Tuesday, October 28, 2008

Contrasting polling in 2008 to 2004 and 2006

Here is an interesting graph showing the RCP tracking poll average from 2004.
For those like me, who are stressing out big time with the election only a week away, it's reassuring to see that Bush's lead against Kerry was consistently smaller than the lead Obama is sustaining over McCain. I haven't seen McCain ahead in a single poll since September. Here is a list of the actual poll numbers leading up to the election in 2004.

With so much divergence from pollsters trying to assess the make up of electorate this year it's also fascinating to take a look at the projections for the 2006 senate races, compared to the final results. Here are the numbers.

In the 2006 election in Arizona the Republican performed 0.4% better than the RCP average. In Maryland the Democrat performed 5.3% better. In Michigan 1.5% to the Democrat, Minnesotta 3.4% to the Democrat, Missouri 0.5% to the Democrat, in Montana 2.4% to the Republican, New Jersey 1.6% to the Dem, Ohio 2% to the Dem, Pennsylvania 5.5% to the Dem, Rhode Island 5% to the Dem, Tennessee 3% to the Dem, Virginia 1.2% to the Republican, and Washington State 5.7% to the Dem.

It seems as though in the traditionally Democratic states, Democratic turnout considerably exceeded expectations. I think we're going to see something very similar this election. I live in California, and even though this state isn't competitive everyone I've been in contact with is extremely enthusiastic about casting their vote as if it was potentially decisive. One wonders whether the same can be said in traditional red states like Texas, Utah, Oklahoma, etc, where Obama doesn't have a shot. For that reason, based on the numbers above you could safely argue that Obama will poll at least 1-2% better in the nationwide popular vote. Of course undecideds are probably going to break narrowly for McCain so this might end up being a wash.

Additionally, it's good to see that in Ohio and Pennsylvania, Democrats also out performed their polling in 2006. I think this bodes well for Obama in both of those states, but especially Pennsylvania where he currently has a double digit lead in the polls.

The bad news is that in the competitive red states, Montana, Virginia, and Arizona, the swing went in the opposite direction. As Republican's rallied towards the end of that election it was noticeable just how close both Democrat, Jon Tester in Montana (who had been widely favored to beat Conrad Burns) and Jim Webb (who had opened up a decent lead in some polls against George Allen) came to losing their races. From this you could argue that Republican turnout in Virginia, North Carolina, Indiana, Montana, North Dakota, Colorado, New Mexico, and Arizona might be much higher than is expected. Florida is anyone's guess.

The geographic diversity of these trends is what I think has to be taken into account with the current polling. For example Pew has Obama up by 15, Gallup by 7 in their expanded model, Hotline by 8, and Research 2000 by 7, nationally. By comparison Gallup's traditional polling model has Obama up by only 2, and TIPP, Zogby, and Battleground have him up by 3-4 points. Their turnout models are probably all correct, but geographically relative. This is a big country, some parts of which have been harder hit by the economy. I think it's fair to assume we're going to see different things happen in different parts the nation come November 4th.

Obama had a magnificent week by any estimation last week. From Colin Powell's endorsement, to the $150,000 spent on clothes for Palin, the infighting between McCain and Palin, it helped Obama reverse a trend that was heading in McCain's favor. State polling is a lagging indicator of the national mood and I think the extent of the leads you're currently seeing for Obama in Ohio, Florida, Virginia, Colorado, etc, are as a result of his successes last week.

By contrast, Obama had a poor weekend, and a terrible Monday this week, in my opinion. His campaign have a habit of losing their focus and reverting back to the very general bi-partisan optimism that doesn't translate well into soundbites on the evening news, and it doesn't come close to controlling the national conversation. Obama needs to be on the offense in making focused attacks on McCain. Right now the McCain campaign has free reign to throw their ridiculous bombs at Obama because he doesn't have to compensate for any of Obama's attacks. As far as I'm concerned every sentence that comes out of Obama's mouth from now until election day should include the words Bush and McCain. Obama has earned 50% + support, but now he has to ensure that the right wing attack machine doesn't have the leverage to re-brand Obama as a Marxist, terrorist, muslim, manchurian candidate that is only going to make the economy worse (as if that were possible).

However, today the right were unable to sustain that momentum. A McCain campaign source calling Palin a "whackjob" certainly didn't help. That is a good sign for Obama. Zogby's latest tracking poll will be out very shortly and while this should generally reflect Monday's narrative, if there is no trend for McCain that will be a good sign for Obama. Likewise, watch Gallup tomorrow, the most consistent indicator of the direction of the national mood, in my opinion, this campaign season. McCain's only hope is day in, day out, incessant negativity against Obama. Hopefully Bill Clinton campaigning with Obama in Florida tomorrow and the 30 minute prime time advertisement that airs tomorrow evening will assist Obama in preventing that from occuring.

UPDATE: Zogby shows Obama gaining 0.4%.