On an aside, this bit was useful for me in particular, because it helped me clear something up.
In my "final" analysis
here, I concluded that "it's hard to discern a strong impact of newly mobilised conservative Christians in this list [of states where Bush won most and least]".
But one of the three PS's I never got round to posting was going to note that I did come across one counter-indication that I couldn't properly explain.
The counter-indication in question was the
striking regional pattern that showed how the last state-level opinion polls "had on average been pretty much on-target in 13 of 16 non-Southern battleground states, but had Bush's support considerably underestimated in all six contested states in or near the South."
This article now suggests that this makes total sense, since in the South there
was in fact a clearly boosted turnout of weekly churchgoers. So the theory that Bush profited from a last-minute mobilisation surge of devout Christians who would not have shown up in the polls (if only because those who didn't vote last time were often not counted as "likely voters")
did hold true for the Southern states.
Its just that because the same did not happen elsewhere - because, in fact, elsewhere the share of weekly churchgoers in the electorate went
down because other groups were mobilised even more strongly - this didn't show up in the national exit polls. The numbers simply cancelled each other out.
A surge in turnout of devout Christians in the South might thus be said to have helped make the difference in states like AR, NC, perhaps MO or VA. But because their numbers if anything made
less of an impact than in 2000 elsewhere, the role of other crucial voting blocks like Catholics, seniors or national security-focused seculars was more important in most of the real battleground states, the ones in the Midwest and West that attracted all the journalistic-analytic attention.