Thanks for the detail, but why not just throw out a survey like AOL's whose findings would be considered statistical outliers compared to any other legitimate survey? If the "trending" is by some fortunate coincidence similar, the AOL poll becomes unnecessary. I doubt, for example, that a poll (like AOL's) that has Kerry losing his home state of Massachusetts -- an absolute impossibility since he leads there by 20-30% -- would reflect a similar trending upward for Kerry as we've seen in the Harris and Pew polls this week.

I've read that any online poll generally skews conservative regardless of the issue. I'm not smart enough in my sociology to venture why this might be, but I think that the discussion about how conservatives dominate short-form, quick-view media like talk radio or "political news" like Fox might apply to the Internet as well. Blogs are a good example of this: people can present stark viewpoints without editorial review with a modicum of research accompanied by flaming rhetoric. Certainly, it seems that conservatives (for whatever reason) are more motivated to vote in online polls -- I also think AOL's sample pool would skew significantly more conservative than the country as a whole.

You give this nonsense poll too much credit, especially for someone who's linking articles disputing poll methodology. AOL's is clearly the least scientific and least representative poll you've cited -- hence the 30%+ right-ward skew. That's why you'll never hear any legitimate conservative politician or analyst cite the AOL poll -- besides you, who are these conservatives you mention (bloggers don't count) that are touting the AOL electoral map? It might as well be drawn in crayon.

That is all.

Pollsters are in an interesting position, because unlike direct marketers or tele-marketers, they don't actually have to deliver. They just make all that shit up based on spurious data and methodologies, get paid, and skip off to fleece the next sucker, which usually is the previous sucker.

They do have the circle jerk going with the media conglomerates and PR firms, so it's an evergreen business. It's a great deal when you don't have to be accountable for results, but get paid up front.

Let's just go on and face the most basic fact. Our correspondent is relying on a fallacy:

. . .I think it does show some trending, particularly because over 300,000 people "voted" in it over a one week period beginning September 1.

We don't know that at all. The lack of controls means 300,000 votes may have been generated for the period, but we don't know why or how.

Bob has addressed the other analytical problems.

Bottom line: Give it up! The AOL straw poll is entertainment. The Right has latched onto it as anything but what it is. Going to the trouble of making up fanciful graphs based on phony data is worthless. Surely, one has something better to do with one's time. With a nod to Jim, masturbation would be a better pastime.

Wake up, everybody.

I think it's all pretty interesting - thanks Jeremy!

I am convinced that Bush will win every electoral college vote other than DC, which always goes Democratic because that's where the politicians live and everyone knows all politicians vote Democratic even if they are Republican, just like the journalists.

And immediately following that most vertiginous of landslides, I anticpate the pope and Queen Elizabeth will mud wrestle for a pack of condoms and a Milky Way bar.