Show me the histogram!: Best practices in Peer Instruction


Late on a Friday afternoon, I stood in front of 30 exhausted graduate students in a social science course I teach at Teachers College, Columbia University. Knowing that I had the perfect thing to perk them up, I excitedly informed them it was “time for a clicker question!”

As they ruffled through their book bags searching for their clickers, I displayed a beautifully crafted ConcepTest that I had taken much care to prepare and validate with some colleagues. The ConcepTest included what I considered to be an artfully constructed stem as well as three realistic misconceptions and one correct response built into the answer choices.

While my students were thinking silently before entering their first vote, I picked up on several non-verbal and visual cues in the classroom that led me to believe it was a great question and was going to generate excellent discussion and debate. When their clicker responses started to come in, a smile spread across my face as I observed around 60% had selected one of the most plausible misconceptions and 40% of the students had responded correctly.

Eagerly awaiting the loud chaos that was about to erupt through their conversations about the question, I asked my students to turn to their neighbor. And then, I inadvertently pressed the wrong button and displayed the histogram from their first vote for all to see.

Wrinkling my forehead in distress, I lamented out loud, “Oops, sorry about that.”  I shook my head at myself and thought, “I ruined Peer Instruction!” And at least for that question, I think I had.

Last week, a Peer Instruction Network member from Ohio asked, “should I show the results of the first vote, before asking students to engage in discussion?” The majority of Turn to Your Neighbor readers who responded (n=93) voted no, but 1/3 voted yes (see Figure 1).

Figure 1. TTYN User Poll 3/23/2012

As Peer Instruction Network member Steve Pollock from Colorado aptly commented, “the correct answer [to the question in Figure 1] is, as is almost always the case regarding any educational practice, ‘it depends’ (which was not included as an answer!) For instance, there are (admittedly only occasional) times when the histogram can *stimulate* conversation in very productive ways (e.g. when it’s an impressively close tie, or broad split).”

Brian Lukoff, a PI-Network member and math instructor at Harvard and Boston University, agrees. He suggests that while giving students information about other students’ first votes can change the context of their discussions, it also provides him unique opportunities to purposefully direct student conversations. “Sometimes, it can be useful to focus my students’ discussions on a particular set of candidate responses. And if I ask students a question in a free-response format, selecting a few popular student answers is one way for me to guide and focus the debate.”

On the other side of the debate, PI-Network member Stephanie Chasteen, also from Colorado, pointed to a research paper on this very topic by Perez et al. 2010, which reports that observations of “the most common response can bias a student’s second vote on a question and may be misinterpreted as an increase in performance due to student discussion alone.” To avoid both biasing students’ responses on the second poll and potentially biasing our before and after data, we recommend waiting until the second poll to display histograms. However, if you are not trying to measure the influence of student discussion on shifts to the correct answer this matters less.  And Steve is right, of course, decisions to display the histogram cease to be an issue when the first poll reflects no clear majority, or if your intention is to drive discussion based on the results of the first vote.

Another variation in Peer Instruction implementation occurs when instructors skip displaying the histogram resulting from the second vote, and simply move on to the next ConcepTest. We think there are some benefits to displaying the histogram on the second vote for the class, including that it helps signal the correct answer to students and diagnoses their understanding compared to the rest of the class.

So, Eric? Did I ruin Peer Instruction? On revealing histograms, Mazur says: “Generally, I think displaying the histogram on the first vote biases student discussion, so I almost never do that. But I almost always display the histogram on the second vote, because it helps me bring closure to the concept–I can explain the correct answer to the ConcepTest in addition to reflecting with the students about the misconceptions that would lead them to select one of the incorrect responses.”

1 Comment

Comments RSS
  1. Erin Malone, UMN College of Veterinary Medicine

    I agree :When two answers rise to the top fairly evenly, it almost always gets a laugh and generates fantastic discussions. I often do show those graphs prior to asking them to turn to their neighbor. The students can very quickly see the need for more discussion and that the answer isn’t necessarily as straight forward as they might have thought. (I do turn off the histogram during the voting; seeing it in action definitely influences them … and they like to make the graphs change).

What do you think?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: