Did the marshmallow test really predict who would succeed in life?

Only weakly, and much less than the legend says. In the famous studies, preschoolers offered one marshmallow now or two later were followed for years, and those who waited longer tended to do better as teenagers. That made the test look like a crystal ball for self-control.

Then researchers tried a conceptual replication with data on more than 900 children from a geographically diverse national study. Among children whose mothers had not completed college, each extra minute waited at age 4 predicted about a tenth of a standard deviation more achievement at 15. But that link was only half the size of those in the original studies, and shrank by two thirds once family background, early cognitive ability and the home environment were controlled for. Links with behavior at 15 were much smaller and rarely statistically significant.

Two other findings help explain why. Most of the gain came from children who could wait at least 20 seconds, so the difference was between not waiting at all and waiting a bit, not between tiny and heroic patience. And in a small study, children who had experienced a reliable experimenter waited far longer than those who had experienced an unreliable one, which suggests that waiting also reflects whether a child believes waiting will pay off. The test measures something, but not a fixed trait that decides a life.