The Pew Sample: It All Makes Sense, Really

I see that the Pew poll's likely voter sample skews Republican 36-31 (scroll down to page 35, questions 68 and 69; too bad for Chicago fans, the band not the city, that it wasn't questions 67 & 68!).

But no, I'm not going to grouse about a pro-GOP sample. This is just how party ID works in polls. I said it when Obama led polls, and I'll say the same thing now when Romney leads. Pollsters ask respondents what they consider themselves to be. Usually, more say Democrat, because there are more Democrats than Republicans generally speaking.

But events influence what people say they consider themselves to be. Obviously, a lot of people watched that debate and thought, "Hey, I liked that Romney guy, I guess I'm a Republican."

The shift from September is remarkable. The September Pew party ID numbers: Dem 39, Rep 29. October: Rep 36, Dem 31. Indies were 30 both times.

That's a massive shift. And maybe it holds through Election Day, in which case Romney will most likely win. But this whole business is also a sign of the volatility of this particular question. Scroll further down to page 36 and you will see that the "normal" party ID distribution leans Democratic. Again, as with those New York Times numbers I discussed last week, this doesn't mean Pew is biased toward Dems. It means that that's how more people normally ID themselves.

The core point is that party ID is event driven. So Obama needs to create an event, or an external one needs to happen, to shift those folks back to thinking of themselves as Democrats.