Peer Instruction and Student Guesses

I’ve been meaning to post this for a couple of days now: I think I’m going to have a lot to chew on in the most recent Science – more than just one post. So here’s just a little bit, on an article by Smith et al. on peer instruction. I think this might be the only research-on-teaching article in the whole issue!

Smith and co-authors discuss whether students learn anything when they discuss their answers to conceptual questions with their peers in class. In the Smith et al. classroom (a genetics class), the questions are multiple choice, and students use clickers to register their answers. Typically, students answer a question individually, discuss the answer in small peer groups, and then answer again – a variation on the Think-Pair-Share model (see Eric Mazur’s Perspectives article from the same issue). They then answer a second, similar question individually. In normal instruction, students would be shown a histogram of responses before they discuss the question with their peer group. In the Smith et al. experiment, the instructor did not show the histogram of student responses until after the groups had discussed and the students had registered their second responses.

In short, the authors find that yes, students do learn something. The number of correct responses to the first question increases after the group discussion. The number of correct responses to the second question is higher than the number of initial correct responses to the first. Students who didn’t initially answer the first question correctly got it right after discussion, and then went on to answer the second question correctly. Most surprisingly, some students who didn’t ever answer the first question correctly went on to a correct answer on the second one.

It is this surprising last group – the ones who only answered a question correctly after having worked through a similar problem with their peers  – that caught my eye. I don’t buy one of the explanations that Smith et al. give for these responses (that students needed a “fresh question” to be able to apply the knowledge gained through peer discussion). If I did the calculations right, there should be about 43 students in this anomalous group, so I doubt it’s an issue of the statistics of small numbers. Instead, I wonder if it might be a case of “the dinks.” In an episode of Season 4 of The Wire, a new middle school math teacher is surprised when a struggling student gives a correct answer to a multiple choice question. The student points out the chalk marks – “the dinks” – near the right response. According to Smith et al., some students said in course evaluations that discussions help because “you explore all the options and eliminate the ones you know can’t be correct.” Could the students be “looking for the dinks” in that they’re looking at aspects of the questions that might give them the right answer for the wrong reasons?

I tried to make an “educated” guess about the right answers based on wording, etc. My last biology course was AP Intro Bio in high school in the early ’90s, so my genetics knowledge is about as rusty as that of a poorly-prepared student. I might be able to guess the answers to two of the questions, but maybe my knowledge is better than I give myself credit for. I couldn’t find enough clues in the “intermediate” question (and don’t remember my Punnett squares well enough to answer). In any case, it’s hard to guess, since the answers are given in the supplementary material along with the questions. I’d like to hear some student explanations along with answers.

None of this is to say that students aren’t learning something from the clicker questions – it’s actually to the authors’ credit that the article brought up a further question. But maybe comparing the responses to random guessing (which the authors do) doesn’t give students enough credit. Most are rational people, albeit in ways that might surprise us.

In summary, spend some class time to let students discuss questions among themselves. The results might surprise you.