As Al Giordano at The Field explains, in PA it really comes down to the delegate math problem. Giordano predicted back on March 12th a 59/44 split of pledged congressonal district delegates in favor of Clinton.
But things are changing quite rapidly, and it's still too early to assess the impact of:
(1) Casey's endorsement of Obama
(2) The Casey/Obama bus trip together from Pittsburgh to Philly
(3) Clinton's loss of credibility because of Tuzla, NAFTA, Ireland
(4) Leahy's call to Clinton to withdraw
(5) Backwash from rich spoiled donor threat to Pelosi
So let’s take a look at the "starting picture" based on Giordano's predictions, qualified by jlkenney’s informative preliminary analysis of the PA congressional districts from March 6.
I've put together some cartograms (schematic maps where the area of district reflects the number of delegates it has) to illustrate what is at stake where. Since the geographically huge central PA districts have many fewer delegates than the geographically tiny urban and suburban districts, the cartogram gives a clearer picture of the balance of strength.
First, take a look at the predicted strengths of Clinton vs. Obama, prior to recent developments:
Obama's strong area (blue) is clearly the southeast of the state; Clinton leads in the west (pink). The fuchsia pink areas are predicted to give at least 75% of their delegates to Clinton, while the dark blue area (Philadelphia west) is predicted to give more than 75% of its delegates to Obama. The rose (Pittsburgh/Allentown) and light blue (rest of Philadelphia and suburbs) are predicted to give more than half of their delegates to Clinton and Obama respectively.
The lavender districts are potential ties. As Giordano points out, of the 19 districts in PA, 13 have an odd number of delegates (most bad for Obama), with only 6 having an even number, so that they can be fought to a tie. Of these six the districts containing York, Lancaster, Harrisburg and Reading are the only likely ties; Obama is unlikely to get more than half in the lavender districts and so they are probably not going to be hit very hard.
It is instructive to compare the map above with ones showing the distribution of key demographics from the 2000 census: race, poverty, and urban/rural.
First, percentage of population declaring themselves "white":
Compare the colors here with those on the horse-race map. Is race really so important, or are the prognosticators (knowingly or not) ascribing too much weight to race? And what is going to be the best strategy to change this picture?
Second, percentage of population in poverty:
Not too surprising here: high rate of poverty in Philadelphia; somewhat less but not inconsiderable levels in Pittsburgh; rural white poverty in the center.
Finally, percentage of population living in an urban area:
The correlation between Obama's strength and the urban/rural divide is also very strong. All districts which are "strong Clinton" are less than 84% urban; all districts which are "strong Obama" or "lean Obama" are at least 90% urban; of the highly urbanized zones, only Pittsburgh may lean to Clinton (it ain't over yet!)
Earlier I adopted the conventional Philadelphia-area "wisdom" that the only hope of winning would be to capitalize on the urbanized southeast. But I now see this is clearly wrong. GOTV in Philadelphia, while clearly important, will give Obama at most 4 delegates more.
The question remains, though, where can Obama make a dent? Pittsburgh and district 15 (Allentown) are worth watching and might turn to Obama; Reading (district 6) with its growing Latino population must not slip away. Whether 5 and 10 (State College and Williamsport) can be fought to a tie is also relevant.
There are plenty of indications that rural PA voters are feeling neglected. Can Obama bring them around? He will have to for the general, and now is not a bad time to start.
As the primary approaches, the horse race colors will change. Keep posted.