The marshmallow test: what it predicts, and what it doesn't
The short version: The marshmallow test is famous for a reason: self-control in childhood is linked to how people fare later in life. But a single waiting test at age four tells you much less than its fame suggests. These skills develop over time, and they can be practiced.
The studies
Moffitt, T. E., Arseneault, L., Belsky, D., et al. (2011). A gradient of childhood self-control predicts health, wealth, and public safety. Proceedings of the National Academy of Sciences, 108(7), 2693–2698. doi:10.1073/pnas.1010076108 · free full text
Who was studied: About 1,000 children in Dunedin, New Zealand, followed from birth to age 32, with self-control rated repeatedly between ages 3 and 11. A separate analysis followed about 500 pairs of British fraternal twins from age 5 to age 12.
Design: Long-term cohort (observational)
What they found: Children with more self-control grew into adults with better physical health, less substance dependence, better finances, and fewer criminal convictions. It wasn't only the least self-controlled children who did worse — each step up in self-control was linked with better outcomes. The link held after accounting for intelligence and family social class, and in the separate British twin sample, the twin with lower self-control had less favorable outcomes at age 12, even though both grew up in the same family.
Watts, T. W., Duncan, G. J., & Quan, H. (2018). Revisiting the marshmallow test: A conceptual replication investigating links between early delay of gratification and later outcomes. Psychological Science, 29(7), 1159–1177. doi:10.1177/0956797618761661 · free full text
Who was studied: Children from a large US study who did a waiting task at about age 4½ and were followed to age 15; the main analyses focused on children whose mothers had not finished college.
Design: Longitudinal replication (observational)
What they found: Waiting longer at four did predict better achievement at fifteen — but the link was about half as strong as in the original 1990 marshmallow studies, and it shrank by about two thirds once family background, early cognitive ability, and the home environment were taken into account. Among children whose mothers had not completed college, much of the achievement difference was associated with waiting at least 20 seconds rather than less. Links with behavior at fifteen were much smaller and mostly not statistically significant.
How strong is the evidence?
Strong for the general claim that self-control matters (the Dunedin study followed about 1,000 children from birth with very few dropouts, and the separate twin comparison helps). Weak for the popular claim that one marshmallow test at four predicts a child's future. Both studies are observational: they show links, not that self-control itself causes the outcomes or that training it would change them.
What these studies don't show
- They don't show that a child who can't wait at four is set on a worse path. In the 2018 replication, much of the achievement difference was associated with waiting at least 20 seconds rather than less, and much of the link faded once family background and early ability were taken into account.
- They don't test any program. Whether practicing self-control changes adult outcomes is a separate question — some preschool interventions improve specific self-regulation skills, although effects vary by program and measure.
What it means for parents
- Treat waiting as a skill, not a trait. Children wait better when they have a trick — turning their back, singing, keeping their hands busy — not when they're told to try harder.
- Make waits visible and short: a sand timer or "when the song ends" is easier than an open-ended "later."
- Don't read too much into one bad day or one test. The Dunedin finding is about years of everyday self-control, not a single moment.
What is the marshmallow test?
The marshmallow test comes from experiments run by psychologist Walter Mischel and colleagues at Stanford in the late 1960s and early 1970s, with preschoolers, mostly 4- and 5-year-olds, at Stanford's Bing Nursery School.
The setup was simple. A child was shown two rewards — for example, one marshmallow and two marshmallows. The researcher explained that if the child waited until the researcher came back, they would get the bigger reward; if they rang a bell, the researcher would return right away, but they would get only the smaller one. Then the adult left the room, in many versions for up to 15 minutes, and the child was observed.
The original aim wasn't to predict anyone's future. Mischel wanted to understand how children manage to wait: what they look at, what they do, and which conditions make waiting easier. The famous predictions came later, when follow-up studies in the late 1980s and 1990 reported that children who had waited longer tended to do better as teenagers — including on college entrance tests. Those follow-ups, based on a small group of children from a university community, are what made the test a household name.
How the children managed to wait
The most useful part of the marshmallow test for parents is not who waited, but how.
The researchers noticed that children who waited longer often avoided looking at the treat. They covered their eyes, turned their chairs around, sang, talked to themselves, played with their hands and feet, or invented games. In one experiment, children also waited longer when the treats were covered than when they were in plain view.
Mischel's team also found that how children thought about the treat mattered. Children told to imagine the marshmallow as a fluffy cloud, or to picture it with a frame around it like a photograph, could wait longer than children who thought about how sweet and chewy it would be. Mischel later described this as a shift from "hot" thinking about the reward to "cool" thinking about it.
