Daily Kos has been hyping early voting returns as a sign of excellent Democratic trends, using Montana, New Mexico and Arizona as examples. Indeed, in the latter case, the margins are so strong for challenger Jim Pederson that Kos thinks he has a better chance of scoring an upset than "golden boy" Harold Ford in Tennessee. Given that many Democrats had completely given up on Arizona, this would be a shocker.
I don't know anything about how well early returns predict future outcomes. Does anyone know what type of salt I should be taking this data with?