The theory that believing ability can be developed, rather than being fixed, improves achievement. It became one of the most widely adopted ideas in education, and the size of its effects has been sharply contested since large replications began appearing.
Carol Dweck's work distinguishes two beliefs about ability. A fixed mindset treats intelligence as a trait a person has in some quantity. A growth mindset treats it as something that develops with effort and instruction.
The predicted consequences follow from how each belief frames failure. If ability is fixed, difficulty is evidence of its limit, so effort is threatening and challenge is best avoided. If ability grows, difficulty is the mechanism of growth, so effort is worth spending and challenge is worth seeking.
The applied finding most often cited concerns praise. Dweck and Claudia Mueller reported in 1998 that children praised for intelligence after success subsequently chose easier tasks and performed worse after a setback than children praised for effort. That study shaped a great deal of practice.
The original effects were large and came from small samples, which is the pattern that later replication difficulties usually follow.
A 2018 meta-analysis by Victoria Sisk and colleagues covering over 350,000 participants found the average correlation between mindset and achievement to be very small, and the average effect of mindset interventions on achievement to be near zero. Effects were detectable in two subgroups: students at academic risk, and students from low income backgrounds.
The National Study of Learning Mindsets, published in Nature in 2019, ran a short online intervention with over 12,000 American ninth graders. It found a small but real effect on grades for lower-achieving students, around a tenth of a grade point, and no effect for higher achievers. It also found the effect depended on the school: it appeared where peer norms supported challenge-seeking and not where they did not.
A 2022 replication in Norway with over 5,000 students found no effect on grades.
Both camps accept these numbers, and disagree about what they mean.
Critics point out that the effects are far smaller than the popular presentation implied, that they are absent in many samples, and that an intervention with an average effect near zero cannot support the policy attention it received.
Defenders point out that a brief online intervention producing any measurable change in grades is not a trivial result, that the theory predicts effects concentrated among students who are struggling and that is where they appear, and that the school-context finding is a specific prediction confirmed rather than a post hoc rescue. Dweck has argued separately that much of what schools implemented was a false growth mindset: telling children to try harder without changing what happens when they fail, which the theory does not endorse.
Whether the effect is worth the attention is a matter of judgement rather than data. A small effect delivered cheaply to many students can be defensible; the same effect used to justify large training programmes is harder to defend.
Whether mindset can be durably changed is open. Most successful interventions are brief and their measured effects fade, and it is not clear that a belief about ability, formed over years, is much moved by a short exercise.
And there is a risk the theory can be turned against the students it was meant to help, by locating the cause of poor outcomes in a child's attitude rather than in their schooling or circumstances. Dweck has warned about this directly.
Growth mindset is not debunked. A real effect exists, it is small, it is concentrated among struggling students, and it depends on context in ways the original presentation did not convey.
The trajectory is a common one: a striking finding from small samples, rapid adoption at scale, then large replications returning a modest and conditional version of it. The modest version may still be worth having; it is not the version that was sold.
