Not Too Late: Improving Academic Outcomes Among Adolescents (Saga/Match high-dosage tutoring, two Chicago RCTs)
Guryan, J., Ludwig, J., Bhatt, M. P., Cook, P. J., et al. · 2023
grade Arctindependentmixednumbers spot-checked
Sample
Study 1 N=2,633; Study 2 N=2,710 (internal replication)
Population
Low-income minority 9th-10th graders, Chicago public schools, 2-5 years behind in math.
Design
Two large individually-randomized trials; 2:1 tutoring, 55 min/day as a scheduled class period (~140-165 hrs/yr); paraprofessional tutors (~$17k stipend). Published AER 113(3) — numbers below are the published version (the working paper's were slightly lower).
Key findings
The best evidence in the tutoring literature: replicated large RCTs, independent standardized outcomes, and measured persistence (+0.23 SD 1-2 years later). Daily in-school 2:1 tutoring roughly doubles-to-triples a year of normal high-school math learning at $3,200-4,800/student (~1/5 of per-pupil spending). Also proves the 'too late after early childhood' claim wrong.
Genetic confound
Randomized. Gains are taught-domain math skills; low-performing-sample SD units modestly inflate d vs population.
Replication notes
Internal replication succeeded. Independent scale-ups shrink (Nashville 0.00-0.09; UK NTP ~0.03-0.09).
Cost
$3,200-4,800/student/yr during the study (later ~$1,800 with AmeriCorps subsidy); CPS spends ~$17,000/pupil.
DOI / URL
10.1257/aer.20210434
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Math, Study 1 | SD (ACT EXPLORE/PLAN) | ITT 0.09 / TOT 0.18 | standardized | end of year 1 | business-as-usual | end-of-treatment | domain-skill |
| Math, Study 2 (replication) | SD | TOT 0.40 (abstract range 0.18-0.40 across tests/norming) | standardized | end of year 1 | business-as-usual | end-of-treatment | domain-skill |
| Persistence: 11th-grade math tests / math GPA | SD / GPA pts | +0.23 SD / +0.25 (1-2 years post — rare persistence evidence, ~70% retained) | standardized | 1-2yr follow-up | business-as-usual | 1-2yr | domain-skill |
| On-time graduation | pp | +1.3 (imprecise, ns) | standardized | 2yr+ follow-up | business-as-usual | over-2yr | attainment |
Cited by
- Tutoring — the honest effect, the Bloom 2-sigma myth, and what survives scalestrong supportconf: highgc: low