This academic year will be one for the history books. College for the class of 2024 will be more social distancing than tailgating, more COVID testing than keg stands, and a world of unknowns in an unusual time.
Faced with existential financial losses, American colleges and universities face immense pressure to reopen quickly and safely. Closing in-person classes and activities has been costly: some schools estimate that the costs of reimbursing student fees and tuition, cutting athletic seasons short, and scrambling to develop the infrastructure for online education may cost upward of $100 million. And the impact will drag on, with some concerns that broader economic decline, joblessness, and uncertainty will lead students to forego pricier universities in favor of staying closer to home.
Schools that have reopened have faced immense criticism for their mistakes, and there have been mistakes. Faced with COVID outbreaks on campus, early re-openers like Notre Dame University and the University of North Carolina have moved classes back online. Professors and students have publically confronted school leadership at Pennsylvania State University and Georgia Institute of Technology for pursuing coercive reopening plans without proper public health consideration or buy-in from faculty and students. Others have called to "cancel college", citing the impossibility of convincing college age students to avoid socializing.
Many schools have developed COVID-19 dashboards, designed to keep students and their families abreast of any new diagnoses. We decided to take it one step further and evaluate the impact of re-opening on the neighboring communities. Are newly opened colleges changing the epidemic in their cities and states?
How many schools are reopening?
As of August 21, 2020, 42 four-year colleges have (or intend to) hold the fall semester fully in person, according to the College Crisis Initiative (C2i), a tracking project housed in Davidson College. An additional 416 intend to reopen "primarily in person," with the remainder offering courses either primarily online, fully online, or using some hybrid model.
The larger the university population is relative to its surrounding community, the greater the impact an on-campus outbreak of COVID-19 will be. With that in mind, we have limited our analysis to schools that have either opened completely in-person or where classes are primarily in-person and the school's total enrollment accounts for at least 10% of the surrounding county's population.
Per-capita new infections, before and after re-opening
Notably, we find that all of the colleges reopening in person were experiencing stable incidence rates at the time of reopening. Indeed, no school seemed to have made the decision to reopen against a background of rapidly changing COVID-19 infections.
However, we find that a dramatic increase in per-capita rates of COVID-19 diagnoses in the days following the start of classes. As of the end of August, significant growth in infections were observed in the counties housing the University of South Dakota, the University of Iowa, Mercer University, Iowa State University, Georgia College & State University, East Carolina University, and Georgia Southern University.
What about my school?
Only a minority of universities have planned to reopen fully in-person. However, many colleges pursuing a primarily online semester have nonetheless opened dormitories, creating a risk for outbreaks. To flag where COVID-19 diagnoses are quickly increasing, we analyzed per-capita diagnoses for all universities. Since university-specific data are not universally available, we used data for each school's county.
Since we are using case data from the surrounding county, outbreaks in schools where the student body makes up a big proportion of the county will be more visible in county-level data. In addition, not all students with COVID-19 get tested on campus, so we would miss anyone that travels home and is diagnosed there. However, this type of analysis can nonetheless quickly flag schools where the county is experiencing spikes in diagnoses.
New infections per 10,000, by school's county
Source: The New York Timesrepository of coronavirus cases and deaths in the U.S. Data as of October 26, 2020. Incident cases are displayed as a five-day moving average of new cases per 10,000 population. County population are from the U.S. Census Bureau (2019). University data are extracted from the U.S. Department of Education's College Scorecard data. Counties are assigned from Universities' ZIP codes, using crosswalks between ZIP code tabulation areas (ZCTA) and counties and ZIP code and ZCTAs. Negative values, that occur when a state or county corrects historical data or move cases from one county to another, are displayed as zero. Universities displayed are limited to primarily four-year colleges and excludes schools that primarily grant certificates and associates degrees. Circle size is proportional to the total undergraduate enrollment.
This analysis is limited by several factors. First, we do not have the reopening plans of all universities, so we have only focused on the small subset reporting reopening fully or primarily in-person. Second, we are showing the initial planned date of reopening as of August 30, 2020. Any school that later decides to shut down in-person learning, with the intention or reopening at a later date, would still be shown with their initial opening date. And finally, we are using county-level incidence data, making outbreaks in small schools in large counties less likely to be detected.
Nonetheless, these data provide a useful look at both the impact that colleges reopening can have on surrounding communities and the rapidly-changing epidemiology on campuses as the fall semester begins.
Want to see for yourself? All our data are on GitHub.