The Evidence Base
MASTRii is built on decades of peer-reviewed learning science — and we think you should be able to check it yourself. This is the research behind how we teach, with an emphasis on the people we build for: adult and health-professions learners. We mark what's strongly supported, what's promising, and what we deliberately don't claim.
d, g, SMD) are standardized differences between groups — roughly, how large the gap is in standard deviations. A rough guide: ~0.2 small, ~0.5 moderate, ~0.8 large. They are not percentages, not grade improvements, and they can't be added together. Where a study reports raw group scores, we show those as reported.The strongest card for our audience
Most learning research is run on undergraduates memorizing word lists. This one wasn't — it's a large randomized study in continuing medical education, measuring both learning and later knowledge transfer.
Randomized · continuing medical education
Natesan et al. (2025), Academic Medicine 100(1), 94–102 — American Board of Family Medicine Continuous Knowledge Self-Assessment.
Why it matters for us: this is a professional, adult, health-sciences population — the closest published evidence to the people MASTRii serves — and it measured transfer, not just recall. It's the backbone of our case for spacing material over time instead of cramming it.
doi.org/10.1097/ACM.0000000000005856Construct 1
Spreading the same work across separate sessions beats massing it into one — and the best gap grows with how long you need to remember. This is one of the most replicated findings in the field.
| Study | Population | Finding | Effect |
|---|---|---|---|
| Cepeda et al. 2006 Psychological Bulletin | 839 assessments, 317 experiments | Spacing beats massing for later recall; optimal gap scales with the retention interval. | meta |
| Donovan & Radosevich 1999 J. Applied Psychology | Adult / workplace training | Distributed practice advantage in grown, working learners — not just students. | d = 0.46 |
| Mawson & Kang 2025 Behavioral Sciences | Applied classrooms, 3,000+ learners | The effect holds in real classrooms, outside tightly controlled labs. | d = 0.54 |
| Cepeda et al. 2008 Psychological Science | Adults, controlled experiment | Moving the gap from 0 to 1 day lifted recall ~34% in the reported condition. | ridgeline |
Construct 2
Recalling something from memory — rather than rereading it — strengthens it more, especially after a delay. This is the single most robust lever we build on, and it replicates in medical learners.
| Study | Population | Finding | Effect |
|---|---|---|---|
| Adesope et al. 2017 Review of Educational Research | 217 studies, broad | Reliable retrieval advantage over restudy; larger on delayed tests. | meta |
| Roediger & Karpicke 2006 Psychological Science | College students | Recall (61%) beat rereading (40%) a week later — but rereading won right after study. Lab prose, not grades. | d = 1.26 |
| Larsen, Butler & Roediger 2009 Medical Education | Medical learners (RCT) | Repeated testing beat repeated study for long-term retention in a clinical population. | RCT |
| Butler, Karpicke & Roediger 2007 JEP: Applied | College students | Feedback improves retention; engaging with the correction matters more than its speed. | d ≈ 0.9 |
Construct 3 — the combination we lead with
Spacing and retrieval together are the most defensible design for durable learning. One honest caveat we build in: spaced retrieval clearly beats massed retrieval, but "expanding" intervals are not proven better than evenly-spaced ones — so we don't market an expanding-interval gimmick.
| Study | Comparison | Finding | Effect |
|---|---|---|---|
| Latimier, Peyre & Ramus 2021 Educational Psychology Review | Spaced vs massed retrieval (29 studies) | Spacing your retrieval is strongly better than bunching it. | g = 0.74 |
| Latimier et al. 2021 (same) | Expanding vs uniform intervals | No reliable advantage for expanding intervals. Consistent spacing is the real lever. | g = 0.03 |
Construct 4 — our setting
Evidence that this works through a screen, for clinicians. It's encouraging and directly on-point — but much of it carries risk-of-bias limits, so we treat it as support, not a guarantee.
| Study | Outcome | Finding | Effect |
|---|---|---|---|
| Martinengo et al. 2024 JMIR | Knowledge (spaced vs massed digital) | Spaced digital education helped health professionals learn. | SMD 0.32 |
| Martinengo et al. 2024 JMIR | Clinical behavior change | Larger effect on what clinicians actually did — though 83% of studies had unclear/high bias risk. | SMD 0.67 |
| Kang et al. 2023 Systematic review | Distributed/retrieval practice | 43 of 63 experiments in health-professions education found a significant benefit. | 43 / 63 |
| Davis et al. 2019 JMIR mHealth | Mobile learning vs comparison | Mobile delivery can help — but effects are heterogeneous. Mobile is an enabler, not the mechanism. | SMD 0.43 |
Construct 5
A hard idea lands better when a clear explanation and the right visual are aligned — our mechanism maps. The caveat we respect: this is about coordination, not "more media is always better," and not decorative images.
| Study | Finding | Note |
|---|---|---|
| Mayer & Moreno 2003 Educational Psychologist | Learning improves when words and relevant visuals are presented together and coordinated. | About alignment — not a "+130% from images" claim. |
Just as important
A credible page says what it won't claim. These are popular lines we've left off on purpose.
Everything on this page is published research anyone can read — and we want you to. What MASTRii adds is the part you can't look up: how we turn this evidence into an adaptive schedule that decides what to show you, when it returns, and when you've actually shown you know it. The findings are the field's. The system built on them is ours.