Two weeks ago, we looked at whether taxi alternatives like Uber and Lyft may be reducing drunken driving, a plausible scenario given that the app is particularly popular among young-adult users most likely to go bar-hopping. We focused on DUI arrests in a couple of admittedly crude analyses. But one thing we — and more sophisticated researchers — don't have is usage data from Uber itself.

How are people using the app? Is there reason to believe that they often use it in conjunction with drinking, or that they're more likely to call a car from a neighborhood full of bars? Uber itself collects all kinds of geographic and temporal data in the process of tracking rides that might address some of this.

The company (as well as its competitor Lyft) has an interest in answering this question. If these apps contribute to public safety on the drunk-driving front, that's one point in their favor against other criticisms that they're unsafe because their drivers and vehicles don't face the same regulations as traditional taxis.

And so Uber is offering a veiled look at some of its data in Pennsylvania today (that's the state with the fourth-highest DUI fatalities, it points out). Below, the company has plotted the frequency of trip requests averaged over the past six months of UberX and UberBlack usage in Philadelphia and Pittsburgh (with a y-axis that's no doubt intentionally unlabeled):

Zooming in, here's Saturday night in Pittsburgh:

These patterns are pretty consistent with the idea that consumers are disproportionately using the app to go out at night — and, even more specifically, to get home afterward. Just eyeballing it here, Uber's peak traffic in Philadelphia on Saturday night looks to be about five times as high as its peak traffic on Monday night. People in Pittsburgh, on the other hand, don't seem to want to go anywhere until they want to go out on the weekend.

We still don't know the counterfactual here: Before Uber entered these markets, how were these same consumers getting around? In the absence of Uber, would they drive their own cars home at 2 a.m.? Or take taxis? Has Uber turned potential drunken drivers into passengers? Or DDs into drinkers?

Uber tells me that these usage patterns vary across its markets. Still, it's a rare brand laying claim to the argument that its consumers may be drunk.