I am not really a numbers guy, though I did study and conduct factor analyses with Michael Greenburg at Rutgers--long ago. I am a data guy, however, and I can do basic math. So here is a look at real data and probable projections. I will add notes on my assumptions as I go along.
Obama beat McCain by 10 million votes in 08. 133 million Americans voted in 08. I don't assume the number will be that high in 2012, but it could hit 130 million. 18-29 year olds will probably show the largest drop. White males and Latinos the largest increase.
Jim Messina claims that the minority vote will hit 28% in 2012. If this election were to follow the historical trend line, that would be true. White vote was 74% in 08. If it followed the trend line it would be 72% in 2012.
McCain got 56% of the white vote, 20% of the minority vote. If I discount Messina's calculation as spin and assume white vote will increase to 75% of the total [a huge swing, somewhere between 2010 and 2008] and give Romney 58% of it, that gets him to 43.5%. Assuming Romney gets 20% of the minority vote, that gets him to 48.5%. That is unlikely to happen because all other candidates besides Obama will get about 1%, mostly out of Romney's hide. But let's not go there for now.
President Obama got 44% of the white vote in 08, but now we'll give him 42%. That gets him to 31.5%. 80% of the minority vote [assuming Latinos are only 10% of the vote and AA vote gives 6% to Romney] goes to Obama, giving him 20% of the total vote. Adding 31.5 and 20 = 51.5%.
As far as popular vote, some are spinning that Romney could win the popular vote but lose the EVs. By my calculations, that is highly unlikely.
McCain won Texas by 11%. I'll say Romney will win by 20% and voting will increase by 1 million there. I'll also assume Obama's margins in most of the blue states will fall off, his margins in the purple states will be cut in half, and only in CA will he do as well in a blue state where voting increases. States like Alabama and Georgia, which McCain won by 400k and 200k will stay about the same, and assuming Romney wins Florida by 200k, rather than Obama winning by 200k [a 400k turnaround], you have Obama winning the popular vote by 7 million.
Those are the numbers. I'm willing to entertain any other math, such as Gallup saying that the white vote will leap to 78% or CNN saying Romney will get 59% of the white vote. If any of you think I'm wrong and Gallup is correct, make your case.