A New Zealand cohort tracked from birth to age 32 found that childhood self-control predicted adult health, wealth, and criminal record along a gradient, though a 2018 marshmallow test replication found family background explained much of what looked like individual willpower

A cohort of just over a thousand people born in Dunedin, New Zealand, between 1972 and 1973 has been followed by researchers for more than five decades, and one finding from that study keeps resurfacing whenever the subject of childhood character comes up. In a 2011 paper in the Proceedings of the National Academy of Sciences, Terrie Moffitt and colleagues reported that self-control measured in these individuals between the ages of three and eleven, using assessments from parents, teachers, observers, and the children themselves, predicted their physical health, personal finances, and likelihood of a criminal conviction at age 32.

The finding was not simply that children with very poor self-control did worse than children with very good self-control. The researchers described a gradient: each incremental step up in a child’s measured self-control corresponded to somewhat better adult outcomes, across the entire range of the sample, not just at the extremes. Children who ranked in the bottom fifth on self-control were, decades later, more likely to have health problems, credit and savings difficulties, and criminal records than children who ranked just above them, who were in turn more likely to struggle than children ranked above them again.

What was actually being measured

It is worth being precise about what “self-control” meant in this research, because it is doing a lot of the work behind the popular idea of raising a child with strong character. The measure combined observations of impulsive aggression, lack of persistence, inattention, and low tolerance for frustration, gathered from multiple people who knew the child in different contexts, rather than a single test or a single rater’s impression. That breadth is part of why the finding has held up as one of the more credible pieces of evidence connecting early childhood behaviour to adult outcomes decades later.

This is one long-running study, not a universal law about every child in every context. The Dunedin cohort was drawn from a single city in New Zealand across a narrow birth window, and the researchers themselves note that translating the finding into practical advice for any individual family requires caution about how far a population-level gradient applies to one particular child.

Why the marshmallow test needs a different reputation than it has

The other study that shaped the popular idea of childhood character, the marshmallow test, deserves a more complicated reputation than it usually gets. The original experiments, from Walter Mischel and colleagues, found that preschoolers who could wait longer before eating a marshmallow tended to do better later in adolescence.

A 2018 paper in Psychological Science, by Tyler Watts, Greg Duncan, and Haonan Quan, retested this claim with a larger and more socioeconomically diverse sample than the original studies used. They found a similar pattern at first glance: children who waited longer at age four did somewhat better at age fifteen. But once the researchers statistically accounted for family background, including the mother’s education, household income, and the home environment, the size of that relationship shrank by roughly two thirds. Much of what looked like an effect of a child’s willpower turned out to be better explained by the economic and family circumstances the child was already growing up in.

This is one replication, using one dataset and one set of statistical controls, and it does not erase the original finding so much as complicate what it actually shows. Waiting for the second marshmallow was never purely a measure of an innate trait separate from a child’s environment. Family stability, financial predictability, and the reliability of a caregiver’s past promises all shape how much sense it makes for a child to wait for something better later rather than take something certain now.

A theory about how character gets built, not yet a settled fact

If self-control in early childhood really does predict adult outcomes this durably, and if it is shaped substantially by circumstances rather than being purely innate, the natural next question is what parents or caregivers can actually do that helps. Here the evidence gets thinner and more theoretical.

A widely cited 1994 paper in Developmental Psychology, by Joan Grusec and Jacqueline Goodnow, proposed a framework for how children come to internalise values rather than simply comply with rules to avoid punishment. Their model argues that discipline built around explaining the reasoning behind a rule, rather than relying purely on punishment or reward, is more likely to be accurately understood by a child and accepted as the child’s own value, rather than experienced as an external demand to be dropped the moment enforcement disappears.

Grusec and Goodnow were explicit that this was a theoretical reconceptualisation of existing research, not a new experiment with its own dataset, and the framework itself was offered as a set of testable mechanisms rather than a confirmed causal chain. Later researchers in this area have continued to test and refine pieces of it. The honest summary is that induction-based discipline, explaining reasons rather than only enforcing consequences, is a well-regarded theory about how internalised character forms, not a settled experimental result with the same evidentiary weight as the Dunedin cohort’s decades of follow-up data.

What this leaves out, and what it does not

None of this supports treating “strong character” as something a child either has or lacks by nature, nor does it support the opposite claim that character is entirely a product of circumstance with no role for individual behaviour. The Dunedin data shows a durable, graded relationship between early self-control and later outcomes across a large population, tracked with unusual rigour over an unusually long time. The marshmallow replication shows that a substantial share of what looks like individual willpower in early childhood is entangled with the family and economic conditions a child is already living inside. The Grusec and Goodnow framework offers a plausible, still-developing account of the parenting mechanism that might connect the two, without yet being confirmed with the same weight of evidence as either study.

Put together, the research supports something narrower and more honest than either “character is built by strict parenting” or “character is mostly circumstance.” It supports treating self-control as a real, measurable, and consequential pattern in childhood, one that current evidence suggests is shaped by both a child’s early environment and, plausibly, the reasoning a caregiver offers behind the rules that child grows up with.