Starting in the spring of 2020, school boards and superintendents across the country faced a dreadful choice: Keep classrooms open and risk more COVID-19 deaths, or close schools and sacrifice children’s learning. In the name of safety, many districts shut down for long periods. But researchers are now learning that the closures came at a stiff price—a large decline in children’s achievement overall and a historic widening in achievement gaps by race and economic status.
The achievement loss is far greater than most educators and parents seem to realize. The only question now is whether state and local governments will recognize the magnitude of the educational damage and make students whole. Adults are free to disagree about whether school closures were justified or a mistake. But either way, children should not be stuck with the bill for a public-health measure taken on everyone’s behalf.
I am part of a team from the American Institutes for Research, Dartmouth College, Harvard, and the educational-assessment nonprofit NWEA that has been investigating the impact of remote and hybrid instruction on student learning during the 2020–21 academic year. We have assembled testing results from 2.1 million elementary- and middle-school students in 10,000 schools in 49 states and Washington, D.C., and combined those with data on the number of weeks schools were in-person, remote, or hybrid during 2020–21. Our team compared student-achievement growth in the period before the pandemic, from fall 2017 to fall 2019, with the period from fall 2019 to fall 2021. For years, districts have regularly been using NWEA tests to measure how students’ performance in reading and math changes during a school year; in a typical week of in-person instruction before the pandemic, the average student improved 0.3 points in math (on the NWEA’s scale) and 0.2 points in reading.
During the spring semester of 2020, though, nearly all schools went remote. Distractions, technical glitches, and the many other pitfalls of online education made it far less effective than in-person school.
One-fifth of American students, by our calculations, were enrolled in districts that remained remote for the majority of the 2020–21 school year. For these students, the effects were severe. Growth in student achievement slowed to the point that, even in low-poverty schools, students in fall 2021 had fallen well behind what pre-pandemic patterns would have predicted; in effect, students at low-poverty schools that stayed remote had lost the equivalent of 13 weeks of in-person instruction. At high-poverty schools that stayed remote, students lost the equivalent of 22 weeks. Racial gaps widened too: In the districts that stayed remote for most of last year, the outcome was as if Black and Hispanic students had lost four to five more weeks of instruction than white students had.
By our calculations, about 50 percent of students nationally returned in person in the fall and spent less than a month remote during the 2020–21 school year. In these districts where classrooms reopened relatively quickly, student-achievement gaps by race and socioeconomic status widened a bit in reading but, fortunately, not in math. And overall student achievement fell only modestly. The average student in the quicker-to-reopen districts lost the equivalent of about seven to 10 weeks of in-person instruction. (That losing just a quarter of a typical school year’s academic progress is a relatively good outcome only underscores the dimension of the overall problem.)
What happened in spring 2020 was like flipping off a switch on a vital piece of our social infrastructure. Where schools stayed closed longer, gaps widened; where schools reopened sooner, they didn’t. Schools truly are, as Horace Mann famously argued, the “balance wheel of the social machinery.”
Like any other parent who witnessed their child dozing in front of a Zoom screen last year, I was not surprised that learning slowed. However, as a researcher, I did find the size of the losses startling—all the more so because I know that very few remedial interventions have ever been shown to produce benefits equivalent to 22 weeks of additional in-person instruction.
High-dosage tutoring—which educators define as involving a trained tutor working with one to four students at a time, three times a week for a whole year—is one of the few interventions with a demonstrated benefit that comes close, producing an average gain equivalent to 19 weeks of instruction. One of those leading the charge on tutoring is Tennessee Department of Education Commissioner Penny Schwinn, who is offering matching funds to encourage school districts to launch tutoring initiatives. Tennessee’s goal is to provide high-dosage tutors to 50,000 students a year for the next two years. School systems elsewhere have similar ambitions. The educational-policy think tank FutureEd, at Georgetown University, reviewed the pandemic-recovery plans of thousands of districts and found that a quarter had tutoring initiatives in the works.
The obvious challenge with tutoring is how to offer it to students on an enormous scale. To eliminate a 22-week instruction loss would require providing a tutor to every single student in a school. Yet Tennessee’s plan would serve just one out of 12 Tennessee students in the targeted grades.
