After heroic efforts to keep school doors open this Fall, schools are yet again shutting down and returning to distance learning as COVID-19 cases spike across the country. With no choice but to return to remote learning, schools have struggled to support their students and provide them with the resources and education they need to succeed. Already the disease has taken a toll. A preprint study of data from the Netherlands conducted during the pandemic shows significant learning losses sustained from March through May, compared with learning gains observed during the same two-month period last year—with a particularly severe drop in achievement for students from less educated homes. In the midst of a pandemic, how do teachers prevent a generation of students, especially those from disadvantaged backgrounds, from falling further behind?

One solution is tutoring. I and my colleagues recently analyzed 96 randomized evaluations of different tutoring models and found that 80 percent of the studies led to markedly improved outcomes, with more than half of the studies reporting large gains as a result of these programs. In education research, such consensus is a rarity, and the consistency and magnitude of the results are both remarkable and encouraging. In Chicago, for example, a two-on-one high school tutoring program empowered students to learn one to two years ahead in math, compared with what they would typically learn in a school year.

Studies show tutoring is a highly effective tool. The question is how to implement it when any in-person interaction poses a risk of COVID-19 infection. Fortunately, preliminary evidence suggests that virtual tutoring models may produce the same benefits as in-person tutoring. During the pandemic, researchers conducted a randomized evaluation of a tutoring program in Italy. In the study, which has not yet been published, middle school students who received three hours of online tutoring a week—over a computer, tablet or smartphone—from trained university students saw a 4.7 percent boost in performance in math, English and Italian. With six hours of tutoring support, improvement doubled. What’s more, the tutors seemed to play a mentoring role as well. Parents, teachers and the students themselves reported an increase in well-being, goals and social and emotional skills.

While rigorous evaluation of their impact is still needed, we’ve also seen similar virtual tutoring programs reach sizable student populations in the U.S. during the pandemic. One such program, introduced by the Tennessee Tutoring Corps, trained 430 college students to tutor more than 2,000 students across the state in math and English for seven to eight weeks.

A small-scale study described in a recent report tracked the results of a program that combined one-on-one remote tutoring with an online math game. The investigation found that the tutoring component was essential in helping students to improve. The results suggest that tutoring is a key tool—more than educational games or other learning sites—in keeping students engaged and combatting the growing COVID-19 learning loss.

Many affluent families are already employing private tutors to help their children keep up. Some large firms are also subsidizing tutoring for professional parents, underscoring the inequities between those who can afford the help and those who cannot. Individuals without such advantages are falling through the cracks. To close the gap, we need to direct tutoring resources to students with the highest chance of falling behind, especially those attending cash-strapped schools or growing up in low-income households.

To begin, we must address the biggest barrier to access—financial resources—by expanding federal funding to subsidize tutoring programs. The British government, for example, is launching a national program designed to help disadvantaged students make up for learning losses. The U.S. could follow its lead by expanding funding for programs, such as AmeriCorps––which is currently being considered in a proposal in Congress––that have experience running successful tutoring initiatives and could shift their in-person tutoring operations online. With recent college graduates and young professionals struggling to find work, these programs could also add jobs and address record levels of unemployment.

Amid the growing demand, there is also talk of creating a centralized market for tutoring services that would enable students and parents to specialize their tutoring and target particular areas of study. In this vein, Khan Academy founder Sal Khan’s new initiative, which aims to create an online marketplace for students seeking free virtual tutoring on specific topics, could be a model to democratize access.

At a local level, school leaders will have to prioritize tutoring for students who need it the most. School and district leaders can use data to determine which of their students are most likely to fall behind and offer these individuals extra support through direct or mandated services. Our research, for example, found that reading tutoring tends to be more effective for students in preschool through first grade, while math tutoring tends to have the greatest benefit for second through fifth graders. Younger children gain more from one-on-one tutoring while older elementary school students see better results with customized learning alongside peers. These insights can guide education leaders’ decisions about how to deploy limited tutoring capacity or expand certain programs while minimizing other costs.

Cash-strapped schools may not have the resources for tutoring services, but they can invest in computer-assisted learning programs that mimic tutors. And in-person tutoring programs struggling to survive in the pandemic can deliver their services virtually and supplement their sessions with software designed to allow students to work at their own rate—a surprisingly effective strategy with successful learning outcomes.

Teachers can incorporate computer-assisted learning, especially free programs such as the Khan Academy, into their lessons and utilize adaptive learning software to replicate a tutor that provides feedback on homework. To ensure that all students can benefit from these programs, we must simultaneously remediate the lack of Internet connectivity in low-income and rural households.

Although tutoring will look different this year, we have reason to believe that any kind of support could make a difference. While our research found that tutoring programs led by teachers or trained individuals––such as service fellows and teachers in training––were more effective than volunteer programs, students still benefited from volunteer tutoring.

And although virtual tutoring may not be quite as effective as in-person sessions, the positive results point to a potential solution to the growing educational gap. If we act now and use our resources wisely, online tutoring may just help us safeguard the futures of the next generation.