New research: Done right, virtual tutoring nearly rivals in-person version

This story first appeared at The 74, a nonprofit news site covering education. Sign up for free newsletters from The 74 to get more like this in your inbox.

High-dosage, in-person tutoring gets results, recent research suggests. But as federal funding for remediation dries up and schools struggle to raise students’ post-COVID skills, educators have been hoping for a lifeline in the form of live, online tutoring.

While virtual tutors still work directly with students in real time, they can work from anywhere, expanding the potential talent pool and lowering costs.

Until recently, virtual tutoring had scant evidence that it works very well, with few rigorous studies of its effectiveness. But new findings, including two recent studies from Johns Hopkins University’s Center for Research and Reform in Education, are beginning to offer a different narrative: Done well and with the same safeguards as traditional in-person tutoring, the virtual version can be nearly as good.

“I was always one of those people who was so skeptical — ‘it’s never going to work,’” said Amanda Neitzel, an assistant professor at Hopkins and the research center’s deputy director. “And then I did these studies, and I was shocked, because it did work.”

In a study published in December, Neitzel and her colleagues found that first-graders in Massachusetts who used Ignite Reading, a one-to-one virtual tutoring program, made substantial progress in reading, with the percentage of students reading on grade level rising from just 16% in the fall to about 50% by spring.

The share of “struggling” readers also dropped, from 64% in the fall to 28% by the spring.

The study tracked about 1,900 students in 13 high-poverty Massachusetts school districts in the 2023-24 school year. The students showed nearly five-and-a-half months’ more progress on a key reading test than those who didn’t get the tutoring. And they improved across the board, with English learners, students with disabilities and low-income students all gaining ground.

Ignite tutors work with students for 15 minutes every day, typically during “literacy blocks” in class or in separate, staff-monitored rooms.

In a separate study published in September, Neitzel and her colleagues found that students who got online Air Reading tutoring outperformed their peers by about two points on NWEA reading assessments, a “significant” change that would raise the average student slightly to the 55th percentile in the class, or just above average.

While researchers saw no difference in impacts for English language learners or those with special needs, they found that first-graders got more out of the tutoring, meaning that the hypothetical 50th-percentile student who got tutoring would rise to the 58th percentile.

Six elementary schools in a district in Texas took part in the randomized controlled trial evaluating Air Reading for 418 first-through-sixth-grade students during the 2023-24 school year. The small-group tutoring ran for just a few months in the spring, from late January through April.

Neitzel said the effect sizes in the two new studies aren’t necessarily as large as those of the most effective in-person models, but the new evidence provides some of the most compelling evidence yet for schools wondering whether they should offer virtual tutoring. 

“It’s really exciting that every month or two there’s another good study out,” she said. “And there are more in the field right now too. So I think in the next couple years, we’ll be able to answer that question better.”

Matthew Kraft, an associate professor of education and economics at Brown University, agreed, saying several recent studies from 2022 to the present amount to “emerging evidence on the efficacy of virtual tutoring programs,” suggesting they hold promise.

He noted that randomized control trials generally find that virtual tutoring has positive effects, but often of smaller magnitude than those found in meta-analyses of in-person tutoring programs. “However, the devil is in the specific program design details,” he said. For instance, several studies find that one-on-one virtual tutoring is more effective than programs that use small groups.

Jennifer Krajewski, director of outreach and engagement for Proven Tutoring, a clearinghouse for research-proven tutoring models housed at Johns Hopkins’ Center for Research and Reform in Education, noted that both Air Reading and Ignite Reading employ well-trained live tutors and a “highly structured” program, with ongoing coaching for tutors and a clear instructional process that addresses students’ individual needs. These characteristics, she said, are often part of in-person tutoring programs that have been found effective.

Both programs work hard at getting students to actually attend, she and Neitzel said. 

Reviewing the Ignite study, Neitzel said the percentage of students actually receiving tutoring when they were supposed to was “shockingly high,” topping 85% for the vast majority of students. That suggests implementation is key in a field where attendance isn’t always tracked very well. 

“You could have the best model in the world, but if the kids aren’t actually there, it’s not going to move the needle,” she said.

Attendance remains one of virtual tutoring’s biggest challenges, she said. “When it’s a physical person in the building, they can pull you out of class. It’s harder to avoid. Whereas if it’s on a computer, you just don’t log in — or you log off, or [you say], ‘Oh, it’s not working.’ ”

Krajewski said that for the study, Ignite worked with a local funder in Massachusetts to hire on-the-ground workers who ensured that students were showing up. It also held regular virtual meetings with educators “to make sure everyone understood the milestones and the goals,” ensuring that the program would be launched consistently across several districts. “Everyone was really on the same page because of these meetings,” she said.

Ignite and the local funder also appointed paid school and district “champions” to supervise implementation. Each school champion worked about three hours weekly to troubleshoot problems that arose. And they required that schools review student achievement data weekly, moving students out of tutoring when they succeeded and filling those seats with struggling students. 

Neitzel said one of the keys to Ignite’s success, at least in the study, was that it paired students with tutors who spoke the same language, offering “a little connection” between them, even if tutoring took place primarily in English.

If schools can’t find enough bilingual teachers locally, she said, “maybe virtual tutoring is the best option you have.” In-person tutoring programs might be slightly more effective, she said, but virtual programs offer flexibility on hiring and other challenging aspects of implementation. 

In the Air Reading study, Neitzel said, company representatives met with schools every other week, focusing closely on attendance and which students weren’t attending sessions.

On occasion, she said, Air Reading teams flew out to schools “to make sure stuff was happening and getting set up or trying to troubleshoot what’s going on. I was impressed with just how well they knew the schools they were working with.”

In one case, she recalled an Air Reading worker who was so attuned to the school he oversaw that he knew an attendance monitor’s father had died. “That’s how involved they are with this,” Neitzel said. “When it works well, there are these tremendous relationships with people in the district to make it work.”

Krajewski, who was not an author on either study, said researchers haven’t yet seen evidence of effectiveness for tutoring using AI agents working directly with students. “We’ve seen that the most effective models use human tutors,” she said. 

Hopkins researchers are working on an evaluation of an AI-assisted tutoring model developed by Carnegie Mellon University and predicted there’d be noteworthy data by the end of 2025. “But even then, it’s not that the tutoring is replaced by AI,” she said. The AI, she said, is helping human tutors be more effective.

“These studies show how important that human tutor continues to be,” she said. “We’re learning that that human tutor, virtual or in person, is driving the instructional process.”