# Peer Instruction Fail

(This is not a criticism of Peer Instruction. It is just a tale of one of those times where the right idea failed to gain traction)

Yesterday we tackled a classic parabolic motion conceptual question in class, “three ships”, shown below.

They answered this question in their pre-class “reading” (actually I’m using smartPhysics so they watched a multimedia presentation). I am teaching two sections (35 students each) and will just combine their data here. On the first in-class vote, they were 51% correct. After discussion (5 minutes in one section, just about 10 minutes in the other), they revoted and were only 59% correct, which is a learning gain of 0.16 for those who like to use learning gains on clicker questions.

And the discussion was super animated. It seemed like it was so productive, but I guess it was really more about the two sides (B vs. C) digging their heels in.

And this does happen on occasion with Peer Instruction. You see 50% correct on the initial vote and smile to yourself thinking that the subsequent peer discussions is going to be a good one, but then sometimes the number of correct answers barely budges on the revote.

In the case of this question, it seems like I need some scaffolding clicker questions or other activities leading up to this question. Perhaps this scaffolding won’t improve the initial vote, but will perhaps give them more points of reference and examples to use in their discussions, helping the peer instruction really do its job.

Before coming to this question we spent 20-30 minutes developing position and velocity graphs in the horizontal and vertical directions starting from a motion diagram of a basketball shot, but did not talk explicitly about time at all during the sequence. Then we discussed the clicker question for the horizontal projectile vs ball drop demo  and they were nearly unanimous in getting that question correct. But it seems like more scaffolding is still in order.

### 6 Comments on “Peer Instruction Fail”

1. Are you saying this is a “fail” because only 59% got this question correct? In my experience (with the algebra-based class), I would be happy getting as high as 59%. This is a tough question.

• Joss Ives says:

Hi Rhett,

I’m saying that going from 51% correct after the solo vote to 59% correct on the revote after more than 5 minutes of very animated student discussion was a fail.

2. I think that some scaffolding questions after the first clicker question are essential to shaping the discussion. One of my main beefs about peer instruction is that it requires a substantial fraction of the students to know why their initial answer is correct and convince the others—when the initial results are close to random, the final result of peer instruction is not going to be much better. I’d want to see closer to 75% correct before trusting undirected peer instruction to do anything useful.

• Joss Ives says:

I definitely see my fair share of “failed” peer instruction for votes that were initially near 50%, but have also seen the results of the revote over 90% for questions where it wasn’t just a simple misunderstanding that lead to the initial wrong answer. There’s a bio-ed paper that shows that groups of students with wrong answers will often get to the correct answer during peer discussion despite none of them initially coming into the discussion with the correct answer. But for this to happen there usually needs to be at least a couple of different incorrect answers coming into the discussion so that they have something to discuss, which then can lead them to the correct answer that none of them had.

But I do have to say that if I see anything lower than 50% correct on the initial solo vote, I take a moment or two to consider how to proceed instead of just launching into peer discussions and a revote.

3. KNA says:

I’ve had those exact same results on that question…and I think you’re right, it turns into people who are convinced it has to be the same (I often do it shortly after the horizontal projectile vs dropped object example) vs others.

I usually “scaffold” at that point by tossing two erasers (or whatever I have handy) up in the air from my two hands to different heights, and asking which one will return to my hand first. Of course they unanimously get that. I then ask the battleship question a third time…with better success.

• Joss Ives says:

I did exactly what you did and tossed two batteries up in the air to different heights. And then revisited the question. And that is the kind of scaffolding I was thinking about bringing in to lead up to that question instead of giving them that question when they haven’t already had a chance to work through some of the important logical and conceptual pieces that go into the reasoning for that question.