Changing Mindsets: Effectiveness trial — Evaluation Report

101 English primary schools (50 intervention, 51 control), 5,018 Year 6 pupils aged 10-11; analysis sample 4,584 after six schools withdrew; primary-outcome attrition 11.2-11.6%

England; Year 6 (age 10-11) pupils in state primary schools.

The single most decision-relevant record in the growth-mindset literature, and the reason is measurement. Two-arm cluster RCT randomised at school level by an independent NIESR consultant, block-randomised on prior attainment and geography, registered as ISRCTN14702744, with a pre-specified analysis plan. Waitlist control. The intervention was one day of teacher training from the University of Portsmouth Growing Learners team plus an 8-week pupil programme of up to 2.5 hours per week delivered by the pupils own teachers. Crucially the primary outcomes are KS2 national assessments — externally set and externally marked — so this is a growth-mindset trial measured on an independent standardised test rather than on teacher-assigned grades. Macnamara and Burgoyne identify it as one of only TWO studies in the entire literature that is both high quality and free of conflict of interest. MDES 0.19 at randomisation. Cost about GBP 4 per pupil.

Exactly zero, three times over. KS2 maths effect size -0.01 (95% CI -0.04 to 0.01, p = 0.37, n = 4,454), KS2 reading -0.00 (-0.02 to 0.02, p = 0.72, n = 4,437), KS2 grammar/punctuation/ spelling -0.00 (-0.03 to 0.03, p = 0.90, n = 4,448) — nought months additional progress on all three. The disadvantaged (FSM) subgroup, where the theory predicts the largest effect, was also zero on all three outcomes. Every self-reported non-cognitive secondary outcome was null - intrinsic value 0.09 (-0.10 to 0.28), self-efficacy -0.05 (-0.21 to 0.10), test anxiety -0.01 (-0.09 to 0.06), self-regulation 0.05 (-0.11 to 0.21). The process evaluation notes that control schools were already using growth-mindset ideas, so contamination pushes toward the null, but the confidence intervals are tight enough to rule out anything of practical size.

Low. School-level randomisation with block balancing on prior attainment.

This trial IS the scale-up of the EEF's own 2015 Changing Mindsets pilot, whose "+2 months, evidence of promise" on 286 pupils evaporated when the sample grew seventeen-fold and the outcome became a national test. That pilot-to-scale pair is the cleanest illustration in this literature of a small promising result vanishing under scale and better measurement.