Monday, December 24, 2007

Multiple Choice: Expected Value



When we discuss how our students answer exam questions, a common comment is that since they seem to just pick multiple choice answers at random, they might be better off just picking the same choice for all the answers, and get the same expected value outcome. However reasonable this seems, none of us had ever seen a student actually try this tactic, until this year, when a student tried it in StephenC's class.

When StephenC showed me that exam paper, I commented that maybe the usual practice of giving points for correct answers but not taking off points for incorrect ones actually encouraged students to try such a tactic. To try and preempt that, I set my exam to give 2 points for a correct multiple choice answer, and -0.5 points for an incorrect one.

I spoke too soon. Take a look at this exam paper. Apparently the possibility of getting points deducted isn't strong enough. And yes, the result is right at the expected value.

No comments: