Why do people still labor under the impression that "all of the above" makes for a good selection in multiple-choice questions?
In one question bank, I have 42 questions where "all of the above" is an option. In 36 of them, that is actually the correct answer. So there are 36 questions that are easier to guess than the others.
Multiple response questions are a much better indicator of knowledge retention. Sure, you still have the issue of computerized testing instead of someone evaluating your answers, but it's less likely to be guessed.
I often put in a section with "For these questions, between zero and four of the choices are correct. Choose all answers that are correct; if none are correct, write 'none'."
This sort of question must be graded as if it were n true/false questions, where n is the number of choices.
I once had a final exam that had multiple correct choices for its multiple choice questions, but because it was being corrected by an old scantron, you either got full marks or no marks on the question. Missing just one of the correct answers or being incorrect about an option being true led to a mark of zero on that question. Combined with the topic of the exam being the professional code, the exam was absolutely brutal.
I got about 68% percent on that exam. The highest mark on the exam was about 75%, and I was in the top six on it. That course was bell curved like you wouldn't believe (average is made to be a C).