As we look for ideas to improve our SimUText Ecology chapters, I’ve been digging through literature on how students learn from reading. Last week I found an interesting experiment on the effect of using embedded questions. Quite a few authors have shown that giving students immediate feedback on their understanding of a topic is a great aid to learning. It turns out that story gets a little more complicated depending on what you do after providing feedback.
In a study that is now a bit old (1996) but still seems very relevant, David Reinking and colleagues from the University of Georgia used an old e-text system (Hypercard, for those of you that remember it—quite a cool program back in the day) to give students a passage to read. Every 2 – 3 paragraphs, the software presented the student with a question about the preceding paragraphs. If the student answered correctly, they simply kept reading. If they answered wrong, the program gave feedback in one of three ways. In some cases, the students were simply told they were wrong, given the correct answer, and allowed to continue reading (“Feedback Only”). In a second treatment, students were told they were wrong, asked to review the preceding material, and then asked to answer the question again. This repeated until the student got the answer right (“Review / Same Question”). A third condition was similar to the second, but after the student had a chance to review the material they were asked a different question (“Review / Different Question”). At the end, all students were given a post-test.
Making the student answer questions again until they get them right has an unexpected effect. The student will do better on the material addressed by the question, but in some cases, they do worse on other material from the same passage that wasn’t tested. The students tend to go back and skim through the reading again just looking for the correct answer to the question they were asked. That reinforces the correct answer to that particular question, but de-emphasizes other material which the students then apparently forget.
The authors of this study were particularly interested in whether the students would do a more thorough review job on all the material if, after review, they knew they would have to answer a different question than the one they initially got wrong. The results were mixed, but interesting. On questions that required making inferences, the students did better when they knew the questions would change. But on questions where they were being asked for direct recall, they did worse. The authors propose that this indicates the students were re-reading the passages more holistically when they didn’t know what the new review question would be. The students memorized details less, but spent more time grasping the overall ideas.
While this result was found for reading, it fits my intuition for other kinds of teaching as well. I think it suggests we mix up the questions we use in all kinds of teaching situations—one-on-one tutorials, clicker questions in lecture, as well as virtual biology labs and interactive chapters—whenever there is more than one important idea or fact contained in a bit of material. While the authors didn’t test this, I wonder if the technique to use in more direct teaching would be to ask the same question again after review, and then follow that with a new question (which is probably what many of us do intuitively already). The results also suggest to me that in our own work at SimBiotic, we should place questions in our modules for every important point we want students to remember, otherwise they may skim over crucial bits looking just for the information needed to answer the next question.