Indeed - but, as I said, with MCQ's the handling/processing of the results has become very sophisticated, and I've been somewhat involved in the planning of such processes in my time. To give a few examples ... if a 'higher than expected' proportion of candidates give the same 'wrong' answer, the question (and its answer) will, as one would expect, be revisited. 'Poorly discriminatory' questions will either be excluded or given a low weighting - that basically relates to the situation in which correctness of response to a particular question correlates poorly with candidates' overall performance across all questions (and hence will include those questions which most people get right, or most people get wrong, regardless of their overall performance). However, it can then get much more clever in terms of looking at patterns of responses across questions and groups/types of questions.