Well-designed survey results are objective and reproducible. There are margins of error as with all measurements, and maybe these margins are wider than in some fields and narrower than in others, but asking people for their opinions isn't necessarily bad science just because it happens to involve opinions.