Short answer
The marshmallow test offered preschoolers one treat now or a bigger one after waiting. A 1990 follow-up linked longer waits to higher teen SAT scores, but mainly in one version of the task and a small Stanford campus sample. Larger replications found the link much weaker and largely explained by family background and early ability. Trust in the adult also shapes how long a child waits.
Most of us know the story. A small child, one marshmallow, an agonizing wait. The kids who held out supposedly grew up to ace the SAT. Then a lab task became a parenting parable.
Say your 11-year-old spends their birthday money within the hour. It is hard not to wonder what that predicts. The actual research is more interesting and far less fatalistic. Below, study by study, we keep three things apart: the evidence, our read of it, and a small experiment to try at home.
What the marshmallow test actually did
The original experiments ran at a nursery school on the Stanford campus. A preschooler could have a small treat right away or a larger one by waiting. The children were roughly three and a half to six. Nobody in these studies was 8, let alone 15.
The prediction story came later, from a follow-up of children tested between 1968 and 1974.
What the research says
- Looking at the treat makes waiting harder. In a 1972 study, 50 of 56 recruited preschoolers completed the task. They waited 3.09 minutes on average with the treats visible and 8.90 minutes with the treats hidden (Mischel, Ebbesen and Zeiss, summarized by Simply Psychology).
- The famous follow-up was small. Of 653 children tested, only 185 had follow-up data as teenagers (average age at testing: 4 years 4 months). Parents of 94 reported SAT scores, and 134 had coping-survey data. All attended a school the authors described as serving "mostly middle-class children of faculty and students" (Shoda, Mischel and Peake, 1990).
- The link depended on the version of the task. It showed up mainly when the treats stayed in view and no coping strategy was suggested. There, waiting correlated r = .58 with a rating of self-control in frustrating situations, among 43 children. In the other versions, only 3 of 42 correlations were statistically significant.
- The SAT link split by sex. In that key version, the verbal SAT correlation was .74 for 18 girls and .02 for 17 boys. The authors flagged the gap themselves.
Most of these limits are stated in the paper itself. The retellings dropped them, and a condition-specific finding from a few dozen campus kids became a law of life.
Was the marshmallow test debunked? What replications found
Two later studies checked the claim with a much bigger sample: children from the NICHD SECCYD, a U.S. study that followed the same kids from early childhood into adulthood.
What the research says
- Bigger sample, smaller link. Watts, Duncan and Quan (2018) studied about 918 children, with the wait capped at 7 minutes, and focused on those whose mothers had not finished college. Before any adjustment, the link between waiting at age 4 and achievement at 15 was β = .236. Controlling for family background and the home environment cut it to .081. Adding measures taken at the same age, including early cognitive ability, left .050, which was not statistically significant (Watts, Duncan and Quan, 2018).
- No link to behavior problems. The same study found no significant association between waiting and behavior problems.
- Adulthood looked similar. A 2024 follow-up reached 702 of those children at age 26. Links were weak (r = .17 with education, r = −.17 with BMI) and mostly vanished after full controls. In the paper's words, "Marshmallow Test performance does not reliably predict adult outcomes" (Sperber and colleagues, 2024).
Our read
"Debunked" goes too far, and "proven" goes much further. The original association was real, but it came from one version of the task and a small, narrow sample. In the best-powered replication, the raw link was roughly half the original size and lost statistical significance once family background and early ability were counted.
So a preschooler's wait seems to say a little about their circumstances and early skills, and very little about their future on its own.
| Study | What it did | What it found | What it does not show |
|---|---|---|---|
| Mischel, Ebbesen & Zeiss (1972) | 50 preschoolers, treats in view or hidden | 3.09 vs. 8.90 minutes on average | That a short wait means low willpower |
| Shoda, Mischel & Peake (1990) | Followed 185 of 653 campus preschoolers into their teens | SAT and self-control links, mainly in one version of the task | A general rule: 94 SAT reports, one school, r = .02 for boys |
| Watts, Duncan & Quan (2018) | About 918 children, 7-minute cap, followed to 15 | β = .236 unadjusted; .050, not significant, with full controls | That waiting predicts achievement beyond background and early ability |
| Sperber et al. (2024) | 702 of the same children, surveyed at 26 | Weak links, mostly gone with controls | That the test reliably predicts adult life |
Why kids wait: trust, an audience and a cooler view
If the test is not mainly measuring destiny, what is it picking up? Several studies point to the situation, and especially to the adult in it. From a child's chair, waiting is a bet that only pays if the bigger treat really shows up.
