This diary was first posted as a series of comments, but I decided to expand them and write a diary so that people could actually see it.
As you probably know by now, brownsox has posted the latest Dailykos/R2000 poll numbers, this time from WV-02. Taken as they are, the numbers have Capito up by 13 points, and Obama trailing by 7. Not a set of very appealing numbers, considering that Ann Barth was supposed to be a strong recruit, and also considering the ARG poll that had Obama+8 in the state overall.
However, there is a very obvious problem with the R2000 poll, and I'm surprised that nobody else seemed to have notice it.
The poll methodology claims that
Quotas were assigned to reflect the voter registration of distribution by district.
However, this most obviously is not the case. The poll reports party identification to be D- 38%, R-33%. In contrast, the latest numbers from the SOS office had the statewide party breakdown to be D - 56%, R-29%. In 2004, it was D- 50%, R-32%. WV-02 is slightly more Republican than the state as a whole, but there's no way its even close to 22 points that way.
The West Virginia secretary of state office does not provide a breakdown of party registration by CD (at least if it does, I couldn't find it.) However, it does provide party registration by county. Using the National Atlas congressional district map, WV-02 consists of the following counties: Mason, Putnam, Jackson, Wirt, Calhoun, Roane, Kanawha, Clay, Braxton, Lewis, Upshur, Randolph, Pendleton, Hardy, Hampshire, Morgan, Berkeley, Jefferson.
Brief overview of breakdown by party ID in those counties (skip if you don't like number-crunching). Note that 'total' includes independents/others as well:
Mason: R-5,969 D-10,232 Total-18,339
Putnam: R-14,349 D-15,974 Total-35,048
Jackson: R-8,560 D-8,957 Total-20,429
Wirt: R-1,460 D-2,212 Total-4,251
Calhoun: R-1,228 D-3,481 Total-5,429
Roane: R-3,316 D-4,395 Total-9,293
Kanawha: R-37,826 D-71,943 Total-130,096
Clay: R-1,307 D-4,856 Total-6,666
Braxton: R-1,559 D-6,516 Total-8,800
Lewis: R-3,427 D-4,872 Total-9,441
Upshur: R-6,934 D-5,023 Total-13,730
Randolph: R-3,718 D-10,416 Total-16,526
Pendleton: R-1,496 D-3,021 Total-5,160
Hardy: R-2,078 D-4,660 Total-8,033
Hampshire: R-4,426 D-5,902 Total-13,231
Morgan: R-4,919 D-3,511 Total-10,897
Berkeley: R-23,400 D-23,103 Total-61,232
Jefferson: R-10,034 D-13,666 Total-32,902
Overall, I'm getting that the breakdown by party is R-33.2%, D-49.5%, I-17.3%.
Readjusting the poll to reflect these party breakdown, we get the following numbers:
Congressional race: Capito - 46.7%, Barth-46.2%
Presidential: Obama-46.5%, McCain-42.6%
And for a bonus number, if we assume that the rest of the state splits the same partywise, we get an Obama victory 50-39.
Obviously, these numbers look a lot more appealing than the ones posted on the front page. One reason why I might be wrong:- Voters may be registered for one party, but identify with another. This has been the case in a lot of states, but I don't think it makes a significant difference here, as even in 2004, WV voters reported themselves to be overwhelmingly Democratic; the exit poll split had D-50%, R-32%. In contrast, a state like KY has a 21% Democratic registration advantage, but that only translated into a 4% self-identified advantage at the polls in 2004; similar things happened in states like OK.
And by the way, if you're wondering if the district doesn't exactly follow county boundaries, the national atlas map is a bit confusing in that department, but I checked with a much clearer WV county map, and amazingly enough, it does seem to follow exactly along the boundaries of each county - no gerrymandering here!