"Surely their internal polling must be showing similar trends."

I totally misread that at the beginning and thought it said "international polling". And the sad thing is that it didn't even alarm me.

I'm not challenging the poll (I hope it's true) but I would beware the media intentionally inflating the GOP post convention bounce in the coming week(s). As the bounce will ultimately decline thus giving the media the opportunity to pronounce the end of Palinmania and a new Obama surge.

You can just hear the welping from the Obama crowd, "But, we're entitled, he's The One! Arrrrrrrrrrrrrrrrrrrrgh!!!"

Hee, hee.

I can see the Obamamessiah choking on his waffle.

Just remember it is a poll of 1,000 people and I've been an executive in market research and polling (now retired).

This is a tiny sample; it would be considered not even accurate or worthwhile to anyone who knows anything about the size a sample has to be to make it meaningful.

Poll a couple million people and I'll be impressed with the results......

Explains the indignant tone the community organizer has had the last few days. Probably why he and the plagiarist were on the Sunday shows.

The fly on the wall reported in from Obama's campaign headquarters:

"Quick! There's not a second to lose! Start leaking the adultery story! Call all the journalists you know! Muffy! Get me the New York Times on the horn! Bruce! Get me the National Enquirer! We're losing ground here! Everybody, see Nkwama for today's denial sheet! Hop hop, people!"

Over 2 hours and not a single comment from any of our resident Trolls or trolls-come-lately?

There must be an emergency meeting at KosHQ for new talking points.

Diane,

I guess you must have outsourced your statistical advice to the NYT or something. We poll millions, but that's called election day.

Otherwise a sample starting roughly at 600 people gives you a pretty good idea of what you are trying to estimate. I'm going from memory (not having done this for a few years) but you roughly get +/- 5% at 95% confidence with sample size of 600.

My guess is you can get up to 99% with +/-5 or 95% with +/2 or so as you increase the sample size to over 2000. After that you really see hugely dimishing returns.

A sample size of millions would just be rediculous. Would you as "former executive in marketing research" actually pay 1000 times the cost necessaery (2 million samples vs 2000 sample - what is a large sample when the means are between 0.3 and 0,7) to get information that is only marginally better? LOL, only if you work for the NYT I guess.

It's not so much the poll that's important, it's the effect of folks hearing those numbers.

It can lull McCain supporters into complacency, but it can also demoralize Obama's support (as in "why vote at all?").

Let's hope (and change) it's the latter.

Seems there may be a good reason to fire the "odd couple".

Ed and Mitch at Hot Air interviewed Doug, a volunteer driver who worked for the Republican National Convention Committee in St. Paul. The summary ...

-- According to Doug, MS-NBC apparently took no chances on questions from the crowd. Rather than get caught with a question that might make Republicans look good, their producer pre-screened questioners, and Chris Matthews pretended it was random.
-- Republicans were good tippers. MS-NBC stiffed the drivers.
-- Media people talked in the cars about how effective the Republican convention turned out to be, while publicly saying something else entirely.

"Just remember it is a poll of 1,000 people and I've been an executive in market research and polling (now retired).

This is a tiny sample;..."

Diane, the size of the sample is effectively irrelevant: instead, the purpose of all such samples is to capture the makeup of the larger population. Polling companies go to enormous lengths to try and ensure that their samples are representative. As a former exectutive in market research I'd have thought you'd know this.