Jessica Cohen, Sara Bleich, Joseph Allen and Benjamin Sommers are professors at the Harvard T.H. Chan School of Public Health.

As Election Day approaches, the electoral map isn’t the only thing divided into red, blue and purple. Our schools have similarly fallen into partisan divisions, with Republican-leaning areas largely following in-person schooling models and Democratic-leaning areas mostly using remote or hybrid models. This highlights a sobering fact: School policy decisions across the ideological spectrum are being made largely devoid of data, threatening the well-being of our children, economy and society.

As public health professors, we have spent our careers researching and analyzing public health policies. Our country’s current approach to schools is failing four basic principles of evidence-based policymaking.

First, policy should not be driven by emotion or anecdote. There are tragic stories about teachers who have died from covid-19 and viral images on social media of teens ignoring safety guidelines. There are painful portraits of children in unsafe, unsupervised and isolated settings because of remote learning. These stories might help build empathy at a time of deep division, but as the cliche goes in medicine and public policy, the plural of anecdotes is not data.

Second, cases in school must be reported in context — both in terms of the population size in question (not just case counts) and whether infections were actually transmitted in schools rather than in the broader community. We’ve grown accustomed to population-level statistics at the city, state and national levels (e.g., 14-day cases per 100,000 people); we should insist on the same rigor for school-based measures. Data on covid-19 clusters within schools, when they occur, should also be reported alongside information on which risk mitigation measures (e.g., masks, ventilation, etc.) were in place.

Third, any policy analysis must ask, “Compared with what?” While remote learning is often called the “safest option,” that view assumes children are at home, safely distancing from others. But policymakers must consider that in many households, parents have to work, meaning children are often in teaching “pods,” nanny shares or group hangouts at local playgrounds. This leads to a series of mixed interactions between children and adults.

Given this reality, it’s possible that hybrid models and remote learning may not reduce infection risk relative to in-person schooling that requires masks and keeps kids in smaller, contained groups. But we need better data to make that determination. In addition to tracking cases in schools, as the New York Times is doing, districts should collect anonymous survey data from parents on how children are spending time outside school and what risk-mitigation strategies they use. Contact tracers should assess whether these informal arrangements are leading to outbreaks, and should share that information with schools so that we can understand the risks of alternatives.

Finally, policy choices should not focus on just one outcome. We need a richer accounting of the costs and benefits of schooling models beyond just virus transmission. Key metrics should include student learning, engagement and well-being. How many students are consistently attending school (either in person or remotely), completing assignments and meeting learning objectives? How many have a stable Internet connection and quiet place to work? How many are missing meals or health-care services they typically receive at school?

Many districts and state education departments are creating scorecards and benchmarks to assess transmission risk within schools. Similar benchmarks are needed for these other measures of well-being, with close attention paid to equity issues. Schooling decisions risk widening existing education achievement gaps by socioeconomic status, race or ethnicity; tracking these effects on equity is critical.

We also must consider economic costs associated with school policy decisions. While many districts are weighing the costs of school policies with respect to building upgrades, new staff and new technology, school policy decisions have much broader economic implications. School closures have led to billions of dollars in lost productivity, disproportionately among women, who are leaving the workplace in record numbers. These are not decisions school leaders or teachers unions should be weighing directly; instead, city leaders, legislators and governors should be considering the broader impacts of school closures on their local economies and working families.

Looking at the national pattern of school choices, it is clear that evidence-based policy is not driving these decisions. In states with high and increasing rates of community spread, many schools are open for full in-person learning — putting economic and political considerations before safety and health. In states with lower community spread, many schools are still remote, allowing a single factor — fear of covid-19 — to drive decision-making, despite low absolute risks of school-based transmission and ignoring the real-world risks of hybrid or remote learning.

So far, our society’s decision-making on schools doesn’t deserve a passing grade. Let’s start evaluating school policy related to the pandemic with the attention, balance and evidence it deserves.

Read more: