AP's poll story was misleading not only in its claim, but also what it didn't reveal

[ProSense]

Josh Marshall

The AP ran a poll story yesterday painting a pretty dark picture for the Democrats based on big defections among Obama voters. A number of readers noted that the piece seemed a little misleading since the poll actually showed the roughly the same amount of Obama voters were defecting from Dems as McCain voters were bailing on Republicans (obviously the absolute number of Obama voters was greater -- but then again no one is expecting the Dems to kick the GOP's butts in two weeks.)

But now they've released the Congressional Generic ballot number. And it actually has the Dems up by 5 points.

Now, that number is way off the average of other recent congressional generics. And it was taken more than ten days ago. So I don't think there's much to read into this one particular poll. But what the AP drew out of it seems pretty odd in its proper context.

[Pirate Smile]

1. Josh had that AP article as the lead story on his website for a long time. Seemed very weird.

Somebody wasn't paying attention for a while. :)

[ProSense]

3. The headline should have been a big clue. n/t

[DFLforever]

7. That happens frequently at TPM , especially with polls.

It's like they outsource their headline writing to Politico or NRO.

[Pirate Smile]

8. Yeah, it was on the weekend and they seem to check-out on the weekends - not quite as bad as

MSNBC Lockout marathons but close.

[sharp_stick]

2. Anytime I see the byline Liz Sidoti

on an AP story my radar goes up. Her ability to mangle the simplest statistics is astounding, I'm amazed she managed to graduate from high school never mind Journalism school.

Totally pathetic story made even worse by the really horrible job of editing. Any editor that allowed that mush to be published needs to be laid off in the next round.

[FBaggins]

4. The part that it "didn't reveal" isn't much.

The poll was not, in fact, "way off the average". It was a poll of "adults" - not even "registered voters" (let alone likely ones).

It was also a mailed poll with requested responses online or by phone (semi-self-selected), so you can't really compare it to the bulk of the polling that's out there.

And it was taken more than ten days ago

It ended ten days ago. It started over a month ago.

I have no idea whether the poll is valid re: issue preferences, etc. (I would guess "no") ... but it could never be used as a measure of candidate electoral chances.

[ProSense]

5. Regardless

As Marshall states: "So I don't think there's much to read into this one particular poll. But what the AP drew out of it seems pretty odd in its proper context."

The entire poll is completely irrelevant to the election. So why leave out this information while intentionally distorting the data they selectively released?

[JamesA1102]

6. Thanks ProSense nt