Slate's Mickey Kaus suggested Saturday that "the normal tendency of voters to 'go with the winner' [might] be magnified this year because many voters who don't feel strongly about the candidates do feel strongly that they don't want it to be a close election with all the attendant Florida style recount madness. They will cast their vote to try to give the front-runner a 5 percent win instead of a 1 percent win."
Now, there may be some small bandwagon effect, but a) Bush tried this idea in 2000 and it certainly didn't work and b) there's no reason to think almost any voters are going to try to maximize the winner's margin because of Florida, especially when the election is probably going to come down to the wire and it won't be clear who is going to win. For now, the default public opinion is that Bush is going to take it, but that should change as the press starts to (finally) do the Electoral College math.
I put Kaus's speculation in the same category as Peter Beinart's story a few weeks ago claiming Howard Dean would be doing better than Kerry in the race against Bush, which is just ridiculous. Kerry is a terrible candidate who is crippled by his record and senatorial speaking habits, but candidates who set themselves up as anti-war liberals from Vermont don't win national elections. The problem is the journalistic emphasis on always having to say something new and counter-intuitive leads to a lot of dumb stories.