Pennsylvania Reality Check

Horse race junkies are hyperventilating about the importance of Pennsylvania to the candidates' chances on Tuesday. In particular, some polls there have shown a 4 point lead for Obama compared to the double-digit lead that most other polls have shown. This has naturally led to cautious optimism among Republicans and anguished bed-wetting among some Democrats. There is broad consensus that McCain absolutely needs to turn Pennsylvania red for him to have any chance of winning the election; this is a state that has been one of the reliably blue states for a while (both Kerry and Gore won here).

Not to extinguish such entertaining anxiety on both sides, but, it may be worthwhile to compare current polls with those from 2004 when John Kerry carried the state by 2 points. First, the current polling trend from Real Clear Politics, where the spread of the averages for the week leading up to the election is 8.5 points:


Now, from 2004, my compilation of RCP data from September through November:


Not too difficult to spot the amateur graphic, right? However, I tried to use the same scale on the left, so the comparison should be clear: there is none. Kerry's highest lead during the period was never more than 4 or 5 points, and in some polls he was down by as many as 4 points to Bush (RCP average for the final week before the election was Kerry leading by 0.9 points!). Yet, the final result (the yellow tick mark above) of a 2 point victory was pretty much in line with the polls. Obama, by contrast, has averaged nearly a 10 point lead, and has not even trailed in a single poll since April!

Yes, anything is possible, there are worries about a Bradley Effect, etc. But, I would just like to say the obvious: if Obama loses Pennsylvania, he will have lost in a landslide to McCain, not in a squeaker. Also, I would expect much wailing and gnashing of the teeth among pollsters for years and decades afterward to figure out what the heck happened. This would be a cataclysmic event in the science of polling, falling into the "damn lies" category of how Benjamin Disraeli described statistics.

But, please, keep the anxiety going. It is very entertaining.