They interview only 1000 people? I mean I was looking to gay acceptance what if they happen to be all homophobes or pro-gay? :dry:
Surveys that are properly done attempt to sample a diverse population set, and then to recognize the possible bias and adjust for it. For example, if you are trying to take a look at sentiment nationwide about gay marriage, you might choose phone numbers all over the US. But then, during the call, you might also ask age, income level, political leaning, and sex, so that you can, at the end of the survey, ensure that the number of people you talk to roughly mirror the population of the US in terms of political leanings, sex, income levels, etc. Statistical sampling and data analysis is a pretty sophisticated science. The person that probably knows and understands it better than almost anyone else in the context of surveys is Nate Silver. He is the person who has predicted the outcome of the last two elections, both presidential and congressional races, with uncanny accuracy. You might check out his site, fivethirtyeight.com, and read some of his articles. It's very interesting reading (to me, at least) in that he explains all of his methodologies and then contrasts his results to those of some of the more established pollsters.
