Incentive bias identified in ridesharing platforms

The research, conducted by Jorge Mejia of Indiana University and Chris Parker of American University, found that in addition to demonstrating that racial biases persist, similar phenomena were also documented against people who show support for the LGBTQ+ community.

For the study, Mejia and Parker analysed data from a major rideshare application in Washington DC between early October to mid-November 2018. The experiment manipulated riders' names and profile pictures to observe drivers’ behaviour patterns in accepting and cancelling rides.

To illustrate support for LGBTQ+ rights, the researchers used a rainbow profile picture filter. Furthermore, times or ride requests varied to determine how peak and non-peak price periods impact bias.

“We found under-represented minorities are more than twice as likely to have a ride cancelled than Caucasians - that’s about 3 per cent versus 8 per cent,” Mejia said. “Along with racial bias, LGBTQ+ biases are persistent, while there is no evidence of gender bias.”

The researchers also found that peak timing has a moderating effect, with lower cancellation rates for minority riders, although the timing didn't appear to change the bias for riders that signal support for the LGBTQ+ community.

“Data-driven solutions may exist wherein rider characteristics are captured when a driver cancels and the platform penalises the driver for the biased behaviour,” Mejia explained, adding that one possible way to “punish drivers” is to move them down the priority list when they exhibit biased cancellation behaviour, so they then have fewer ride requests.

“Alternatively, less-punitive measures may provide ‘badges’ for drivers that exhibit especially low cancellation rates for minority riders,” Mejia concluded.