Interleaved Practice: When Mixing Problem Types Helps Learning

Interleaving changes the order of practice

Blocked practice completes several examples of one type before moving to the next: A A A | B B B | C C C. Interleaved practice alternates among the types: A C B | B A C | C B A. The content can be identical; the order changes.

This guide is about mixing related categories or problem types that learners need to distinguish. It is not evidence for multitasking or jumping randomly between unrelated subjects.

What the evidence supports—and what it does not

A meta-analysis of 59 studies found a moderate average advantage for interleaving, but the results varied substantially. Benefits were larger for some visual-category tasks, smaller for mathematics, unclear for expository text, and studies using word materials favoured blocking on average. Similarity among categories also mattered.^[1]

A systematic review of concept learning likewise found benefits for memory and transfer, especially when differences among examples were subtle. Most of that evidence came from short laboratory studies, often with university students and visual materials, which limits direct claims about every classroom or subject.^[2]

The defensible conclusion is conditional: interleaving can help, particularly when later performance requires telling related cases apart, but it is not universally superior to blocking.^[1][2]

Mixed sets can add a strategy-selection step

In a blocked worksheet, the heading or previous problem may reveal which method to use. In a mixed worksheet, the learner may first need to identify the problem type and select a method. That can practice discrimination as well as execution.^[4][6]

Category-learning experiments show why the qualification matters. Interleaving helped when categories had high within- and between-category similarity, while blocking helped with low-similarity categories where noticing common features within a category was more useful.^[3]

Interleaving also spaces repeated examples of the same type. Experiments designed to separate these factors suggest that distributed practice can contribute to mathematics benefits. Strategy selection and spacing are plausible, non-exclusive explanations.^[5]

For the timing side of that distinction, read the spaced repetition guide.

Practice performance can be a poor judge of later learning

In one study of children learning four types of prism problems, interleaving reduced accuracy during practice but improved performance on a test one day later, even when spacing was held constant. That is useful evidence from one narrow task, not a promise about every mixed worksheet.^[4]

Lower practice accuracy does not by itself show that a method is failing. But difficulty is not proof that learning is working either. Use a delayed, unlabelled test to compare what learners can later choose and execute.

Blocking still has a legitimate place

Kaboosh’s practical approach is to use explanation, modelling, and focused practice when a component skill is unavailable.

Blocking can also focus attention on what members of one category have in common.^[3]

In a study of 107 low-achieving adolescents learning second-language word pairs, interleaving alone created excessive difficulty, while a hybrid sequence with initial blocking produced more robust long-term retention. That result applies to a specific population and task; it supports considering a transition, not a universal rule that every novice must block first.^[7]

Kaboosh’s practical advice is to establish enough understanding for a genuine attempt, then mix related types with feedback. Return briefly to focused examples when errors show that the underlying method is unavailable. The ideal transition point remains uncertain.

For teachers: separate selection errors from execution errors

After modelling the area formulas for triangles, trapezoids, and parallelograms, give enough supported practice for students to attempt each method. Then provide an unlabelled mixed set. Before calculating, ask students to write the formula they selected and the feature that justified it.

Checking both decisions reveals whether a learner chose the wrong method or chose correctly and then made a procedural error. Give solutions and require correction; the large classroom mathematics trial that favoured mostly interleaved practice also included feedback and corrections, so “just shuffle everything” misses part of the implementation.^[6]

For independent learners: name the rule and the cue

For differentiation practice, first confirm that you can attempt the product, quotient, and chain rules with support. Build a set that mixes all three without chapter labels. Before each solution, write “rule” and “cue,” then check both the selection and the calculation.

If the same rule repeatedly fails, use a short focused block to repair it and then return it to the mixed set. If you are using flashcards, retrieval practice concerns producing an answer; interleaving concerns which kind of prompt appears next. A single session can use both without treating them as the same method.

For the retrieval step, see the active recall guide.

Learners can also build a mixed set from Kaboosh’s public flashcard decks.

The realistic promise

Interleaving is a useful way to practice distinguishing related cases and selecting among methods. It is not evidence that blocking is fake learning, that confusion is desirable, or that unrelated subject switching improves memory.^[1][2]

The optimal number of categories, the best point to move from blocked to mixed practice, and the balance between discrimination and spacing remain uncertain. Judge a schedule by delayed, relevant performance—not by how smooth or difficult the worksheet feels.^[1][2][5][7]