The Evidence on Teaching

Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention

Cepeda, N. J., Vul, E., Rohrer, D., Wixted, J. T., & Pashler, H. · 2008

grade Bquasi-experimentindependentunreplicatednumbers spot-checked
Sample
1,354 online participants; 32 facts each
Population
Adults, online panel; obscure trivia facts (proximal aligned recall/recognition), not curriculum.
Design
Large parametric experiment — the best source for the optimal-gap parameter. Crossed 26 gap x retention-interval cells; final tests to 1 year.
Key findings
Best quantitative source for spacing parameters: the optimal inter-study gap is non-monotonic (inverted-U) and scales to the target retention interval — roughly 10-20% of the desired retention interval. The spacing benefit is smaller (but reliable) at short intervals and large at long delays (recall gain peaks ~111% at a 70-day interval).
Genetic confound
Low: within-subject gap manipulation isolates scheduling from ability.
Replication notes
Numbers corrected per audit (earlier draft mislabeled recognition gains as recall and aggregate as 350-day). The 10-20%-of-RI rule is widely cited.
DOI / URL
10.1111/j.1467-9280.2008.02209.x

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Optimal gap vs 0-day gap at 350-day (1yr) retention interval% gain; drecall +77% (d=0.9); recognition +60% (d=2.1)researcher-designeddelayed 1yr; lab/onlineactive-alternativeunder-1yrdomain-skill
RECALL gain by retention interval (7/35/70/350 days)% recall gain10 / 59 / 111 / 77% — NON-monotonic, peaks at 70 daysresearcher-designedimmediate-to-1yr; labactive-alternativeunder-1yrdomain-skill
Optimal gap as proportion of retention interval% of RI~20% at 1-week RI declining to ~5-10% (7%) at 1-year RIresearcher-designeddelayed; labactive-alternativeunder-1yrdomain-skill

Cited by

Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention · The Evidence on Teaching