New York Daily News-What Zogby tells me:Kerry WINS

[Richardson08]

Pollster John Zogby, in a telephone interview with me yesterday, predicted that John Kerry will win the election. "It's close," he said, "but in the last couple of days things have been trending toward Kerry - nationally and in the swing states. Between this and history, I think it will be Kerry."

When Zogby talks, politicians listen. He made his bones in the Bill Clinton-Bob Dole election of 1996, when he came within one-tenth of a percentage point of the final tally.

Bet me that when the Bushies read what Zogby told me, not just the rhetoric will rise, but so will the fever.

Particularly since one of their favorite columnists, Robert Novak, reported in yesterday's Washington Post that Zogby called the race for President Bush in a conversation he had with the pollster on Monday.

Zogby was jocular about the Novak column, although he has decided not to post a comment on his Web site. Here's what he told me: "I said Bush was winning, I didn't say I thought he'd win. On Monday, he was indeed looking good. But on Tuesday, things changed. Kerry, in that one day, picked up 5 points."

:toast:

http://www.nydailynews.com/front/story/247447p-211694c.html

[Quetzal]

1. what was his prediction in 2000?

[NorthernSpy]

7. Zogby (alone) predicted that Gore would get the most votes

I can't think of another major poll organization that got that one right. As far as I know, the others all called the popular vote for the shrub.

Most polling methods do tend to oversample Republican constituencies, for a number of reasons.


Mary

[Kahuna]

9. Yep. And Zogby alone had Gore winning, Florida.

nt

[robbedvoter]

20. Zogby alobe stopped cheating in W's favor - because Village Voice caught

him doing that.
Amongst his more recent predictions: Edwards will win NY in the primaries....
Sorry, I KNOW Kerry wins, but if Zogby sez it I get nervous

[sweettater]

2. I have a nagging question

If and I say IF * pulls this off and "wins" with all these scandals just in the last few days, are we not going to ever get this country back on track? I mean it will be constant investigations, etc. This stuff will not just go away.

[Brucey]

8. Yes, IF shrub wins.....

It will be a very low day, maybe the lowest in American history. Shrub is an awful candidate, has a terrible record, can barely speak, is arrogant and clumsy, and has made many, many errors. Kerry is tall, experienced, well-spoken, has a good past and good experience with no scandals... in other words, a good candidate. If an awful Repug can beat a good Dem, what the hell hope is there???? What does this tell us about the US?

[Almost_there]

26. Kerry is... tall?

I didn't put that down on my list of reasons to vote for him, but, what the hell, right? I'll go with the "tall" candidate. Although, there truly are people who vote for the candidate based on looks. Hence, Edwards being a good choice.

~Almost

[JoFerret]

33. I agree

...in the highly improbable event that Bush were to win then either we will know that the election was stolen again or that we are all truly doomed. Bush has been a worse president and a weaker candidate than we could have predicted while Kerry has risen to the occasion magnificently and given every sign of being a strong and competent leader. And with the issues so stark and so deeply important in every area I can imagine no other outcome than a Kerry landslide. Take any ONE of the huge issues with which we are confronted - Iraq, security, economy, social security, education, environment the supreme court and its implications for civil rights etc etc. and BushCo is either weak or incompetent or headed down a dark and dismal road to ruin.It is hard to imagine living in a nation that would choose four more years of this. Hard -or at least very depressing - to imagine a world that could survive it.
but then - it is not going to happen. So there we have it. Kerry in a landslide.

[JSJ]

3. work both sides- tell both what they want to hear

And they'll buy your shit til doomsday. Zogby's got this America thing pretty well worked out.

[non sociopath skin]

4. So that's official, then. Zogby thinks that either Kerry or Bush will win.

Man, we can't pay these people too much!

The Skin

[DoYouEverWonder]

5. Zogby said the same thing on Jon Stewart last night

Kerry is going to win!

[Celebrandil]

6. I made a comparison between Zogby and other pollster...

and came to the conclusion that Zogby gives Kerry a 0.9 percentage points higher score (and thus Bush 0.9 lower) than other pollsters on average. On the other side of the spectrum you have Strategic Vision (-0.7) and Mason-Dixon (-1.1). Rasmussen was in the middle of the pack and significantly more stable than all the others. I got the results by calculating the trends using the last 9 polls for a large number of states and then measuring the deviation from the trends for each pollster. So... the question is who you'll trust the most?

