The Marshmallow Test Was Wrong. Your Kid Isn't 'Just Impulsive.'

In 2018, three researchers reran the most famous experiment in parenting psychology on about ten times as many kids as the original. The famous effect shrank by half. Then they factored in family background and early thinking skills, and it shrank by two thirds more, down to almost nothing.

If you have ever watched your kid grab the cookie, interrupt the story, or melt down in minute one, this study is for you. Because the lesson you were told to take from that moment is wrong.

The story you already know

The marshmallow test works like this. A researcher puts one marshmallow in front of a four-year-old and makes an offer: eat it now, or wait alone for a while and get two. In the late 1960s, psychologist Walter Mischel ran versions of it at a Stanford preschool. Years later, his team reported something striking: the kids who waited longer had grown into teens with better test scores.

The pop version of that finding got simpler and meaner with every retelling. The wait became willpower. Willpower became a fixed trait. And the test became a crystal ball that tells you which kind of kid you have.

That phrase — “fixed trait” — is worth slowing down on, because the rest of this turns on it. A fixed trait is something your kid just has, in about the same amount on Tuesday as on Saturday, at the birthday party as at the dinner table. Eye color is a fixed trait. The marshmallow story quietly says patience is one too: that the test reads something permanent in your child, and the number it comes back with is who she is.

The alternative isn’t that self-control doesn’t exist. It’s that self-control depends on the situation. It’s real, but it rises and falls with the room, the day, who else is there, and what waiting has gotten her before. Same kid, different day, different answer.

That difference decides what you do on a hard afternoon. If patience is fixed, you get two options: be grateful or be resigned. If it depends on the situation, you have something you can actually change.

The bigger study told a different story

In 2018, a research team reran the test at scale in Psychological Science . The original Stanford follow-ups tracked fewer than 50 children, most of them professors’ kids. The new team used 918 children from a large national child-care study. Their headline test focused on 552 kids whose mothers had not finished college.

The wait did predict something. An extra minute of waiting at age four went with a small bump in achievement at age fifteen. But the link was about half the size of the original.

And when the researchers controlled for family background, early cognitive ability, and the home environment, it shrank by another two thirds, to nearly nothing. Links to behavior problems at fifteen? Small, and they rarely held up at all.

The marshmallow was never measuring something permanent inside your child. It was largely measuring the world your child came from, and how much reason she had to trust that a second marshmallow was actually coming. The authors say almost exactly this: hidden factors behind a child’s ability to wait, not the waiting itself, likely drove the long-run results.

Twenty seconds, not twenty minutes

In the same paper, most of the long-run link for those 552 kids came from being able to wait just twenty seconds. Past that first moment, extra minutes of waiting barely moved the needle. “It surprised us,” the authors admitted.

Think about what that does to the mental movie in your head. “My kid caved at minute three, and my neighbor’s kid held out until minute six.” The data says that distinction barely matters. Whatever the test picks up shows itself in the first few breaths. It is not a deep well of grit that some kids have and yours lacks.

One broken promise cut waiting time by 75 percent

If wait times track trust, you should be able to move them by changing the trust. In 2013, researchers at the University of Rochester did exactly that .

Before the marshmallow task, an adult promised each child better art supplies. For half the kids, the adult came through. For the other half, the adult shrugged and broke the promise.

Then came the marshmallow. Kids who had just met a reliable adult waited about 12 minutes. Kids who had just been let down waited about 3. Same age, same task, same marshmallow, and a fourfold difference in “willpower,” built in one interaction.

The study was small (28 preschoolers in one lab), so treat it as proof of the mechanism, not a precise national number. But the mechanism is hard to unsee.

The kid who grabs the treat may not be failing the test. She may be reading the room correctly. If adults in her world break promises, eating the sure thing now is the smart bet.

Even the inventor’s own data said no

In 2020, a team that included Mischel himself went back to his original preschoolers (113 of them, now middle-aged) and checked about a dozen adult outcomes: net worth, debt, smoking, diet and exercise. One of Mischel’s co-authors on that paper put it flatly: “With the marshmallow waiting times, we found no statistically meaningful relationships with any of the outcomes that we studied.”

Mischel spent his last years pushing back on the myth his test spawned. His words: “the idea that your child is doomed if she chooses not to wait for her marshmallows is really just a serious misinterpretation.”

Now, self-control is not nothing. In the 2018 data, a separate self-control measure still explained about a quarter of the achievement link, and a patient kid may still have an edge over an equally situated impatient one. It’s substantially about circumstance, not all about circumstance.

Nothing in this research shows that any activity trains willpower, either. The authors say the opposite: programs that only teach kids to wait will probably change very little. Anyone selling you willpower training, including a martial arts gym, is ahead of the evidence.

Labels change how you parent

This isn’t just a debate about words. Treating an ability as fixed changes behavior — your kid’s, and yours.

In a classic set of six 1998 experiments, Carol Dweck and a colleague praised fifth-graders in one of two ways after a puzzle task: “you must be smart” or “you must have worked hard.” After a later failure, the kids praised for being smart persisted less, enjoyed the task less, and performed worse. They had also come to describe intelligence as a fixed thing you either have or don’t.

“He’s just impulsive” works the same way. The kid hears a verdict about who he is. And you stop scanning the situation (hungry, tired, new place, adults who flaked on him twice this week) because the verdict already explained everything.

What actually helps

If self-control depends on the situation, then the useful question is never “what kind of kid is this?” It is “what is this situation telling my kid?” Kids are always running the math: does waiting pay off here, do these adults keep their word, is trying again worth it? That is good news. You can change a situation. You can’t change a verdict.

It is also why we run the kids room the way we do at Journey . We aim to be sturdy adults: warm and connected, but the grown-up stays in charge and keeps their word. That’s the style we laid out in our take on gentle parenting .

Kids classes follow the same shape every time: a short warmup, technique, drilling, then sparring. Belt promotions come when the skill is there, with no testing fees and no surprise rules.

We won’t tell you mat time trains willpower. What a steady room full of reliable adults gives a kid is simpler: a place where waiting, trying, and losing all pay off the way an adult said they would.

In our coaches’ experience, the “impulsive” kid usually looks different within a few weeks of that. Not because he changed — because the room did. It’s one reason kids tend to stay in this sport after quitting others.

If you want to see that room for yourself, our two-week kids intro is $49 and includes a free uniform. Madison, WI parents can also read what most parents worry about before a first class . The worries are more common than you think.

Try this week

Keep one small promise loudly. Say “five more minutes, then we leave,” and leave at five. Then name it: “I said five, and it was five.” You are depositing into the exact account the Rochester study measured.

Swap “what kind of kid is this” for “what’s going on right now.” When your kid melts down, skip “why is he like this” and ask what is different today. Hungry? New place? A canceled playdate? You will find a lever more often than not.

Praise the retry, not the trait. “You went back and tried it again” beats “you’re so smart.” The Dweck study is thirty years of receipts on this one.

The four-year-old who grabbed the marshmallow was not showing you her character. She was showing you her odds. You can’t argue a kid out of her odds. But you can change them.