What the research says
- Trust in the adult. In the "rational snacking" study, 28 preschoolers first had a reliable or an unreliable experience with the experimenter. Then 9 of 14 in the reliable group waited the full 15 minutes, versus 1 of 14 in the unreliable group (Kidd, Palmeri and Aslin, 2013).
- Trust they only watched. Among 34 children aged 3 to 6, some first watched an adult act untrustworthily toward someone else. They later gave up waiting at nearly three times the rate of children who watched trustworthy behavior (Michaelson and Munakata, 2016).
- Who will know. Among 273 preschoolers aged 3 and 4 in China, some were told a teacher would learn how long they waited. They waited nearly twice as long as children with no audience (Heyman and colleagues, 2020).
- Kids now wait longer, not shorter. A reanalysis of 840 U.S. children aged 3 to 5, from three birth cohorts, found wait times rising. Children in the 2000s waited about 2 minutes longer on average than those in the 1960s (Carlson, Shoda et al., 2018).
- Hot and cool. A theory paper, not an experiment: Metcalfe and Mischel proposed a fast, emotional "hot" system and a slower, reflective "cool" one. Waiting, in their account, depends on shifting attention from a treat's arousing features to cooler, abstract ones (Metcalfe and Mischel, 1999).
Our read
Together, these studies make waiting look less like a fixed trait and more like a sensible reading of the setting. Three things seem to matter: whether the promise is reliable, whether the treat is in plain view, and whether someone the child respects will notice. A family can shape all three.
Two cautions: these were short lab tasks, some with small samples, so no single effect should be stretched. And the reputation study shows that waiting is social, not that pressure helps.
How to teach kids delayed gratification, age by age
From here on, these are our suggestions: grounded in the studies above, but not tested on kids 8 to 17.
Keep the promises you make
If you say Saturday, make it Saturday. In the trust studies, an adult's track record shaped whether waiting seemed worth it. At home, your follow-through is that track record.
Cool the temptation instead of fighting it
Staring at the prize is waiting on hard mode, as the treats-in-view condition showed. For a younger child, cooling can mean putting the item out of sight. For an older one, it means thinking about the goal in calmer, more abstract terms.
Make a long wait visible
This one is our suggestion, not a research finding. Say your 10-year-old wants a $60 skateboard. Six $10 boxes, each marked when saved, turn one long wait into six short ones.
Our guides to breaking a big goal into steps, saving for one goal and visible progress versus rewards go further. The same idea drives DreamBig, our app coming soon to iPhone and Android. A child picks a dream, its price splits into equal squares, and each saving reveals part of the picture.
Let the child choose the strategy
Our bet: a strategy your child picks is one they will actually use. Offer two or three, in the spirit of raising a motivated child without pressure.
| Age | What to try | What it might sound like | Keep in mind |
|---|---|---|---|
| 8–10 | Playful cooling: pocket or put away the tempting item. Short, visible waits. | "Let's put the gift card in the drawer and color in a box every Friday." | Days or weeks, not months |
| 11–13 | Name "hot brain" and "cool brain." Let your child pick the strategy. | "That sounds like hot brain. Let's hear from cool brain." | Their odd strategy may beat yours |
| 14–17 | Explicit trade-offs: name the later goal and track why waiting is worth it. | "Concert tickets now or the car fund later. Your call, and I'll keep my side of the deal." | Spending is sometimes the right call |
A two-week waiting experiment at home
A family experiment, not a research protocol, and the only comparison is with your child's own first try.
Try this at home · 14 days
- Offer it once, as an experiment. If your child says no, drop it.
- Pick one modest want they already have, about two weeks away. Say your 12-year-old wants a $30 game.
- Agree on the deal in writing: what they are waiting for, until when, and what you will do. Then keep every part of your side.
- Split the wait into equal steps on paper and let your child mark each one.
- Let them choose one cooling strategy, such as keeping the item out of sight or describing it like a dull product listing.
- On day 1, day 7 and day 14, ask them to rate how hard the wait feels, from 1 to 5. Compare only with their own day 1, never with a sibling or a friend.
- If they spend early, treat it as information, not failure. Ask what made it hard and adjust the setup.
What does not work, and why
Treating one moment as a verdict. Eating the marshmallow at 4, or spending the gift card at 12, is not a forecast. The largest replications found the prediction small and mostly explained by other factors.
