In LV, they have D 39, R 29, I 31 split. So, it’s D +10, and McCain running behind the Party ID gap.
Regarding black vote - they simply allocated undecideds to Obama. 97-1 is still a stunner, but 96-3 is entirely possible. +50 in Hispanic vote is very suspect, but I would buy a +35 estimate.
So, assume McCain is +14 in white vote (twice the Gallup margin, and btw R2000 poll by Kos shows him +14 there) and -60 in non-white vote (not -70). Then he loses by 5% or so.
Gallup chose to double down on its bets, though.
We're both wrong. It's D39, I31, R29.
http://www.gallup.com/poll/111703/Final-Presidential-Estimate-Obama-55-McCain-44.aspx
"In the final poll, 38% of U.S. adults identified as Democrats, 34% as independents and 26% as Republicans. Among likely voters, the figures are 39%, 31% and 29%, respectively."
So, assume McCain is +14 in white vote (twice the Gallup margin, and btw R2000 poll by Kos shows him +14 there) and -60 in non-white vote (not -70). Then he loses by 5% or so.
Closer to three points. But even under this scenario, you need to keep in mind that much of this non-white vote is concentrated in big blue non-competitive states such as New York and California. Additional big chunks of blacks and Hispanics are in Texas (which McCain will win), Illinois (where Obama will win), New Jersey (where Obama will win), and various Southern states where McCain will win or should win even if not by as big a margin as Bush four years ago. And the Hispanics in Florida are Cuban and far more GOP-friendly than Hispanics in other states.
And even more importantly, with a more realistic party distribution, McCain's share of the white vote will be closer to 16 or 17 percent, with each additional point of white voter support added to his total or subtracted from Obama's total adding approximately 0.77% to the top line numbers.