Teachers who give tests on a daily or weekly basis--often at the expense of their popularity--can take solace in a new study out of Washington University in St. Louis. Researchers found that tests help students remember what they've been taught--including the material that doesn't appear on the exam. The findings appear in the November issue of the Journal of Experimental Psychology: General.

The research team, led by psychology graduate student Jason Chan, designed three experiments to determine whether testing can enhance the long-term recall of studied material. In the first, 84 students, split into three groups, read an essay on the toucan bird. The first group took a short 22-question test on the information they'd just read but received no feedback on their performance. The second one reviewed 22 statements culled from the longer passage. The third group went home without being tested. The next day, all the students took a 44-question test--with the same 22 questions on the first day's test and 22 additional ones. The researchers found that the first group did significantly better on the questions that only appeared on day two (performing at least 9 percent better than the other groups).

In a second experiment, 72 students studied two articles on different topics. Immediately after the students read the articles, they took a 12-question test on one of the pieces. The next day, the students took a 48-question test with the same 12 questions they saw the day before, 12 more from the same article and 24 from the other article. The students did significantly better on the second set of 12 questions of related material than they did on the 24 questions about the second article. In the third study, Chan manipulated the recall methods of students--asking them, during the first round of testing, either to think of all the information related to test questions or to home in solely on the answer. On the second test, the students who thought more broadly the day before performed much better on related questions.

The findings, Chan says, indicate that courses should proceed via "a study-test-study-test schedule" rather than studying, reviewing and then being tested. "Restudying a subset of the learned material will not produce enhancement for the remaining material--presumably because restudying is a more passive learning process than is testing," Chan explains. "Most people probably do not test themselves until they feel comfortable with the material, but that probably isn't the optimal way to study. Frequent testing, not restudying, is the key to long-term retention."

Michael Anderson, a psychologist at the University of Oregon, begs to differ. He notes that over 80 published articles in the field claim that testing actually harms retention, a phenomenon called "retrieval-induced forgetting." He says Chan's findings likely only apply to the specific set of conditions in his study. "This claim may actually be dangerously misleading," Anderson warns. "It is a mistake to make the generalization that testing enhances memory for related material."

Chan acknowledges that the literature is against him, but he argues that his study approximated a college course because it used "textlike narratives" rather than word lists for subjects to memorize, which he says most of the research arguing against him employed. "One thing we know about retrieval-induced forgetting is that it is a short-lived phenomenon that typically does not last for a day," Chan points out that his study mimics students cramming for examinations by allowing participants a 24-hour delay between tests. Previous studies only allowed 20 minutes.

"One last implication for education is that it is a good idea to give short-answer exams as opposed to multiple-choice exams," Chan says. "It is unlikely for students to try to recall related information during a multiple-choice exam because they tend to answer a question by first looking at the choices, instead of trying to recall information from what they know." Sorry, Scantron fans, looks like blue books make for better learning.