The Scaling Dynamics and Causal Effects of a District-Operated Tutoring Program (Metro Nashville)
Kraft, M. A., Edwards, D. S., & Cannata, M. · 2024
grade Bquasi-experimentindependentreplicated
Sample
>6,800 tutored students over 2.5 years (pilot RCT + cluster RCT + DiD)
Population
Metro Nashville grades 1-8, high-poverty urban district.
Design
The best-implemented district-run program: paid district teachers as tutors, standards-aligned, but modest dosage (12-18 hrs/yr vs Saga's ~150).
Key findings
Even with unusually good implementation, a district program at ~$1,500/student/yr and 12-18 delivered hours produced reading-only 0.04-0.09 and math zero. Dosage is the binding constraint — the efficacy numbers assume 100+ hours that district programs rarely deliver.
Genetic confound
Minimal.
Replication notes
Consistent with the Kraft-Schueler-Falken 1000+-student bin and with the Carbonari district nulls.
Cost
~$1,500/student/yr; tutor stipends = 80% of cost.
DOI / URL
10.26300/zcw7-4547
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Reading | SD | +0.04 to +0.09 (DiD 0.09 sig) | standardized | end of year | business-as-usual | end-of-treatment | domain-skill |
| Math | SD | ~0.00 (null in all designs) | standardized | end of year | business-as-usual | end-of-treatment | domain-skill |
Cited by
- Tutoring — the honest effect, the Bloom 2-sigma myth, and what survives scalestrong supportconf: highgc: low