March 25, 2006
Frequent Tests Increase Memory Retention When Studying

The potential time savings from this report is enormous. The economic value ditto. Repeated testing improves longer term memory retention.

"Our study indicates that testing can be used as a powerful means for improving learning, not just assessing it," says Henry L. "Roddy" Roediger III, Ph.D., an internationally recognized scholar of human memory function and the James S. McDonnell Distinguished University Professor at Washington University.


In two experiments, one group of students studied a prose passage for about five minutes and then took either one or three immediate free-recall tests, receiving no feedback on the accuracy of answers. Another group received no tests in this phase, but was allowed another five minutes to restudy the passage each time their counterparts were involved in a testing session.

After phase one, each student was asked to take a final retention test presented at one of three intervals — five minutes, two days or one week later. When the final test was presented five minutes after the last study or testing session, the study-study-study-study (SSSS) group initially scored better, recalling 81 percent of the passage as opposed to 75 percent for the repeated-test group.

However, tested just two days later, the study-only group had forgotten much of what they had learned, already scoring slightly lower than the repeated-test group. Tested one week later, the study-test-test-test group scored dramatically better, remembering 61 percent of the passage as compared with only 40 percent by the study-only group.

The study-only group had read the passage about 14 times, but still recalled less than the repeated testing group, which had read the passage only 3.4 times in its one-and-only study session.

"Taking a memory test not only assesses what one knows, but also enhances later retention, a phenomenon known as the 'testing effect,'" says Roediger.

"Our findings demonstrate that the testing effect is not simply a result of students gaining re-exposure to the material during testing because students in our repeated-study group had multiple opportunities to re-experience 100 percent of the material but still produced poor long-term retention. Clearly, testing enhances long-term retention through some mechanism that is both different from and more effective than restudy alone."

This strikes me as an important result with obvious and very valuable practical applications. Problem: Labor costs for testers are too high. But that can be solved by use of computer programs. Picture online books with associated online tests. You could read each section of a book and then click your way into a test about it and do the test.

I'd like to see technical computer books come with associated tests. Someone tell Tim O'Reilly, New Rider Publishing, and similar tech book publishers.

There's an obvious implication to this result: Most tests should be done to improve memory retention, not for grades. Tests delivered around the time of learning some material should be seen as drills to exercise the memory rather than for scoring to assign grades. Tests for grades could come much later after memory formation has become well advanced.

I've always thought that tests for a subject given right after learning the material (e.g. the material taught during the last couple of weeks of a college course) aren't testing for permanent memory formation. Well, look at the results above. Two groups can score close to the same level of knowledge at one point but due to differences in how they learned the material they can have very different longer term patterns of memory retention.

I've long advocated for tests one can take to earn college credits for most college courses (particularly the subjects with clearer objective bodies of knowledge such as the hard sciences, math, and engineering subjects) without having to enroll in and attend an entire course. Such tests should require that a person pass the same tests in two more more separate sessions several weeks apart. Someone who can pass a test and then pass it again 4 and 8 weeks later will retain the information far better than the average person who crams to take college course finals.

Also see my ParaPundit posts Accelerate Education To Increase Tax Revenue, Reduce Costs and Walter Russell Mead For Standard National Tests.

Share |      Randall Parker, 2006 March 25 02:02 PM  Brain Memory

Nobody said at March 25, 2006 2:37 PM:

Personal experience certainly reflects this, even though data is not the plural of anecdote.

I've taken many standardized tests to measure proficiency at various things in my lifetime, both for academic (such as SATs) and professional licensure or qualification purposes.

In each case, I prepared myself by first studying the material, then taking an exemplar test. Wash, rinse, repeat, until I felt prepared to take the actual test. In every case, I found my performance increased markedly after taking a few tests, and the time spent studying material decreased markedly as well.

After the first iteration of the study-then-test cycle, study time diminished rapidly to only that required to check a few answers -- to determine why I had gotten some answer correct or incorrect.

Bob Badour said at March 25, 2006 5:16 PM:

Sounds like English 3200 has the right approach. The question is how to adopt that to other fields of study.

Doug said at March 26, 2006 6:21 PM:

Most textbooks have numerous "review questions" at the end of each chapter. Wouldn't simply completing the questions after completing the chapter earn a lot of the benefits? Most of my math/science/engineering professors assigned problems after every reading assignment - and graded them.

James Bowery said at March 28, 2006 1:09 AM:

Actually this was the basis of a drill and practice algorithm developed at the PLATO project during the 1970s as part of a program to get prison inmates to pass the GED. The basic idea was to drive memory from short term to long term memory by queuing stimulus response pairs in such a way that recently missed pairs would come up sooner and recently successful pairs would come up later and later. Not too sophisticated when you think about it but it worked very well. I was receiving royalties from a vocabulary trainer until the very last gasp of the PLATO system well into the 1980s.

I tried a similar training algorithm on neural networks and it did seem to raise the rate of training but then it also drove prior learning out of the network. I'm sure something could be done to better train the networks using this technique and I'm sure someone has probably published a paper that I haven't read on just this topic.

Gregg Dippold said at May 1, 2006 8:23 AM:

"Picture online books with associated online tests. You could read each section of a book and then click your way into a test about it and do the test."

I purchased a program for doing roughly that, called Supermemo. It takes a while to understand their approach and to put the information in but once it's there it handles the testing and reviewing nicely. The interface is suboptimal and window resizing is somewhat erratic but I am overall satisfied.

Sajeesh said at October 27, 2007 7:12 AM:

Comment on the last paragraph of the article: I do not see merit in what the author is saying because it is not possible to prevent students from learning between tests. And if students are allowed to learn the same material between tests, then the results may not match with the experimental studies described earlier.

Loarn Austin said at October 5, 2012 9:13 AM:

Holy Skinner Box Batman

gejala stroke said at August 29, 2016 7:39 PM:

Gejala Storoke ringan dan cara menanganinya secara alami harus diperhatikan selalu secara serius , karena saat ini kasus penyakit stroke banyak terjadi dikalangan masyarakat dari mulai stroke ringan hingga stroke berat.
Meski yang dialami masih gejala stroke ringan tapi penanganannya harus secara tepat karena bisa berakibat fatal jika tidak diatasi dengan obat stroke yang ampuh. Sebelum kita membahas lebih lanjut mengenai pengobatan stroke, maka kita harus mempelajari dulu tentang apa itu penyakit stroke, penyebab penyakit stroke, gejala stroke,
serta bahaya apa yang bisa muncul? Setelah itu baru kita bisa memilih obat stroke yang tepat yang akan kita gunakan.

Post a comment
Name (not anon or anonymous):
Email Address:
Remember info?

Go Read More Posts On FuturePundit
Site Traffic Info
The contents of this site are copyright