The Evidence on Teaching

Research on Direct Instruction: 25 Years Beyond DISTAR

Adams GL, Engelmann S · 1996

grade Cmeta-analysisdeveloper-ledmixed
Sample
~34 studies / ~173 comparisons
Population
K-12, general + special education; reading/math/language/spelling.
Design
Proponent-authored (Engelmann created Direct Instruction) — strong allegiance/publication-bias risk. Superseded in scope by Stockard 2018, which is recorded separately. A book (Educational Achievement Systems, Seattle) with no DOI; the stable locator is the ERIC catalogue record ED413575. The DOI previously recorded here belonged to Stockard 2018, not to this book.
Key findings
Reports large DI effects (~0.87), frequently quoted as headline DI evidence, but proponent-authored and ~0.3 SD above the more inclusive Stockard 2018 (0.54). Use Stockard as the primary estimate.
Genetic confound
Domain-achievement gains; no durable g.
Replication notes
Same direction as Stockard 2018 but ~0.3 SD larger — treat as a proponent UPPER BOUND, not the primary estimate.

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Overall DI vs comparison instructiond~0.87 (large; ~0.3 SD above Stockard 2018)mixedmostly immediatebusiness-as-usualend-of-treatmentdomain-skill

Cited by