Here's a fascinating reply from the BattlegroundWatch website to those who predict the race based on state-by-state polls:
The votes may be counted by state. But they are cast by race. White Americans overwhelmingly reject the Obama presidency. The Obama campaign expects white to provide 72% of the electorate - low enough that the Obama campaign can offset its big deficit among whites with super-majorities among non-whites. But if whites provide 75% of the electorate … it's lights out in Chicago. On those 3 points of turnout will turn the decision. Which is it?
The white share of the vote has been declining for many years, as shown in this Battleground chart.
But looking at the image, you see that the decline between 2004 and 2008 was unusually steep. Will it recur - or be reversed?
The doubling of the average decrease in White participation was a combination of 2 competing factors: first, non-Whites were excited over the prospects of the first viable non-White Presidential candidate and White voters of the opposition party were unenthusiastic over their candidate and did not participate in the election. Without the combination of these factors the White vote percentage of the electorate would still have declined but the decline would not doubled. Compared to the recent rate of change of -0.4pp, the change in the decrease of White composition from 2004 to 2008 was -1.4pp, 3.5x greater than the modern trend.
The folly of the Obama campaign’s election assumptions is the 2008 perfect storm that doubled the election-over-election decrease in White participation at a pace 3.5x as great as the norm will repeat itself with another -1.4pp rate of change resulting in a -4.3pp decline to 72%.
Or to sum up:
David Axelrod’s entire campaign is predicated on the above assumptions that expect a “White flight” that exists in no poll nationally or in any state. At a state level, it is due to differences of opinion like the above that both campaigns are reportedly seeing dramatically different electorates in Ohio with each campaign completely confident they will win the state. One of them is very wrong.
National polls often use 74% as the representative White vote in this election, but from a historic stand-point 75% is the more reasonable level which would be a -1.3% decline from 2008. With polls today consistently showing Obama’s support between 36-38% with this segment of the electorate comprising 75% of voters, it is easy to see how a tight race can turn into a blowout rather quickly. As for David Axelrod’s turnout model, he is talking his book when every ounce of data says he blowing smoke. If Axelrod is right on the racial make-up of the electorate, President Obama probably wins re-election in a close race. But there is little evidence that the 76.3% of White voters in 2008 when combined with a probable return of the missing 1.7 million whites will make up only 72% of the electorate Team Obama needs to avoid a sizable Romney win on November 6.