Ever Failed, Try Again, Succeed Better: Results from a Randomized Educational Intervention on Grit

More than 3,200 fourth-graders (about 3,500 including attrition) in 52 state elementary schools in Istanbul, across two independent samples

Turkey; deliberately low-socioeconomic-status state primary schools; children around age 10.

The only trial located anywhere that raised a grit-like construct AND raised objective achievement, so it deserves careful reading. School-level randomisation, registered in the AEA trial registry (AEARCTR-0003317 and -0003327), permutation inference, balanced attrition (p = 0.935), and long-run follow-ups at 1.5 and 2.5 years. Two independent samples make it an internal replication. It is developer-led - the authors designed the twelve-week curriculum of animated videos, goal-setting and effort-productivity beliefs. And it is arguably not a grit intervention at all but an effort-productivity and mindset curriculum, which moved a behavioural task and a self-report index rather than a validated trait.

On a researcher-administered standardized mathematics test the effect was +0.31 SD in the short run (permutation p = 0.008), +0.19 SD at 1.5 years (p = 0.026) and +0.23 SD at 2.5 years (p = 0.044). Verbal Turkish was +0.13 SD short-run and not significant. TEACHER-GIVEN GRADES showed no significant effect in either sample - the reverse of the growth-mindset pattern, and evidence against a grading artefact. On an incentivised real-effort task treated students were more likely to choose the challenging rewarding option, less likely to give up after failure, and 6 to 8 percentage points more likely to make the payoff-maximising choice. Self-reported grit rose 0.29 and 0.35 SD and self-reported mindset 0.35 and 0.33 SD. No effect on self-confidence, risk tolerance or patience. The authors flag that dropping their rich covariate set makes the sample-2 long-run maths effect fall below conventional significance.

Low. School-level randomisation.

Internally replicated across two independent samples; no independent external replication exists. Its existence is the strongest single argument against a blanket dismissal of non-cognitive curricula, and its developer-led status and construct ambiguity are the reasons it cannot settle the question alone.