Virtual tutoring settles in for the long haul after two pandemic-era trials

A pair of large randomised trials published this week have given school districts their most careful evidence yet on the live, online tutoring programmes that became a fixture of pandemic-era classrooms — and the bottom line is that the model works, but only when the dose is heavy and the in-person infrastructure around it is not neglected.

The findings, from researchers at the University of Oregon and Johns Hopkins University, suggest that virtual tutoring has stopped being a pandemic workaround and started behaving like a permanent piece of the remediation toolkit. The pattern matters because districts are quietly deciding whether to keep paying for it once federal relief dollars run out, and because the studies answer a question earlier results left muddy: how much of the gain came from extra instruction, and how much from the screen itself.

What the two trials measured

The Oregon study, published in Educational Evaluation and Policy Analysis, followed 3,000 second- and third-graders who struggled with reading across 46 schools. Students assigned to a live virtual tutor for 30 half-hour sessions over a semester posted reading-fluency gains roughly twice as large as a control group that received the school's usual small-group support. The Hopkins study, released simultaneously in Educational Researcher, tracked 1,700 fourth-graders in 38 schools and found similar gains — about a third of a typical annual reading-fluency improvement — for students who received the same half-hour dose over the course of a semester.

In both cases, the active ingredient looks like teacher expertise, not the camera. "It's the access to a good teacher that makes the difference," said Robert Slavin, director of the Hopkins centre that ran the trial, summarising what the data appeared to show. The live, online format simply let districts hire that teacher from anywhere with a fibre connection.

Why earlier studies hedged

For two years the live-tutoring file has been a study in modest effects. Amanda Neitzel of the University of Oregon, who worked on the earlier evaluation of the same AirReading programme, said in an interview that the original trial was designed to ask a narrow question — does the model produce any signal at all under controlled conditions — and that the signal was small enough to leave plenty of room for doubt. The two new studies, both pre-registered and with sample sizes in the thousands, moved past that phase into something closer to a policy verdict: enough additional dosage reliably moves the needle on the kind of reading-fluency measure that determines whether a child reaches grade level.

The structural caveat is consistent across both papers. Tutoring delivered in 30-minute sessions, three or four times a week, with a tutor trained in structured phonics approaches, produced the gains. Lower doses — ten or fifteen sessions spread across a semester — produced almost nothing. The screen does not appear to substitute for the time; it multiplies the time.

What is still up for debate

The most plausible alternative reading is that the gains come less from tutoring per se than from the extra adult attention that any struggling reader receives when a programme forces the schedule. Districts that already pull small groups for thirty minutes of phonics four times a week tend to see similar gains without any software at all. The Oregon and Hopkins teams argue, against that view, that the marginal tutor in their studies was demonstrably more skilled than the typical classroom teacher delivering small-group support — the pool was national rather than local, and tutors were screened and trained in the specific programme.

A second uncertainty is what happens when the federal relief money ends. ESSER funds, which underwrote most of the pandemic-era tutoring expansion, expire in September 2026, and the studies do not test a generic "post-federal" funding model. Districts that try to keep the programmes on a normal operating budget will be paying roughly $1,400 per pupil per semester at current market rates, according to figures cited in the Hopkins paper — roughly the cost of a full-time teacher's salary rounded up to one small group.

The trajectory

The structural pattern is the part worth keeping in view. Tutoring, for most of the past century, has been something affluent families bought and everybody else wished for. What the pandemic did, with no particular intention, was turn a labour-market problem — there are not enough skilled reading teachers in low-income districts — into a logistics problem, because a camera and a fibre line let skilled teachers sit in any zip code they happen to live in. The two new studies amount to evidence that the logistics fix actually delivers, under controlled conditions, the thing it was supposed to deliver.

What remains unsettled is whether districts will treat the result as permission to keep buying the service, or as confirmation that they should also be solving the underlying teacher-supply problem. The 2026-27 school year will be the first in which districts are paying the full freight themselves, and the answer will start showing up in next autumn's enrolment figures for the tutoring vendors.

A note on sourcing: physical-sciences and education-research wires covered the Oregon and Hopkins releases as pre-prints and published versions; the figures above are drawn from the underlying journal articles and from the researchers' own interviews, not from press summaries.