Key Results
Note that reliability here refers to the certainty of the result (terminology likely to change in a future site update).
| Result Name | Effect Size | Reliability | |
|---|---|---|---|
| Spaced vs. Massed (Final Test — Correctness of Solution) |
 Medium |
 Low |
|
| Low Initial Performance — Correctness of Solution |
Become a Member
|
Become a Member
|
Become a Member
|
| Average Initial Performance — Correctness of Solution |
Become a Member
|
Become a Member
|
Become a Member
|
| High Initial Performance — Correctness of Solution |
Become a Member
|
Become a Member
|
Become a Member
|
| Spaced vs. Massed (Final Test — Correctness of Steps) |
Become a Member
|
Become a Member
|
Become a Member
|
| Low Initial Performance — Correctness of Steps |
Become a Member
|
Become a Member
|
Become a Member
|
| Average Initial Performance — Correctness of Steps |
Become a Member
|
Become a Member
|
Become a Member
|
| High Initial Performance — Correctness of Steps |
Become a Member
|
Become a Member
|
Become a Member
|
Study Characteristics
| Learning Domain | Procedural Learning |
| Study Topic | Massed Practice, Spaced Practice |
| Study Type |
Membership Required
|
| Task Studied |
Membership Required
|
| Subject Type |
Membership Required
|
| Quality Score (%)* |
Membership Required
|
* Study quality assessed using JBI RCT Checklist.
Reviewer Notes
I’ve been looking for studies that test whether experience changes the benefit of spacing, so I was happy to see this classroom experiment in 5th-grade mathematics. It compares spaced practice across sessions with a single massed session on a complex procedure: solving linear equations. What’s useful here is that the authors examine how the effect varies with initial performance: beginner, average, or advanced.
First, as usual, spaced practice beat massed practice on the final test. For solution correctness (is the answer correct?) the advantage was medium (Cohen’s d≈0.51; note the p=.018 comes from the model, while d is from unadjusted means). For step correctness — the steps are done correctly even if the final number is off — the advantage was medium to large (d≈0.78). Groups saw the same content for the same total time; only the schedule differed, and 61 students took part. The paper did not report a main-effect p-value for steps; evidence there comes from simple-slope tests within the same model (low and average initial performance, both p=.001). Effect sizes had to be calculated from other reported stats.
Sub-analyses showed a consistent pattern: spacing helped most when students hadn’t fully mastered the material. For solution correctness, effects were medium at low (d≈0.74, p=.004) and average (d≈0.62, p=.015) initial performance, but small and uncertain at high performance (d≈0.20, p=.433). Step correctness showed large, certain gains at low and average performance (d≈0.88 and 0.86; both p=.001) and a small, uncertain gain at high performance (d≈0.44, p=.085). The formal interaction terms were small and not statistically significant, so treat these subgroup patterns as informative but cautious.
Interpreting this study does take a minute. Could it just be that when a task is already easy there’s less to gain? The authors checked for ceiling effects (one way of checking on that) and found no evidence. My take is that once proficiency is high, the challenge shifts: you start working trickier examples and connecting to other skills, which is a different problem than consolidating the base procedure. That shift may be where spacing helps less.
On that quality score (58%): methods were solid for a classroom trial (identical materials, no attrition, appropriate analyses), but the randomization method and assessor blinding weren’t described.
The practical takeaway is that when learning multi-step procedures, space practice across days, and expect the biggest retention gains while you’re still consolidating the skill. If the task is already too easy, it’s time to step up the challenge!