Surveys, if done correctly (random sample taken from a well defined population) can correspond to the population very accurately, even when the sample is just 600 people. The statistical error can get extremely low. The margin of error when the sample is 1000 is just 3%.