On September 22, the Rasmussen poll indicated a dead heat: 46% to 46%.
I checked Intrade, the Irish betting site. It said the odds favoring Obama’s victory are at a high point in 2012: 70.2%.
This discrepancy will be a good test retroactively of sampling theories. Rasmussen polls people who are eligible to vote. Intrade just reports the odds based on actual bets.
I think Obama will win. I think so because of Intrade. We will see in November if my trust in Intrade was justified.
Of course, the best test will be on the day before the election: traditional polling vs. betting.
If Obama wins handily, and other betting sites confirm this, traditional polling firms had better report an Obama victory, too. If they don’t, word will get out: they are as useless as horoscopes. But horoscopes are always popular.