***Attention Poll Watchers***

[JohnnyCougar]

Polls move around by random chance. In fact, a poll with a MOE of 3% means that 95% of the time the poll numbers will be correct WITHIN THE MOE. 1 out of 20 polls will show an error greater than the MOE, and therefore the poll is worthless.

Let me put this in perspective. If Bush is up by 3% in a poll, statistically it means next to nothing, just like if Kerry is up by 3%. Any results within 3% vary by random chance, even if NOTHING HAS CHANGED 95% of the time. Take a coin toss, for example. The chances of getting heads is 50%, tails the same. If you flip a coin 100 times, you will not get 50 heads all of the time!!! Sometimes you will get 53 heads, sometimes you will get 53 tails. Just because you get 53 heads once DOESN'T MEAN THE CHANCES OF GETTING A "HEAD" HAVE MOVED TO 53%!!! It just means you got 53 by random chance that particular time!

The same thing goes for the polls. If the nation has 50,000,000 Kerry voters and 50,000,000 Bush voters, you can survey 1000 voters randomly, and you won't question the same amount of Bush and Kerry voters every time. Even if the race is a dead heat, sometimes Kerry will have a 3% lead, sometimes Bush will have a 5% lead, because random chance says you will not get 50% all the time. In fact, a MOE of 4% means that 95% of the time, each candidate's results will be correct WITHIN 4 percentage points! The other 5% of the time, the spread could be off by 8 points or more!

So every time you see a Bush gain of 1 or 2 percent, or a Kerry gain of 1 or 2 percent, it's not like things are changing!! It is polling error due to random chance!!!

The recent polls show this race is a DEAD HEAT!!! Instead of proclaiming doom, phone bank for Kerry. Canvass for Kerry. Donate $$$ to the DNC, DSCC or the DCCC!! The more time you spend watching the polls, the more time you give up helping the Kerry campaign!

As Terry McAuliffe said to me personally, "Look at this race as if we're going to win it by one vote, and that vote is the one person you have to talk to and persuade!"

[The Velveteen Ocelot]

1. Thanks for the great analysis.

Sometimes we get our panties in a huge wad over these polls and take them way too seriously. If they were really that accurate we could dispense with elections altogether.

[JohnnyCougar]

3. You're welcome.

I for got to mention, everybody kick :kick: this thread when you are fed up with the poll-based fearmongering on here. This just plays in to the Rethug agenda. If we are behind, we are discouraged, if we are ahead, we are complacent.

It's a tie everyone, and we have to work harder than ever!!

[DemocratSinceBirth]

2. The More Polls You Have The More Robust Your Sample Is..

That's what Truth Is All and Professor Sam Wong (sp) at Princeton do...

If twenty polls show Kerry with a 3% lead the probability of him really having a three point lead is much greater than if only one poll shows him with a three point lead...

[JohnnyCougar]

4. Exactly

That's why freaking out over one poll is senseless!

[DuaneBidoux]

12. At one time I had found a prof at Princeton who was

like a really smart guy with all kinds of equations predicting a Kerry win. You couldn't really tell it was a Princeton prof but I followed the link up to find out he was like a physics prof. I've never again found that link but have wanted to forever. Is this the guy?

[Maat]

5. Sounds good. I persuaded more than one.

But I'm still going to be working on it.

[BUSHOUT]

6. Love that last sentance! Talk to people.....pursuade them.

Let them know in CLEAR and NON-PARTISAN terms how bad another 4 years of Bush would be.

[rockymountaindem]

7. Kick!

[DrFunkenstein]

8. So, Do You Think Tails Will Win?

Is that with Nader included?

[ClassWarrior]

9. Awesome fellow Cheesehead!

I'll be out volunteering for JK/JE tomorrow morning!

24.

[endnote]

10. AND when the samples are not random things are even worse

Like it might be the case for Gallup.

[JohnnyCougar]

11. Exactly...

The MoE and the 95% confidence interval all ASSUME that the sample is perfectly random and the methodology is flawless. Factor methodological errors like not sampling cell phone users and oversampling Rethugs will throw any poll off even more.