| So, the PA-12 special election is tomorrow, occurring in the Kerry-McCain district. Ironically, despite the failure of the Pennsylvania dummymander (the GOP having lost the 3rd, 4th,
|6th, 7th, 8th, AND 10th since the 2002 remap and 2004 readjustment), this is yet another district where the GOP's intentions fell significantly short.
It's no secret that the 12th is quite the gerrymander, winding its way from Greene and Fayette County in the Southwest, through Washington County, with an arm through Somerset County, a large section of Johnstown and Cambria County (site of John Murtha Johnstown-Cambria County Airport, no less), another arm to pick up the college town of Indiana, three distinct sections of Allegheny and Westmoreland counties each (!!), and part of Armstrong County.
The 12th, however, is somewhat ancestrally Democratic - this was Joe Hoeffel's 4th best district in 2004, and turned out strong for Bob Casey. If you average the four federal statewide races since 2004 (Kerry v. Bush, Hoeffel v. Specter, Casey v. Santorum, and Obama v. McCain), the district's returned on average a Democratic performance of 51.3% to 46.2%.
We can visualize this as follows (click on images for larger versions):
The Democratic strength is concentrated in the southern bulb around Washington and along the Monongahela River south of Pittsburgh, as well as in Johnstown. Of course, connecting the two areas does require passing through some significantly Republican areas.
At risk of falling victim to the "Republican Heartland" fallacy, playing around with my new GIS toys, we can pull some NYT-style map goodness, with graduated circles:
In this map, the sheer Democratic dominance of Johnstown and the Monongahela River towns becomes even more evident.
So what does this all mean for tomorrow? Critz needs to do well in Murtha's old base in Cambria County, and hopefully stanch some of the Democratic bleeding in the southwestern half of the district.
As with Martha Coakley and Scott Brown, I also made an election-night model to predict results as they come in last night. It uses the similar uniform-swing assumptions (as compared to the 2004-2008 Democratic average) and accounts for possible variation within a given jurisdiction (this is necessary since counties here are much larger than towns in Massachusetts). I'm still fine-tuning the specifics, but expect that online sometime tomorrow afternoon!