Polls exaggerate McCain Gains

[tannybogus]

The media is likely to focus on the near-term trendline -- one that shows movement toward John McCain within the last month. The last set of Quinnipiac polls were conducted near the peak of Obama's post-primary bounce, and there is no doubt that he has lost a little bit of ground since then.

We like looking at trendlines too. But focusing on only the last month risks failing to see the forest for the trees. Fundamentally, the news is that Obama is ahead in all three states -- two of which are states that Democrats have made a habit of losing. Moreover, if you compare his performance not just to the most recent number, but to all other instances of the Quinnipiac polls -- this is how our model looks at things -- the results are pretty decent for him:

This is a weaker performance for Obama than in June, but a better performance for him than in any month but June. Our model weights those two factors, and concludes that the status quo has more or less been preserved. As of last night, our model gave Barack Obama a 68.0 percent chance of winning the election in November. With these polls rolled in, he has a 67.7 percent chance.

Here's the other type of spin to watch out for. Quinnipiac's Peter Brown implies that the movement in the polls reflects a negative reaction to Barack Obama's overseas trip:

"The $64,000 question is whether Sen. John McCain's surge is a result of Sen. Obama's much-publicized Middle Eastern and European trip, or just a coincidence that it occurred while Sen. Obama was abroad," said Peter A. Brown, assistant director of the Quinnipiac University Polling Institute.

"While Obama was on tour, trying to show voters he could handle world affairs, voters were home trying to fill their gas tanks," Brown added.
This might be a perfectly valid point of view -- if Quinnipiac had conducted polling 10-14 days ago, immediately before Barack Obama embarked on his trip to Europe and the Middle East. But it didn't; the last time the Quinnipiac polls were in the field was six weeks ago. In the period intervening mid-June and Obama's Iraq trip, a number of different things happened: Obama took a lot of criticism for flip-flopping, the McCain campaign began to champion offshore drilling as a wedge issue ... the campaigns really picked up their advertising spending. Our model sees some decline in Obama's numbers over this period. But it also thinks that the decline has halted -- and has possibly begun to reverse itself -- since that time.

http://www.fivethirtyeight.com/2008/07/todays-polls-am-edition-731.html

[MJW]

1. polls don't mean crap this far out

anyway