Takeaways from research on tutoring to address coronavirus learning loss

Many educators and policymakers are worried about low-income children falling woefully behind in math, reading and other subjects while schools are closed during the coronavirus pandemic. One proposal is to give them personal tutors. Normally, the idea of giving every poor child a professional tutor would seem too expensive but extreme circumstances have put big ideas on the table and policymakers are suggesting cheaper, if not exactly cheap, ways to do it.

Tennessee is taking the lead. Earlier in May 2020 former Gov. Bill Haslam announced that he’s personally going to pay 1,000 college students up to $1,000 each to tutor children in grades kindergarten through six over the summer. Meanwhile, there are a number of proposals in Congress to dramatically expand AmeriCorps, a network of national community service programs, with recent college graduates tutoring children across the country.

Prominent members of the education research community are enthusiastically pushing for a dramatic increase in tutoring. The University of Michigan’s Susan Dynarski wrote an OpEd in the New York Times on May 7, arguing for a multi-billion dollar investment in tutoring both to offset shutdown learning loss and to give jobs to unemployed college students and graduates. Robert Slavin of Johns Hopkins University wrote a May 14, 2020 blog post calling for a “Marshall Plan for education, to recruit, train, and deploy thousands of tutors in schools across America.” He also suggests using new college graduates and having them work with small groups of students. Brown University’s Matthew Kraft is advocating for all students at low-income schools to receive a daily dose of tutoring, either individually or in pairs of students for each tutor, for a full class period during the normal school day for an entire year. He calls it “high-dosage tutoring.” Yes, there’s already an acronym: HDT.

I was curious to learn more about how effective individual and small group tutoring are and how much learning loss tutoring might be able to offset. And just as importantly, I wanted to know what researchers know about using novices to do this kind of work. Can academic tutoring be done properly by those who aren’t trained in pedagogy?

It isn’t easy to get quick or simple answers. Many studies have been done on tutoring but they vary in quality. And tutoring can be structured in so many different ways: in school or after school; once a month or every day. It can adhere to a specific scripted curriculum or tutors themselves can decide what or how best to teach. Some tutoring programs are laser focused on math or reading; some help students with a wide variety of subjects.

One effort to sum up the high-quality research on tutoring was a Harvard study from 2016. The study sorted through almost 200 well-designed experiments in improving education, from expanding preschool to reducing class size, and found that frequent one-to-one tutoring with research proven instruction was especially effective in increasing learning rates of low-performing students. But less frequent tutoring, such as weekly sessions, under-performed many other types of educational interventions. According to the study, high-dosage tutoring was 20 times more effective than low-dosage tutoring in math. In reading, high-dosage tutoring was 15 times more effective than low-dosage tutoring.

Slavin, the director of the Center for Research and Reform in Education at Johns Hopkins, has also looked at high-quality tutoring studies with a focus on specific instructional programs and curricula. He shared two meta-analyses, submitted for publication at an academic journal, in which tutoring programs generally rise to the top, above most classroom-wide approaches and performing considerably better than high-tech interventions that use educational software.

Slavin found that several tutoring programs posted effect sizes of 0.4 standard deviations or more, a statistical unit that he compared to five additional months of school over and above what students would have learned without the tutoring. (It’s worth noting that none of the underlying studies achieved anything close to Benjamin Bloom’s oft-cited result of “2 sigma” or standard deviations, which would be five times these gains.)

Five months of catch-up might be not enough to offset the learning losses that some experts are predicting from school closures. A preliminary estimate publicized in April 2020 by NWEA, a nonprofit organization that sells assessments to schools, said that students are likely to return to school in the fall of 2020 with less than half of the learning gains that students usually make in math during a school year. In reading, the learning losses weren’t predicted to be as dire, perhaps a loss of just 30 percent of the school year, based on the organization’s analysis of historical testing data and summer learning loss calculations on the organization’s Measures of Academic Progress or MAP tests.

