Effectiveness of Mastery Learning Programs: A Meta-Analysis
Kulik, C.-L. C., Kulik, J. A., & Bangert-Drowns, R. L. · 1990
grade Cmeta-analysisindependentreplicatednumbers spot-checked
Sample
108 studies (72 college PSI + 36 group-based LFM)
Population
Mostly college (91/108); K-12 LFM concentrated in upper elementary-high school.
Design
The standard 'mastery d=0.5' citation — but mostly college self-paced PSI on instructor-made finals; 75/103 nonrandom. Its own moderator table is the tell.
Key findings
The pro-mastery meta that CONFIRMS Slavin despite itself: overall 0.52, but its own within-study comparisons show standardized-test effects of ~0.04-0.09, and 'standardized' is a significant negative moderator in their regression. Also documents the fast-learner cost: mastery gains concentrate in low-aptitude students while high-aptitude students gain little and wait.
Genetic confound
Variance compresses (ratio 0.77) mainly on aligned tests — redistribution of instructional time, not reduction of underlying differences.
Replication notes
Numbers verified. On shared K-12 studies, Kulik and Slavin nearly agree; the headline gap is inclusion rules (college PSI, brief studies) and test type.
DOI / URL
10.3102/00346543060002265
Effects
| Outcome | Metric | Value | Measure | Timing | Vs | Horizon | Class |
|---|---|---|---|---|---|---|---|
| Final exam, all studies | ES | 0.52 (SE .033) | mixed | end of instruction | business-as-usual | end-of-treatment | domain-skill |
| By test type | ES | locally-developed 0.57 vs STANDARDIZED 0.29 (K-12 LFM standardized: 0.10 from 2 studies) | mixed | end of instruction | business-as-usual | end-of-treatment | domain-skill |
| Within-study standardized-vs-local (4 studies) | ES | standardized 0.04/0.09/0.07/-0.05 vs local 0.64/0.41/-0.19/0.09 — 'trivially small' | mixed | end of instruction | business-as-usual | end-of-treatment | domain-skill |
| By aptitude | ES | low-aptitude 0.61 vs high-aptitude 0.40 — gains concentrate at the bottom | mixed | end of instruction | business-as-usual | end-of-treatment | domain-skill |
Cited by
- Mastery learning (teach → test → reteach to criterion → advance)mixedconf: highgc: low