Sometimes when your theory doesn't fit the data, you change the theory. Other times, you need to reject the data. For example, I
simply don't believe that John Edwards would beat Rudy Giuliani in Kentucky or that Giuliani would beat Barack Obama in Massachusetts.
UPDATE: It should be said, though, that one can't dismiss out of hand the possibility that Giuliani would win Democratic strongholds like New York and New Jersey.
This article available online at:
http://www.theatlantic.com/politics/archive/2007/05/things-i-dont-believe/42578/