I tried an experiment this quarter in my Human Factors class. I set up a SocialText wiki, extended invitations to all of my students, and told them that 10% of their midterm grade would be the quality of their submissions to the wiki. Everyone was expected to submit a minimum of 10 points worth of questions to the wiki, and I promised that once the submission deadline had passed, only questions on the wiki would appear on the exam.
Overall, the quality of the questions provided by the students wasn't great. Many were poorly worded, and they didn't cover the full range of topics we'd covered in class. However, I was able to extract a sufficient number of questions (some with wording changes to clarify them) to create a full exam, which they took today. Here's what the grades looked like:
That's just about a perfect textbook (I didn't mean to imply that as a teacher I want my students to not all do consistently great work...) curve from a pedagogical standpoint. And the grades matched my expectations, for the most part, in terms of how individual students would perform. So I'd call this experiment a success.
I will modify the process for the final exam. First of all, I'll provide a little more structure to the page before they start adding questions. I'll create specific topic headings associated with lectures and readings, so that they can see which areas need to have questions developed. I'll provide a few questions to "seed" the page, so they can see examples of well worded questions. And I'll moderate the page a bit more, marking questions that I'm likely to use in some way. I may even add a few questions of my own to supplement theirs. The key thing is that I don't want them wasting time studying for questions that are poorly worded, since that's not a good use of their time.
I'll report back after the final exam in February.
(Update: I've made the wiki publicly readable, though only students can edit it.)
Simply amazing. I wonder how this would work with a physical science course.