Missed this post on accident.
But as a whole, you're thrown them out, which is why it's strange that you'd ask about information which is largely obtained by their aggregate.
That depends on your definition of "throwing them out". I think they represent elements of truth, but nothing that is accurate enough for me be wager anything. There's a reason why these political polls are done so much, because everyday it changes. After events such as the debates, the hurricane, Bengazi, unemployment, etc., the polls will change. So, if they will change every day until the day of the election, why should I even care enough to make a statement of who is going to win?
If that were the case, we wouldn't be polling everyday. We would just ride off our initial predictions, but since they can change at any given moment, they are only good for snapshots of the race. "What if the election were held today". If anything, that helps the candidates more than anyone.
You insisted on this before, and it's so very odd. Why do you think this is supposed to matter?
Because the main benefit of statistics is to get make educated predictions using math, you aren't mathematically proving anything. It's relevant because people are taking these estimations as proof as if it's impossible for the predictions to sway the other way.