Rasmussen And Gallup vs The Rest, Ctd

Andrew Sullivan —  Oct 27 2012 @ 10:03pm

First off: a slight correction to this post. Gallup does not put much weight in its sample by party identification, as I wrote. To quote Frank Newport:

We don't find that party identification is stable enough to be of much use when it comes to comparing sample-to-sample variations, or sample to exit poll differences.

But it does measure such identification, and its results are way different than all the others, except Rasmussen. In fact, in yesterday's Gallup poll column by Jeffrey M. Jones, he said that Gallup found that the

key elements of President Obama's electoral coalition, such as racial minorities, women, young adults, and postgraduates will likely turn out at rates similar to those in 2008

And yet, they still give Romney a clear 5 point lead national lead which would easily translate into a Romney victory. WTF? Then we read about the largest shift in their polling:

The largest changes in the composition of the electorate compared with the last presidential election concern the partisan affiliation of voters. Currently, 46% of likely voters identify as Democrats or lean Democratic, compared with 54% in 2008. But in 2008, Democrats enjoyed a wide 12-point advantage in party affiliation among national adults, the largest Gallup had seen in at least two decades. More recently, Americans have been about as likely to identify as or lean Republican as to identify as or lean Democratic. Consequently, the electorate has also become less Democratic and more Republican in its political orientation than in 2008.

There's some confusion here. The 2008 Gallup numbers do not include independents, while the new ones do, and push them toward lean-Dem or lean-Rep. So this is an apples to oranges comparison. For the long and biggest view, here's the poll of all 705 polls on party identification since 2008 including Rasmussen and Gallup:

Screen shot 2012-10-27 at 9.00.31 PM

Blue is Democrat; red Republican; and Green Independent. Now remember Gallup's core conclusion:

the electorate has also become less Democratic and more Republican in its political orientation than in 2008.

But the full data currently show that the party identification is basically exactly the same as 2008. And you can see the decline in independents seems to be benefiting the GOP and the Dems pretty much evenly. They've both gained an extra two points or so. That's why Gallup and Rasmussen really are outliers. As Josh Jordan notes, in September 2012, when Romney was tanking

Rasmussen had Republicans … edging Democrats by 2.6 percent, with a split of 34 percent Democrats, 37 percent Republicans, and 29 percent independents.

In mid-September 2012, the poll of polls found Democrats ahead with 35 percent, Independents at 32 and Republicans last with 28 percent. So Rasmussen at the same time found Republican identification a full 9 points higher than everyone else. So what's really going on here is that Gallup and Rasmussen see Independents breaking massively toward the GOP since 2008, while the rest see it as pretty even. If Gallup and Rasmussen are correct, we are on the verge of a Romney landslide, due to independents. Or as Jordan puts it

anytime you have two surveys with such comprehensive data showing the same trend it is impossible to ignore.

But not as easy to ignore than all the other polls combined.