I don't know, but the key is to properly account for the biases in responses of those who do answer. Substantial percentages of people do answer surveys. Don't make the mistake of thinking along the lines of "i would never answer one, therefore nobody would"