Student test scores are on a path to recovery nearly four years after the pandemic wreaked havoc on K-12 academics, extensive new data analyzed and released by a group of education researchers this week indicate.
Two years of uneven instruction, school closures, and online learning dragged math and reading levels down to their lowest levels in decades. Some after-effects, like chronic absenteeism, remain a persistent problem. But the last year of academic recovery efforts have partially reversed that trend, according to the data, released by the Center for Education Policy Research at Harvard University and The Education Opportunity Project at Stanford University. They reflect standardized test scores from 8,000 school districts in 30 states.
In a single school year—between spring 2022 and 2023—students on average gained back one-third of their original loss in math, and one-quarter of the original loss in reading. The improvements amount to more than what students would have learned in a regular, pre-pandemic academic year.
Those are notable gains considering just how far student learning had fallen: In spring 2022 students in grades 3 through 8 had lost the equivalent of half a grade level in math and a third of a grade level in reading.
These gains are also large, the researchers said, compared to the annual rate of change in test scores between 1990 and 2019 on the federally administered National Assessment of Educational Progress.
In response to the academic sliding, schools poured billions of dollars in federal relief aid—an unprecedented investment from the federal government known as ESSER—into recovery initiatives like tutoring, after-school programming, and hiring academic interventionists.
The return on ESSER investment has been substantial. “The recovery was larger than we would have expected based on spending alone,” the researchers wrote.
Recovery efforts made by districts, like high-intensity tutoring, contributed to the gains, the researchers surmised.
The good news on student achievement, though, is marred by how unevenly these gains were distributed. In all states except for Arkansas, Pennsylvania, and Virginia, poorer districts remain further behind their 2019 levels of achievement than their wealthier counterparts—an indication that learning gaps between rich and poor districts now are wider now than they were before the pandemic.
In some states, like Connecticut and Massachusetts, that pattern emerged because students in wealthier districts caught up faster.
The widening achievement gap should be concerning for districts: It shows that some students had more access to coaching, tutoring, or high-quality instruction than others. It also suggests that further improvement may hinge on doubling down on academic recovery—a challenge as federal ESSER funding winds down this year.
“Spending close to $2,000 per student isn’t going to be sustainable anymore. We need to understand how to scale an intervention like high-dosage tutoring. We need research to uncover what kind of tutoring is most impactful. But to invest in research, school leaders need significant cover,” noted Christina Grant, the superintendent of education for the District of Columbia, at an event held in the city to discuss the findings.
States’ gains varied; poor districts lag behind everywhere
Students in four states—Alabama, Illinois, Mississippi, and Louisiana—exceeded or met their pre-pandemic levels of achievement in 2023. All other states still showed a net loss in achievement.
But even within states where the test scores exceed pre-pandemic levels, the variation between poor and wealthier districts is apparent.
Alabama, for instance, had a net gain in math achievement between 2019 and 2023, but the gap between its richer and poorer districts widened. Districts like Montgomery, Mobile, and Birmingham lost more than half a grade level in math, while higher-income districts, such as Hoover and Shelby counties, lost little ground—or improved.
Among the 30 states analyzed, data from Massachusetts, Connecticut, Ohio and Indiana show the widest gaps in reading and math scores between wealthy and poor districts.
Education Week spoke to Thomas Kane, a professor of education at the Harvard Graduate School of Education and one of the authors of the analysis. The interview has been edited for length and clarity.
What does the data telegraph for principals and district leaders in terms of what they need to do about the uneven recovery?
One, parents can’t advocate for their child if they are misinformed. Schools should let every parent know before the federal money runs out if their child is below grade level. Schools may think, “well … that’s what the state test is for.” But the results won’t be coming back till August or September. And yet, most schools don’t have to wait for state assessment to come back to figure out who’s proficient and who’s not.
Second, to the extent that school leaders have authority over the budget on how the remaining [recovery] dollars are used, I’d encourage them to shift the spending as much as possible to the summer of 2024 and things they can contract. They can spend it on tutoring and instructional coaches, summer learning, after-school programs. If they have a choice between pushing money to a nonacademic use to just get it spent before the deadline runs out, and holding onto it for academic use, I’d encourage them to hold onto it. It may be the only way for leaders to extend funding into next year.
It takes time and effort for school leaders to find the right intervention for their students. Are you saying principals should speed up the contracting process before summer?
Principals should be looking at their results now and ask, where were we before the pandemic? If we are still behind, they should be allocating more dollars to academic catchup because there won’t be another federal package. Hopefully, though, more states will pick up [the tab] as federal dollars run out.
Are there resources or guidlines for principals as they think about spending or contracting this money?
We have worked with several districts who ran summer learning programs during the pandemic. We find that over six weeks, a 90-minute block each day on academic instruction produces the equivalent of one quarter of a year’s learning. Not every district has been doing six weeks; some do four weeks and maybe not always 90 minutes per day. But we’ve seen that has effects in pre-pandemic research and pandemic-era research. The progress is linear.
One thing I’d encourage districts to learn about is the Boston after-school and beyond model. There, rather than have the school district plan academic intervention and enrichment, why not have organizations like zoos and summer camps plan the enrichment? Schools can send teachers to provide an hour-and-a-half of academic instruction on-site. In Boston, the school provides an incentive to summer programs that make the hour and a half available for enrichment.
There are significant achievement gaps between wealthy and poor districts. Some of these gaps have widened even within districts. What do you think happened there?
This is fascinating from a research perspective. From 2019 to 2022, when a district lost ground, almost all the subgroups in the district lost the same amount of ground. There was a widening of achievement gap, but it was because high-poverty districts lost more ground than low-poverty ones and part of that was a result of longer school closure rates. But when a high-poverty district lost ground, even non-poor kids lost the same amount of ground in the district.
But during the recovery, the non-poor kids within lots of districts grew more than poor kids.
We don’t have direct evidence yet, but one explanation was that it was a mechanism of the loss and the mechanism of the recovery. The loss was due to community-level factors like how long schools were closed, what was the quality of hybrid instruction, how quickly was the school able to shift to remote instruction. What was going on in the community … the economic and social disruption meant that everyone was in the same boat and lost the same amount of ground.
During the recovery the community-level factors were less important. Now, it was about which kids had more resources at home to help catch up.
Is it also linked to how recovery measures like tutoring were implemented across districts?
There is no question that there was variable quality in the implementation. In districts where we’ve been evaluating efficacy [of tutoring], the typical dosage, instead of three sessions a week for 36 weeks, so roughly 108 sessions … it’s more like 20 to 25. In those districts, we’re seeing a quarter of the effect sizes.
There have been implementation challenges. We saw that during the 2021 and 2022 school years. We thought maybe the challenges would ease in 2022-23, but the dosage did not dramatically increase. It points to the challenges with scheduling the times and all the problems people are talking about when it comes to implementation.
What can your field do to support school leaders, especially now that they may have to speed up the help they provide?
One promising practice is outcomes-based contracting. Some districts go into agreements where tutors are paid per kid assigned to them, but the key problem is to get kids to show up. Instead, pay them by sessions attended. The next step is to pay them based on the growth that kids show on the assessment the school is using. If the contract is at the district level, you can make these payments contingent on student growth.
