Wednesday, September 16, 2009

Group Work

I spent a lot of in-class time doing group practice exercises during the class I taught this summer (for ruminations on summer classes in general, see this paired post). I just got my teaching evaluations back, and it looks like the students and I agree - group work was a success. To paraphrase my students, having to explain your thought process to others helps you to learn the material, plus the skills and experiences necessary to work in groups will be useful to have in the workplace. It also breaks up the in-class time so that everyone (teacher included) can use their brain in a different way and avoid that lecture class stupor. And while it does take more thought to prepare this kind of class, I'm not convinced that it actually takes more time.

So what is this group work thing I'm talking about? I did group work in three forms with my class this summer:

  1. Short, straightforward exercises reinforcing what we just talked about, done on their own or with the people sitting near them, usually lasting about 5 minutes. After introducing an algorithm and doing a sample run of it on the board, I would often give another input instance and ask the students to run the algorithm again.
  2. Longer, more difficult exercises applying the topic of the day with the aim of making sure all students can see a typical (easy) problem on this topic through from beginning to end, done in groups of about 4 at the boards around the classroom, usually lasting about 20 minutes, with me walking around and talking them through it when they get stuck. For example, proving that a problem is NP-complete.
  3. Out of class group project - I'm putting off a full explanation of this project until a later post, but in short, they were asked to work in groups of 2 or 3 on a 3-week long (out of 6 weeks) in-depth project.

Practice exercises in class - isn't that what discussion sections and homework are for? Yes (though this class didn't have a discussion section). But making sure the students understand the material is important enough to do multiple times, don't you think? In fact, most algorithms classes take the time to run the algorithm on sample input, it's just usually done by the professor at the board. So it doesn't actually take much more time to have the students work on it together, but it seems to me that they're much more likely to understand it if they had to do it themselves than if they watched you (and maybe a few of the more outgoing members of the class) do it on the board. Also, as the teacher, it gives you a chance for some mental breathing space, and a chance to find out what your students are struggling with instead of just guessing.

So, are you convinced? Already a fan? I'd be interested to hear other folks' experiences and suggestions.


Erin said...

I like the idea of having them run the algorithm on a sample input, and will definitely incorporate that next time.

Instead of having a 20 minute problem in a class, I spend the last class day on a particular topic as an in class exercise day. I give them 3-4 "easy" homework problems, with the assumption they will finish at least 2 of them in class. They work in groups, and I wander around giving hints or listening to their answers. The nice thing is, these usually can then be slightly modified to be easier midterm or final exam problems, since they're about the right level of difficulty and will be familiar but not identical to something they've seen before.

UIUC (where I did grad school) has "optional" extra credit session like this once a week which are run by TAs, but I found that I unfortunately can't commit that much evening time (since I don't have TAs or evening daycare). I like the once a week format better, but just can't give up that much class time in a 3 credit hour course.

I'm curious - do you find having a problem every lecture or two works better? I can see it being a nice change of pace, but I like have a whole day devoted to it also. Maybe I'll play around with both formats the next time I teach it.

Anonymous said...

I've used the following format in teaching a number of classes and have found it works well:

At the beginning of lecture, I give a short exercise which forces students to recall whatever we covered in the previous lecture.

In the ideal case, I can use it to set up my lecture for that day, but I do this even when the lecture is on completely new material.

Student evaluations have usually been positive about this aspect of the class.

The main problem with this format is it cuts into class times that are already very short. 50 minute lectures, for example, are hard to make good use of as it is. Cutting them down to 40 minutes makes it worse.

I suspect the optimal strategy depends a lot on where you're teaching. Students at top-flight universities tend to be better at reviewing material out of class, and they would perhaps benefit more from longer lectures with only occasional review.

Anonymous said...

Doesn't this idea only work when class times are rather long (as they are over the summer)? I can't see this working in a 50-minute class period, and it would even be a stretch in a 75 minute class.