The strong tutoring outcomes cited by Slavin were achieved when students were tutored one-to-one or in small groups. Students received daily tutoring sessions for four to eight months.

Slavin found that many types of people were effective tutors. College-educated teaching assistants produced learning gains through tutoring at least as high as certified teachers did, sometimes larger. Even paid volunteers, such as AmeriCorps members working as tutors, were able to produce strong results, Slavin said.

“One of the things that makes it easier to be a tutor than to be a classroom teacher is that you’re doing a specific task with kids with materials that are designed to support that,” Slavin said. “You start with computerized testing to figure out where kids are, and then you put the kids into a sequence and use proven strategies with them day in and day out. A much broader range of people can do that. They couldn’t trade places with teachers. Classroom teaching is a much more complicated job.”

The important thing, according to Slavin, is to have some training and a tested reading or math program for the tutors to follow. “One of the things that characterizes all successful tutoring programs is that they’re very structured,” said Slavin. “There are curricula that work better than others. But there are several that get good results.”

That’s not to say that every college student would make an effective tutor. Slavin says it’s important to vet and interview candidates. He says an expressed desire to work with children, along with experience as a camp counselor, a Sunday school teacher or a babysitter are important.

“I think they could be terrific tutors and many studies have shown just that,” said Slavin.

Unpaid volunteers haven’t worked as well, Slavin said. Paying them makes it more likely that the tutors themselves show up so that a single adult can work with a child consistently. Fostering a relationship between tutor and tutee is important.

“Being an effective tutor is probably more about the structure of the tutoring program itself than the type of training tutors receive or whether the tutor is a recent college grad, teacher’s aid, or full-time teacher,” said Brown University’s Kraft, who has been publicly advocating for a national tutoring corps since 2015. “The best training is on-the-job through consistent feedback from peers and supervisors, not a week-long crash course and then being left to sink or swim on your own.”

There are small disputes among researchers about how many students a single tutor can work with effectively. Slavin sees evidence that small groups can be as effective as one-to-one. Kraft is concerned that some studies have shown poor results with small groups, such as a four-to-one experiment in New York City, where students’ reading achievement didn’t improve.

Kraft points to other more successful experiments as a model, such as a 2013-14 trial in Chicago high schools, where students learned between one and two extra years of math, over and above what the typical American high school student learns in one year. Students who were behind in math were randomly selected to receive daily tutoring by recent college graduates, who worked with students two at a time. But it was expensive. Each tutor received a stipend of $19,000 for an entire school year. The tutoring program cost $3,800 for each high school student.

It remains unclear whether small tutoring trials like this can be expanded into large programs and still achieve such impressive results.

A large study of City Year AmeriCorps tutors across almost 30 school districts and more than 300 schools was released on May 20. It found that 34,000 students in grades 3 through 10 gained between two and four months of academic growth from tutoring. That’s a much smaller result. And, unfortunately, there was no control group, so we don’t know how the students would have fared without the tutoring. It was a modest dosage of tutoring, between 12 and 20 hours over the course of a school year, which included a social-emotional curriculum in addition to academic instruction in math or English. Some of the students worked individually with a tutor, others in small groups and the results were averaged together. One of the things Robert Balfanz of Johns Hopkins, one of the study’s authors, noticed was that the more hours of tutoring a student received, the more academic improvement a student posted. The benefits of tutoring didn’t taper off or diminish, Balfanz told me in a telephone interview.

That’s a sign that AmeriCorps might get even larger results if it added more tutoring hours. But many more tutors and cities would need to be added to AmeriCorps’ current portfolio if we want to make a dent in correcting school shutdown learning loss on a national level. “We can’t do this entirely through AmeriCorps,” said Johns Hopkins’s Slavin. “A lot of different groups that are doing tutoring are going to have to scale themselves up.”

This story about one to one tutoring was written by Jill Barshay and produced by The Hechinger Report, a nonprofit, independent news organization focused on inequality and innovation in education. Sign up for the Hechinger newsletter.