Saturday, May 3, 2008

Testing In Mind





Adam CarleAdam Carle

Department of Psychology
adam.carle@unf.edu

Why do we bother to give tests? Our students hate them. We often hate them. And, let’s be honest, both students and instructors sometimes come away feel a little angry at each other once the process ends. So, what point do tests serve? Before you read ahead, pause for a few moments. Reflect. Ask yourself, “Why do I give tests? What do I think they accomplish?”

Welcome back. I bet your answer included the notion that we give tests to measure memory. In other words, we give tests to examine what students learned. We write tests to see whether our students learned anything during our course. We implicitly (and sometimes explicitly) assume that testing doesn’t play an active role in learning. Tests don’t help students learn, do they? Class assists learning. Studying for tests helps students learn. But, does taking a test facilitate learning? Recent work by Karpicke and Roediger (2008) sheds some new and interesting light on the learning process and testing’s role in it. Believe it or not, taking tests adds to the learning process.

Karpicke and Roediger wanted to know several things. First, they wanted to know whether repeated studying following correct recall leads to improved long term retention. Once students learn information, does repeatedly studying it lead to long term retention? Second, they wanted to know whether testing plays a role in retention. Once students learn information, does requiring them to recall the information lead to increased retention? Lastly, they wondered whether they might find a joint effect; do repeated studying and testing uniquely lead to long term recall? Do they interact?

To examine this, they worked with students learning 40 Swahili/English word pairs and manipulated two things, one: whether word pairs remained in a study list following correct recall on a test, and, two: whether the test covered the entire set of words or only incorrectly recalled words. They randomly assigned students to participate in one of four conditions and tested them repeatedly. In the first condition, students studied the word pairs and then took a test to examine how many of the words they recalled. After the test, the students resumed studying the entire list, regardless of which pairs they’d remembered correctly. At the end of the new study period, they took a new test covering the entire set of 40 word pairs. They did this a total of four times. In the second condition, students also studied the list and then took a test. In this condition though, the students resumed studying after the test, but only studied words they’d missed. They no longer studied pairs they’d remembered correctly. At the end of the new study session, however, they took a test covering the entire set of 40 word pairs. They too did this four times. In the third condition, students received four study and testing sessions as well. These students studied the entire list in each study session, but each new test only examined words the students missed on an earlier test. In the last condition, students studied the pairs and took a test on them four times. In this condition, though, once they correctly identified a word pair, they stopped studying it and stopped getting tested on it. Essentially, as soon as students correctly recalled a word pair, they no longer studied it or received a subsequent test on it. At the end of the sessions, all students took a test covering entire list. Nearly all of the students could recall all of the pairs correctly. Thus, regardless of the type of learning, students had nearly perfect recall before leaving. Great. Wonderful. Right? Not exactly, I hope my students remember things later. I hope they show long term recall. I bet you do to. Not surprisingly, Karpicke and Roediger (2008) mostly wanted to examine how the different learning phases would affect long term recall. And, now, folks, it gets interesting.

One week after the learning phases, they tested the students again. Guess what they found. Seriously. Take a second and think about what you would expect. Think about the tests you give. Do you continue testing your students on the same material across the semester? Or, do you give a test once and move on to new material? You response gives your expectation. So, with your current practices in mind, who do you think showed the best long term recall? Let me tell you.

Students who received repeated testing on the entire list showed markedly better retention than students who did not receive repeated testing on the entire list. Students who repeatedly received testing on the entire set of items recalled about 80% of the words. Students who only got tested on words they’d missed barely recalled 30% of the words. Moreover, this did NOT depend on the type of studying students did during the learning phases. Regardless of whether students studied the entire list or whether they only studied words they’d missed, one week later students who’d received repeated testing on the entire set of items showed much better recall than students who only tested themselves on words they’d missed earlier. Repeated retrieval in the form of repeated testing increased retention, regardless of whether students repeatedly encoded (studied) words. Testing makes a difference. A big difference.

What does all this mean for us? Among other things, it suggests that we might want to reconsider how we construct our tests. We should probably put more thought into what we test. If we want students to remember important concepts, we should make and include mainly these concepts on the test. We should test our students on these concepts repeatedly. If we want students to retain more esoteric aspects of the course, we should make sure we cover that material on the test. And, again, we should require our students to retrieve those concepts from memory repeatedly. At the least, we should let ourselves feel less guilty for “forcing” our students to take tests. We can take comfort that a series of well-designed tests should help our students remember the material later, above and beyond “just” studying. As Karpicke and Roediger conclude, the “benefits [of studying]… clearly depend on the process learners engage in during repetition.” I don’t know about you, but I feel responsible for creating and maintaining this process. I know I’m going to spend some time between now and the start of the next semester thinking about my tests. You?

Karpicke, J. D., & Roediger, H. L. (2008). The Critical importance of retrieval for learning. Science, 319, 966 - 968.

2 comments:

Dan Richard said...

Adam,
Thanks for the informative post. The research you cite and the recommendations you give seem to match my own intuitive feeling about testing. More and more, I have been feeling that students need to be tested more. From my reading of your blog, the researchers are suggesting that the act of recall provides practice and long-term retention.

I guess this is why homework assignments (as opposed to reading assignments) can be so effective. Reading allows the student to study the information, but recall is not required to "cover" or read text. For homework assignments, students must practice recalling what they have learned and apply what they have learned to a novel situation.

I test quite a bit, but I will give more thought to repeating information in multiple tests in the future.
Thanks,
Dan-

Anonymous said...

Dan,
Good evening. You hit the nail on the head with your summary. Although I don't discuss it in the blog, these researchers have a long and deep program of research that investigates this topic in the lab and the classroom. Their work shows how robust this finding is. Moreover, in their work, they highlight the importance of teaching students how to best study. We should remind (teach) our students that self-study involves self-testing. It has a powerful effect.
Best,
Adam