Heuristic

Interleave when the learner must discriminate, not by default

Interleaving is strongest when choosing the right rule is part of the skill.

When it fits

  • You are deciding whether to mix several problem types, categories or skills during practice instead of blocking them separately.

When to avoid it

  • The meta-analysis found substantial moderation by material type. Interleaving is not globally superior, and the best schedule can change as expertise grows.

Why it matters

Use interleaving when the learner must notice differences between confusable cases and select the right method, not merely repeat one procedure. A large meta-analysis found a moderate overall effect but strong dependence on the material: benefits were larger where categories were similar across groups and needed discrimination, while word-learning studies in that synthesis favored blocking. Design the mix around the decision the learner must make.

An example

Mixing several SAP defect types can train diagnosis if the real job requires selecting the right check. Randomly mixing unfamiliar transaction steps before any one flow is understood may add noise instead.

Check your result

The practice mix forces a meaningful method/category choice, and delayed performance is compared with a sensible blocked baseline.

Keep this limit in mind

  • The meta-analysis found substantial moderation by material type. Interleaving is not globally superior, and the best schedule can change as expertise grows.

Connected ideas

Useful with
Make difficulty earn its keep

Evidence and sources

Supports

A meta-analysis of 59 interleaving studies found a moderate overall benefit (g=0.42) but strong moderation by material: interleaving helped most when between-category discrimination mattered, while blocking outperformed interleaving for word materials in the analyzed studies.

The average effect does not justify interleaving every curriculum or practice set; task structure and material type materially changed outcomes.

Similarity matters: A meta-analysis of interleaved learning and its moderators · Abstract

All sources (1)