A study of race and Uber found the company’s drivers in Boston canceled pickups of riders with African-American-sounding names at double the rate of those with names more likely to belong to a white person.
The study, published by the National Bureau of Economic Research, tracked 1,500 rides on Uber and competitors such as Lyft in Boston and Seattle. The goal was to determine whether customers with African-American-sounding names experienced discrimination in the form of longer wait times for rides and more cancellations of their requests.
According to the study, Boston showed “significant evidence of racial discrimination” through canceled rides. Research associates in Boston created two separate profiles to order Uber rides — one with a name the researchers determined would sound like that of an African-American customer, the other with a name that seemed white.
Passengers in Boston with seemingly African-American names had 10.1 percent of their rides canceled after being initially accepted, compared to 4.9 percent for white names. The discrepancy was even greater for male African-American names than for female ones.
One of the authors, Stanford University professor Stephen Zoepf, said the results reflect more on the drivers who work for Uber in Boston than on the company itself. Drivers are considered subcontractors, not direct employees of Uber, though they must undergo a background check.
“This does not show that Uber is discriminatory, or Lyft,” Zoepf said. “This is not about the actions of the companies themselves. This is the action by individuals using their platform. I think this does say something about the cultures of the cities.”
Uber said in a statement that its service has brought improvements to low-income and minority neighborhoods in Boston, where taxis have historically been less likely to go. But “studies like this one are helpful in thinking about how we can do even more,” said Rachel Holt, head of North American operations.
The study dovetails with similar findings in the transportation industry, and in other fields. The taxi industry in many cities has been criticized repeatedly over the years for providing less service to black communities.
Indeed, state Senator Linda Dorcena Forry of Dorchester noted that during the drafting of new regulations for the ride-hailing industry this year, Uber specifically pointed to its record in minority neighborhoods.
“It’s fascinating, because Uber during the whole argument about regulation said, ‘Listen, we’re serving communities of color, and traditional taxis are not,’ ” Dorcena Forry said Monday. “And then we see this.”
Other research has shown companies are less likely to call back a job applicant with an African-American name than one with a white-sounding name.
More recently, researchers have seen the same dynamic at another powerhouse New Economy company, Airbnb. Prospective lodgers with African-American names were less likely to be accepted by homeowners. Airbnb acknowledges the problem, issuing a report on the subject and asking users to pledge not to discriminate.
Ben Edelman, a Harvard professor and coauthor of the 2015 study of Airbnb, said the companies need to take responsibility for any racial bias of providers who use their networks — whether drivers or home renters. Uber and Airbnb are vulnerable to discrimination, Edelman said, because they give “an incredible amount of discretion” to those who provide services through the platforms.
“Both Uber and Airbnb have decided to structure their affairs in a way that entails less oversight over the people actually doing the work,” Edelman said. “It leads everybody to their worst instincts.”
The researchers did not find the same cancellation rates with Lyft, however. They theorize that’s because of the different ways Lyft and Uber drivers learn about their riders. Uber drivers don’t learn the names of their passengers until after they accept the fare, while Lyft drivers receive their names and photos before accepting the rides.
As a result, Lyft drivers may simply ignore requests from African-American names, so that “any discrimination occurs prior to accepting the initial request,” the study says.
Lyft spokesman Adrian Durbin said the company does not tolerate discrimination.
“Because of Lyft, people living in underserved areas — which taxis have historically neglected — are now able to access convenient, affordable rides,” he said.
Damon Cox, director of economic development and entrepreneurship at The Boston Foundation, said the study called attention to a bigger problem: the intrusion of bias and racism into digital systems that were supposed to be bastions of equality. He pointed to online shopping algorithms that offer products at different prices to different users, based on demographic assumptions.
“The Internet and technology were supposed to be this democratizing force,” Cox said. “But now we know it’s not a level playing field.”
Earlier this year, the online retailing giant Amazon.com was sharply criticized for not offering same-day delivery in Roxbury when it rolled out the new service in the Boston area. The company relented a week later.
Uber itself is involved in other racial controversies. The company refuses to include a tipping function in its app, saying in April that unconscious racial biases would result in white drivers receiving more tips than black drivers.
In October, Boston labor attorney Shannon Liss-Riordan filed a federal discrimination complaint against Uber, alleging the rating system for drivers is racially discriminatory because it is based on the same unconscious biases of passengers.
The Uber-Lyft research was conducted over an eight-month period that ended in March. In Boston, the economists tested 451 rides with Uber and 460 with Lyft. In Seattle, they tested 208 rides with Uber, 222 with Lyft, and another 143 rides with another competitor, Flywheel. The study was led by researchers at MIT, the University of Washington, and Stanford University.
The tests in Boston and Seattle differed in some ways. For instance, researchers did not test for cancellation rates in Seattle. In Seattle, researchers found African-American Uber riders experienced longer wait times, compared to white riders, by about 30 percent.
But in Boston, the study uncovered another anomaly:
Female riders were more likely to be driven for longer periods of time than male riders on similar routes.
That finding comes amid continued concern about the safety of women using the ride-hailing technology, as Uber drivers have in several instances sexually assaulted female passengers in Boston and elsewhere. In some cases, the researchers say, the longer rides could be the result of drivers “flirting to a captive audience.”
In a statement Monday, Boston Mayor Martin J. Walsh said that all services in the city should be offered “fairly, with the utmost respect and free of any sort of discrimination.”
Dorcena Forry and Edelman called on Uber to use its technology to identify drivers susceptible to discrimination and educate them. Dorcena Forry added the company should consider removing violators from its network.
Zoepf and his colleagues also suggest another solution: giving Uber and Lyft riders and drivers a “unique passcode” to summon and confirm their rides, instead of using names.