Fall semester is still months away, but at campuses around the country, a make-believe semester has already begun.
Researchers are building computer models of varying complexity to predict how Covid-19 might spread on a college campus. The models provide an opportunity to test the effectiveness of such mitigation strategies as large-scale randomized testing, limits on class size, and face-mask requirements.
You’re just sort of rapidly quarantining 100 people a day. And that gets pretty big, pretty fast.
“Universities really need to be taking their planning and preparation very seriously,” said Alex Perkins, an infectious-disease expert at the University of Notre Dame who is building a model focused on his campus.
Colleges are under intense pressure to announce their plans for the fall semester — a high-stakes decision that pits the safety of remote learning against students’ clear desire for an in-person experience: Stick with online classes and risk a precipitous drop in enrollment? Or reopen the campus and face the dual challenge of fundamentally altering operations while trying to change the behavior of students who are hard-wired to be social?
At the Universities of Michigan, Richmond, California at San Diego, and elsewhere, researchers are modeling the potential impact of Covid-19 on their workplaces, sometimes independently of campus administrators.
Colleges that don’t have a customized model will be reopening with a sort of informal “mental model” in their heads, Perkins said. But a computerized one can be more precise.
“The value of the model is that it formalizes all those assumptions and tries to translate them into a quantitative answer,” he said. “In the absence of modeling, you’re just going to be dealing with a lot of sloppy assumptions.”
While useful, such models remain somewhat rare. Most colleges lack the in-house expertise to create a customized one. College presidents, meanwhile, aren’t making their development a top priority.
But almost every college can benefit from paying attention to the limited modeling that is taking place.
As colleges and universities have struggled to devise policies to respond to the quickly evolving situation, here are links to The Chronicle’s key coverage of how this worldwide health crisis is affecting campuses.
Fall reopening looms as the biggest challenge to colleges since ... well, ever. The logistics are daunting. The expenses are enormous. And no matter how carefully colleges prepare, some coronavirus cases will very likely happen anyway.
Computer models can offer a glimmer of hope. They also provide a statistical basis for deciding which preventive measures are most important to take.
At UC-San Diego, for example, the model predicts a dramatic reduction of coronavirus transmission when classes are capped at 50 students or fewer.
Class-size caps are “going to be a good thing,” said Natasha Martin, an associate professor and infectious-disease modeler who is part of the university’s Return to Learn pandemic-preparation program.
Martin normally creates models that analyze data for an entire country — in recent years, her primary research focused on modeling hepatitis C and HIV transmission among high-risk groups. Now she is focused on fine-tuning her coronavirus model so that it can account for even the smallest details of campus life, such as how often student social groups meet, and the higher risk of virus exposure among student-athletes.
“The deeper that I dig, the more nuanced the questions become,” Martin said.
The Danger of Large Classes
Not every computer model uses a real brick-and-mortar campus as its backdrop.
Philip T. Gressman, a professor of mathematics at the University of Pennsylvania, teamed up with Jennifer R. Peck, an assistant professor of economics at Swarthmore College, to create a hypothetical large research university. The institution, called “University” in the modeling code, enrolls 20,000 students and employs 2,500 instructors. Each run of the model has the students and faculty interact daily for 100 days.
The researchers posted their findings publicly this month. They found, like Martin did, that keeping large classes online is critical to keeping infections from spreading. Although any individual college might be a bit different from the generic “University,” this result came up so often in Gressman and Peck’s runs that the pair felt confident it would be true for most institutions.
“The largest classes are extremely dangerous,” Gressman said.
Meanwhile, a combination of random testing, contact-tracing, quarantining, and requiring masks helps keep things contained.
Gressman and Peck did find one big problem: Covid-19 tests can sometimes create false positives. Each positive result can lead to the quarantining of 10 to 20 students. The researchers warn that some coronavirus tests have been shown to have a false-positive rate higher than 1 percent, “which could lead to the unnecessary quarantining of literally thousands of students over the course of a semester.”
Even if testing-accuracy issues are managed effectively, student quarantines are going to be a reality on college campuses in the fall.
“If it’s not done carefully, it could cause a crisis, in the sense that way too many students will be quarantined,” Gressman told The Chronicle in an interview.
Peck, his co-author, chimed in: “We were surprised … how quickly the quarantine population gets out of control, even with what seems like a really low false-positive rate.”
“You’re just sort of rapidly quarantining 100 people a day,” she said. “And that gets pretty big, pretty fast.”
In their paper, the two researchers recommend that “all instructors need to prepare for extended student absences due to quarantine. In our best-case scenario, classes spend an average of 1-2 weeks of the semester with at least 10% of the students absent due to quarantine.”
Even the best-designed models aren’t perfect forecasts, and their accuracy is determined by the depth and reliability of data. With Covid-19, those data have some serious limitations, because so much about the virus — such as its rate of transmission among the college-aged population — is not yet fully known.
Students tend to have a larger-than-average social network, and they also interact with a different group of classmates in each course. Social distancing can put a dent in those challenges, but it doesn’t erase them.
The Late-Fall Threat
In one recent study, the sociologists Kim A. Weeden and Benjamin Cornwell, both of Cornell University, found that most students there are connected by two steps — they don’t share a class, but they are each in a class with the same third party. Ninety-two percent of students are connected by three steps, and 97 percent, within four steps. And that’s considering only classes.
In the fall, cooler weather creates the risk of a second wave of coronavirus infections. And campus protests could erupt over the issue of police brutality — or because of the tense run-up to the presidential election. Those protests could fuel even higher infection rates.
“If the whole nation is boiling over, then I don’t think colleges are going to be cool as a cucumber,” said Andrew Noymer, an associate professor of public health at the University of California at Irvine.
Almost two-thirds of the nearly 1,000 colleges tracked by The Chronicle are planning for in-person instruction in the fall.
Eastern Washington University is one of the colleges bucking that trend. The institution announced last month that it would be an “online first” campus in the fall, although it will still hold a small number of classes in areas where physical contact is required, such as physical therapy or dental hygiene.
David May, the provost, did not have the luxury of customized models when making his decision. But he did have public-health experts on his faculty, and he remembers one suggestion they gave him:
Make a decision, ahead of time, for what the key milestone will be — the moment that signifies in-person learning has become too dangerous, and the college must fully retreat back to online.
Is it the first coronavirus case? The tenth? Or maybe the first coronavirus-related death? Perhaps the fifth?
“I don’t have an answer that feels acceptable to me,” May said.
But one thing is certain, he said: Faculty have been appreciative of the decision to remain online.