Nate Cohn analyzes the polling data on race:
[S]ince the first presidential debate, Obama’s support among white voters has fallen beneath the range consistent with reelection, even if minorities vote at the same rate that they did four years ago.
He notes that polls hint Obama may do better among Latinos than he did in 2008, suggesting a strategic miscalculation by Republicans:
There probably isn't anything the Republicans could have done to significantly improve their standing among black voters so long as they faced Obama, but there are plenty of Latino swing voters and 40 percent voted for Bush in 2004. If Romney makes a comeback in Ohio and Obama regenerates Latino turnout and wins Latino voters by as much or more than he did in '08 in states like Colorado, Nevada, or Florida, there's a chance we look back on the Republican decision to oppose comprehensive immigration reform (not to mention the DREAM Act) as the moment that ultimately cost them the 2012 election.
It’s hard to say how much Obama’s relatively low support levels among white voters is attributable to his race. But we are well on the way to caring a lot less than we used to, and viewing non-white voters as central, not ancillary, to political coalitions. So if Obama is re-elected with a historically low share of the white vote, it may prove as a useful Copernican reminder that we’re not the center of the political universe any more.
If you look at every state, you will find that the proportion of whites that voted for Obama varied from 10% in Alabama to 86% in the District of Columbia and 70% in Hawaii. Even if we exclude the most extreme cases the middle thirty states range from 33% (Idaho and Alaska) to 53% (Minnesota and Delaware). This is nothing like the cross state racial uniformity imposed by the calculator. The implicit assumption of the racial bloc voting calculator is that racial proportions are consistent across states and this is clearly untrue.
(Chart from Safiya Merchant)