Stanford Prof. explains Polling on NPR

[Cattledog]

Stated that 1000 is the minimum # you need to survey to get a 3% MOE. So todays Zogby 1 day poll number is way outside any reliable model. When you look at his 3 day average of 1200 O is up +5.

[fishnfla]

1. Pollyanna, I know, but I've NEVER trusted polls

Its human nature to say onething and do another...the only poll I trust is on Nov4th, the vote , not an exit poll

[stoge18]

2. What's a Phd from Stanford know???

He / She is an elitist living in an ivory tower. Definitely not from the real America. :sarcasm:

[KharmaTrain]

3. Statistics Lie and Liars Use Statistics

First...a this "big news" was a one-day average...everyone else on this planet is using 3 day. And with 3 days til the polls close, even that 3-day will be rough to wipe out unless Gramps kicks ass today and tomorrow. And, remember, it's national numbers...the beauty pagent...meant for the talking heads to babble about and nothing more.

I get a strong feeling once this election is over, we'll see a massive revision of how polling is conducted. The 2004 model, as Gallup already admits is flawed and they're just guessing at what the real populace is. From a statistics perspective a 1500 person survey nationally (about a 2.5 MOE) is the standard for a poll...and smaller numbers proportional to what state they're sampling. The problem of upping that MOE to even 2% is that it would require doubling the sample...and I get a feeling these pollsters are hard-pressed to get 1,500 people who will talk to them.

Cheers...

[fishnfla]

4. I "heard" that the Investors Business Daily poll was the most accurate

of the last 2 elections, and is calling this one close. What makes theirs different?

[KharmaTrain]

5. Depends On Who They're Polling

They may limit it to their subscribers or a specific demo that reads their magazine. With a higher repugnican turn out in the past 2 elections, it would make sense that an organization that polled them would see their results verified, but this election has a far different dynamic.

The confusing thing is the way the Electoral College can make a close election look like a blow-out. Win the right states and you're golden.

These elections have become a big poll suck...it's the lifeblood of the 24/7 cables and the chattering class and a good statistician can dig into even the worst numbers and come up with something positive.

Cheers...