Parents and students opposed to school closings protest in New York on Nov. 14. (AP Photo/Mary Altaffer)

Emily Oster, a professor of economics at Brown University, is the author of “Cribsheet: A Data-Driven Guide to Better, More Relaxed Parenting, from Birth to Preschool.” She is a co-founder of the website Covid-Explained.

As covid-19 surges nationwide, the debate over school reopening has intensified. On one side are arguments that schools are not major sources of spread. On the other, we hear of schools with significant rates of infection. In response, many districts are delaying plans to bring back their students. On Wednesday, New York City announced that the nation’s largest district — one of the first to open for in-person instruction — would join the ranks of cities closing their public schools.

But as the country grapples with how to educate kids while also curbing the coronavirus, the emphasis on transmission in schools may be misplaced. The best available data suggests that infection rates in schools simply mirror the prevalence of covid-19 in the surrounding community — and that addressing community spread is where our efforts should be focused.

At one level, this finding is intuitive. But its implications have been largely overlooked. Part of the challenge is a seeming disconnect between the claim that “schools aren’t superspreaders” and the objective fact that some people tested at schools are found to have covid-19. Understanding why these two realities can coexist first requires understanding how we measure cases in schools.

Several states maintain school-level data dashboards that provide information on infections among students and staff who attend in-person school. But this data generally does not document where covid-19 was acquired. Virus transmissions that took place in schools are not separated from infections that occurred elsewhere — at a friend’s house, at church — but were simply detected in schools. So reports of cases among students and staff should not, in themselves, lead us to believe that attending school is a source of spread.

Because school numbers reflect covid infections that could have been picked up anywhere, our baseline assumption should be that school infection rates would at least mirror community infection rates. The prevalence of covid-19 detected in schools, in other words, shouldn’t be any different than if coronavirus testing were conducted among populations at local grocery stores or restaurants or gyms or public parks.

What should worry us is if we start to see infection rates among staff or students that are higher than in the surrounding community. Along with evidence of large clusters of cases in schools, this would point to the schools themselves as being loci of covid-19 spread.

To determine whether schools are superspreader locations, we need to be able to calculate the infection rates in those schools and evaluate them side-by-side with infection rates in the community. Unfortunately, this is difficult, for several reasons related to limited data-keeping. But New York state has recently started publishing comprehensive data for all schools that allows for this comparative analysis. What it shows should make us optimistic that schools are not sources of superspreading.

This side-by-side comparison of the New York data focuses on the period between Oct. 12 and Nov. 6. For clarity, I’ve divided all Zip codes in the state — 1,071 total — into six “buckets,” based on the prevalence of covid-19 in those communities. The lowest bucket includes all Zip codes with fewer than three covid cases per 100,000 people; the highest includes Zip codes with 20 or more cases per 100,000 people; the rest are distributed in between.

To plot each bucket along the x-axis, I calculated the population-weighted average of covid cases within each grouping. For example, the average for the lowest bucket is two cases per 100,000 people; for the highest, it is 28 cases per 100,000 people.

I then looked at all schools within the Zip codes that constitute each bucket. For these schools, I calculated the prevalence of covid-19 among three different populations: elementary and middle school students; high school students; and faculty and staff at all schools. The average infection rates among these populations determine those points’ placement on the y-axis.

Once these figures are plotted, we can see that, for high-school students and staff, the rates are similar to population case rates. For elementary and middle school students, they are lower — which we would expect, given the generally lower disease rate in younger children.

Covid-19 infection rates in

communities and schools

In-person schooling does not appear to increase the risk of covid-19 transmission for staff and students, according to data from New York state.

Case rate in schools among elementary

and middle school students is lower

than the rate in the overall community.

*Cases per 100,000 people

School

case rate

30

25

20

15

10

Each dot represents

a group of Zip codes.

5

0

5

10

15

20

25

30

Community case rate

While case rate among school staff

and high school students is similar

to the community rate.

School

case rate

30

25

20

15

10

5

0

5

10

15

20

25

30

Community case rate

Source: Analysis of New York state school

data by Emily Oster.

The Washington Post

Covid-19 infection rates in

communities and schools

In-person schooling does not appear to increase the risk of covid-19 transmission for staff and students, according to data from New York state.

Case rate in schools among elementary and middle

school students is lower than the rate in the

overall community.

*Cases per 100,000 people

School

case rate

30

25

20

15

10

Each dot represents

a group of Zip codes.

5

0

5

10

15

20

25

30

Community case rate

While case rate among school staff and high school

students is similar to the community rate.

School

case rate

30

25

20

15

10

5

0

5

10

15

20

25

30

Community case rate

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

Covid-19 infection rates in communities and schools

In-person schooling does not appear to increase the risk of covid-19 transmission for staff and students, according to data from New York state.

Case rate in schools among elementary

and middle school students is lower

than the rate in the overall community.

While case rate among school staff

and high school students is similar

to the community rate.

School

case rate

School

case rate

30

30

25

25

20

20

15

15

10

10

Each dot represents

a group of Zip codes.

5

5

0

5

10

15

20

25

30

0

5

10

15

20

25

30

Community case rate

Community case rate

*Cases per 100,000 people

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

Covid-19 infection rates in communities and schools

In-person schooling does not appear to increase the risk of covid-19 transmission for staff and students, according to data from New York state.

Case rate in schools among elementary and

middle school students is lower than

the rate in the overall community.

While case rate among school staff and

high school students is similar to

the community rate.

School

case rate

School

case rate

30

30

25

25

20

20

15

15

10

10

Each dot represents

a group of Zip codes.

