Its nice to have healthy skepticism for these kinds of things, but I don't think we as laypeople know more about polling than the polling organizations whose job it is to eliminate biases with regards to the verbiage of questions and sampling. A lot of this stuff is based on well established models and prior theories.