Ask a parent whether homework helps and you’ll get a confident answer either way. The honest research answer is less quotable: it depends mostly on how old the child is. Homework shows a real, measurable benefit from around middle school onward — but for elementary-age kids, the evidence that it improves grades is close to zero.

The study nearly every argument traces back to

Almost every claim on either side of the homework debate leads back to one body of work: a synthesis led by Duke University psychologist Harris Cooper, with Jorgianne Civey Robinson and Erika A. Patall, reviewing more than 60 studies on homework from 1987 to 2003, published in Review of Educational Research in 2006. Two decades on, it’s still the most comprehensive analysis available.

The headline finding — a positive but modest link between homework and achievement — is well established. What it hides is how much that average changes by age. Looking at simple correlation between homework and test scores, the relationship was essentially flat in elementary school, rose to a small r = .07 in middle school, and reached a moderate r = .25 in high school. (r here is a correlation coefficient — a way of scoring how tightly two things move together, from 0, no relationship, to 1, a perfect one — so .25 means high schoolers who do more homework meaningfully tend to score higher, not that homework alone explains everything about their grades.) In the smaller set of studies that experimentally assigned homework rather than just observing who did it, the effect nearly tripled: a weighted average effect size of d = .60 across those studies, and d = .53 in the subset using true random assignment — evidence solid enough to call this an established pattern, not a hunch.

Why the effect changes so much by age

Cooper’s team found a likely explanation: younger children are more easily derailed by distractions at home, haven’t yet built independent study strategies, and may get more out of homework’s side effects — practicing time management, building the habit of sitting down and working — than out of the material itself. For elementary kids, the real value may be less about mastering that night’s content and more about building the discipline that makes studying possible later. That’s why Cooper has resisted both ends of the fight: “The bottom line really is all kids should be doing homework, but the amount and type should vary according to their developmental level.” He’s compared dosage to medication — too little has no effect, too much can do harm, the right amount helps.

Is more homework actually better?

Not every researcher accepts that the correlation proves homework is driving the improvement. Education professor Cathy Vatterott has pushed back directly: does homework cause achievement, or do students who are already succeeding simply complete more of it, inflating the apparent benefit without homework actually causing any of it? That’s a real open question the data can’t fully settle — which is why Vatterott has argued for scaling back or dropping homework for the youngest students specifically, where the evidence of benefit is thinnest. Cooper’s own numbers support at least a ceiling: additional homework beyond roughly two hours a night showed no further link to achievement, even for older students.

So how much is actually worth assigning

The most-cited practical guideline — about 10 minutes of homework per grade level per night, roughly 40 minutes in fourth grade and near two hours by senior year — comes from several U.S. education bodies, including the National PTA and the National Education Association; treat it as a reasonable starting rule, not a formula backed by its own controlled study. A 2024 systematic review in Campbell Systematic Reviews, led by Liping Guo and colleagues, focused specifically on experimental studies and found an overall effect size of 0.45 for students who completed assigned homework versus those who didn’t — with the clearest gains in arithmetic computation and problem-solving. In head-to-head comparisons, moderate loads of 20 to 30 minutes outperformed shorter 10-to-15-minute assignments for building real conceptual understanding — though the review’s authors were explicit that no single optimal dose has been nailed down yet, especially for older students.

Put together, the research doesn’t support “homework doesn’t work” or “more homework is always better.” It supports something harder to fit on a poster: the payoff depends heavily on age, is easy to over-dose past a fairly low ceiling, and nobody has pinned down the exact right amount for any given subject or grade — which may be exactly why the argument keeps getting fought in absolutes instead.

Sources: Harris Cooper, Jorgianne Civey Robinson, Erika A. Patall, “Does Homework Improve Academic Achievement? A Synthesis of Research, 1987-2003,” Review of Educational Research, Spring 2006 (assess.ucr.edu); Liping Guo et al., “The Relationship between Homework Time and Academic Performance among K-12: A Systematic Review,” Campbell Systematic Reviews, September 2024 (doaj.org); “Is Homework Getting Too Complicated? A Look at How Homework Policies Are Changing,” Duke Today, March 2006 (today.duke.edu); “Homework: Is It Good for Kids? Here’s What the Research Says,” Time, August 2016 (time.com).