Interleaving: the strongest practice effect you haven't heard of

Interleaving means mixing different kinds of problems in one practice set instead of doing a block of one kind, then a block of the next. In a randomized trial with 787 seventh graders, students who practised interleaved scored 61% on a test a month later against 38% for blocked practice, an effect size of 0.83. Peabody (peabody.kids), free through the end of 2026, mixes subjects and fact types today, while most of its practice rounds still stick to one skill. It feels harder while it happens, which is exactly why most worksheets don't do it.

What interleaving is

A typical math worksheet is blocked: twenty subtraction problems, then twenty more on tomorrow's topic. Every problem on a blocked page uses the same method as the one before it, so the student never has to decide which method to use. They only have to execute.

An interleaved page mixes kinds: a subtraction problem, then a word problem that needs addition, then a times-table fact, then subtraction again. The student has to look at each problem and work out what kind it is before solving it. That discrimination step is the skill a test, and real life, demands, and blocked practice never trains it.

The trial that put it on the map

Doug Rohrer, Robert Dedrick, Marissa Hartwig and Chi-Ngai Cheung ran a cluster randomized trial in 54 seventh-grade classes in Florida, published in the Journal of Educational Psychology in 2020. Classes got the same problems over several months, either interleaved with other kinds or blocked by kind. On an unannounced test a month after the practice ended, the interleaved group scored 61% and the blocked group 38%, an effect size of 0.83 with a confidence interval of 0.68 to 0.97. The What Works Clearinghouse reviewed it and rated it as meeting its standards without reservations, its highest rating.

For a sense of scale: Matthew Kraft's 2020 guidelines call 0.20 on a standardized measure large for a school intervention. An effect of 0.8 is rare in education research.

It feels worse while it is working

Kelli Taylor and Rohrer (2010), in Applied Cognitive Psychology, gave fourth graders practice on finding the faces, edges, corners and angles of prisms. Children who practised the kinds mixed got fewer right during practice than children who practised them blocked. The next day the result flipped: 77% for mixed practice against 38% for blocked.

This is what Robert and Elizabeth Bjork call a desirable difficulty: a condition that makes practice feel harder and look worse in the moment while producing more learning later. Their 2011 chapter lists spacing, interleaving, varied practice and testing, with one condition attached: the difficulty must be one the learner can overcome. A child who cannot yet do any of the problem types gains nothing from mixing them.

It is also why interleaving is rare. A teacher, a parent and a child all watch practice scores. Blocked practice makes those scores look better, so it feels like it is working.

The limits

• Most evidence is in math. The strongest trials are math problems where the hard part is choosing the method. Interleaving matters less when every item uses the same method.
• It came after first teaching. In the Rohrer trials students were taught each kind of problem first, blocked, before the mixing began. Interleaving is for practice, not for the first lesson.
• It took longer. Mixed assignments took students more time per problem. Some of the gain may be time on task.
• One district, one grade. The 2020 trial was large and well run, in one place. Effects this size are rarely replicated at full size.
• Interleaving is not spacing, though the two travel together. Mixing kinds automatically spaces out each kind. A 2025 meta-analysis of children's math by Murray, Horner and Göbel found spacing reliable (g = 0.28 across 27 studies) while the testing effect for math procedures was not, so the evidence is uneven across techniques.

How Peabody mixes a round

Peabody does part of this today, and the rest is planned. Here is exactly what the code does, so the gap is visible.

Two related rules come from the same research family. A new item is taught first and then rehearsed several questions later, not straight away (in a 30-question round with nothing else being learnt, about seven questions later). A missed question comes back three questions later. Neither delays the first real try to another day yet; that, like mixing whole skills inside one round, is in Peabody's plan and not built.

Whether a kindergartner should see a sum next to a take-away in the same round is an open call. Mixing tests better and feels harder, and Peabody's rounds are built so most answers are wins, so the plan is to mix skills once both are open and leave the youngest to Aaron's decision.

What a parent can do

Mix the homework. If a worksheet has twenty of one kind, do five, then five from last week's sheet, then five more. Ask "what kind of problem is this?" before "what's the answer?" Expect a few more mistakes during practice, and judge it by what your child remembers next week, not by tonight's score. Start free or read a story from the library first.

Learn more elsewhere

• The Learning Scientists: free posters and bookmarks on six study strategies, from cognitive psychologists.
• Bjork and Bjork on desirable difficulties: the researchers who coined the term explain why harder practice often works better.
• IES practice guide: organizing instruction and study: seven US government recommendations on spacing, quizzing and mixing, each rated for evidence.

For about seventy more, checked and sorted by who they're for, see where to learn about AI.

Questions parents ask

What is interleaving in simple terms?

Mixing different kinds of problems in one practice session instead of doing all of one kind and then all of another. It forces a child to work out which method each problem needs.

Is interleaving better than blocked practice?

For math practice after a skill has been taught, the trials say yes: 61% against 38% a month later in a 787-student trial, and 77% against 38% the next day for fourth graders. For a brand-new skill, teach it on its own first.

Why does my child get more wrong when practice is mixed?

Because mixed practice is harder in the moment. Taylor and Rohrer (2010) found practice scores were lower for mixed practice and test scores the next day were twice as high. Lower practice scores are part of how it works.

Does interleaving work for reading?

The strongest evidence is in math. Reading practice that mixes kinds of words (a sound, a rhyme, a sight word) has not been tested the same way, though the reason interleaving works, deciding what kind of item each one is, applies.