Be prepared for the media to do everything it can to suppress Republican votes, including calling as many states for Obama as early as they can and dragging their heels calling any states for Romney that aren’t blatantly obvious. That means guys out West will be voting thinking Obama is ahead and folks in the East going to bed with the illusion that Obama is going to win. When they wake up the next morning, they’re going to assume that Republicans stole the election again with dirty tricks when, in reality, the media vastly overstated Obama’s victories and downplayed Romney’s until the inevitable truth could no longer be avoided.
This theory is closer to reality than I want to think. I tend to agree. I’ll be flipping between CNN and Fox. I think NBCABCCBS will fall for this, CNN and Fox, not so much. As biased as CNN is they get huge numbers for their election night coverage because they are pretty good at it.
They definitely did so in 2000, esp in Florida. I remember when they called Florida for Gore. I pulled in the county by county data from Florida from 1996 and 1992 and started projecting final totals for Florida based on percent of precincts in -- and my conclusion was "this is too close to call".
Around 2 a.m. they called Florida for Bush (largely due to a clerical error), but I looked at the numbers again, and it was clear that it was too close to call.
The lesson to me was that these people rely far too much on exit polls and guesswork and not enough on the actual numbers. Since then I've set up spreadsheets with county by county results for key states and pull in the results as they come in and see where things are headed. It's easier when it's a one-key-state election (like the Scott Walker recall), but it's really the only way to get a clear picture.
The real key, especially for 50-50 counties, is how far ahead or behind the candidate is running compared to his same-party counterpart in the most recent close election. If Romney, for instance, is running within 1 point of Bush 2004 in various closely contested Ohio counties, that's a good sign (because then it comes down to turnout and margin in Cuyahoga and Obama's going to have to come close to his 2008 turnout *and* margin).