I've noticed several where the average was below random guessing. Most of the time it is because the question is just horribly written - usually because in the attempt to leave out key clues such that you have to "know more" to get to the correct answer, they left the question with multiple equally or nearly so correct choices. Or, at least, because of the way the information was left out the remaining pieces almost exactly describe another condition that was unintended in the original question.
Sometimes it is because the mechanism is very obscure, and people tend to guess something that seems a little more familiar even if they have no clue what the correct answer really is. Those are fair, but probably very low yield for the actual exam.
Occasionally I just disagree with the question, either because it is so badly written that you can't expect people to consistently figure out what the heck they are getting at, or because the stem is so tangential to the answer that I'm not convinced it is actually correct.