Spaced (distributed) practice

Verdict

If you have a fixed amount of study time, spreading it out beats cramming it — for durable retention, reliably, across ages and materials. This is the most robust single finding in the learning sciences (over a century of evidence; in the anchor meta only 12 of 271 comparisons were null or negative), it is Dunlosky's other "high utility" technique alongside retrieval, and like retrieval it is genetic-confound-clean (the core studies manipulate schedule within learners, so ability can't explain it). It is a genuine free lunch: same total effort, more durable learning, purely from when you schedule.

The scope: the benefit is delay-dependent (it's small when you're tested immediately, large when you're tested weeks or months later), the real-classroom magnitude (d≈0.54) is well below lab estimates, and it can go negative in the primary grades or at very short delays. And it moves domain retention, not general ability.

What the evidence shows

Timing is the whole game. The spacing benefit is smaller (though usually still reliable) when you're tested soon after, and grows with the retention interval. The parametric data show the optimal gap is non-monotonic — it scales to how long you want to remember something: roughly a day to retain for a week, a week to retain for a month, a few weeks to retain for a year. As a rule of thumb, space at ~10-20% of the target retention interval. (Note: an earlier reading of the Cepeda data conflated recognition with recall; the corrected picture is that recall gains are large but non-monotonic, peaking around a 70-day interval, while the 10-20% rule holds.)

Hereditarian-lens assessment

Risk: low. Within-subject scheduling manipulations hold ability constant, so this is not a selection artifact. On the ability-gap question, the genetics review found spacing is neutral-to-narrowing: it helps low and average performers but can be absent for already-high performers on complex procedural material (Ruitenburg et al. 2025) — so Dunlosky's "benefits all abilities" is slightly too broad, but the direction is anti-Matthew, not gap-widening. No effect on g; this is a retention-efficiency tool.

Lab-to-classroom & durability

The effect survives in classrooms but shrinks (d≈0.54 vs lab ≈0.85) and gets heterogeneous. Two practical translations from the classroom-only meta: fixed ~7-day gaps are more reliable than "expanding" schedules (the popular expanding-retrieval prescription is weakly supported in real settings and sometimes negative), and ~3 spaced exposures suffice ("less is more" — piling on more re-exposures dilutes the advantage). Durability is the point of the method — the benefit is defined at delay.

Boundaries & what critics say

• Primary grades: the classroom meta found the largest cluster of negative effects there — spacing is not a universal prescription for young children on all content.
• Short retention intervals: at short delays massing can match or beat spacing.
• Standardized magnitude: nearly all effects use aligned recall tests; a standardized-achievement equivalent is likely smaller (~half).
• Students won't do it themselves: the metacognitive illusion favors massing (cramming feels productive), so spacing has to be built into the curriculum, not left to student choice. Added 2026-07-30 — now measured, not just asserted. A citation-graph audit surfaced Bjork, Dunlosky & Kornell 2013, which reports the cleanest quantification of this illusion available: across Kornell's (2009) experiments, 90% of learners performed better after spacing — and 72% of them reported afterwards that massing had worked better. People are not just wrong about spacing; they are wrong about their own just-completed experiment, because fluency during study is mistaken for learning. This does not change the verdict or confidence (grade D review, no new effect size); it converts a hand-waved "students won't do it" into a documented near-inversion of belief and outcome, and it is the strongest argument in this file for making spacing a scheduling decision rather than a study tip. Caveat: college samples, not the 6–18 band this topic covers.

Practical guidance

• Design spacing into the curriculum: cumulative reviews, spiral scope-and-sequence, distributed homework, cumulative (not unit-only) exams. Don't rely on students to self-space.
• Use fixed ~1-week gaps and about 3 spaced re-exposures of key material.
• Combine with retrieval practice — spaced retrieval is the strongest version of both methods.
• Match the gap to how long it must last (~10-20% of the target retention interval).

Open questions

• Long-interval (semester/year) classroom benefit is extrapolated from sparse lab data (few studies test beyond ~6 weeks).
• Why spacing goes negative for some young children / short delays isn't fully characterized.