Should we really be using multiple choice tests at all?

This is factored out from a web page with instructions on how to use scantron sheets for multiple choice testing at the University of Delaware.

I used to dismiss this kind of testing, taking the attitude that essay and short answer questions are "educationally superior". Well, that may be true, but when you have more than 200 students, some compromises have to be made.

Most courses contain some set of "critical thinking skills" or "problem solving skills" that need to be mastered, and also some "body of knowledge" of facts relevant to a particular discipline. For example in a networks course, being able to describe the pros and cons of link-state vs. distance vector routing algorithms is a good essay question. Testing whether the students have figured out which one has the "count to infinity problem" is a question that can easily be tested via multiple choice.

So, my philosophy is that a 50/50 or 60/40 split between multiple choice and short answer/essay/problem type questions is a reasonable compromise. It allows me to spend more time doing a thoughtful job grading the problem solving, critical thinking problems, while also testing a larger proportion of the "body of knowledge" material.