In 2008, most knowledgeable observers on both sides of the aisle thought that Barack Obama was going to win the election against John McCain handily. In fact, out of the 15 big name pollsters, bloggers, and columnists who called the election beforehand with some specificity, 14 called it for Obama and 13 predicted he'd win in a walk with more than 300 electoral votes.
This year there has been a lot more disagreement because of discrepancies in the polling data. Long story short, the state polls and national polls don't match up. Moreover, Gallup and Rasmussen seem to be assuming a partisan breakdown that's going to be closer to the one we had in 2010 while most of the other pollsters are assuming the Democrats will have the same kind of enormous advantage they did in 2008 -- if not greater.
In other words, the 2012 election comes down to a simple question: Is the electorate that turns out this year going to look more like the one that voted in 2008 or 2010? If it's 2008, it will be close, but Obama will probably win. This is the assumption that's baked into the numbers of people like Nate Silver. On the other hand, if you've read or heard Dick Morris lately, he's started tossing the word "landslide" around. That's because, like and Gallup and Rasmussen, he believes the partisan voting breakdown will be much more Republican than it was back then.
The truth is, we really don't know who's right yet. That is why we have elections and don't just go by polls - especially not these polls. Believe it or not, there is generally a lot more agreement amongst pollsters than there is this year. Although many conservatives would disagree, in the aggregate, polls are usually a fairly reliable indicator of who's going to win a race. That's not the case this year.
So, the question becomes: Which polls do you believe? Although there's no way to be sure yet, I believe Gallup and Rasmussen. Not only do I think Mitt is going to win Ohio, I think he's going to win by a large enough margin that Ohio doesn't matter. Here's why I say that: