I'm no statistician, but since the underlying population in this case (the set of all politicians) is much smaller than the general population, shouldn't that mean that a smaller sample size is more representative?