Yeah NM exit poll is hard to figure out. I can't find the results from 2000 right now, but 32% seems a little low. Why couldn't the 44% be correct for 2004? That was the national average according to CNN exit poll. That would be a 12% swing, but nationally hispanics swung 9%, so it's not that outlandish. Couple that with a slight swing among white voters to Kerry and it could add up.
In 2000, Bush got 47.9% of the vote in NM, same as nationally. In 2004, he got 49.8%, 0.9 points less than nationally. So (unless we're assuming a massive increase in Latino and Indian turnout) the Whites would have to swing Dem almost as much as the Hispanics swung Rep.
Here's a localized look, btw...presentation follows the following scheme -
New Mexico 42.1% Hispanic 49.8% Bush in 2004 +1.9 change of Bush percentage over 2000 26.3% increase in vote cast 41.6% votes cast in 2004 as percentage of 2000 census population
Spanish North New Mexico
Rio Arriba 72.9% 34.3% +5.4 24.0% 36.4%
Taos 57.9% 24.7% -0.5 36.1% 49.5%
Santa Fe 49.0% 27.9% -0.3 33.8% 51.2% - these two counties' gringo population is pretty left wing too, o/c
Colfax 47.5% 51.6% +4.0 9.4% 42.1%
Mora 81.6% 32.8% +2.3 29.0% 54.6%
San Miguel 78.0% 27.3% +3.1 32.2% 40.2%
Guadalupe 81.2% 40.3% +7.1 37.4% 48.4%
Little Texas
Union 35.1% 77.3% +5.0 7.1% 45.1%
Harding 44.9% 59.0% -2.7 8.7% 79.5% safe to say Harding County has a ballotstuffing problem
Quay 38.0% 64.6% +5.0 7.3% 40.6%
Curry 30.4% 74.5% +5.1 19.3% 31.7%
Roosevelt 33.3% 69.9% +3.3 26.7% 39.6%
DeBaca 35.3% 71.1% +8.8 1.0% 44.3%
Lincoln 25.6% 67.3% +1.5 33.0% 46.4%
Chaves 43.8% 68.1% +5.4 19.4% 35.4%
Lea 39.6% 79.4% +8.2 27.0% 32.8%
Eddy 38.8% 65.5% +7.4 14.2% 39.2%
Otero 32.2% 67.6% +4.4 28.0% 33.3%
Center of the state
Bernalillo 42.0% 47.3% +0.7 25.7% 46.1%
Sandoval 29.4% 50.8% +2.2 40.2% 49.5%
Los Alamos 11.7% 51.9% -3.1 9.6% 61.0%
Torrance 37.2% 61.9% +3.6 31.2% 38.5%
Valencia 55.0% 55.6% +5.1 21.7% 39.3%
Socorro 48.7% 47.1% +0.6 15.0% 43.4%
Western New Mexico, which is fairly diverse with counties not resembling each other much
San Juan 15.0% 65.6% +3.8 29.9% 39.5%
McKinley 12.4% 35.6% +3.7 29.8% 27.6% overwhelmingly Navajo (74.7% Native American)
Cibola 33.4% 46.4% +7.6 5.5% 29.3%
Catron 19.2% 71.6% -2.8 16.5% 56.3%
Grant 48.8% 45.8% +1.7 31.3% 43.2%
Hidalgo 56.0% 55.0% +3.0 7.1% 33.1%
Luna 57.7% 54.8% +3.5 14.8% 30.4%
Sierra 26.3% 61.3% +2.0 12.5% 38.9%
Dona Ana 63.4% 47.7% +2.1 32.8% 35.5%
Other counties with sizable Indian populations are
Cibola 40.3% (Zuni and Navajo)
San Juan 36.9% (Navajo)
Sandoval 16.3% (diverse Pueblo)
Rio Arriba 13.9% (diverse Pueblo)
Socorro 10.9% (diverse Pueblo)
Taos 6.6% (Taos Pueblo)
Otero 5.8% (Mescalero Apache)
And maybe I should have added data on population size as well... DeBaca and Harding and Hidalgo are
tiny. Bernalillo is huge in the NM scheme of things...but that would have made the table even more confusing...
Certainly the really heavily Latino counties swung as a rule a little more Rep than the average, and Reps lost in heavily Anglo Los Alamos (very white collar, you know why) and Catron (Mormon IIRC). Turnout also was up much less than averagely (or maybe population growth since 2000 simply lagged) in Little Texas (the Hispanic populations there are mostly first-generation immigrants from Mexico, and most of them probably don't vote. In fact many are Illegals. Notice most of the region has low participation rates. In the Indian areas the same phenomenon is probably due to massive demographic growth, ie a large proportion under 18) But I certainly don't see anything allowing for a
12 point increase in the Rep share of the Hispanic vote. An increase from about 35% to 39%, 40% is more like it.