Given the magnitude and breadth of the losses, educators should not see tutoring as the sole answer to the problem. School systems need a patch big enough to cover the hole.
Many district leaders I know are considering three additional measures. One option is voluntary summer school, which, according to prior research, has yielded about five weeks of instructional gain per student. Another option is an extra period each day of instruction in core subjects. A double dose of math over the course of an entire school year has been shown to produce gains equivalent to about 10 weeks of in-person instruction, although the evidence on reading is weaker. (Our team will be working with districts to measure the efficacy of these and other catch-up efforts over the next two years.)
Like tutoring, double-dose math will be hard to scale up. Staffing the additional sections of math requires hiring more math teachers amid a historically hot labor market. Unlike tutors (who can be contractors), districts are hesitant to add permanent teaching staff for a short-term catch-up effort.
Meanwhile, summer school has historically struggled with low student attendance. In a typical pre-pandemic year, only about 6 percent of students attended summer school. Even if districts managed to triple that number, enrollment would still fall far short of the magnitude required to eliminate learning loss.
A third alternative would be lengthening the school year for the next two years. Of course, districts would have to pay teachers, janitors, and bus drivers more, perhaps at time and a half, to work the extra weeks. But unlike with tutoring or double-dose math, districts already have the personnel, the buildings, the buses, the schedules. As long as educators, parents, and students view the extra instructional time as just an extension of the school year—like days added to make up for snow closures—the power of family and school routine will deliver higher attendance than summer school.
The primary problem with a longer school year is political, not logistical. After opposition from the local teachers’ union and some parents, the Los Angeles Unified School District was able to add only four optional days of school next year. This is, to be sure, more make-up time than many other school systems have planned, but quite inadequate given that the nation’s second-largest school district was remote for three-quarters of 2020–21.
I fear that, in areas where classrooms remained closed for long periods, school officials are not doing the basic math. High-dosage tutoring may produce the equivalent of 19 weeks of instruction for students who receive it, but is a district prepared to offer it to everyone? Alternatively, suppose that a school offers double-dose math for every single student and somehow convinces them to attend summer school, too. That, educational research suggests, would help students make up a total of 15 weeks of lost instruction. Even if every single student in a high-poverty school received both interventions, they would still face a seven-week gap.
Educational interventions have a way of being watered down in practice; many superintendents and school boards may tell themselves that they are taking a variety of steps to help students make up lost time. And yet most district plans are currently nowhere near commensurate with their students’ losses.
I understand the many practical challenges of implementing any of these measures—much less implementing all of them quickly. Yet speed is essential. State and local school agencies received $190 billion in federal pandemic relief, much of which remains unspent. Districts have more than two full school years in which to spend the aid. But if they do not get started at sufficient scale during the coming school year, they risk using the aid for other purposes and running out of time and money later.
Last year, Miguel Cardona, the secretary of education, urged schools to return to in-person learning. Now the department that he leads should be encouraging (or requiring) districts to update their recovery plans in light of achievement losses. Local school boards should have to show that their recovery programs are plausibly sufficient to cover those losses. When Congress approved federal aid packages for schools, the magnitude of the achievement losses were unclear, and many analysts were preoccupied with potential shortfalls in the state and local tax revenue used to fund schools. Thus, the law required districts to spend only 20 percent of their COVID-relief money on academic recovery; the rest could go to the day-to-day needs of a school district—salaries, curriculum materials, teacher training, facility improvements. But many districts, especially those that were remote for much of 2020–21, will have to spend nearly all of their federal relief funds on academic recovery if they want students to catch up.
Reversing pandemic-era achievement losses will take aggressive action over the next several years. And yet the problem also presents an opportunity for any governor or mayor or superintendent looking to make meaningful improvements in children’s education. Federal aid is available. No obvious partisan roadblocks stand in the way. Most communities just need leadership—and a sufficiently ambitious recovery plan. In Tennessee, Schwinn has at least recognized the enormous scope of the problem. Which other state and local leaders will join her?