UK National Tutoring Programme: Year 2 and Year 3 Impact Evaluations (NFER/DfE)
National Foundation for Educational Research, for the Department for Education · 2024
grade Cquasi-experimentindependentreplicatednumbers spot-checked
Sample
National Year 6 and Year 11 cohorts (£1B+ programme; 232k+ pupils in year 1)
Population
England state schools; KS2 and KS4 national exams.
Design
National Pupil Database matched/regression analyses. Dilution and negative-selection biases push estimates down; residual selection could fabricate small effects. Route comparison: school-led tutoring vs external tuition partners vs academic mentors.
Key findings
The largest tutoring dataset in existence: at true national scale, tutoring bought at most ~1 month of extra progress (school-led maths), nothing via external vendors, ~0 at GCSE, and ~no persistence a year later. School-integrated staffing beat vendors. National-scale effects are an order of magnitude below efficacy claims — on independent national exams with zero measure inflation.
Genetic confound
Aggregate national data; selection biases acknowledged in both directions.
Replication notes
Same pattern three consecutive years; Victoria (Australia) statewide tutoring initiative also null.
DOI / URL
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| KS2 maths (year 3, school-led route) | g | +0.063 all pupils (0.041-0.085); +0.094 previously-low-attaining; ~1 month progress | standardized | end of year | business-as-usual | end-of-treatment | domain-skill |
| KS2 English | g | ~+0.03 (sig only for disadvantaged/low-attaining) | standardized | end of year | business-as-usual | end-of-treatment | domain-skill |
| KS4 (GCSE) | g | 0.014-0.016, mostly ns | standardized | end of GCSE year | business-as-usual | end-of-treatment | domain-skill |
| External vendor routes (Tuition Partners/Academic Mentors) | g | no improvement; small negatives down to -0.065 | standardized | end of year | business-as-usual | end-of-treatment | domain-skill |
| Persistence 1 year later | g | ~none (sub-1-month maths trace only; vendor-route negative English effects persisted) | standardized | 1yr+ follow-up | business-as-usual | 1-2yr | domain-skill |
Cited by
- Tutoring — the honest effect, the Bloom 2-sigma myth, and what survives scalestrong supportconf: highgc: low