Article: Smith et al. (2009)
An article in last Friday’s edition of the journal Science on peer instruction, a pedagogy often used with (and enhanced by) clickers, has received some attention. I thought I would comment on it here…
Reference: Smith, M. K., Wood, W. B., Adams, W. K., Wieman, C., Knight, J. K., Gulid, N., & Su, T. T. (2009). Why peer discussion improves student performance on in-class concept questions. Science, 323(5910), 122-124.
Summary: The peer instruction teaching method popularized by Eric Mazur (Mazur, 1997) involves having small groups of students discuss and respond to questions, usually multiple-choice, conceptual understanding questions asked via clickers. There’s strong evidence (Crouch & Mazur, 2001; Fagen, Crouch, & Mazur, 2002) that it is an effective teaching method, but some have questioned the extent to which learning occurs during such peer discussions. Perhaps students “simply choose the answer most strongly supported by neighbors they perceive to be knowledgeable,” write the authors. This study provides fairly convincing evidence that this is not the case and that real learning occurs during in-class peer discussions.
The authors, most from the University of Colorado-Boulder, looked at data from sixteen pairs of clicker questions asked in an undergraduate genetics course. Each pair of questions were considered to be “isomorphic,” that is, questions about the same concept with different “cover stories.”
Students answered the first question in each pair on their own, and, on average, 52% of students answered these questions correctly. Then students discussed the same question in small groups and answered it again, increasing the success rate to 68%. Then students answered the second, isomorphic question on their own without discussion, and 73% of students answered these questions correctly.
Furthermore, the authors point out that of the 48% of students who answered the first question in each pair incorrectly on their own, 42% of them answered the question correctly after discussion. Of these students, 77% were then able to answer the second question correctly on their own, indicating that they had learned something useful from the peer instruction time.
Additionally, the authors categorized the sixteen question pairs as easy, medium, or difficult, depending on the percentage of students who answered the initial question correctly. For the difficult questions, the gains from the first question to the second question were even more dramatic: 16% of students answered the first questions correctly on their own, 32% answered the first questions correctly after discussion, and 54% answered the second questions correctly on their own.
These results indicate that students learned something during peer instruction time, something they could apply toward answering the second question in each pair on their own. The effects of peer instruction appear to be even greater for more difficult questions.
Comments: These are very interesting results and provide fairly compelling evidence that peer instruction has real value in facilitating student learning. I’ve seen some unpublished (to my knowledge) research by Dennis Jacobs, a chemist at the University of Notre Dame, along these lines showing the value of peer instruction. I’m glad to see some published work in this direction, published in Science no less!
The comments on the Inside Higher Ed post on this story point out a couple of important points. One is that knowing how to answer a question on a particular concept when that concept is isolated from other course concepts does not mean that a student will be able to answer a similar question on a test covering multiple concepts. That is, applying a particular problem-solving approach does not imply the ability to select the correct approach when confronted by a new problem. This is a good point, and it also raises the issue of long-term retention. To what extent might learning gains from peer instruction be evident weeks later on midterm or final exams?
It’s also worth noting that in this study, the role of the instructor was minimized. I’m guessing that the instructor gave the students some instructions for discussing the clicker questions, but beyond those instructions, the instructor did not provide any explanations that would affect student performance on these questions. The gains seen here derive from peer instruction alone. Peer instruction supplemented by instruction by the teacher might show even greater gains. That might be something worth investigating in future studies.
Additionally, the authors point out that “there may also be a learning benefit to considering successive clicker questions on the same topic.” This gets at the role of question sequences seen, for instance, in Reay, Li, and Bao (2008). A study using a control group methodology might ask some students to discuss the first question in each isomorphic pair and other students just to spend more time on their own thinking about it. That would provide some potentially useful data.
Finally, one of the comments on the Chronicle‘s Wired Campus blog entry on this study pointed out that the study doesn’t really say anything about the role of clickers in facilitating peer instruction. Would similar gains in student performance be seen were clickers not used? I would argue that clickers enhance peer instruction in a few ways–by creating the expectation that all students should participate and commit to answers thereby increasing engagement and preparing students to participate more actively in peer instruction. Further study of the role of clickers in creating the gains seen in this study would be interesting.