Are College Students Killing Townies?

Are college students killing townies? It seems ridiculous to ask, but that narrative has started to emerge. After schools opened in September, The Wall Street Journal rushed to publicize preprint studies saying that in-person college was fueling a rise in cases. Other articles followed suit, suggesting that campuses are major superspreaders. In The New York Times’s recent telling, college campuses through the early fall festered with the virus, which then spread out into local communities and caused a spike in deaths.

Selfish, hard-partying students make an easy villain, but the evidence linking colleges to coronavirus deaths is not as ironclad as these media outlets suggest.

Let’s focus on The New York Times story from last week. It offers two types of evidence. One is a set of anecdotes describing “signs of a connection” between outbreaks at colleges and deaths in communities. These anecdotes show that genetic strains of Covid-19 found in college students also show up in nearby nursing homes, and are accompanied by quotes from public-health officials who say that they were hit by a wave of cases just after classes began. In a sense — but only a minimal sense — these arguments are true. Colleges aren’t fortresses. Students work, they live with families, they shop for groceries. It would be stunning if there were no cases of the virus moving from campus to nursing homes. (Or vice versa, for that matter.)

The real question is whether colleges have posed an outsize risk to their communities. The central part of the Times’s evidence here is a piece of data journalism, a chart purporting to show a steep rise in cases and deaths in 203 “college counties” starting in August and September. The expectation for the chart is clear: The article asserts that “deaths in communities that are home to colleges have risen faster than the rest of the nation.” And yet the chart itself, given the expectations readers bring to it, is misleading.

As the Times data show, college counties have almost precisely the same coronavirus rate as the rest of the country. Whatever sweeping irresponsibility you think college students or administrators have shown, it needs to be tempered with the knowledge that college towns are no more dangerous, today, than the average city or town in America. (Which is to say: quite dangerous indeed.) What’s unusual about college counties is not how they look today, but how low their rates were compared with the rest of country in the spring and summer.

Still, maybe there’s some blame to give colleges for not staying closed? Maybe college counties — unlike meatpacking counties, or prison counties, or international elite ski counties — saw low rates because their major institutions shut down effectively in March, and thus regressing to the mean was always a risk, with or without colleges’ reopening. While the Times allows that the college-Covid relationship may be “indirect and difficult to document,” their combination of steeply rising rates and tragic anecdotes certainly implies that the reopening — and even just the routine activities of off-campus students — bears blame for the rise in college cases after September.

But even this fuzzy implication is hard to substantiate. The major reason college counties saw low coronavirus counts in the spring and summer relative to the rest of the country has little to do with policy and much more to do with geography. Before August 15, the country’s major outbreaks were first in the New York-New Jersey area, and then in the Sun Belt states of Arizona, Florida, and Texas. The Times defines a college county as having at least 10 percent of its population made up of college students (including grad students). That definition happens to exclude most of these early-outbreak states. There’s no mystery about why college towns did better in this period; it’s because New York is not a college town, because Arizona State University’s huge student population is in the even more enormous Maricopa County, and because Austin, Tex., has grown so quickly that it now has only a 9-percent-student population. Conversely, as the virus spread this fall, it did so especially in the states of the upper Midwest, where land-grant colleges dominate their neighborhoods.

Another complicating factor: The Times’s longstanding and idiosyncratic view of college is firmly rooted in a traditional idea of long, shining lines of station wagons arriving on campus on a single day in September. Most college students stay close to home, and about half of undergraduates attend either a two-year institution or a four-year institution part time.

The real question is whether, as the virus spread through the North this fall, there has been an outsize aggregate impact in college towns. As best I can tell, the evidence the Times is offering doesn’t bear that out. Take the article’s opening story of Phyllis Baukol, who died in October in a nursing home not far from the University of North Dakota. Yes, student-heavy Grand Forks County has gotten much more dangerous since August compared with the national average. But so has the rest of the state.

There are two other counties in North Dakota of about the same size, but with much smaller college populations: Burleigh (home of Bismarck, the state capital) and Ward (home of the city of Minot and an Air Force base). Both are doing much worse than the college county: Burleigh has seen 102 deaths since September 15; Ward has seen 155; Grand Forks has seen just 43. (This data is drawn from the Center for Systems Science and Engineering (CSSE) at the Johns Hopkins University.)

