# What the research supports — and what it does not claim

This page is for buyers who want citeable honesty. No effect size here is a guarantee for your workforce.

### What supports Digital Learning Practice

**Spacing as a scheduling principle, not a content format.** Distributing the same learning time over a longer period beats cramming for durable recall in a large research base. A major review covering 839 assessments from 317 experiments reported an average observed distributed-practice benefit of about **15%** (massed accuracy ~36.7% vs spaced ~47.3% in that synthesis). A newer classroom meta-analysis estimated a moderate effect (**d = 0.54**). Those figures are about retention under study conditions; they are not a workplace ROI promise.

**Active recall over passive reminders.** Guidance from the Learning Scientists line of work: distribute learning over time; use retrieval or elaboration rather than re-reading. Practice should ask people to recall, decide, explain, and apply — then give feedback.

**Retention horizon matters.** In a study of more than 1,350 people, the best gap grew as the final-test delay grew. A gap suited to a one-week horizon was not suited to a one-year horizon. So we ask when competent performance must still be available, and schedule backward. We do not sell one universal forgetting curve.

**Transfer is a separate outcome.** Spaced retrieval has improved later course performance in some education studies; individual transfer results can be mixed or non-significant. We measure transfer separately from retention.

**Professional / health education signal (with bias caveats).** A health-professions meta-analysis reported effects for knowledge (SMD 0.32), skills (SMD 1.15), and clinical behaviour (SMD 0.67), but most included studies had unclear or high risk of bias. Useful to justify a pilot — not an unconditional claim.

**Behaviour before completion.** The UC San Diego Health phishing work (19,500 employees): recent annual training completion was not significantly related to susceptibility; embedded training cut clicking by ~2%. Completion ≠ behaviour.

### What we do **not** claim

- That every topic benefits equally  
- That knowledge gains automatically change behaviour  
- That expanding intervals always beat fixed intervals  
- That passive notifications equal practice  
- That completion, satisfaction, or immediate post-tests prove durable capability  
- A universal forgetting curve for every learner and skill  
- Guaranteed ROI or guaranteed incident reduction  
- That vendor-reported case studies (e.g. large retail safety programmes) are independent causal proof for your setting  

### Competitive honesty

Products such as Qstream, Axonify, and MindMarker already offer reinforcement, journeys, or analytics. “We use spaced repetition” is not a differentiator. Digital Learning Practice differentiates on evidence-led conversion of *your* training, manager observation, workflow-shaped practice, and a readout that keeps outcome layers separate.

### How this ties to the free-site method

The free lessons already refuse end quizzes as completion and certificates as proof. Retrieval openers and spacing without accounts are already in the method page. Digital Learning Practice is the paid organisational form of the same rule: the learner was not behind; nobody taught them — and teaching once is not the same as still being able to do it.

### Sources we lean on (non-exhaustive)

- Cepeda et al., distributed practice / verbal recall synthesis (~15% average observed benefit)  
- Classroom distributed-practice meta-analysis (d ≈ 0.54; high between-study variance; retention intervals in that set did not exceed 42 days)  
- Spacing and retention-horizon work (optimal gap depends on final delay)  
- Learning Scientists materials on spaced practice and spaced retrieval / transfer  
- Health professions spaced digital education meta-analysis (effects with bias caveats)  
- UC San Diego Health / related reporting on phishing training vs behaviour  
- HTBD published numbers elsewhere on the site: Ipsos/Lloyds Essential Digital Skills (8m UK adults); ONS AI in UK businesses (35% of businesses with 10+ staff using AI; 11% say more than half the workforce has had any training) — context for the teaching gap, not spacing effect sizes  

Full reference list available on request for procurement packs. We will not invent statistics for a sales deck.

---