Should you correct your child’s homework yourself?
Mostly, no. Research links direct help to lower achievement, while asking questions and modelling a method build real skill instead of a clean page.
You are looking over a shoulder at a page of sums, and one of them is wrong. The pencil is right there. Fixing it would take four seconds, and the evening would move on. Almost every parent has felt that pull, and almost every parent has given in to it at least once. It rarely feels like a mistake in the moment. It feels like helping.
A 2023 meta-analysis published in Frontiers in Psychology, pooling data from dozens of studies on parents and mathematics homework, found something worth sitting with: what the researchers call intrusive involvement, taking over, correcting, supplying the answer, was negatively associated with achievement (r = −0.153). Supportive involvement that respected the child’s own attempt, by contrast, was positively associated with it, and the strongest effect of all came from what the researchers labelled autonomy support: letting the child stay the one doing the thinking (r = 0.133). The study’s own explanation is blunt: even when a child asks for the answer, giving it ’may lead to a sense of incompetence’. The help lands as a small message that they could not have done it.
Ask before you answer
So what replaces the red pen? Mostly, questions. ’Read the instructions again, out loud.’ ’Walk me through what you tried.’ ’What’s the first thing you’re sure of?’ None of these solve the sum. All of them hand the thinking back to the child, which is where it needs to stay if the sum is going to teach them anything.
Writing in The Conversation, researchers who study family learning describe a method that works even when a parent does not know the subject well: watching an adult think out loud through one problem, then tackling the next one together, then letting the child try the last one alone. The same piece cites a study of Afghan mothers who spoke no English yet still supported their children’s learning simply by sitting with them, asking questions, and encouraging them to talk through what they were doing. Understanding the content was never the requirement. Staying present and curious was.
A wrong answer is a message, not a mess
This is the part that is easiest to forget: a wrong answer left on the page is useful. It tells the teacher exactly where the lesson did not land, which is precisely the information they need to teach it again, differently, tomorrow. Erase it, and that signal disappears. The child looks like they understood something they did not, and the gap quietly grows until it shows up on a test instead of a worksheet, where it costs a lot more.
None of that means leaving a child stuck and frustrated on principle. It means the posture in the room matters: a coach who asks ’what happens if you try it this way’ is doing something entirely different from a corrector who takes the pencil. The coach’s job is to notice where the thinking snags and loosen it with a question. The corrector’s job, without meaning to, is to make the snag disappear. Only one of those teaches anything that survives past this particular sheet of paper.
A coach also knows when to step back in. If the questions keep circling and the child is truly lost, not just slow, that is a fair moment to show one worked example and hand the next one back. The line between coaching and correcting is not about never helping. It is about who holds the pencil while the thinking happens.
What this looks like on an ordinary evening
In practice it is less dramatic than it sounds. It is resisting the urge to jump in for the first thirty seconds of struggle, since that is often exactly when a child works something out on their own. It is naming what they did right before naming what still needs work. It is a plan broken into smaller pieces when a whole page looks impossible, so the child has something to finish rather than something to abandon.
The same Conversation piece points to a large meta-analysis covering more than four hundred studies, which found that parental homework help was linked to higher motivation and engagement, but not to higher achievement, largely because so much of that help slides into doing the work rather than supporting the child doing it. Motivation matters. It is just not the same thing as learning, and homework has room for both if the roles stay clear.
A child who finishes a page with three wrong answers they found on their own has learned more than a page with three answers a parent quietly supplied. The first version is not a worse evening. It is simply an honest one, and honest is what the teacher, and the child, actually need from it.