Over the last few

Start your day with TPM.
Sign up for the Morning Memo newsletter

Over the last few days, I’ve written out two or three lengthy posts about the odd disparity in the polls, only to decide after reading them over that they didn’t work for some reason and then scrapping them.

There is also a spread in survey data — surveys never match up precisely. But here we don’t have a spread so much as two sets of polls showing two markedly different pictures of the race. One shows the race dead even or with a small Bush lead; another shows the president running a commanding margin of roughly ten points.

In the last few days the pattern has repeated itself with Gallup and CBS/NYT showing roughly a ten point race and Harris and Pew showing it roughly dead even.

This sort of clustering can’t be explained by margins of error — not after it shows up so repeatedly. It can only be explained by different organizations using quite distinct polling models. Recently, Ruy Teixeira has been arguing, pretty convincingly I think, that Gallup numbers are skewed because they include a substantially higher percentage of Republicans than have shown up to vote in the last several presidential elections.

As Ruy notes this evening, the Gallup data show independents favoring Kerry by 7 points, while the overall numbers have him down by 8 points (these are Gallups registered voters numbers.) That’s hard to figure since in recent history self-identified Democrats always outnumber self-identified Republicans.

(See this post here for a more detailed explanation of Ruy’s argument. In fact, for all questions like these on the innards of polls I recommend Ruy’s blog.)

So, as I say, I think Ruy’s pretty convincing on Gallup.

But it’s not just Gallup. It’s also CBS/NYT and, if their last poll is any measure, ABC/WaPo as well.

I assume that these other surveys are using models similar to Gallup’s — and I’d like to see analyses of the internals of these other polls to show if that’s the case. Still, I’m not willing to dismiss all these big Bush lead polls too easily.

What Dems should take from this is that there are a slew of polling organizations that say this race is basically a tie and buck up their morale accordingly. On the broader question of what’s going on here, count me as still puzzled and, needless to say, it’s a debate I’d much rather be having on the other side of the 50 yard line.

Latest Editors' Blog
Masthead Masthead
Founder & Editor-in-Chief:
Executive Editor:
Managing Editor:
Associate Editor:
Editor at Large:
General Counsel:
Publisher:
Head of Product:
Director of Technology:
Associate Publisher:
Front End Developer:
Senior Designer: