John Ewing, president of Math for America, skewers the concept of “learning loss” in this article in Forbes.

I have been a fan of Dr. Ewing ever since I read his article “Mathematical Intimidation: Driven by the Data,” in which he eviscerated the idiotic idea of rating students by the test scores of their students. If you have not read it, you should.

In his latest essay, he shows how various interest groups, politicians, and pundits have used the idea of “learning loss” to promote reopening of schools, regardless of local conditions. The same concept is now used to promote the useless requirement of standardized testing in the midst of the pandemic.

You might say that he is the anti-Emily Oster; Oster is the Brown University economist who has written extensively about why schools should reopen and how young children are unlikely to catch or carry the coronavirus.

Ewing disagrees.

He writes:

With a mix of exasperation and despair, many writers (and quite a few politicians) have demanded that schools open for in person instruction during the pandemic. They are exasperated because they believe schools were closed unnecessarily, since children don’t get sick and schools don’t contribute to Covid’s spread. They despair because remote learning has led to disastrous “learning loss” for an entire generation of students. 

But they are wrong about the science— schools do contribute to community spread—and they are wrong about the disaster as well. While remote learning surely affects students, we don’t know yet exactly how or how much. Learning loss isn’t a meaningful answer.

Early in the pandemic, people observed that children didn’t get sick as readily as adults. Children were tested much less often than adults. Asymptomatic spread was unknown or uncertain. Studies focused on sickness in schools rather than transmission, and they suggested that keeping schools open had few costs. 

But this is wrong. A recent article in the German magazine Spiegel International details an Austrian study that shows schoolchildren are infected at the same rates as adults and quite efficiently spread Covid-19 to others. There are now many other studies that draw similar conclusions. A review of studies from a group of scientists and doctors in Sweden (disclosure: my brother is among them) links to 25 studies from around the world and provides summaries of each. The science is clear: Children become infected and spread covid-19 to their parents, grandparents, siblings, and next-door neighbors. Those infected get sick. Some have long-term complications. Some die. Opening schools costs lives. If you believe in science, you have to accept even uncomfortable truths.

Should we open them anyway? What about that disastrous learning loss? As the pandemic plays out, learning loss has become the focus of education policy. Research firms (McKinsey is the best known) publish reports that cite, with great precision, the number of months of learning loss; politicians and pundits hysterically lament a coming lost generation; parents and the public angrily demand a return to in person learning. Learning loss drives all this; it’s become the central educational feature of the pandemic.

But what’s it mean—”five months of learning loss”? What exactly is lost? Do students forget facts?  Skills? Are memories erased? Can they find what’s lost? And what does “five months” mean? Yes, I know, it’s calculated from a mathematical formula, but formulas are only as good as the data and assumptions that go into them. Mathematics is not magic. What are the assumptions? What’s the data? Where does it come from? When people discuss learning loss, they generally don’t know the answers to any of these questions. And if the notion is so vague, how can it be so easily and precisely measured? 

Of course, the term “learning loss” comes from the language of test enthusiasts. For them, learning is a substance that’s poured into students over time. One measures the accumulated substance by the number of correct answers on a test (standardized, usually multiple-choice). By administering two comparable tests at different moments in time, one measures success or failure for learning. An increase in correct responses is gain; a decrease is loss.

Learning loss is usually illustrated by the summer break. We are told that students experience about three months of loss each summer. Again, what’s this mean? If a student does more poorly on a test in September than in May, is learning really lost? Seems doubtful, or at least incomplete. Mathematicians know that stepping away from a topic for a while requires time to recollect the bits and pieces when you return. Those bits and pieces aren’t lost—they only require reassembling, and often the reassembling leads to greater understanding. Similar things occur in every subject, and in other areas of life as well, like riding a bike or playing the piano.

Learning is complicated. Plutarch famously wrote that minds are not vessels to be filled but fires to be kindled. Fires don’t leak. You don’t measure them in months. Learning loss is a calculation masquerading as a concept—a rather shallow, naïve, ridiculous concept.

Of course, those who talk about learning loss might mean the absence of (new) learning. Fair enough. In the spring, remote teaching and learning were novel to both teachers and students. They struggled because remote instruction was an unfamiliar skill. Mastering skills takes practice and requires dedication. In the spring, everyone planned in two-week intervals, hoping the pandemic would soon be over, and dedication was in short supply. But remote instruction improved this fall, and while it’s very far from ideal, remote teaching and remote learning are much, much better … and getting better still. Kids are resilient. We don’t yet know how resilient they will be in the pandemic.

There remains an enormous problem of equity. Students who live in poverty are at a severe disadvantage in remote instruction. No internet, no computer, sometimes little parental support. But while the pandemic exacerbates this problem, it’s not the cause. We need to solve the equity problem permanently, not just in the pandemic. We had an opportunity to do so in the spring by providing free internet access and computing devices to every student in need. It would have been roundoff error in the stimulus package. It would have entailed massive logistical issues, but that’s what responsive governments do in times of crisis. Our politicians chose not to do so.

Should schools open for in person learning? Maybe. But not because of some ridiculous idea of learning loss. If schools choose to return to in person instruction, it’s because, like bars and restaurants, they serve a vital social and economic function. For younger children especially, the socialization afforded by schools is crucial for a child’s development. For parents, the childcare provided by schools allows them to work (or just to take a soul-saving break). The societal cost of eliminating these functions has been substantial. 

We have to balance that cost against the sickness and death that will be caused by the opening of schools. Balancing livelihoods against lives can be agonizingly complicated. It requires clear, precise thinking. Above all, it requires putting the right things on each side of the scale … and learning loss isn’t one of them.