This is a continuation of last month’s post, summarizing the results of a recent literature review of Peer Instruction, Research-Based Implementation of Peer Instruction: A literature Review. In this month’s post, I’ll review the results on how to use peer instruction effectively.
Peer instruction is the recommended use of clickers, following the following cycle:
- Instructor lectures for a short time
- Students vote individually using a clicker or other mechanism
- Students discuss the question together (if the majority didn’t get it right)
- The instructor explains the answer and holds a class discussion.
Here are the key questions addressed about peer instruction (PI) in this review.
1. Does it matter if students vote/think individually first?
Some instructors use a modified version of PI where the “vote individually” step is skipped. I’ve often wondered if this matters, since it seems really important for students to commit to an answer on their own before they can engage meaningfully in conversation. Students seem to agree with me: most students prefer having the individual time to think first, because (as summarized by the study authors), “(this) time forced them to think about and identify an answer to the question; they felt that this led them to be more active and engaged during the peer discussion,” letting them form their opinions without being influenced by others. Starting with the peer discussion led to more passivity. Only one study has directly measured some outcomes associated with students having time to think on their own (as opposed to asking student opinion): They found that, when students had time to think on their own, they spent more time arguing their ideas with their neighbors during peer discussion, suggesting that the conversation was of higher quality.
So, yes, it does seem to be important to give students time to think on their own before talking to their neighbors.
2. Does it matter if you show students the histogram (after the first vote)?
After students vote on their own, the instructor might show students how the class voted, or wait and have them discuss without knowing the class majority. I always advocate not showing the histogram, because I think it gives away the game and reduces the interest in discussion. The research on whether this is true is mixed: One study showed that students tended to converge to the more common answer when they saw histogram before their discussions, because they think that answer is correct. However, another study didn’t replicate those results. The current study authors suggest waiting until after discussion to show the results, to limit bias, but helping the confidence of those who got the correct answer initially.
3. When should you have students discuss the question?
When most of the students get the question right, it may not be worthwhile to have students turn to their neighbors to discuss. Several studies have confirmed this; students don’t learn as much from talking to their neighbors on easy questions, they learn the most when many students got the question initially wrong (during the individual vote). Even when the majority of the class get the question wrong (e.g. less than 35%), students still get benefit from talking about each other about the question. Above 70% correct in that initial vote, however, it may be best to skip the peer discussion.
4. Does peer discussion matter?
Some instructors might think that these gains can be achieved without having students talk to their neighbors; that it’s more efficient to explain the answer to the students rather than having them talk to each other. I have blogged about this before. Overall, the answer is yes, peer discussion does matter. Students are able to put together the correct answer through discussion, even if none of them knew the answer before. That suggests that the right answer isn’t spreading around the room, necessarily; students can construct their knowledge through discussion. Giving students more time to think about the question doesn’t get the same results; learning gains are highest when students talk to one another.
5. How much time should be given for voting?
This is not a very well-studied question, but one study suggested that students take longer to give an incorrect answer to an easy question than a correct answer, likely because those who know the answer can respond quite quickly. Difficult questions, however, take more time to answer, regardless of whether the student eventually votes with the correct or incorrect response. The authors of the current study suggest that when 80% of the students have voted, the instructor should give students a final countdown letting them know that they will end the vote soon.
6. How much does the instructor’s cues and explanation matter?
Since discussion is so important, what role does the instructor have to play? A lot! I’ve written about research on the instructor’s role before. Two studies have found that students learn the most when they discuss with their neighbors and the instructor gives their explanation for the answer; either one alone just doesn’t suffice. How the instructors cue students to discuss with their neighbor is also important; when students are told to discuss their reasons, rather than just their answers, students have much higher quality reasoning in their discussions.
7. Does grading matter?
I have written before about the pitfalls of giving credit for correct answers to clickers. This caution is mirrored in the current study; tying students’ grades to their ability to answer clicker questions correctly changes the dynamics of their conversation so that the discussion is dominated by one student, presumably the one that is more knowledgeable.
This research does have its limitations, such as little study of the relationship with student characteristics such as gender or underrepresented students. However, it can serve as useful guidance for instructors; the recommendations from the study are outlined in the graphic below.
Reposted from my article on the iclicker blog.