5

5

0

5

10

15

20

25

30

0

5

10

15

20

25

30

Community case rate

Community case rate

*Cases per 100,000 people

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

One urgent concern in the debate surrounding covid-19 has been the disease’s disproportionate impact on Black and brown communities. To focus on these communities, I ranked each Zip code in New York state by share of the population that is Black or Latino, and looked only at the top 30 percent of those Zip codes. As a result, there were only four buckets; all of these Zip codes had covid case rates higher than five per 100,000 people, meaning that none fell into the two lowest buckets used for the broader analysis above.

Covid-19 in Black and

Latino schools and communities

Communities and schools with significant minority populations have higher infection rates than the general population. Yet school infection rates still mirror the prevalence of covid-19 in communities.

*Cases per 100,000 people

Elementary and middle school students

School

case rate

30

25

20

15

10

Each dot represents

a group of Zip codes.

5

0

5

10

15

20

25

Community case rate

School staff and high school students

School

case rate

30

25

20

15

10

5

0

5

10

15

20

25

Community case rate

Source: Analysis of New York state school

data by Emily Oster.

The Washington Post

Covid-19 in Black and

Latino schools and communities

Communities and schools with significant minority populations have higher infection rates than the general population. Yet school infection rates still mirror the prevalence of covid-19 in communities.

*Cases per 100,000 people

Elementary and middle school students

School

case rate

30

25

20

15

10

Each dot represents

a group of Zip codes.

5

0

5

10

15

20

25

Community case rate

School staff and high school students

School

case rate

30

25

20

15

10

5

0

5

10

15

20

25

Community case rate

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

Covid-19 in Black and Latino schools and communities

Communities and schools with significant minority populations have higher infection rates than the general population. Yet school infection rates still mirror the prevalence of covid-19 in communities.

Elementary and middle school

students

School staff and

high school students

School

case rate

School

case rate

30

30

25

25

20

20

15

15

10

10

Each dot represents

a group of Zip codes.

5

5

0

5

10

15

20

25

0

5

10

15

20

25

Community case rate

Community case rate

*Cases per 100,000 people

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

Covid-19 in Black and Latino schools and communities

Communities and schools with significant minority populations have higher infection rates than the general population. Yet school infection rates still mirror the prevalence of covid-19 in communities.

Elementary and middle school students

School staff and high school students

School

case rate

School

case rate

30

30

25

25

20

20

15

15

10

10

Each dot represents

a group of Zip codes.

5

5

0

5

10

15

20

25

0

5

10

15

20

25

Community case rate

Community case rate

*Cases per 100,000 people

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

We see similar patterns in this data. Sadly, schools in areas with significant minority populations do have higher infection rates — but the same is true of the communities these schools are located in. Again, there is a noticeably lower rate of infection among younger students than among staff, suggesting that sending children in these neighborhoods to school for in-person instruction would not increase the danger of their contracting covid-19. Given that these are precisely the groups that are losing the most from distance learning, this suggests we should make their return to in-person school a priority.

The data from New York also makes it possible to look at the distribution of cases. During the four-week period studied, roughly 80 percent of schools in the state reported no covid-19 cases at all. And of those schools that did detect covid, nearly 90 percent had only one or two cases across all students and staff. A single case is unlikely to be the result of in-school transmission — meaning students and teachers don’t appear to be catching covid from each other.

Among schools that

had Covid cases...

65% had only

one case

Two cases 22%

Three cases 7.5%

Four to six cases 4.5%

Seven cases or more 1%

Source: Analysis of New York State school

data by Emily Oster.

The Washington Post

Among schools that had Covid cases...

65% had only one case

Two cases 22%

Three cases 7.5%

Four to six cases 4.5%

Seven cases or more 1%

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

Among schools that had Covid cases...

Two cases

22%

Three cases

7.5%

65% had only one case

Four to six cases

4.5%

Seven cases or more

1%

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

Among schools that had Covid cases...

Two cases

22%

Three cases

7.5%

65% had only one case

Four to six cases

4.5%

Seven cases or more

1%

Source: Analysis of New York state school data by Emily Oster.

The Washington Post

This data from New York doesn’t mean that all schools are safe to open in all environments. When levels of community spread are very high, the chance of spread at school, too, is certainly higher. What it does mean, though, is that by far the most helpful thing we can do for schools is to control community spread.

By the same token, if we are looking for ways to control community spread, shutting down schools is not the answer. Other measures, focused more on the locations and activities known to have superspreader potential, would do much more to curb the pandemic. This is where New York City is making a mistake: It is closing schools because city case rates are rising, not because of any evidence that schools are spreading the disease.

Other countries have managed to keep schools open even while locking everything else down. They see the essential need for in-person schooling and have been willing to invest the resources necessary to make sure this continues safely. In the United States, especially as infection rates continue to rise, it’s not surprising that teachers are afraid to return to the classroom. They understandably want and deserve better personal protective equipment, testing, contact tracing and ventilation.

But what the data increasingly shows is that the best way to protect teachers and students isn’t to shut down schools. It’s to focus on all the measures that will keep them — and their families, friends and neighbors — safe outside the classroom.

Watch Opinions videos:

Read more:

Joe Allen, Akiko Iwasaki and Linsey Marr: This winter, fight covid-19 with humidity

Emily Oster: How the media has us thinking all wrong about the coronavirus

Danielle Allen and Ashish Jha: We’ve figured out it’s safe to have schools open. Keep them that way.

Gary Abernathy: Our covid-19 polarization will only get worse. We need to find a balance.

The Post’s View: D.C. needs to figure out how to get students back in the classroom