One of the ways to do this (without outright lying) is to cherry-pick details, report them out of context, and imply a much greater significance than they actually possess. For example, a drug may be shown to reduce cholesterol, but to have absolutely no effect on the rate of heart attacks and strokes.
