1. Averaging out the state by state polls. (This is the base where I start from, but it is useless unless the following points are also taken into consideration)
2. Correcting rampant oversampling of Democrats (starting to get a little less rampant, bi-pollar as another, more famous blogger with the initials AJS likes to say).
3. Factoring in which states the candidates are campaigning AND advertising in down the stretch.
This is my number 1 factor. Their internal polls are more accurate than media polls, and the best way to get a window on the internal polls is to see where the candidates are putting their money where their mouth is, and also where they are putting their boots on their feet.
4. Including an increased Black turnout.
5. An energized youth vote/an apathetic youth vote. (Checking to early voting regularly to see how it is going. Obviously I can not know Election Day early voting.)
6. Voter fraud. ACORN. Walking Around Money. etc...
7. Throwing in a modest Bradley Effect (About 2 to 3 pts, since some of you have asked)...I don't believe it will be zero, but to play it safe I have been using a low number. Many liberals themselves have warned of up to around a 6 pt. Bradley Effect. So I'm using a pretty conservative number I think.
Here is the state of the Presidential Race
Today, Sunday, November 2, 2008: