The Evidence on Teaching

The Impact of Feedback on Student Attainment: A Systematic Review (EEF/EPPI-Centre)

Newman, M., Bird, K., Kwan, I., Shemilt, I., Richardson, M., & Hoo, H. · 2021

grade Bmeta-analysisindependentunreplicated
Sample
Risk-of-bias-coded, RCT-heavy, K-12 attainment-only synthesis
Population
K-12 students; school attainment outcomes only.
Design
The only risk-of-bias-coded, school-age-only feedback synthesis — the best current anchor. Grades effects by study rigor.
Key findings
The honest number for feedback in schools: small-but-real ~0.17 overall, falling to 0.07 in the least-biased studies, near-zero for secondary students, positive mainly in early primary and math. Feedback is not zero — but it is nothing like the 0.5-0.8 folklore.
Genetic confound
RCT-heavy; low confounding.
Replication notes
Confirms the deflation thesis with a clean bias gradient: the more rigorous the study, the smaller the effect (0.62 → 0.20 → 0.07).

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Overall (low/moderate risk-of-bias studies)g0.17 [0.09,0.25]standardizedend-of-treatmentunclearend-of-treatmentdomain-skill
LOW risk-of-bias studies only (n=4)g0.07 [0.00,0.14]standardizedend-of-treatmentunclearend-of-treatmentdomain-skill
Serious risk-of-bias studiesg0.62 — rigor drives the number downmixedend-of-treatmentunclearend-of-treatmentdomain-skill
Secondary school / ages 14-16g0.05 ns / -0.04 (all constituent studies negative)standardizedend-of-treatmentunclearend-of-treatmentdomain-skill
Early primary (KS1) / mathg0.34 (mostly one research group) / 0.25standardizedend-of-treatmentunclearend-of-treatmentdomain-skill

Cited by