Skip to content

Here’s why you need to bring a rubber band to every class you teach, every time.

A student discussion leader in every class period

Recently we’ve been having a student play the role of discussion leader in class. That is, each class period we get a student to volunteer to lead the discussion next time. This student takes special effort to be prepared, and I’ve seen three positive results:

– At least one student has thought hard about the readings, and that alone can take the group discussion to a higher level.

– With student-led discussion, I talk less and students talk more. I think they learn more, and I’m always there to bring up points, answer questions, and guide the discussion as needed.

– The other students in the class—those who are not the discussion leader today—know they’ll have to do it themselves later on, so they have more of a sense of ownership and active participation, compared to the traditional instructor-led session.

But that’s not actually what I want to talk about right now. What I want to do is share something I learned from Ben Levine, our discussion leader today (actually the same class where we had the Reinhart and Rogoff questions that I just blogged).

The rubber band

What happened was this.

As usual, I arrived about five minutes early and I kicked off an informal conversation about statistics as the students trickled into the room. The discussion continued as class began, and I opened up the Jitt responses. At this point Ben raised his hand and reminded me that he, not I, was the discussion leader! So he came up to the front of the room and the discussion continued for a moment.

At this point Ben recognized that the conversation was going all over the place. We were talking about important topics, highly relevant to the course as a whole, but we’d diverged from that week’s topic. So he stopped and said: OK, let’s get back on topic. Let’s first discuss today’s topic, then we can return to the conversation we’ve been having, keeping in mind how it relates to our main subject.

This was really helpful, and I realized I should be doing it all the time in my own. We have great discussions in all my classes, but often we lose the thread. And if the conversation isn’t tied to the main flow of course material, it can be forgotten. Ideas are much more helpful when connected to other ideas we’ve been thinking about.

What Ben did was use a rubber band. Not a physical rubber band; a conceptual rubber band, tied on one end to the day’s scheduled syllabus material and tied on the other end to the class discussion. Digressions are fine, but you have to keep that connection, you have to keep springing back to the main points of the class.

This was great, and I’m gonna try doing this every time I teach. I actually already knew about this when delivering a lecture: when I give a talk, I like to pause from time to time and explain how all the pieces fit together, so the audience can see the details within the context of the larger structure. But, until now, I hadn’t thought of this as a way of keeping class discussions relevant.

Also, as with many risk-limiting tricks, I suspect that the stability attained by the rubber-band technique might well allow discussion to flow even more freely: as a student, you can feel more comfortable moving to a digression, if you are secure in the knowledge that the discussion leader will connect this to the key ideas you’re trying to learn that day.


  1. I learned this technique when I took over teaching our year long research methods class in our M.S./Ph.D. program in computational linguistics. I loved it, but it was hard on the students for reasons I’ll explain below (we ran it pass/fail and told everyone they’d pass if they participated because it would have otherwise been too stressful for students).

    Before I took over, David Evans (the previous professor for the class who designed its structure) had set it up to be a simulation of what students would run into outside of a classroom setting while doing research. This kind of thing would only be possible in a class the size we had—about fifteen students. Any more would’ve been very difficult because we couldn’t have gotten enough stage time for each student; any fewer than ten and it could have been problematic to get discussions going.

    The goal was to have the students learn how to do research by reproducing the whole refereed research process in miniature. The output expected from the class was a paper or a kind of mini thesis. (We didn’t officially have M.S. theses, because that involved all sorts of busy work like binding.)

    Before the yearlong class started in year two of grad school, students had to present literature surveys in their areas of interest (this seems to be a lost art, which is a shame—scholarship among scholars has never been that great in my experience).

    The class itself started with proposals, like mini dissertation proposals or paper abstracts. And we ran them just like that in class. The student first got up and did a short oral presentation of what they wanted to do. Then they got feedback and either a thumbs up to proceed or revisions (they were almost always too ambitious in the sense of being undoable, which is one of the hardest lessons for a grad student to learn). From the other students.

    For every presentation, we made one student the moderator just like they had invited an outside seminar speaker. The moderator was required to introduce the student, moderate questions, and also ask the student questions after the talk. The latter was to simulate the common situation where someone gives a talk and there are zero questions from the audience—the organizer has to step in to break the silence. My experience was just like Andrew’s—this really engages the students like nothing else I ever did in a classroom setting. I was there mostly to meta-moderate to keep things on schedule.

    The next twist was that the other faculty and I didn’t evaluate the proposals, the other students did, just like in real peer review. The written versions went out to other students for feedback, just like a paper. We did that again for the final drafts. This is a great way to get students thinking about other people’s work and wrestling with work in very preliminary stages, which is one of the fundamental skills of doing cutting-edge research. And it sure keeps them engaged in class. But wow, were they ever critical. Much more so than most real-world feedback from NSF or journals (though those can be pretty harsh at times, especially in linguistics, which seems to have more than its fair share of clueless jerks, which I largely chalked up to poor social examples in the field coupled with an aversion to mathematics and logic). So one of my other roles was to moderate the comments from other students and try to help establish tone so that nobody ran out of the room in tears. Putting ideas out there in front of other students for the first time when it feels like it matters is really stressful for a lot of students. I’m still not sure it wasn’t cruel and unusual letting the students evaluate each other. But like I said before, the evaluations didn’t count in the sense that if two student reviewers didn’t like a proposal, I just overruled them and let it go if it was OK after addressing legitimate concerns. And as I said earlier, the whole course was pass/fail.

    Anyway, to make a long story a bit longer, this is definitely worth trying.

  2. Jaellard says:

    “OK, let’s get back on topic.”

    … well, this seems fundamental to the necessary skillset of any professional teacher.

    It’s also basic in conducting any serious meeting anywhere.

    … not a fan of student-led classroom teaching, it’s usually a gimmick time-filler.
    But perhaps it has a place within some advanced graduate level courses, depending upon the specific academic degree field.

  3. Terry says:

    Very helpful post. News I can use. Thanks.

  4. D Kane says:

    1) Can you remind us of the technology you use for Jitts? I think this is covered in your book with Nolan but I can’t find it. Perhaps a copy/paste from your syllabus of how this is used?

    2) It is hard to make a sensible comment without knowing the size of the class.

    3) For me, cold-calling works much better than student discussion leaders. Much, much better.

Leave a Reply to Bob Carpenter