Guide · Patience · last reviewed 8 September 2026

Waiting: How Patience Actually Develops

The marshmallow test does not mean what you have been told it means. When it was repeated with a sample ten times larger, the link between waiting at four and achievement at fifteen fell away once family background and the child's earlier cognitive ability were taken into account. A separate study found that what most changed how long children waited was whether the adult had been reliable beforehand. If you want a child to wait, the highest-leverage thing you can do is keep small promises.

Three different things we all call patience

Parents use one word for three things that behave completely differently, and most of the frustration in this topic comes from mixing them up.

• Waiting for something. The biscuit after lunch, the turn on the swing. Nothing is being done in the meantime. This is the one the famous study measured.
• Tolerating a slow process. The bean that takes nine days to sprout, the weaving that takes four sittings. The child is occupied, but the result arrives late.
• Persisting through difficulty. The lace that will not go through the hole. This is not really about time at all, it is about frustration, and it has its own guide.

A child who is hopeless at the first can be excellent at the second. The middle one is the one this library is actually good at, because it can be built into an object rather than asked for.

The marshmallow test, and what happened to it

The version everybody knows: a four year old is left with a marshmallow and told that if they can wait, they will get two. The ones who waited, the story goes, did better at school, earned more and were better adjusted, and self-control at four therefore predicts how a life goes. That version rests on follow-ups of a small sample of children from the Stanford university nursery.

Watts, Duncan and Quan ran a conceptual replication with a sample ten times larger than the 1990 follow-up study, focusing on 552 children whose mothers had not completed college. The bivariate association between delay time at age four and achievement at age fifteen was beta = 0.24, p less than .001, unadjusted. Adding controls for family background dropped it to beta = 0.08, p = .016. Adding controls for the child's earlier cognitive ability dropped it to beta = 0.05, p = .140, which is not statistically significant.

None of which makes self-control unimportant, and none of it makes the original work fraudulent or foolish. What it means is that most of the predictive power people attributed to a four year old's willingness to wait was carried by the family circumstances and the general ability that also produced the waiting. The waiting was largely a symptom rather than a cause.

The finding a parent can actually use

A child who does not wait is often not a child with poor self-control. They are a child who has worked out, correctly, that promised things do not always arrive. Five years before the replication, Kidd, Palmeri and Aslin showed exactly that. They ran the classic paradigm with 28 children of about four and a half, but first gave each child an environment that was either reliable or unreliable: an adult promised better art supplies and either produced them or did not, and only then went the marshmallow on the table. Children in the reliable condition waited significantly longer, p less than 0.0005. The authors conclude that wait times may not only reflect differences in self-control abilities, but also beliefs about whether waiting will ultimately pay off.

It is a small study, 28 children, and it should be read as one. It is also the most useful thing on this page, because it turns the whole topic around. If the promise is shaky, taking the marshmallow now is the rational move, and the adult is the variable rather than the child.

So the strongest thing an adult can do to build a child's willingness to wait is to be reliable about small promises. Not a tray. Not a timer. Keeping your word about the biscuit after lunch.

What waiting looks like at each age

Two things to hold in mind as you read it. Waiting is situational rather than a trait: the same child will wait ten minutes for one thing and nine seconds for another, and that is normal rather than inconsistent. And the figure you will see everywhere, that a child can attend for two to three minutes per year of age, comes from practice literature rather than from a study. Use it as a rough shape and never as a number.

• At two, the thing that works is a very short, visible gap with something happening in it. Not a promise about later.
• At three, a wait works if it is filled and if it is short. A three year old who is given a job during the wait is not really waiting, which is exactly why it works.
• At four, a wait with a visible end point becomes possible. A timer that can be seen, a jug filling, a queue that is visibly shortening.
• At five, a slow process becomes genuinely interesting rather than merely tolerable. This is the age at which growing something starts to work.
• At six, waiting across days is possible if the thing is being visibly measured. Not otherwise.

These are descriptions of what you will commonly see, not a schedule, and a child who is a year off any of them is not doing anything wrong.

This is where activity marketing does its worst work, so here is the thing that deflates it. Executive control is not one thing, and the measures do not isolate it. As Diamond puts it, almost no executive function measure requires only one executive function. Any page that tells you a specific tray trains a specific mental faculty is claiming a precision the field's own instruments do not have.

Then there is the scale of it. A full prekindergarten curriculum built specifically around self-regulation, delivered by trained teachers over a whole year, has an inconsistent record in trials: the implementation literature on Tools of the Mind states plainly that intervention studies comparing those classrooms with control classrooms reveal inconsistent findings for children's outcomes. If a year of that lands inconsistently, a tray of tweezers on a Tuesday afternoon is not going to land reliably either. Do it because it is a good half hour.

What is worth doing this week

Five things. Four of them cost nothing, and three of them are about you rather than about your child, which is the best news on this page.

1. Make promises small enough that you will definitely keep them. A promise you keep is worth more than a generous one you might not. If you are not sure you can do it, do not say it.
2. Say when, not soon. After we put our shoes on. When this song finishes. A four year old cannot do anything with in a minute, and in a minute is almost always a lie.
3. Make the wait visible. A sand timer they can watch, a jug filling, five buttons moved one at a time from one bowl to another. A wait with a visible end is a different experience from a wait with no edges.
4. Give a job during the wait, not a distraction from it. Carrying the plates is a job. A screen is a distraction, and the wait it fills does not seem to count for anything afterwards.
5. Stop testing. If you find yourself setting up situations to see whether your child can wait, stop. The evidence gives you no reason to run the test and one clear reason not to: it makes you unreliable on purpose.

Activities where the waiting is in the material

The second kind of patience, tolerating a slow process, is the one an object can carry for you. At two, Water Pouring and Transfer is the cleanest example in the library: the pour cannot be hurried without the water going on the tray, so the material sets the speed rather than the adult. Color Shade Grading at three does the same job by eye instead of by hand, and Pattern Card Bead Lacing makes the wait countable, because there are eleven beads left and no way to make there be fewer.

Later on the results start arriving on a longer delay. Button Sewing Card at four is the first one where a child has to keep going past the point where it stops being fun, and Paper Folding Precision Challenge at five punishes hurry visibly, which is more persuasive than being told. At six, Soap Carving and Safe Whittling takes forty minutes and cannot be rushed, and Plant Care and Growth Measuring is one of only two activities in the whole library where the wait runs across days rather than minutes. At seven, Cardboard Loom Yarn Weaving is the long one.

There is one activity here that is really about the adult half of this page. Three-Step Instruction Hunt only works if you say the list once and then hold to that. Children work out very quickly whether the rule you stated is the rule you are running, which is the same mechanism as the reliability study above, in a much smaller form.

The full shelf is at Patience, and it lists all twenty-three. That tag is ours rather than the source's. The word patience appears exactly once in the material this library was built from, so every activity on that shelf is there because of what it demands of a child, not because anybody labelled it.

When to ask someone

Children reach the same point at very different times, and the normal range is much wider than most milestone lists make it look. Checklists exist to start a conversation with a professional, not to grade a child. If something is worrying you, that is reason enough to ask a health visitor, a paediatrician or a speech and language therapist. You do not need to wait and see.

For this topic specifically, the thing worth raising is not how long a child can wait. It is a child who cannot recover: distress that goes on for a long time after a small disappointment, most days, in a way that is out of step with other children the same age, or a child who is not able to settle anywhere at all. Start with the health visitor or the GP. A nursery keyworker or a teacher who sees thirty children a day is also a genuinely useful second opinion and costs nothing to ask.

Evidence note

Read next