Later research added a surprise: trust matters too. In a small 2013 study from the University of Rochester, children who had just seen an adult break a promise waited much less than children who had seen the adult keep one. The authors suggest that waiting is a reasonable choice only when the promised reward seems likely to arrive.
For parents, the lesson is practical. Waiting can be easier with strategies such as distraction and thinking about the reward differently, and with trust that the wait will end. In these experiments, simple instructions or changes to the setup made waiting easier.
Has the marshmallow test been debunked?
Not debunked, but its predictive power turned out to be much smaller than its reputation.
The 2018 replication described above, with a much larger and more varied group of children, found the link between waiting at four and achievement at fifteen was about half as strong as the original reports, and much weaker once family background was taken into account.
A 2020 follow-up of the original Stanford participants, then in their forties, found that how long they had waited as preschoolers did not reliably predict a broad set of midlife outcomes, such as education and financial well-being.
At the same time, broader measures of self-control — how children manage themselves across many situations and years, as in the Dunedin study — are linked to later outcomes. The difference is between one short test on one day and a pattern of behavior over childhood.
So the fair summary is this: self-control matters, it varies with circumstances as well as with the child, and a single marshmallow test is a poor predictor of an individual child's future. It remains a good window into how young children cope with waiting.
How self-control develops between ages 3 and 7
Self-control isn't one thing that a child has or lacks. It's a set of skills — stopping an action, holding a rule in mind, waiting, shifting attention on purpose — and they develop fast between about three and seven.
At three, most children can wait only briefly, and mainly with an adult's help. Young children may find it hard to keep game rules in mind: a 3-year-old playing a stop-and-go game often keeps going after the signal.
At four and five, waiting improves noticeably, especially when children have a strategy. Many can wait several minutes for something they want if they can look away or do something else. Game rules become manageable, though still easily lost when excited or tired.
In the early school years, many children increasingly use strategies on their own and can keep more complicated rules in mind, although development varies. Some talk themselves through hard moments, first out loud and then silently.
Two things seem to help. The first is practice in play: games with rules, turn-taking, pretend play with roles, and activities where waiting is built in. The second is a calm adult who models and prompts strategies — "let's sing while we wait" — until the child can do it alone. Lack of sleep, hunger, and stress pull in the opposite direction, which is why the same child can wait patiently on a Saturday morning and not at all on a Friday evening.
FAQ
What is delayed gratification?
Delayed gratification is choosing to wait for a bigger or better reward instead of taking a smaller one now — saving allowance money for a toy instead of spending it on candy today, or waiting for dessert until after dinner. In young children it depends on tricks for waiting and on trust that the reward will really come, and it tends to improve between about ages 3 and 7, though children vary.
How can I practice delayed gratification with my child?
Build small, real waits into ordinary life, each with a visible end and a trick: a sand timer before a snack, saving coins in a clear jar for something they chose, baking where the oven sets the pace. Keep early waits short and always keep your word, so the reward really arrives. Over weeks, stretch the waits slightly.
Why is delayed gratification important?
Because many good things in life require waiting or effort now for something later — learning to read, making friends, saving money, staying healthy. Studies such as Dunedin link better self-control in childhood with better outcomes later. But one test at age 4 doesn't set a child's future: the skill grows with age and practice, and family circumstances matter too.
Is the marshmallow test a good way to test my child?
Not really. A single waiting test at home tells you little: results change with hunger, tiredness, trust in the adult, and even how much the child likes the treat. Many researchers now see it as a window into waiting strategies rather than a predictor of an individual child's future. Watching how your child copes with everyday waits tells you more.
At what age can kids delay gratification?
Most children start to manage short waits for a reward around three or four, especially with help, and get much better between four and seven. In the original Stanford studies, many 4- and 5-year-olds could wait several minutes, and some for the full time, often 15 minutes. Waiting keeps improving through childhood, and it depends a lot on strategies and trust.
Can you teach delayed gratification?
Yes, in the sense that the strategies behind it can be taught. Children wait longer when they have a trick — looking away, singing, keeping their hands busy — when the end of the wait is visible, and when they trust the reward will come. Practicing short, real waits with those supports, and keeping your promises, is likely to help more than lectures about patience.
Boss of My Want (lesson 13) helps children practice waiting with a sand timer and a set of strategies; the Pause Button (lesson 4) helps children practice pausing in the moment.
How to Teach Kids Patience: Waiting Is a Skill, Not a Virtue
How to teach kids patience at ages 4–7: why "be patient" doesn't work, the three parts of a wait a child can manage, and waiting games that actually help.
Read next →