Monthly Archives: November 2008

Undecideds in Pennsylvania – Did They Break for McCain?

Because it’s pertinent to our recent discussion on undecideds, I want to provide a brief overview of the movement of undecideds in Pennsylvania in the last two weeks. In looking at this data, don’t forget the ecological fallacy issue – we can’t be sure of individual level movement based on aggregate data.  With that in mind, Muhlenberg College conducts the only daily tracking poll in Pennsylvania. Since Oct. 14, Obama has gone from 51% to 52%, a gain of 1%.  McCain has gone from 38% to 44%, a gain of 6%. The number of “undecideds” or “others” has dropped from 11% to 5%.   This is consistent with – but not proof of – undecideds/others breaking for McCain at about 5-1.  Even if this is true, however, and the remaining 5% break for him at this rate, Obama will still win the state.  For McCain to spring the upset, he needs to see some slippage in Obama’s support rather than simply relying on winning the undecideds.  And this holds true at the national level as well.

Keep in mind that this is all based on one tracking poll with all the requisite warnings I have given you regarding relying on a single poll, random variations, margin of error, etc.  Moreover, Pennsylvania’s demographics do not mirror the nation’s, so one should not necessarily extrapolate trends from that state to the overall race.

What’s Happening in Pennsylvania – A Third Look

In yesterday’s posts, I made two arguments. First, given the demographics of the undecideds, more of them will likely break toward McCain (see the comments section for a good discussion on whether this is true and if so what might be the magnitude of that movement).  Second, that Obama was making a mistake by not focusing his resources on the battleground states in order to snuff out any glimmering hope McCain might have of winning this election.  Both arguments are pertinent to what is happening in Pennsylvania right now.  Simply put, the race continues to tighten there as McCain and particularly Palin have practically moved into the state, with positive results for their campaign.  Palin (who is now on the stump with “Iron Mike” Ditka) is drawing huge crowds, particularly in areas dominated by white, lower-income, less educated voters – the people who make up the bulk of the undecideds.  If you look at where she is traveling in Pennsylvania, it is largely in smaller, rural towns – neither she nor McCain have gone to Philadelphia at all, although Palin has spent time outside Pittsburgh. But she has been focusing her stump speeches in the so-called “middle-T” of the state which is more rural and conservative.

I have said all along that I did not see how McCain could win Pennsylvania, given the size of his polling deficit and the fact that it has gone Democrat in recent presidential elections. And yet McCain continues to defy my “wisdom” and concentrate his efforts there. Now it is clear that these efforts are paying off, although I am still not convinced he can win there.  Here is the latest polling data (all polls in last 5 days) showing McCain’s gain in the most recent two polls by each pollster in that state (most recent polls at top, Obama number first in each poll listed):

Latest Poll Previous Poll McCain gain

Rasmussen       51   47             53-46   (3 days ago)     +3

Morning Call    52   44             54-41 (2 days ago)       +5

Strategic Vision 49-44  50-43   (7 days ago)     +2

NBC/Mason Dixon 47-43        (No previous poll within last month)

Marist              55-41               53-41   (1 month ago)  -2

In the last two weeks, RCP’s average of the polls in Pennsylvania shows Obama’s lead dropping from13.4% to 8.5%.  Now, let’s be clear here. Obama continues to lead by margins that are for the most part outside the margin of error for most of these polls, and there’s very little time for McCain to make up the remaining deficit. However, the trend lines clearly favor McCain. More worrisome still, McCain’s gain appears to result from a combination of slippage in Obama’s support and the undecideds breaking heavily for McCain.  (Note that we have to be very careful in attributing individual level movement based on aggregate data [the ecological fallacy].  In this case, we see trends in the overall data, but we don’t know for certain how much is driven by “decided” voters switching support, and undecideds becoming decideds.  So take my generalizations with a very big dose of salt.)

Given this, I’m hard pressed to understand why Obama is spending money on advertising in Arizona, rather than here unless he simply has so much to go around that the marginal cost of spending resources in Arizona is negligible.  But if I’m an Obama supporter, I would feel much more comfortable seeing him pour resources in Pennsylvania to build the get-out-the-vote infrastructure and fund additional advertising.  Let’s remember: given the current state-level polling trends, almost every road to victory for McCain leads through Pennsylvania.  If Obama wins there, McCain has to put together a more difficult electoral map based on winning several smaller states from a combination of Nevada, Colorado, and New Hampshire – and he’s doing poorly in all of them.  So Pennsylvania is truly a “keystone” state in this election.