« Iraq and the Democratic Candidates, or Let's Fact-Check Frank Rich | Main | More Fact Checking about New Hampshire: The Famous 1968 Result »

Does Huckabee Have a "Catholic Problem"?

A couple days ago, Philip Klinkner posted this analysis of the county-level vote in Iowa and found that Huckabee did worse in counties with more Catholics. (Henry Farrell and Matthew Yglesias also noted Klinkner’s results.)

Some individual-level data would be nice here, if only to avoid (as Klinkner noted) ecological inference problems. Fortunately, the New Hampshire exit poll provides some. And the answer suggests that Huckabee does do a bit worse among Catholics than among other religious groups:

religrepsnh.PNG

Huckabee did better among Protestants (12% of whom voted for him) and “other Christian” (26%) than among Catholics (7%). Catholic support was evenly split between Romney and McCain. But it remains to be seen whether this “Catholic gap” is (1) big enough to be problematic and (2) durable, if Huckabee works to appeal to Catholic voters.

[Original post analyzed Romney, not Huckabee, for some reason. Thanks to commenter Sam for pointing this out. Clearly, I need to blog less about the election.]

Comments

Wait, I thought it was Huckabee who had the "Catholic problem." I have to go re-read...

County data in Iowa (see below) also show that Catholic counties went for Obama (particularly over Edwards), but the individual level data in New Hampshire show Catholics going for Clinton over Obama: http://www.msnbc.msn.com/id/21225995/. Could be that Catholics are quite different in these two states - and will differ by state even more so as the primaries move to places like New Jersey (Italian Catholics) and Nevada (Latino Catholics).

County-level Iowa data, standardized coefficients from multivariate model:

Catholic: Clinton (-0.02); Obama (0.28*); Edwards (-0.30*).

Education: Clinton (-0.43*); Obama (0.36*); Edwards (-0.09).

Income: Clinton (0.09); Obama (0.35*); Edwards (-0.45*).

County size: Clinton (0.05); Obama (-0.09); Edwards (0.06).

Given the ecological inference problems, interesting to note these county-level findings are similar to the individual level data for income and education in the entrance polls: Obama got the more educated, richer caucus voters.