Run the survey, do the experiment, crunch every number — and if the question at the top was vague, all of it was wasted.

The question is the part that decides what the answer is worth.