There's an instructional technique that's very popular but rubs me the wrong way. I'm talking about "pair-share".
Basically, you ask a question of the class, have them turn to a partner and share their answer or discuss the problem, and then you call on some people to share their answers with the rest of the class.
Does it work?
That's a question I've wondered quite a bit about. And to find answers to such questions, I tend to look to science. Because well... science works better than anything else.
A search for evidence for pair-share's effectiveness yielded pretty much nothing but descriptions of what pair-share is, why it's (supposedly) so great, and when you should use it. So basically, heresay and dogma.
I did, however, find one study out there that describes itself as 'quasi-experimental'... which I take it means 'un-scientific'.
Carss, W. D. (2007). The effect of using Think-Pair-Share during guided reading
lessons. Unpublished thesis, University of Waikato, Hamilton, NZ.
This is a master's thesis for a masters in education. If this were for a masters in psychology or cognitive science, that would lend the study at least some heft. But in the US, a masters in education is about as impressive as a masters in underwater basketweaving.
So I don't have high hopes for this study.
Basically, a group of students was split into 4 groups with 3 of those groups doing some sort of pair-share-enhanced reading program. The fourth group was a control.
While a lot of text is used describing the pair-share program, the methodology for the control group remains pretty much a mystery.
The results were that the pair-share group did better on comprehension tests.
One of the problems with this study is that it's not clear that this is an apples-to-apples comparison. We don't know if the pair-share and control sessions were of the same durration, for example. Possibly, the greater time required by the pair-share is responsible for the difference.
We also don't know if the control group did similar activities. We know that the pair-share groups were asked to predict certain things as they read, but what about the control group?
A good study would try to isolate the act of sharing with a partner as much as possible. So the control and pair-share groups should do essentially the same thing. The only difference should be that students in the control group 'share' information with no one but themselves, while the experimental groups share information with a peer.
Unfortunately, this isn't that kind of study. It's admittedly, only quasi-experimental.
We can't say that the results are from sharing something with a peer, spending more time on the subject matter, or any of the other differences between the control and experimental groups. In fact, we don't even know what those differences were because the lessons in the control group were not adequately described.
This doesn't mean that pair-share is a bad idea. If it works, go ahead and use it. I'd just like to see it tested in an actual experiment.