Agreed. You can also survey groups likely to give the response you want, or use a method like phone dialing (as many polls use) with substantial non-response (which usually introduces bias). Long list... getting at the true state of affairs is hard!
I've been a statistician for many years, and I'm appalled by how many surveys and polls are based on poor science. Most have sample sizes which are too small, rely on overly simplistic assumptions about variance, and are deliberately and/or unintentionally biased. Where are the details about...