There is one number behind almost every claim that coding makes children better at maths. A meta-analysis of 105 studies and 539 effect sizes, published in the Journal of Educational Psychology, found a moderate overall transfer effect from learning to program, g = 0.49. Mathematical skills sat among the areas that gained the most.
The same body of work carries a condition that gets dropped every time the number is quoted. The researchers state that "the direct transfer of what is learned through coding is complex and does not happen automatically" and that transfer "must be facilitated and trained explicitly." That is not a footnote. It is the difference between a child who codes for two years and gets better at maths, and a child who codes for two years and does not.
This is a guide to which maths coding actually exercises, which it does not, and what has to be true in the classroom for the first group to include your child.
Does learning to code make a child better at maths?
At specific maths reasoning skills, measurably yes. At school maths marks, the evidence is weakest of all. Those two findings come from the same analysis, and holding both is the whole of an honest answer.
The published figures are specific. Overall transfer came out at g = 0.49, with a confidence interval of 0.37 to 0.61. Near transfer, meaning gains on tasks that look a lot like programming, was g = 0.75 but with a very wide interval of 0.39 to 1.11. Far transfer, meaning gains on tasks that look nothing like programming, was g = 0.47, interval 0.35 to 0.59.
Within far transfer, the summary is worth reading in the researchers' own grouping: the largest effects were for creative thinking, mathematical skills, metacognition, reasoning and spatial skills, in a band of roughly g = 0.37 to 0.73. Then the sentence that almost never gets repeated: "school achievement and literacy, however, benefited the least from learning to program."
So the realistic promise is not better marks next term. It is a child who reasons differently about quantity, structure and rules, which is a slower and more durable thing. A separate systematic review of 11 studies, published in Computers & Education, reached a compatible conclusion: children learning to code also picked up mathematical problem-solving, critical thinking, self-management and social skills.
The condition the headline number leaves out
Transfer is not a property of coding. It is a property of how coding is taught, and the analysis shows this twice over.
What happens when the comparison group is also busy
The moderator analysis found "significantly larger transfer effects for studies with untreated control groups than those with treated (active) control groups." In plain terms: when you compare children who learned to code against children who did nothing in particular, coding looks excellent. When you compare them against children who spent the same hours on another structured, demanding activity, the advantage shrinks.
That matters for a parent deciding between options rather than deciding between coding and an empty afternoon. Much of what the literature firmly establishes is that structured reasoning practice beats no structured reasoning practice. Coding is one good way to buy that, not a uniquely magic one. Anyone selling it as the only route is overselling a real finding.
Why school achievement gained the least
School maths marks measure a syllabus, a pace and a way of showing work. Coding exercises the reasoning underneath it. Those overlap less than they should, because a child can reason beautifully about a rule and still lose marks for method, which we wrote about in why a right answer still loses marks.
The honest expectation, then: watch for a change in how your child talks about problems months before you watch for a change in grades. If the grade is the only thing being measured, the effect that is actually there will be invisible.
Not sure which level your child should start at? A free trial class with a Codeyoung teacher shows you exactly where they are and what they are ready for next, before you commit to anything.
Book a Free Trial →Which maths ideas does writing code actually exercise?
The clearest way to see the overlap is to put code next to the official description of mathematical reasoning. The United States Common Core sets out eight Standards for Mathematical Practice, which describe how a mathematically proficient student thinks rather than what they are taught. Three of them describe programming almost word for word.
A repeat block is the standard about repeated reasoning
Standard 8 asks students to "look for and express regularity in repeated reasoning" and to "notice if calculations are repeated, and look both for general methods and for shortcuts." A child who drags out the same four blocks five times and then replaces them with a repeat block has done precisely that. They noticed the repetition and wrote the general method. The maths lesson version of this is spotting that 7 added five times is a multiplication.
A variable is the standard about abstraction
Standard 2, reasoning abstractly and quantitatively, describes "the ability to decontextualize, to abstract a given situation and represent it symbolically and manipulate the representing symbols as if they have a life of their own." That is what a variable named score is. It stands for a quantity that is not fixed, it can be changed without re-reading the game, and the child manipulates the name rather than the number. Children usually meet this idea in code years before they meet x in algebra, and they meet it with immediate feedback, which is the part a textbook cannot offer.
A custom block is the standard about structure
Standard 7 asks students to "look for and make use of structure" and to "see complicated things, such as some algebraic expressions, as single objects or as being composed of several objects." Building a custom block called draw house out of twelve movement blocks is exactly that move: a dozen things treated as one thing, named, and then reused. It is also what factorising an expression asks a child to do.
