The iconic Marshmallow Test, a simple experiment where children are asked to delay gratification for a bigger reward, has long been touted as a window into future success. For decades, research suggested that children who could resist the temptation of eating a single treat immediately, in favor of a larger reward later, were more likely to enjoy better academic, social, and health outcomes as adults. However, a new study published in Child Development challenges this assumption, casting doubt on the predictive power of the Marshmallow Test when it comes to adult achievement, health, and behavior.
Researchers from Teachers College, Columbia University, and the University of California, Irvine, set out to reexamine the long-term implications of the Marshmallow Test, specifically whether a child’s ability to delay gratification at age four could reliably predict adult outcomes at age 26. Their findings suggest that, contrary to popular belief, a preschooler’s response to the Marshmallow Test does not strongly predict their success or well-being in adulthood, especially when other factors are accounted for.
The Marshmallow Test has been a cornerstone in psychological research since the late 1960s, largely due to the work of psychologist Walter Mischel. His studies famously linked the ability to delay gratification with a variety of positive life outcomes, such as higher SAT scores, better social skills, and lower body mass index (BMI). These findings have influenced both scientific thought and public policy, with early childhood interventions often designed to enhance self-control and delay of gratification, under the belief that these traits are critical for later success.
However, more recent studies have begun to question the robustness of these findings, suggesting that the predictive power of the Marshmallow Test may have been overstated. This new study aimed to provide a more rigorous analysis by using a larger and more diverse sample than the original studies and by controlling for a variety of factors that could influence both a child’s ability to delay gratification and their later outcomes.
“We were interested in this topic as an extension of previous work from our team (Watts et al., 2018), which examined the association between the Marshmallow Test and achievement and behavior at age-15,” said study author Jessica Sperber, a doctoral candidate in developmental psychology at Columbia University.
“With the present study, we sought to ask ‘does the Marshmallow Test predict important measures of well-being and functioning at age-26?’ So, we extended the previous work to examine longitudinally if a childhood measure of delayed gratification could directly predict a wide range of measures that reflect adult functioning – including achievement, health, and behavior.”
The study analyzed data from the National Institute of Child Health and Human Development (NICHD) Study of Early Child Care and Youth Development (SECCYD), a long-term project that tracked over 1,300 children from birth into adulthood. For this specific analysis, researchers focused on 702 participants who had completed a version of the Marshmallow Test at 54 months old (around four and a half years) and provided data on various adult outcomes at age 26.
In the Marshmallow Test, each child was presented with a treat and given the option to either eat it immediately or wait until the experimenter returned, at which point they would receive a larger reward. The researchers then collected data on the participants at age 26, including their educational attainment, annual earnings, levels of debt, BMI, symptoms of depression, substance use, and behaviors related to risk-taking and impulsivity.
The study employed a series of statistical models to assess whether the time each child waited in the Marshmallow Test predicted their adult outcomes. These models also accounted for various demographic factors, such as socioeconomic status, as well as early childhood cognitive and behavioral characteristics.
While there were modest correlations between waiting time on the Marshmallow Test and certain outcomes, such as educational attainment and BMI, these associations largely disappeared when the researchers controlled for other factors like family background and early childhood environment. For example, the initial link between delaying gratification and higher educational attainment was no longer significant after adjusting for these variables.
“We were initially surprised by the lack of bivariate correlations between delay of gratification and adult outcomes,” Sperber told PsyPost. “Surprisingly, the Marshmallow Test did not predict adult behaviors that one might expect would be highly related to the early ability to delay gratification, like impulse control and risk-taking behaviors. When we did see some significant correlations, they were diminished entirely by controlling for basic demographic and background information of the participants.”
In addition, the researchers found no significant associations between Marshmallow Test performance and other adult outcomes, such as annual earnings, depressive symptoms, substance use, impulsive behavior, or risk-taking behaviors. Even when the researchers examined whether the Marshmallow Test predicted adult outcomes differently for children from lower socioeconomic backgrounds or based on sex, no consistent patterns emerged.
The findings also raise important questions about the construct validity of the Marshmallow Test. If the test primarily reflects other factors, such as early cognitive abilities or family environment, rather than a unique capacity for self-control, its utility as a predictor of long-term success may be limited.
“Performance on the Marshmallow Test has been framed as a key predictor of life-long well-being and success. With this study, we found that this doesn’t appear to be the case: the Marshmallow Test in early childhood does not reliably predict adult functioning,” Sperber said.
One of the few exceptions was BMI: children who waited the full seven minutes tended to have a lower BMI at age 26 compared to those who waited less time. However, this finding was somewhat isolated, as no other health-related outcomes, such as substance use or depressive symptoms, were linked to Marshmallow Test performance.
“I think we have learned that all measures of ‘self-control’ are not created equally,” added co-author Tyler Watts, an assistant professor in human development at Columbia University. “There has been other convincing work showing that very broad assessments of self-regulation in childhood can predict important life outcomes. So, our conclusions seem to be restricted to the Marshmallow Test.”
Despite its rigorous design, the study does have limitations. The sample, while larger and more diverse than Mischel’s original cohort (which was limited to children from the Stanford University community), was still not nationally representative, as it was predominantly White and middle-class. Additionally, the maximum waiting time was capped at seven minutes, a shorter duration than the original studies conducted by Mischel.
“It is important to point out that the version of the test used in our study was capped at seven minutes, which is a much shorter ceiling than what other studies have often used,” Watts noted. “We looked into this issue as best we could and did not see evidence that it was a major problem, but it is certainly a limitation worth noting.”
The study provides a critical reappraisal of the Marshmallow Test, challenging the long-held belief that a child’s ability to delay gratification is a strong predictor of adult success. The researchers suggest that interventions aimed at improving children’s self-control should take a broader approach, focusing on enhancing overall cognitive and behavioral skills rather than narrowly targeting the ability to delay gratification.
“Our lab is broadly interested in the ways that interventions targeting child cognitive and behavioral skills affect long-term trajectories,” Watts explained. “Longitudinal studies, like the famous Marshmallow Test work, often suggest that targeting early skills should generate important long-term effects. In other words, if we were to see a strong correlation between early delay of gratification and later educational attainment, would this mean that a program that targeted early delay of gratification would also have long-term effects on educational attainment? Such predictions are at the heart of early intervention research.
“To the contrary, the field has accumulated evidence of fadeout for long-term intervention effects. Our team recently published a meta-analysis of a large set of educational intervention studies showing that interventions targeting social- emotional and cognitive skills usually have fading effects in the medium term. In other words, intervention effects on skills like self-regulation do not tend to last very long in most studies. We have been trying to understand why this happens, and whether we can better design programs that will have long-lasting effects on developmental trajectories.”
The study, “Delay of gratification and adult outcomes: The Marshmallow Test does not reliably predict adult functioning,” was authored by Jessica F. Sperber, Deborah Lowe Vandell, Greg J. Duncan, and Tyler W. Watts.
