Educational Outcomes of Tutoring: A Meta-Analysis of Findings
Cohen, P. A., Kulik, J. A., & Kulik, C.-L. C. · 1982
grade Cmeta-analysisindependentreplicatednumbers spot-checked
Sample
65 studies (52 with tutee achievement outcomes)
Population
US elementary/secondary; tutees skew low-ability; mostly peer/paraprofessional structured programs.
Design
Published two years BEFORE Bloom's essay; Bloom did not engage with it. Moderator table verified from full text.
Key findings
The contemporaneous honest number Bloom ignored: typical school tutoring = 0.40 SD, 3x larger on author-built than standardized tests, and inflated by short study durations. Bloom's 2.0 sat at the far tail of an already-published distribution.
Genetic confound
Low-ability restricted samples shrink SD denominators, inflating d vs population units.
Replication notes
Direction/magnitude replicated by modern RCT metas (Nickow 0.29-0.37).
DOI / URL
10.3102/00028312019002237
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Tutee achievement, pooled | ES | 0.40 (45/52 positive) | mixed | end of treatment | business-as-usual | end-of-treatment | domain-skill |
| Measure alignment | ES | instructor-developed tests 0.84 vs standardized 0.27 — a 3x gap | mixed | end of treatment | business-as-usual | end-of-treatment | domain-skill |
| Duration | ES | 0-4 wks 0.95; 5-18 wks 0.42; 19-36 wks 0.16 — short hothouse studies inflate | mixed | end of treatment | business-as-usual | end-of-treatment | domain-skill |
| Two-sigma frequency | count | 1 of 65 studies >=2.0 (a 32-student dissertation) | mixed | end of treatment | none | end-of-treatment | domain-skill |
Cited by
- Tutoring — the honest effect, the Bloom 2-sigma myth, and what survives scalestrong supportconf: highgc: low