None of this happens because the software is clever. It happens when somebody says out loud, at the moment the child builds the repeat block, that this is the same thinking their maths teacher calls finding a general rule. That sentence is the transfer. Without it, the child has learned Scratch.
What this looks like by age
Readiness matters more than enthusiasm here, because a child who cannot yet hold the underlying idea will copy blocks instead of reasoning with them, and copying teaches very little maths.
| Age | Maths idea the coding actually exercises | What it looks like going well |
|---|---|---|
| 6 to 7 | Sequence, counting on, position | Predicts where the sprite will end up before pressing go |
| 8 to 9 | Repetition as a general rule, coordinates, negative numbers | Replaces repeated blocks with a repeat block unprompted |
| 10 to 11 | Variables as changing quantities, conditions, comparison | Explains what the variable holds without pointing at the screen |
| 12 to 13 | Functions, parameters, structure and decomposition | Builds a custom block and reuses it with different inputs |
| 14 and up | Algorithms, efficiency, formal notation in text code | Compares two working solutions and argues which is better |
The right-hand column is deliberately about explanation rather than output. A working project proves the blocks are in the right order. Only the explanation proves the maths is in the child's head. If your child is somewhere behind this, that is ordinary, and is my child behind in maths is a more useful place to start than a coding class.
Where the transfer breaks down at home
The research names the conditions under which transfer happens, and they read less like a product description and more like a list of things that can each quietly fail: explicit facilitation of the transfer, sufficient teaching time, student-centred and cognitively demanding tasks, a supportive and motivating environment, and trained teachers.
Three failure modes account for most of what we see.
Project collecting. The child finishes a game a week and can rebuild none of them from scratch. Output is high, reasoning is thin, and nothing names a general rule. The fix is to build one project twice, the second time without the tutorial.
Blocks without language. The child uses variables fluently and cannot say what a variable is. This one is easy to miss because the projects work. Ask the explanation question and it surfaces in thirty seconds.
Debugging treated as failure. A child who reads an error as a verdict on themselves stops reasoning and starts guessing, and guessing is the opposite of the practice standards above. Teaching the opposite habit is its own skill, covered in how to teach children to debug.
Maths anxiety interacts with all three, and it travels into the coding class with the child. If numbers already feel like a threat, the reasoning practice will not land until that does, which is the subject of maths anxiety in children.
Four questions to ask before you pay for both
If you are considering a coding class, a maths class, or both, these four separate the providers who produce transfer from the ones who produce certificates.
- Who connects the two subjects out loud, and when? Ask for the specific moment in a lesson where a coding idea is named as a maths idea. A provider who has never thought about it will answer in generalities.
- What happens when my child's project works but they cannot explain it? The answer you want involves going back, not moving on.
- How much of the hour is my child thinking versus following? Cognitively demanding tasks are one of the named conditions. Watching a screen share is not one.
- What is the plan when they get stuck for a whole lesson? Getting stuck is where the reasoning is, and a curriculum built to avoid it has removed the active ingredient.
If you are weighing the maths side specifically, our online maths classes for kids are built around explaining the why before the method, and the age-by-age guide to choosing a coding course covers the readiness question in more detail. For a no-screen option that exercises the same reasoning, the best maths books for kids list is a cheaper first experiment.
Where this leaves you
Coding and maths do reinforce each other, and the effect is real enough to have been measured across 105 studies. It is also conditional, smaller against an active alternative than against an idle one, and slowest to appear in exactly the place parents look first, which is the report card. The reasoning shows up earlier than the grade.
What turns the two subjects into one is not the curriculum or the software. It is somebody naming the connection at the moment the child builds it, often enough that the child starts naming it themselves. Ask any provider how they do that, and listen to whether they have an answer.
Codeyoung runs 1:1 live online classes for children aged 6 to 17, with a teacher who adapts the pace to your child rather than a fixed syllabus. The first class is free, so you can see how they respond before deciding.
Book a Free TrialSources checked 26 October 2026: Scherer, Siddiq and Sánchez Viveros, "The Cognitive Benefits of Learning Computer Programming: A Meta-Analysis of Transfer Effects", Journal of Educational Psychology, 2019; the same authors' follow-up discussion of transfer conditions in Frontiers in Psychology, 2021; Popat and Starkey, Computers & Education 128; and the Common Core Standards for Mathematical Practice.