Using the American Community Survey, I’ve recreated the Times list of 203 counties with a college-student population of over 10 percent. (If you wish to check it out for yourself, the code for this article is available on GitHub.) I’ve then automatically paired each “college county” with the county in the same state with the closest population that doesn’t qualify as a “college county.” Another example that comes up in the Times article is Ingham County, Mich., home to Michigan State University. College students make up more than 18 percent of its population, and like Grand Forks, it has seen a surge in cases this fall. But compare it with similarly sized Ottawa County, Mich. (9-percent college students), and a different story emerges: Ottawa County has seen almost twice as many cases and twice as many deaths as MSU’s county has since the beginning of October.

Here are aggregate charts with each of the 203 college counties paired with another in the same state. (There are a few complications in an automatic pairing like this: If a non-college county would appear twice, I went to the next-closest county in the state, and since Washington, D.C., is a “college county,” I treated it as part of Maryland, which led the algorithm to pair it with Baltimore.)

This data tells a different story. The spike in cases in September, as students returned to campus, is evident. This is most likely because colleges engaged in a huge program of testing unlike almost any other industry — something that they should be proud of. But since the end of October, the number of confirmed cases is less in college towns than in non-college towns. And while deaths have risen in the college counties, the rates in other counties in the same states started from almost as low and have consistently remained higher than their student-heavy equivalents.

Looking at this data, an opposing narrative could be told. Perhaps aggressive testing in September did uncover Covid-19 on college campuses, but despite early fears (promoted in no small part by the Times itself), the worst has not come to happen, and colleges have played merely an incidental role in the pandemic. Anecdotes could also be found to support this theory. The University of Illinois at Urbana-Champaign, for example, implemented a vast testing campaign in the early fall, was mocked for its naïve assumptions about student parties, then went on to administer over a million tests, resulting in a positivity rate well below 0.5 percent.

None of this is the last word in the great campus Covid-19 debate of 2020. Legions of epidemiologists, public-health experts, and social scientists are plunging through the data right now. Are demographic, economic, or political differences clouding the “college county” data? (There are probably higher percentages of elderly in non-college counties than in them.) What of regional cultural differences, mask-wearing tendencies, and community-mobility data? Whatever story emerges will be complicated. How are we to factor in counterfactuals, like a world in which college students live with their high-school siblings and middle-age parents, visit their grandparents on the weekends, and sneak out to see their friends at night?

Despite all the unknowns, the narrative about college and the pandemic is hardening. Since the Times launched a college-case tracker with much fanfare in August, it was only a matter of time before they followed it up with stories on how colleges were irresponsibly seeding outbreaks. It’s no surprise that they were able to find locals who agreed that the loss of their loved ones might partially rest on the maskless Frisbee games they saw outside. Town-gown tensions are as old as the university.

But underlying this framing is the idea that students are reckless, and that higher education falls firmly on the inessential side of our national accounting. One of the more depressing second-order effects of the pandemic has been the way that people have successively blamed activities that they view as decadent. This has led to a perfidious tendency to choose a villain to blame, without which we wouldn’t all be in this mess.

Pick your villain depending on what you don’t mind losing: the Chinese wet markets filled with bats and pangolins, the unnaturally crowded subways of New York City, Trump voters and their maskless rallies, restaurants, bars, high schools, beaches. If you think college kids are spoiled brats, or college administrators are malevolent bean-counters, why not throw them on the offering as well?

None of this addresses the underlying problem. The nature of an uncontrolled national pandemic is that everyone has been spreading the virus to everyone. Singling out colleges ignores the remarkable measures they have taken compared with other industries. Blaming students reinforces an idea that college is a decadent luxury, one we might be better off without. The blame game exacerbates tensions in divided communities without making anyone safer. The reality is that there’s no easy villain. We’re all blundering along in the same mess together. And while we look forward to rebuilding after a vaccine, we should make sure we’re not unnecessarily undermining institutions we want to keep strong.