that riders drive themselves, they increase total vehicle miles traveled in cities. About half of trips taken using ride hailing service would either have not occurred or would have been taken by walking, biking, or public transit instead.
2. This leads to a focus on single occupancy vehicle congestion rather than on building and improving a multimodal transportation system that better serves not only drivers, but also pedestrians, bicyclists, and transit riders. The ready availability of single occupancy vehicle data should not serve as the only guide for transportation planning and infrastructure decision-making.
The data may have equity and demographic biases. Ride hailing drivers may intentionally choose not to drive in certain neighborhoods they perceive as dangerous or to discriminate against certain passengers. Further, demand for ride hailing services may be lower in lower-income areas and fewer residents may own the smartphones required to access them.
The impacts of ride hailing services on transit are also a concern. The same University of California Davis study found that ride hailing service users are less likely to take public transit other than commuter rail (which actually showed a small 3 percent increase among ride hailing service users). Ride hailing service users reported a 6 percent reduction in bus trips and 3 percent reduction in light rail trips.
In San Francisco, one study found that the majority of ride hailing trips are in a part of the city that is the most congested but also the part of the city best served by transit, bike lanes, and walkable streets. In order to counter the adverse effects of ride hailing on other modes, some cities, such as Chicago, have responded by instituting per-ride fees on ride hailing trips to raise funding to improve public transit.
Ride hailing companies, for their part, continue to be protective of their data, and negotiations with public agencies will be required to create data sharing agreements. Cities are a long way from the widespread usage of private data, and the data itself is not without its own biases.
Brian Lutenegger is a Program Associate at Smart Growth America.