The strong, positive effects of high-dose tutoring for boys and (girls)

Why it matters

Many educational disparities between boys and girls appear early and persist through their educational journeys. They are less ready than girls upon school entry, are more likely to be diagnosed with learning disabilities—including in reading and language—earn lower grades, and are more likely to drop out before completing high school. These gaps matter and are reflected in lower rates of college enrollment, declining labor-market outcomes, and fewer stable pathways into adulthood for young men.

Given these longstanding patterns, what actually works to improve academic outcomes, especially for boys?

The honest answer from decades of U.S. education research is sobering. A huge range of educational interventions have been tried, but the results are often underwhelming. Many interventions have no significant effect on the outcomes they try to influence while others show great promise in pilot studies but end up with much smaller impacts when scaled. Given widely varying effectiveness, federal legislation has placed increasing emphasis on interventions with rigorous empirical support (see the American Rescue Plan funding for schools following the pandemic as a recent example).

High-dose tutoring, and tutoring more broadly, is one of the few interventions that shows large, replicable effects. In a recent meta-analysis of randomized controlled trials (RCTs), tutoring shows the largest average effect size (figure 1).

Figure 1

High-dose tutoring—which tends to be defined as frequent (at least three days per week), one-on-one or with small groups of students, and from a teacher or other instructor that supplements regular classroom learning—is one of the few that shows strong, positive effects. And its theory of action is straightforward:

• When it’s not replacing normal class time, it gives students extra instruction that they otherwise would not get.
• It’s personalized. With just a few students or one-on-one, instructors can provide tailored help aimed at remediating a student’s specific skill gaps.
• With less divided attention, instructors can provide quicker feedback.
• One-on-one and small‑group settings may nurture mentor‑like relationships.
• It allows regular classroom teachers to work with students at more similar skill levels, which has benefits for both tutored students and their classmates who remain in the classroom.

Finally, the positive effects of high-dose tutoring may be particularly salient and important for boys since the most rigorously tested tutoring interventions tend to focus on early-literacy, boys are more likely to struggle in this area, and reading proficiency in elementary school strongly predicts later education outcomes. In one of the largest known tutoring interventions targeting struggling first-grade readers, 60% of the sample were boys. Nationally, boys account for 57% of students identified with a specific learning disability (SLD), a category that includes reading-related disabilities such as dyslexia, and they are more likely than girls to score below the NAEP “basic” level in 4th-grade reading (44% versus 36%). Using state-level NAEP data combined with enrollment counts, we find that boys constitute a majority of below-basic readers in every state, ranging from 112 boys per 100 girls in Kansas (53% boys) to 149 boys per 100 girls in West Virginia (60% boys).

Figure 2

The evidence in brief

On average, tutoring raises achievement by roughly 0.29 SD (about four months of extra learning for a struggling elementary-school student), though impacts vary based on several factors. They tend to be lower for older students than younger students (0.13 SD for 6-11 graders vs. 0.41 for pre-K and kindergarteners), mostly due to the declining impact of literacy interventions with age. Effects also shrink as programs scale beyond about 1,000 students, either because high-quality tutors become harder to find, implementation fidelity declines, or programs enroll students that need less help.  Conversely, researchers find that effects are strongest in programs with certified teachers or trained paraprofessionals, low student-teacher ratios (less than 3:1), high dosage (greater than 60 total hours), and in-school delivery. Notably, even the most comprehensive meta-analyses (like figure 2) do not analyze effects by gender, likely because there are not enough individual studies to estimate an average effect.

Where results for boys and girls are broken out separately, the gains for both tend to be similar. However, boys do make up a larger share of struggling readers, and assuming similar benefits on average, tutoring can potentially narrow early gender gaps without tailoring the curriculum by sex. More research is needed, but what we do know is that when tutoring is frequent, personalized, and run by trained adults, it delivers some of the largest, most reliable learning returns in the education literature.

Figure 3

Implications, gaps, and next steps

Overall, high-dose tutoring interventions show large, positive effects on average, especially compared to alternatives. Tutoring interventions focused on reading and literacy tend to show larger effects in earlier elementary grades and those focused on math find larger effects in middle school. We found relatively few studies that break out effects by gender although those that do tend to find similar impacts for both boys and girls.

As mentioned earlier, it’s important to note that while boys and girls may see similar benefits from tutoring interventions, when it comes to reading, boys are more often the ones falling behind. For example, in one evaluation of a literacy tutoring intervention, almost 60% of students receiving the intervention were male. Simply by virtue of who is more likely to receive it, wide-spread literacy tutoring would both raise reading achievement overall and reduce the gender gap in reading.

Finally, high-dose tutoring is no magic bullet. It works because it tends to be relatively intensive, and costly. Identifying ways to capture the most important elements that make high-dose tutoring effective while reducing cost is imperative. Programs like Reading Recovery can cost an estimated $6,000–$13,000 (all amounts in 2025 dollars) per student. Saga Education’s math tutoring program, described in the “Not Too Late” RCT, cost an estimated $3,200-$4,800 per student, though the authors note that the price has since been lowered to $1,900 per student.

Making use of volunteers and virtual options are two promising options to reduce costs. A recent (and positive) evaluation of OnYourMark virtual tutoring reported costs of approximately $1,400 per student. Reading Partners, which has also been rigorously evaluated, uses volunteers and costs schools about $1,000 per student (though total program costs are higher). Another recent RCT of an in-school math tutoring program found that high-dose computer-assisted tutoring was nearly as effective as in-person tutoring, and at only two-thirds the cost.

While there has been a significant amount of research on the effects of high-dose tutoring, important research gaps remain. We encourage researchers and policymakers to:

• Consistently break out intervention effects by gender. A few studies do this, but many do not, and understanding average effects by gender is crucial for tailoring interventions.
• Conduct longer-term longitudinal studies of intervention effectiveness. Some tutoring interventions have been effective over the short-run but long-run impacts can be marginal or negative.
• Continue to evaluate tutoring options that can be scaled at lower cost while maintaining effectiveness, including those that can be deployed virtually and that take advantage of recent rapid advances in artificial intelligence and machine learning.

Despite some of these gaps, the results so far are compelling, and worth acting upon.

Do you know of any studies that we’ve missed? Please reach out at [email protected]!