Ranger REG said:This should be a purely democratic poll, one man or one household = one vote.
Actually, this sort of thing is part of the problem. You see, when people are given the opportunity to give information on what they watch, they apparently don't give accurate information. They forget to mention shows they do watch. They note that they watched a show, when what they really did was turn on the TV and leave the room, and so on. They are starting to develop new technologies that might help combat this problem.
Technically, there's nothing wrong with taking a subset of the population and using it to model the whole. So long as that subset is large enough, and chosen without bias, the statistics can be pretty solid.