[Justice]

10. Between You and Zogby - Zogby of course, cuz

you can't back up what you say with any proof.

It is Gallup that weights its polls in favor of Bush.

[Celebrandil]

11. I'm sorry...

but you obviously didn't understand what I did. I was not trying to get my own estimate in an attempt to compete with Zogby. I was comparing pollsters head to head, without judging which pollster to trust.

By the way, I didn't include Gallup because they had too few state polls among those 700 polls I looked at.

[cubsfan forever]

13. And you base this "weighting" of polls

on.............what? Remember, there are three kinds of lies: Lies, Damned Lies, and Statistics.

Professor 2

[Celebrandil]

15. I don't weight the polls...

they have equal weights, but once a low-dimensional curve is fitted to the data, the deviations between polls and the curve are measured.

However, after this it's possible to apply an M-estimator and curve fitting once and then you use different weights for different polls, but the weights are set automatically from the initial deviations. This is standard robust statistics.

Nor professor yet... just PhD in computer science

[cubsfan forever]

16. Are you talking standard deviations?

Because these are determined after applying the instrument and crunching the numbers, not before.

Professor 2
(Ph.D. completed)

[ProfessorGAC]

18. LOL!

The Professor

[Celebrandil]

24. No...

the mean error between samples and curve, for each respective pollster, with curves fitted to data from all pollsters. The M-estimator I used only to get a better prediction of trend, but reducing the influence of outliers (shitty polls).

[ProfessorGAC]

17. You Sure About That?

Standard Robust Statistics, huh? With no knowledge of the integrity of the dataset? Kind of blows the robustness out of the water.
The Professor

[cubsfan forever]

19. LOL! Professor 2 n/t

[cubsfan forever]

25. Seems to me you are right back where you started from

Perhaps you should read post #17

Professor 2

[Celebrandil]

27. I don't see the problem...

You dont need any robust methods to draw the conclusion that different pollster give significantly different results.

[cubsfan forever]

28. Then why all of the "robust method" minutiae?

Besides, the first rule of quantitative research is to adjust the hypothesis to fit the data (if necessary), not the other way around.
"And the circle remains unbroken..."
Nice talking to you. Bye!

Professor 2

[Celebrandil]

29. Because you asked me what weights I used...

which I never used, but hypothetically could have to make the trend less sensitive to "odd" polls.

[cubsfan forever]

31. Once again, circular reasoning. Bye!

Professor 2

[ProfessorGAC]

12. Rasmussen's Methodology Is Flawed

Their stratification techniques are so rigid that there are essentially no swing voters in their sample. So, of course, they would be more stable. They are essentially constantly polling the same people.

If i call 500 hard and fast Kerry people, and 500 hard and fast Bush people, i'll get 50:50. If i use the exact same stratified random sampling technique the next day, i'll get the same result.

If i were to do this, i would automatically toss out Gallup and Rasmussen. ARC, Pew, Zogby, and ORC are the best methodology firms.
The Professor

[cubsfan forever]

14. Well said, Professor! n/t

Professor 2

[robbedvoter]

22. I don't trust any of them - Zogby has shilled for W in 2000 and now

and contributed to the "it's close" BS.

[robbedvoter]

21. Zogby, March 2004: Edwards wins in NY!

I am getting nervous hearing this - much as I am aware of Kerry's landslide.
Only because Zogby is such an a*hole.

[cubsfan forever]

23. GOTV n/t

Professor 2

[wicket]

30. Zogby: Novakula is lying

heh :D

[Bandit]

32. I think the major news story after the election is how far off the polls

are and have been. There are close to 200,000,000 cell phones now in america and a good portion of those are the only phones owned. Cell phones are not polled at all. I strongly believe this election will shake the pollsters right down to their socks. If early turnout is any indication of how the voting will go the Democrats are turning out in far larger numbers than Republicans. I think Zogby has hinted at this very thing. He has said he really doesn't have a clue but thinks it will swing toward Kerry in the last couple of days. I agree with him.

[JoFerret]

34. Told Jon Stewart the same thing.

.