It is a firm principle in my household that we will not, under almost any circumstance, get in the car after sundown on Friday or Saturday night. We won't pick you up at the airport, or drive to dinner at your house. We won't just run out to the grocery store, or partake of social events unreachable by foot or bike, or a short Uber.

We live off H Street in Washington with its bars and restaurants and performing arts, and if we drive away in the evening, when we get back there will simply be nowhere to park. We would behave, no doubt, a lot differently if parking were not an issue. We would probably take more trips.

Scale up this logic, and it's reasonable to think that parking on a much larger scale induces more driving across cities. But this is an incredibly hard thing to prove: When cities pave more parking lots, does it make people drive more? When you're sitting in traffic hemmed in by other cars, is easy parking in part to blame?

There's a lot of evidence that the two go hand in hand. Past studies have found that parking availability at home is strongly associated with car ownership and use. And more parking at the office is correlated with more employees driving to work alone. Commuters who work in Manhattan, for instance, are also more likely to drive in when they have parking to return to at night.

It's a provocative argument, though, that parking causes driving, and if this were true, a lot of city policies would look sort of backwards. When cities think they're merely accommodating all the driving we do — by, for starters, requiring apartments and businesses to build parking lots — they're actually encouraging that driving in the first place.

"Is there a reason parking could affect driving?" asks Chris McCahill, a senior associate at the State Smart Transportation Initiative in Madison, Wis. He was presenting new research on the question this week at the Transportation Research Board annual meeting in Washington. "On a city-wide scale it does make sense that the overall cost and convenience of parking in that place affects driving habits, as anyone who’s lived in a parking-restricted place knows."

Now, McCahill and three researchers at the University of Connecticut, Norman Garrick, Carol Atkinson-Palombo and Adam Polinski, think they've found solid evidence that parking is a "likely cause" of increased driving. Their case is the strongest yet.

It's based on historic data in nine mid-sized American cities going back to 1960, including parking counts painstakingly tallied in each city using archival aerial photos. The cities, roughly equal in size but with varying auto use, include Albany, N.Y.; Berkeley, Calif.; Cambridge, Ma.; Hartford, Conn.; and Silver Spring.

The researchers found, to begin with, that as these cities added more parking over the years, the share of commuters who drove to work increased. In this chart, as a city goes from having about 20 parking spaces to 50 spaces per 100 people, the share of commuters driving rises from 60 percent to 83 percent:

Now that's just a correlation. To go a step further, the researchers borrow from a criteria in epidemiology used to establish more causal links between, say, smoking and cancer. Parking is not that theoretically different. Does a change in the environment (more parking supply) influence the frequency of an undesirable event (more driving)? As the "dose" of parking goes up, does the likelihood of driving, too?

Epidemiologists would note that the relationship between these two factors is strong (as the above chart demonstrates) and consistent (it recurs in a lot of different cities and at different moments in time). Parking also emerges as a potential cause when there are no other clear explanations for an increase in driving.

In one study led by the University of Pennsylvania's Rachel Weinberger that the authors cite, commuters in the Jackson Heights neighborhood of Queens were more likely to drive to work in the central business district than commuters in Brooklyn's Park Slope. Income levels, car ownership rates, commuting times and transit access would suggest the opposite. But there was another key difference between these two neighborhoods: Commuters in Jackson Heights had a lot more off-street parking to return to when they got home at night.

Epidemiologists would also ask about the sequence of events. A treatment (smoking) must come before an outcome (cancer) and not the other way around. And so we'd expect that more parking would predict more driving, to a stronger degree than driving predicts parking. The researchers find that here as well: Cities where parking increased a lot between 1960 and 1980 saw much larger increases in driving in the following two decades: