Uber is once again dipping its toe into the world of innovative social science. Folks may recall that earlier this year it commissioned Alan Krueger, one of the country’s leading labor economists and formerly President Obama’s chief economist, to do an analysis of Uber drivers’ pay. While Uber shared data with Kreuger on drivers’ gross receipts, it did not share data on miles driven. This meant that Krueger was left comparing the gross receipts of its drivers with the net income of cab drivers in the incumbent taxi industry. The gross receipts do not deduct costs borne by the driver, such as gas, depreciation on the car, and insurance.

Not surprisingly, the gross receipts of Uber drivers were higher than the net income of drivers for the incumbent tax industry. It’s not clear if this comparison would hold up if Krueger had done an apples to apples comparison where he deducted expenses for Uber drivers, but he couldn’t do this, since Uber didn’t give him the miles data.

In keeping with this approach to social science Uber has commissioned a new study that purports to show that it provides better service to minorities than the incumbent taxi industry. The test was to have someone order an Uber car in a heavily minority community on their smartphone, and compare the time it takes to get their pickup with the time it takes someone calling for a taxi from an incumbent company. Uber found that its service was markedly faster than the service of the incumbent industry.

Before anyone celebrates over this finding that Uber has eliminated or at least reduced discrimination in taxi service, a bit of thinking is required. To order an Uber car it is necessary to have both a smart phone and a credit card. A substantial portion of the low income and minority populations lack one or the other.

The Uber study effectively asked the question of whether Uber provides better service to a screened portion of the minority community, using a screening mechanism that is likely to weed out the poorer portion of this community. Furthermore, Uber knew of this screening, since it is how their cars are summoned. The incumbent taxi companies in its study did not know of the screening.

If we think that discrimination against minorities is a mixture of race, ethnicity, and class, the Uber study effectively used a screening mechanism that largely eliminated the class aspect of the matter, at least for the Uber drivers. In this context, the result is not very surprising.

CEPR is proposing that Uber finance a study where we compare the amount of time it takes people in minority communities to get an Uber car or a taxi ordered from an incumbent service, where the passenger does not have a credit card and orders over the phone. It will be interesting to see what we find.