"Hey, Professor Mulder...is this an opinion question?"
A few semesters ago, I had a student taking a test raise her hand to call me over with this concern. She was in the midst of of the test, doing her best to answer carefully, and the thought must have struck her that there were multiple "correct" answers to the question I was asking.
Not every question I ask on a test is cut-and-dried. Some are. Some questions are convergent: there is clearly one correct answer. Convergent questions are usually best for assessing relatively low-level knowledge and understanding. Can the students recall the facts? Have they mastered the vocabulary? Do they have an understanding of the basic concepts? Convergent questions are good for these sort of course material. By asking a convergent question on an exam, I am verifying that my students have mastered a particular concept. And this is valuable in it's way; there are concepts that I want all of my students to learn, and a convergent question is a way of focusing in on their knowledge of a particular concept.
However, I don't think that convergent questions are always the best questions, even on a test. I want my students to provide evidence of thinking, not just rote memorization. How will they use the basic concepts they have learned? I've written before about Bloom's taxonomy of cognitive objectives. Bloom's taxonomy is one way of thinking about different levels of thinking. Here it is in a nutshell:
