Reading First Impact Study: Final Report (NCEE 2009-4038)
Gamse, B. C., Jacob, R. T., Horst, M., Boulay, B., & Unlu, F. · 2008
grade Bquasi-experimentindependentunreplicated
Sample
~250 schools (regression discontinuity)
Population
US grades 1-3 in high-poverty schools; the $1B/yr federal phonics-centric program.
Design
Regression-discontinuity on the funding-assignment cutoff. Largest at-scale implementation of NRP-style systematic phonics.
Key findings
At-scale NRP-aligned phonics raised grade-1 decoding (+0.17 SD) but produced NO significant effect on reading comprehension in grades 1, 2, or 3. The strongest single at-scale datum that phonics-centred instruction moves decoding, not comprehension.
Genetic confound
RD design gives clean local causal identification.
Replication notes
Cleanest large-scale demonstration of the decoding-yes/comprehension-no pattern.
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Grade-1 decoding (TOSWRF) | SD | +0.17 (significant) | standardized | end-of-treatment | business-as-usual | end-of-treatment | domain-skill |
| Reading comprehension, grades 1-3 (SAT-10) | SD | no significant effect | standardized | end-of-treatment | business-as-usual | end-of-treatment | domain-skill |
Cited by
- Systematic/explicit phonics vs whole language and balanced literacystrong supportconf: highgc: low