Can Technology Facilitate Scale? Evidence from a Randomized Evaluation of High-Dosage Tutoring (Saga 4:1 variant)
Bhatt, M. P., Guryan, J., Khan, S. A., LaForest-Tucker, M., & Mishra, B. · 2024
grade Brctdeveloper-ledunreplicated
Sample
~4,000 students, Chicago + NYC
Population
9th graders, low-income urban districts.
Design
Cost-reduction variant: 4 students per tutor alternating between human tutor and computer platform, preserving daily in-school dosage. COVID-disrupted later cohorts.
Key findings
Doubling the ratio to 4:1 with a tech rotation retains most of the effect at ~30% lower cost (~$2,500-3,000/student). The binding ingredient looks like daily scheduled in-school dosage with an accountable adult, not the exact ratio.
Genetic confound
Randomized.
Replication notes
Conceptual replication of Guryan 2023 by the same developer team; independent-team scale-ups remain the weak point.
DOI / URL
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Math achievement | SD | +0.23 (vs 0.18-0.40 for the 2:1 model) at ~30% lower cost | standardized | end of year | unclear | end-of-treatment | domain-skill |
Cited by
- Tutoring — the honest effect, the Bloom 2-sigma myth, and what survives scalestrong supportconf: highgc: low