I have been involved in an interesting project lately giving advice to an outside agency on how to set A, B, C, D, and failing grades on a multiple-choice sociology test. I have never used a multiple-choice test, and am not likely to. Nonetheless, I am glad I took part in this research project, because it helped me articulate why, exactly, I can't use a multiple-choice test.
My standard for B (good) work is that students show mastery of the assigned material. If they tell me back what I told them or assigned them, that is good. If they can do it in detail, that is very good (B+).
A (excellent) work requires B work plus something original. Their addition does not have to be absolutely original - not even Weber could do that every time. Rather, I want them to make their own connection between what we are studying and something else. I urge them to draw from other courses, their personal experiences, or at least material that we studied earlier in the term.
As a rule of thumb, I tell students that mastering the assigned material is a high school A and a college B.
A multiple-choice test only gives students room to show that they have mastered the assigned material. Even if they were able and ready to add original work, the format of the test gives them no place to show it. Thus, I can't tell A from B work on a multiple-choice test. And so should not use them. Which I don't.