On to the Keystone State.
Electoral Votes: 21 (23 in 2000)
2000 Results:
Gore: 51%
Bush: 47%
Nader: 2%
Buchanan: 0%
Pennsylvania, with its 23 electoral votes, was the third-largest state to go blue in 2000 - only California (54) and New York (33) gave us more. Thanks to post-census redistricting, PA had a couple of Congressional seats shaved from its delegation, but its importance to the Democratic nominee is still immense. Nearly everyone agrees that if the Dem candidate cannot win PA, then the odds of him winning the Presidency are virtually nil. It's not just because those 21 EVs would be almost impossible to make up elsewhere, but rather because it would signal that we're out of the running in multiple swing-states. My understanding is that this scenario would come about only if there were a broad surge of pro-Bush sentiment, tipping a wide swath of yellow and green states into red. I don't really plan to explore this potential disaster situation because a) it doesn't pertain to any specific swing state and b) it would rely on utterly unpredictable, large-scale events, such as a turn-around in the economy or rapid progress in Iraq.
So what factors should we look at?
A lot of people have commented about
the presence of Ed Rendell, PA's new Democratic governor and former DNC chairman. The conventional wisdom about governors says something like, "A governor can add 1-2% to his party's vote total." I'm a little skeptical of this because I'm not sure what this really means - or rather, how to analyze the validity of this proposition. Since every state has a governor (of course), how can you know what the real baseline is and how much on top of that the governor has added? Comparing polls to actual election-day results seems unsatisfactory to me because not only are polls often incorrect, but there's no reason to believe polls can filter out this alleged "governor effect", either. (You can't exactly ask, "Would you vote Democrat even if Ed Rendell wasn't governor?") This problem aside, Rendell's good connections and long experience may indeed help on the ground here. How much is, of course, impossible to say.
Then there's the Senate race: "Moderate" Republican Arlen Specter, who is running for re-election, is also facing a primary challenge from the right. People often suggest that a contested Senate (or gubernatorial) race can drive more people to the polls. Again I think it's hard to filter out the effects of state-wide races because so many other factors can affect voter turnout. Furthermore, a bitterly contested race may bring out partisans on both sides.
I will say this: If Specter starts faring poorly in the polls, then that is doubtless a good sign, sort of a reverse canary-in-the-coalmine. This may seem ultra-obvious, but I mention it because I think Specter's already showing some weakness. In a recent Quinnipiac poll, Specter is leading "someone else" (the pollster's phrase, not mine) by a mere 5% - 46% to 41%. That's pretty bleak, considering that 74% of respondents say they don't know enough to for an opinion of Specter's likely opponent, Rep. Joseph Hoeffel. And Specter's primary battle is only likely to drive up his negatives. (I don't think anyone ever emerges from a primary fight looking cleaner than when they went in.)
A quick note here on Bush's popularity: Like almost everywhere else, it's been plummeting - and plummeting sharply of late. Bush's immediate post-war approve/disapprove numbers (again from Quinnipiac) were 67/28 back in April. By August, they stood at 60/35, but in the last two months alone, Bush's approval sunk by a whopping 9 points to 51%, while his negatives rose by the same amount. In baseball, they talk of the "Mendoza line" when a player's batting average sinks below the pitiful figure of .200. If 50% approval ratings are the Mendoza line of the political world, then Bush is hovering awfully close to extreme mediocrity in PA.
Finally - and I think most importantly - I'd like to take a look at employment figures, which I did not examine with regard to New Hampshire. I confess this is where my analysis is weakest because my background in quantitative social science is, to be charitable, limited. But I'll forge ahead with a few thoughts in the hope that wiser souls can offer some more guidance in the comments section. (Also, a tip o' the tam o' shanter to MBW at Wampum who first inspired me to look at this topic.)
The seasonally-adjusted employment rate in PA stood at 4.2% in January, 2001 when Bush took office. In August of this year (the last month for which figures are available), the unemployment rate was a full point higher - 5.2%. If we believe the seemingly sensible proposition that this can only be bad news for the incumbent, then naturally this is "good news" for us (though obviously terrible news for PA residents and the country as a whole). I don't think this tells the whole story, though. Take a look at the maps here. (You can view them embedded into the story at SSP.)
The top two maps are from the Bureau of Labor Statistics, which is linked to above. They show unemployment by county in Jan. '01 & Aug. '03. As you can see, unemployment has actually decreased in most PA counties since the start of the Bush administration. (It is possible that the unemployment rate has been "artificially" lowered by people who have ceased looking for work, but I am unable to assess that.)
But take a look at the third map (which is courtesy of Dave Leip's Atlas of U.S. Presidential Elections, a fantastic resource). If you do a quick visual comparison, you can see that the Gore counties are the ones which have been hardest-hit. (Just be aware that Dave uses red for the Dems and blue for the GOP.) I've done a more precise number-crunching, and I can tell you that 10 of 18 Gore counties have suffered an increase in unemployment, while only 12 of 49 Bush counties have. Furthermore, nearly all the sharpest declines (those over 1%) have been in the Gore counties, particularly in the Philadelphia region in southeastern PA.
While I can't imagine that voters in the Republican counties are crediting Bush with a booming economy, it does seem safe to say that anti-incumbent hostility can only have increased in the Democratic counties due to the economic situation. Turnout in these areas (primarily Philadelphia and Pittsburgh) will be crucial, and I think that the sour economy will propel more disaffected voters to the polls on election day. Obviously, a quick turnaround in the economy can negate this analysis, but time is running out for Bush on this count. As you may recall, the economy had already started to recover from the '90-'91 recession when Bush, Sr. was running for re-election in 1992, but the effects of the recovery failed to reach many voters by November of that year.
Ok, so there is one last thing: Fred in the comments section observes that PA has the second-highest proportion of elderly citizens of any state (only Florida has more). (Census Bureau PDF, page 17.) If you saw the recent NYT article, you know that Bush's popularity is declining among older voters. And if you've ever observed politics, you probably also know that the senior citizen demographic tends to vote in very high proportions.
So, all in all, I think we have a pretty decent shot at keeping PA. It's been said that the Pennsylvania is "Philadelphia and Pittsburgh with Alabama in between," and I think the election will come down to voter turnout in those two urban poles at either end of the state. As I noted, I think some factors (like Specter's race & Rendell) are just too fuzzy to gauge, but the anger generated by joblessness is very real - and present. And I think this has the potential to be a major motivating force in the Democratic strongholds of PA and across the country.
But the bottom line is, this is a must-win for us. If you hear Jeff Greenfield calling Pennsylvania for Bush on election night, grab the Jack Daniels, because it's going to be a long four years.
(Also, I'd like to thank everyone who has been contributing in the comments section. Your insights have proven very helpful, even if I haven't been able to cite everyone by name.)