Sunday, March 26, 2006

Testing causes learning

Randall Parker over at FuturePundit points to a study out of St. Louis' Washington University about the benefits of testing for learning retention:
"Students who self-test frequently while studying on their own may be able to learn more, in much less time, than they might by simply studying the material over and over again," he adds. "Incorporating more frequent classroom testing into a course may improve students' learning and promote retention of material long after a course has ended..."

In an experiment in which students either took quizzes or were permitted to study material repeatedly, students in the study-only group professed an exaggerated confidence, sure that they knew the material well, even though important details already had begun slip-sliding away. The group that took tests on the material, rather than repeatedly reading it, actually did better on a delayed test of their knowledge.

The group that spent all its time studying the material initially had slightly better recall of it. But tested just two days later, the group that had only looked at the material three times and then been tested fared much better than the group that had seen the material fourteen times (61% to 40%) but hadn't yet been tested on it.

The results are not surprising. Kaplan and other test-prep companies succeed by basically getting a hold of old tests and having participants take several of them. Kaplan offers a full refund if test scores do not improve. If this works for standardized tests that are pretty g-loaded, it should work even better for material that lends itself more to being learned.

I find studying to offer a higher return when I use notecards and draw terms randomly or when I replace numbers and tweak wordings from accounting textbook questions than when I just pore over my notes. But the first method requires more mental exertion than the second, so I have to be driven to do it.

How many students are self-driven? Certainly not all of them. The study shows that if they are prodded by instructors who employ tests/quizzes more frequently then they will perform better and retain more information over the long-term. So an emphasis on ambigious projects instead of on traditional lecture/demonstration, sample problems, and testing is probably detrimental. In other words, instructors need to force students to internalize the information by giving them pop quizzes and thorough tests. Sure, that's harder for the students than the "creativity"-driven classrooms that encourage students to "explore different answers" through out-of-class, vague assignments, but it's also better for them.

Unfortunately, the KU business school is becoming increasingly obsessed with the former. Only one of my three courses in the school is strictly testing-based. The other two determine grades primarily by group projects that are very open-ended. Not surprisingly, when I was looking over material during spring break this week, the beginning of semester stuff in the first course came right back to me. In one of the other two, I remembered virtually nothing.



FuturePundit said...

I wonder if you could find automated tests on the web for some of the material you are learning. Just a thought. Might be worth an hour of Googling.

crush41 said...

Yes, lots of universities offer them online. Googling "accounting test" yields lots of them. I keep material from all my classes, but the online tests are particularly useful for assessing how well prepared I am for upcoming classes and graduate school should I choose that route.