Mix several kinds of problems; do not do twenty of one kind in a row

Doing a long string of the same kind of problem feels smooth on the spot. Mixing several kinds works better over the long run, by a moderate amount. But this method is picky about material: picture tasks such as recognizing a painter's style work best; math problems a bit less; vocabulary is worse than blocked practice.

Cost

No money. Change doing the workbook by problem type into mixing the types. The hard part is that mixed practice has lower accuracy on the spot and feels like going backward.

Benefit

A multilevel meta-analysis of 59 studies, 238 effect sizes, and 158 samples. Overall, interleaved practice versus blocked practice has a moderate effect (Hedges' g = 0.42). By material, painting-style tasks are best (g = 0.67); other visual materials are also better. Math tasks have a smaller effect (g = 0.34). Expository text and taste materials show no significant overall effect. For vocabulary, blocked practice is better (g = −0.39). Regression shows: the more similar the categories, the less similar items within a category, and the more complex the material, the larger the gain from interleaving.

Notes

The review cited in item 16 (do not treat highlighting as the main study method) rated interleaved practice moderate, because research was still thin then. Taken together, mix practice when you need to "recognize which kind this is and which method to use," such as math problem types, cases, or which rule of law applies. Use blocked practice for rote content, such as vocabulary or term definitions.

Sources · 1

1. Brunmair M, Richter T (2019). Similarity matters: A meta-analysis of interleaved learning and its moderators. Psychological Bulletin, 145(11), 1029–1052. doi.org/10.1037/bul0000209