A new study suggests that drivers for the popular ridesharing platforms Uber and Lyft may be discriminating against riders of color. The study indicates that black men, or men with “black sounding names,” faced a higher rate of cancellations and longer wait times for rides. Additionally, the study suggests that female drivers were more likely to be taken on longer rides with “chatty drivers.” These are troubling allegations, if true. But do these claims hold up to scrutiny?
Full disclosure: I have driven part-time for Uber and Lyft in Portland, Oregon, since April, 2015. I have provided upwards of 4,300 rides, so I have personal experiences that inform my analysis of this study.
The study was conducted by researchers at MIT, Stanford, and the University of Washington for the National Bureau of Economic Research. The abstract states:
Passengers have faced a history of discrimination in transportation systems. Peer transportation companies such as Uber and Lyft present the opportunity to rectify long-standing discrimination or worsen it. We sent passengers in Seattle, WA and Boston, MA to hail nearly 1,500 rides on controlled routes and recorded key performance metrics. Results indicated a pattern of discrimination, which we observed in Seattle through longer waiting times for African American passengers—as much as a 35 percent increase. In Boston, we observed discrimination by Uber drivers via more frequent cancellations against passengers when they used African American-sounding names. Across all trips, the cancellation rate for African American sounding names was more than twice as frequent compared to white sounding names. Male passengers requesting a ride in low-density areas were more than three times as likely to have their trip canceled when they used a African American-sounding name than when they used a white-sounding name. We also find evidence that drivers took female passengers for longer, more expensive, rides in Boston.
As the author notes, discrimination in traditional taxi hailing settings is well established. Indeed, the very concept of ridesharing is designed to eliminate the “hailing a taxi in Harlem” problem. If drivers—who are independent contractors under IRS rules and individual business owners—are more likely to discriminate, that would indicate that stricter standards would be necessary for drivers who accept the Terms of Service agreements from Uber and Lyft.
Bloomberg provided a bit more detail on the study:
The study, conducted in Seattle and Boston, included almost 1,500 rides. Four black and four white research assistants—split evenly among men and women—ordered cars over six weeks in Seattle. All used their photos on the ride-sharing apps. A second test was held in Boston with riders “whose appearance allowed them to plausibly travel as a passenger of either race,” although they used either “African American sounding” or “white sounding” names, the researchers said. The study found that Uber drivers disproportionately canceled on riders with black-sounding names, even though the company penalizes drivers who cancel frequently.
The research also observed discrimination in the taxi industry—a well-known, decades-old issue. The paper doesn’t compare the rate of discrimination between traditional drivers for taxis or ride-hailing apps.
Lyft and Uber’s issues were slightly different. While researchers found that wait times were noticeably longer for black men on both services in Seattle, Lyft drivers didn’t cancel on black riders disproportionately. But the researchers said that because Lyft shows riders’ names and faces upfront, its drivers could simply screen out black passengers. Uber doesn’t show names until after the driver accepts the fare.
Other female riders reported ‘chatty’ drivers who drove extremely long routes, on some occasions, even driving through the same intersection multiple times. As a result, the additional travel that female riders are exposed to appears to be a combination of profiteering and flirting to a captive audience.
These results are troubling, but they don’t do a good job of defining how widespread the problem is. The study involved 8 researchers conducting a total of 1,500 rides over a period of several weeks. I’m no statistician, so I can’t tell you if that’s a statistically reliable sample, but it sure seems small. There are over 10,000 Uber drivers in Boston, and almost that many in Seattle. There are thousands of rides every day.
Clearly, if a problem exists, it should be rectified. More internal scrutiny by Uber and Lyft appears to be warranted to ensure that they can identify and correct drivers who may have a propensity for refusing rides based on discriminatory factors. The data certainly exists to make such an internal analysis.
As a driver, I can tell you that my experience does not match this study, though I clearly can’t speak for all drivers. I rarely cancel any ride once I’ve agreed to provide it. The vast majority of ride cancellations are due to rider no-show or confusion about the pick up location. Other factors include whether a rider has a low rider rating—which indicates that previous drivers have had an issue with that rider.
Yes, you read that right. Not only do the riders rate the drivers, but DRIVERS RATE RIDERS TOO. If you’ve blown chunks on a previous ride, or you were rude with the driver, or you were a drunken pain, it can affect your ability to get a driver who is willing to put up with you in the future.
The two-way rating system is designed by Uber and Lyft to be a motivation for drivers to provide excellent service—unlike traditional taxis. If you have an issue with a ride in a cab, there is typically precious little recourse. A driver with a very low rating will see an increased number of cancellations—and in extreme cases can face deactivation by the ridesharing platform. The two-way rating system, and inducements for drivers to maintain a high acceptance rate, also function as a backstop against the type of discrimination described in the study. It’s far from perfect, but it’s a strong motivation.
I’ve actually had conversations with riders about these issues from time to time. In one instance, I gave a ride to a freshman at Portland State University back to her home in North Portland. She was African American, and her home was in a historically black neighborhood. She was effusive in her praise of ridesharing as a means of overcoming the historical bias shown by taxi drivers who too often wouldn’t want to provide such a ride.
There are studies that back this up. One study in Seattle—conducted by one of the authors of today’s study—showed that riders in low income neighborhoods were served faster than those in affluent neighborhoods. Another report in New York City showed that low income and minority neighborhoods were served more than three times better by Uber and Lyft than by traditional taxis. Yet another study showed that it is cheaper and faster to order an Uber in low income neighborhoods of Los Angeles than to call a cab.
It is important to properly identify any problems of discrimination in Uber and Lyft and to root them out. It is also important to note that the very nature of ridesharing helps to eliminate discrimination, and that the services often far out perform their traditional taxicab rivals.