The Evidence on Teaching

Mastery Learning Reconsidered (best-evidence synthesis)

Slavin, R. E. · 1987

grade Breviewindependentreplicatednumbers spot-checked
Sample
16-17 studies meeting criteria (from hundreds screened)
Population
US K-12 group-based mastery learning, >=4-week studies.
Design
Best-evidence synthesis with a priori criteria (>=4 weeks, >=2 classes/arm, initial equivalence, outcomes taught in both arms). Splits results by measure type and equal vs extra time — the decisive analysis.
Key findings
THE cleanest demonstration of the researcher-vs-standardized measure gap in all of education: group-based mastery learning produces ~+0.24 on tests of the objectives taught and ~+0.04 (nothing) on standardized measures, at equal time. The famous mastery effects are substantially an artifact of refocusing instruction on tested objectives.
Genetic confound
Equal-time studies show no leveling on broad measures; group-based mastery is a 'Robin Hood' shift of teacher time toward low achievers, not a reduction of ability differences.
Replication notes
Robustly CONFIRMED, not contested: Kulik 1990's own within-study standardized effects are 0.04/0.09/0.07/-0.05, and the modern at-scale EEF Maths Mastery RCT pooled +0.073. No equal-time standardized-test counterexample was ever produced.
DOI / URL
10.3102/00346543057002175

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Standardized tests, 7 equal-time studiesmedian ES+0.04; NOT ONE significantstandardizedend of treatmentbusiness-as-usualend-of-treatmentdomain-skill
Experimenter-made tests, 8 equal-time studiesmedian ES+0.24researcher-designedend of treatmentbusiness-as-usualend-of-treatmentdomain-skill
Extra corrective time (4 studies)median ES+0.31 immediate -> retention median -0.03 (gone by 4-12 weeks)researcher-designed4-12wk retentionbusiness-as-usualunder-1yrdomain-skill
Anderson/Scott/Hutlock within-study gapES+0.64 experimenter-made vs +0.04 California Achievement Test IN THE SAME STUDYmixedend of yearbusiness-as-usualend-of-treatmentdomain-skill

Cited by