The marshmallow test, in which a child chooses between one treat now and two after a delay, is among psychology's most cited paradigms. A well-known 1990 study reported that greater delay of gratification in early childhood predicted better adolescent outcomes, a finding that hardened in popular culture into the claim that childhood willpower drives later success.
A 2018 conceptual replication by Watts, Duncan, and Quan re-examined this with a larger, more socioeconomically diverse sample of roughly 900 children. The headline result was a substantial attenuation of the original association. The bivariate correlation between delay and later achievement was about half the magnitude reported earlier, and it diminished by approximately two-thirds after controlling for family background and early cognitive ability. For children whose mothers had not completed college, an additional minute of waiting corresponded to only a small predicted gain in achievement, with negligible associations to later behavioral outcomes.
The interpretive upshot is that delay behavior may substantially index a child's environment and resources rather than an isolated trait of self-control. A child in a more materially secure and predictable household may more readily trust that a promised reward will materialise. Methodologically, the case illustrates several recurring pitfalls: over-generalising from a small early sample, popularising a simple causal narrative from a correlation, and neglecting confounders. Controlling for family background does not prove willpower is irrelevant, but it shows that much of the original association is better explained by co-varying socioeconomic factors than by delay ability alone.