On a recent answer, which was quite poor, I observed the following statistics:
My comment stating that the answer is bad and I am downvoting it because of $reasons: 19 upvotes
Answer itself: 8 downvotes, 7 if you subtract mine
Only slightly over 1/3 [1] of people who obviously agreed that the answer needed downvoting, actually went on to vote the answer down.
[1] in all fairness, this assumes that all people upvoted my comment agreed with my assessment of "-1", whereas some people may have liked my critique/reasoning but not the "-1" conclusion I arrived at based on it.
Presumably, the -1 reputation cost is one of the reasons for this glaring discrepancy, though I don't know how to test if it's the main/only reason.
Downvoting bad answers is one of the most important mechanics of what makes SE sites in general, and especially Skeptics, produce good quality content.