TITLE: The Testing Effect AUTHOR: Eugene Wallingford DATE: June 07, 2007 10:14 AM DESC: ----- BODY: In the current issue of the Chronicle of Higher Education, the article You Will be Tested on This (behind pay wall) tells us something researchers have known for decades but which too few teachers act on: People learn better when they are required to actively recall and use knowledge soon after they learn it. This idea was first documented by Herbert Spitzer, who did a study in the late 1930s with Iowa sixth-graders. Students who were quizzed about a reading assignment within twenty-four hours of their first reading the article scored much better than students who had been quizzed later or not at all. The results did not follow from different study habits or from extra preparation, as "students did not know when they would be quizzed, and they did not keep the article, so they had no chance to study on their own. This has come to be called the Testing Effect. Spitzer concluded:
"Immediate recall in the form of a test is an effective method of aiding the retention of learning and should, therefore, be employed more frequently in the elementary school."
As the Chronicle piece points out, the Testing Effect runs counter to conventional wisdom:
"The testing effect cuts against the lay understanding of memory," says Jeffrey D. Karpicke, who recently completed a doctorate at Washington University and will become an assistant professor of psychology at Purdue University this fall. "People usually imagine memory as a storage space, as a space where we put things, as if they were books in a library. But the act of retrieval is not neutral. It affects the system."
This is another case where we rely on a metaphor beyond its range of applicability. Knowing where it fails and why can help us do our job better. In the case of human memory, instructors can help students improve their learning simply by giving a quiz promptly after teaching a new idea. Giving feedback promptly is even better, because it allows students to correct misconceptions before they become too firmly implanted. Note that the Testing Effect does not gets its benefit from getting students to do more or different studying:
The purpose of this quizzing is not to motivate students to pay attention and to study more; if those things happen, the researchers say, they are nice side effects. The real point is that quizzing, if done correctly, is a uniquely powerful method for implanting facts in students' memory.
The value of prompt quizzing isn't from students studying for the quiz. It is from the act of taking the quiz itself, making an effort to retrieve items from memory. As a psychology professor from Washington University in St. Louis is quoted in the article, "every time you test someone, you change what they know." There are a lot of open questions about how the Testing Effect works and the conditions under which it is maximized, such as the role of feedback, immediate or otherwise. One of the major objections raise by some university professors I know is that such frequent, short-answer testing favors the memorization of isolated facts at the expense of broader conceptual learning. Current research is trying to answer some of these questions. Many professors also balk at the idea of writing and grading all these quizzes. There are technological solutions to part of this problem. Many folks use Blackboard to give and grade simple quizzes. For writing code, we might try something like Nick Parlante's JavaBat.com tool. Because the Testing Effect does not depend on motivating students to study more, I don't think that grading the quizzes is all that important. The key is simply to get the students to do active recall and retrieval. My teaching may already benefit from the Testing Effect. I do not give quizzes, but I do begin nearly every class session with an Opening Exercise that asks students to use some ideas we learned the previous session to solve a problem. In courses that teach programming, these exercises almost always involve writing code. In an algorithms or compiler course, the exercise might be a more general problem to solve. But in all cases the exercises require students to produce something, not select true/false or a multiple-choice answer. After students have had time to work on the problem, we debrief answers and discuss possibilities. This is the sort of immediate feedback that seems valuable to learners -- and which I have a hard time providing when I have to grade items. Later in the period, I may ask students to solve another exercise as well. Doing things always seems like a better idea than listening to me yammer on for 75 uninterrupted minutes. I have only once concern that my approach doesn't deliver the Testing Effect. Because I don't grade the quiz, I fear that some students choose not to exert much effort -- and effort is the key! I'm not so concerned that I myself am yet motivated to collect and grade the exercises, but maybe I should be. One thing that doesn't concern me is memorization of isolated facts at the expense of broader conceptual learning. My exercises ask student use the knowledge, not parrot it back, and good exercises cause students to integrate new knowledge into their larger understanding. -----