Election myths, truths and another prediction
The "independents break for challenger" assertion is a total red herring anyway. The concept is based on a VERY small sample of Presidential elections, and doesn't hold true in that many of the few there are. The concept has no valid predictive power. There is only one election that you can really compare this to: 2002. This is the second post-9/11 election, and like the first one Republicans will outperform the polls.
Bush will win 52-47.