Want to know the best way to make pupils recall information? Repeated testing could be more effective than revision, says Biddy Passmore

So you thought testing was just assessment and either neutral or actively harmful to learning? Well, think again. New research from the US suggests that, far from being a recipe for a blighted childhood, repeated testing is one of the best ways to learn. The active retrieval of facts from the memory that occurs during testing is far more helpful for consolidating knowledge than passive studying.

The study indicates that pupils who stop revising a topic after they have correctly recalled it once are doing themselves no favours. They need to keep testing themselves and each other on the same material to make it stick.

The researchers, Jeffrey D. Karpicke of Purdue University, Indiana, and Henry L. Roediger III of Washington University in St Louis, decided to test some central assumptions about learning and memory. These are that “learning occurs while people study and encode material”, that additional study should therefore increase learning, and that testing represents a neutral event that merely measures it.

They gave 40 college students a list of 40 Swahili-English word pairs, asked them to study it for a set time, and then tested them on the list.

The 40 students were divided into four groups. One group repeatedly studied and were tested on the whole list (group 1). In the other three, once a student had correctly recognised a Swahili word and recalled its English translation, it was either: repeatedly studied but dropped from further testing (group 2), repeatedly tested but dropped from further study (group 3), or dropped from both study and test (group 4). All of these study and test periods were back-to-back, on one day.

The results? Repeated studying had no effect on the student’s ability to recall the information a week later. Those who were just tested once but carried on studying did much worse than those who stopped studying but were repeatedly tested. In fact, the “one test, repeated study” students did almost as badly as those who stopped both studying and testing.

Only repeated testing embedded the information in the memory, enabling the student to recall it later. Even the students were unaware that repeated testing would make such a difference.

The experiment overturns several assumptions about learning and memory, say the researchers. Not only does it show the powerful effect of testing; it also shows that repeated study makes surprisingly little difference. It is a basic tenet of human learning and memory research that the repetition of material improves its retention. But what this experiment showed was that it depends on how you repeat it.

It also sheds new light on an old debate: whether you are more likely to remember something you learned quickly (as earlier research suggests) or whether this type of high-speed learning is forgotten quicker. This experiment suggests it is not the speed but the type of learning that determines the rate at which it is forgotten.

It also suggests a new approach to revision. “The conventional wisdom among students and educators is that if information can be recalled from memory, it has been learnt and can be dropped from further practice so that students can start to focus on other material,” say the researchers. This wisdom, expressed in many study guides, is wrong.

These findings do not mean that teachers have to spend all their time between now and the exam period grilling their pupils relentlessly. But it does suggest that they should not shy away from repeated testing. And it strengthens the case for pupils to test themselves and each other when they revise, rather than simply staring at notes - as long as they keep on testing after they’ve got it right once.

Reference

The Critical Importance of Retrieval for Learning, Jeffrey D. Karpicke, et al. Science 319, p.966 (2008)

Correction: Last week’s article on how the brain is changed by music said that apart from amusics, about 17 per cent of adults are tone deaf or unable to carry a tune. The figure should have been 15 per cent.