Evidence
These are the papers we lean on most. Each entry includes the limitation we'd want a student to know before treating a finding as a law.
2006 · Roediger & Karpicke
Students who studied a passage once and then recalled it twice outperformed students who re-read it three times on a one-week delayed test — despite predicting the opposite.
Limitation: Prose recall in a lab; the effect is smaller for very complex material.
2013 · Dunlosky et al.
A review of ten common techniques rated practice testing and distributed practice as high-utility, and highlighting, re-reading and summarisation as low-utility.
Limitation: Utility ratings are generalised across ages and subjects.
2006 · Cepeda et al.
Across 254 studies, spaced practice beat massed practice consistently; optimal gap scales with how long you need to remember the material.
Limitation: Mostly verbal learning tasks rather than problem-solving.
2007 · Rohrer & Taylor
Mixed problem sets produced markedly better delayed test performance than blocked sets, even though blocked practice looked better during the session.
Limitation: Demonstrated in maths; transfer to essay subjects is less well mapped.
2013 · Rasch & Born
Sleep actively consolidates newly learned material; losing it degrades retention independently of total study time.
Limitation: Mechanistic review — not a prescription for a specific bedtime.
Matching teaching to a self-reported 'visual' or 'auditory' style has repeatedly failed to improve outcomes. Match the format to the material instead.
Task-switching costs are real and measurable. Two hours of split attention reliably underperforms one hour of single-tasking.
Time-on-task predicts learning only when the task requires retrieval. Passive hours accumulate fatigue, not memory.
