Spacing and interleaving: how to plan an exam period

An exam period is a scheduling problem. You have a fixed number of days, several subjects, and exams at different dates. Most students solve it by working on one subject at a time, in blocks, until its exam is over.

Two findings from learning research suggest a different plan. Spacing: the same amount of study gives more when it is spread across days. Interleaving: practice gives more when different types of problems are mixed instead of grouped.

Neither costs extra time. Both feel less productive while you do them. That is why they are worth knowing about.

Spacing: same hours, more retention

Cepeda and colleagues (2006) reviewed the research on distributed practice: 839 assessments from 317 experiments in 184 articles. Across the studies that compared the two directly, items whose repetitions came back to back (massed) were recalled 36.7% of the time on the final test, and items whose repetitions were separated by a gap (spaced) 47.3% of the time. Total study time was the same in both conditions.

The review also found that the best gap is not fixed. It grows with how long you need to remember the material.

How long should the gap be?

Cepeda, Vul, Rohrer, Wixted and Pashler (2008) ran a large study to map this out. More than 1,350 people learned a set of facts, reviewed them after a gap of up to 3.5 months, and took a final test up to a year later.

The best gaps for recall were:

Two practical points follow.

First, the gap matters a lot. With the same study time, the best gap produced 64% higher recall than reviewing immediately, with no gap.

Second, too short is worse than too long. As the gap grew, recall rose steeply and then declined only gradually, and the authors note that a gap longer than the best one costs much less than a gap that is too short. Measured as a share of the time until the test, the best gap was about 20 to 40% for a test one week away, and about 5 to 10% for a test one year away.

A caution: these participants learned facts, not how to solve differential equations. The direction of the effect is well supported; the exact numbers are a rough guide for other kinds of material.

Interleaving: mix the problem types

Interleaving is about order within practice. Blocked practice means doing ten problems of type A, then ten of type B. Interleaved practice means mixing A, B and C in one set.

Rohrer and Taylor (2007) taught college students to compute the volume of four unfamiliar geometric solids. One group practised in blocks, one solid at a time. The other group got the same problems shuffled.

During practice, the blocked group did better: 89% correct versus 60%. On a test one week later, the result flipped: the mixed group scored 63% and the blocked group 20%.

Rohrer, Dedrick and Stershic (2015) moved this into real classrooms. Over three months, 126 seventh-grade students worked on two kinds of mathematics problems. Each student practised one kind in the usual blocked way and the other kind interleaved with other problems. On an unannounced test 30 days after the final review, students scored 74% on the kind they had practised interleaved and 42% on the kind they had practised in blocks.

The authors give a simple reason. In blocked practice you know which method to use before you read the problem, because every problem in the set uses the method from the last lesson. In an exam, you have to choose the method from the problem itself. Interleaved practice trains that choice.

It feels wrong, and people misjudge it

Kornell and Bjork (2008) asked people to learn the painting styles of 12 artists from six paintings each, either grouped by artist or interleaved. Then they had to identify the artist of new paintings.

78% of participants did better with interleaving. 78% also said that grouping had been as good or better.

This is the same pattern as in the Rohrer and Taylor practice scores. Blocked practice gives smooth, fast progress during the session, which feels like learning. The payoff of mixing shows up later.

How strong is the evidence?

Dunlosky and colleagues (2013) reviewed ten common study techniques. Distributed practice was rated "high utility", alongside practice testing. Interleaved practice was rated "moderate utility", mainly because its benefits had only begun to be studied systematically.

So: space everything, and interleave where you have several problem types or categories that can be confused.

A plan for three exams

Here is an example. You have three exams:

• Statistics on day 10
• Microeconomics on day 15
• Databases on day 20

The blocked plan is days 1 to 10 statistics, days 11 to 15 microeconomics, days 16 to 20 databases. Each subject gets one long block, and you see microeconomics for the first time in a week.

A spaced plan touches every subject early and returns to each one several times.

Days 1 to 9: all three subjects, weighted by date. Each day has two or three sessions of 60 to 90 minutes. Statistics gets the most sessions because its exam comes first, but microeconomics and databases each appear every second or third day. Every session starts with 10 minutes of recalling the previous session of that subject from memory, before any new material.

Days 10 to 14: microeconomics leads. Statistics is done. Databases still gets a session every other day.

Days 15 to 19: databases leads. Return once to microeconomics material only if you want it for later courses.

Last day before each exam: retrieval only. Past papers, blank-page recall, mixed problem sets. No new material.

The gaps are short because the exams are close. That matches the 2008 data: for a test about a week away, the best gap was around a day.

How to interleave inside a subject

Interleaving helps most when the problem types look similar and the hard part is choosing the right method.

• Statistics: instead of ten t-test problems, then ten chi-square problems, make a set where each problem could be a t-test, a chi-square test, a correlation or a regression, and the first step is to decide which.
• Microeconomics: mix questions on elasticity, market structures and welfare, so you have to recognise which model applies.
• Databases: mix normalisation, SQL joins and transaction isolation questions.

Start interleaving after you have learned each type at least once. In both Rohrer studies, students were first taught each kind of problem; the mixing happened in practice.

Past exam papers are a ready-made interleaved set: the questions come from the whole course, in no particular order.

What to expect

Spaced and interleaved sessions feel slower. You will forget things between sessions and need to retrieve them, and mixed problem sets produce more errors. That is the design, not a failure. In the Rohrer and Taylor study, the group with lower practice scores was the group with three times the test score a week later.

If a session went smoothly from start to finish, consider whether it was too blocked or the gap too short.

The short version

• Spread each subject across the whole exam period instead of studying it in one block.
• For an exam about a week away, review roughly every day or two; for later exams, longer gaps are fine.
• Mix problem types once you know each of them, especially types you tend to confuse.
• Begin every session by recalling the last one from memory. Spacing and retrieval practice work well together.

Learnode's progress view shows which concepts you practised and when, which makes it easier to see what is due for another round.