Breaking small promises. "Five more minutes" that turns into twenty teaches that waiting does not pay here. The trust studies suggest children notice.
Repeating the myth. "Kids who can't wait fail in life" is not what the evidence shows. And preschoolers in the 2000s waited longer than those in the 1960s, not less.
Turning a witness into an audience. The reputation study involved no shaming, and we would not add any. A grandparent who warmly asks how the saving is going is a witness. Announcing a slip at dinner is not.
Using preschool tricks on a teenager. Hiding a 15-year-old's phone is a standoff, not a cooling strategy. Older kids need reasons and a real choice.
Key takeaways
- The marshmallow test studied preschoolers; the famous SAT link rested on 94 reports from one campus school, mainly in one version of the task.
- A much larger replication found the link about half as big before adjustments, and not significant after accounting for background and early ability.
- Not debunked, not proven: real but small, and not a reliable predictor of adult outcomes.
- Children waited longer when the adult was reliable, the treat was out of sight, or a teacher would hear how long they waited.
- At home: keep promises, cool the temptation, make long waits visible, and compare only with your child's own first try.
Frequently asked questions
Is the marshmallow test debunked?
Not exactly. The original link was real but small-sample and condition-specific. The strongest replications found it roughly half as large before controls and not significant after accounting for family background and cognitive ability.
What age was the marshmallow test done on?
Preschoolers, roughly three and a half to six years old. Advice for 8- to 17-year-olds, including ours, is an extrapolation rather than a direct finding.
Can delayed gratification be taught, or is it just personality?
Context matters a lot. Children waited far longer after a reliable experience with the adult, and wait times have risen across generations. That argues against a fixed trait, though no study here tested training for older kids.
Are kids today worse at waiting than kids in the past?
No. In a reanalysis of 840 U.S. preschoolers, children tested in the 2000s waited about 2 minutes longer on average than children tested in the 1960s.
What does delayed gratification mean for a child?
Choosing a bigger, later reward over a smaller one now, like saving allowance for a bigger toy. The choice depends on self-control and on whether the child trusts the promise will be kept.
What is a good way to practice delayed gratification with my child?
No single research-backed game exists. Keep promises, help your child cool a temptation rather than fight it, and make long waits visible. That last idea is our suggestion, not a research finding.
This article is general information, not psychological advice for a specific child. If impulsivity, attention or anxiety is disrupting school or family life, talk to your pediatrician or a licensed professional.
Sources
- Navidad, A. E. (2020). Stanford Marshmallow Test Experiment. Simply Psychology (summarizing Mischel, Ebbesen & Zeiss, 1972).
- Shoda, Mischel & Peake (1990). Predicting adolescent cognitive and self-regulatory competencies from preschool delay of gratification: Identifying diagnostic conditions. Developmental Psychology, 26(6), 978–986.
- Watts, Duncan & Quan (2018). Revisiting the marshmallow test: A conceptual replication investigating links between early delay of gratification and later outcomes. Psychological Science, 29(7), 1159–1177.
- Sperber, Vandell, Duncan & Watts (2024). Delay of gratification and adult outcomes: The Marshmallow Test does not reliably predict adult functioning. Child Development, 95(6), 2015–2029.
- Dewar, G. (2023). Delayed gratification isn't just about willpower. Parenting Science (summarizing Kidd, Palmeri & Aslin, 2013, Cognition, 126(1), 109–114).
- CU Boulder Today (2016). Trust in adults affects children's willingness to delay gratification, CU-Boulder study finds. University of Colorado Boulder (on Michaelson & Munakata, Developmental Science, 19(6), 1011–1019).
- Carlson, Shoda, Ayduk, Aber, Schaefer, Sethi, Wilson, Peake & Mischel (2018). Cohort effects in children's delay of gratification. Developmental Psychology, 54(8), 1395–1407.
- Kiderra, I. (2020). The marshmallow test, revisited. University of California (on Heyman and colleagues' study in Psychological Science).
- Metcalfe & Mischel (1999). A hot/cool-system analysis of delay of gratification: Dynamics of willpower. Psychological Review, 106(1), 3–19.
Ilya Makarov is the founder of DreamBig, a visual savings-goal app for kids and teens. DreamBig Parenting articles separate what research shows from our interpretation and from practical experiments, and link every source. This article is general information, not medical, psychological or financial advice for a specific child.
