A Question For TIA/Exit Poll Gurus Regarding New York State

[althecat]

As is shown in our analysis the exit poll deviation in the East was far higher than in the west or central. So in order to get a bit of a handle on this rBr vs ballot stuffing problem I suggest we focus there for a tic.

In NY State there was a huge gain in Bush voters 2004 over 2000. Somewhere I can dig out the county by county stats they make scary reading...

My first question is what did the NY State exit poll say about new voters in NY State?

If as elsewhere it said that a majority of "new" voters voted Kerry then it seems to me we have a fairly obvious inconsistency.

My second question is what did the Mitofsky rBr report say about New York State WPEs?

Sometimes I think it can help in this kind of discussion to drill down a bit. We might just as well do this with Georgia, NJ and Florida but lets start with NY.

al

[OnTheOtherHand]

1. no thought yet, just answers

(1) The final weighted exit poll results say that Kerry won New York 66-33 among first time voters (10% of all voters), and 58-41 among the others.

http://www.cnn.com/ELECTION/2004/pages/results/states/NY/P/00/epolls.0.html

(2) p. 22 of E/M shows that the "best Geo estimator" (basically exit poll result without considering other info such as pre-election poll results) for NY was off by 13.9 points -- it had Kerry winning by 31.3.

p. 33 shows three different calculations of WPE for New York, ranging from -11.4 to -12.2 -- so, according to E/M, almost all the error in New York was within-precinct error.

[TruthIsAll]

3. Self delete

[TruthIsAll]

4. Note: How voted in 2000 is missing. How convenient. OTOH, a question.

I would love to see the NY "How Voted in 2000" weights to compare them to the 13660 FINAL National, which had an impossible 43% who claimed to have voted for Bush in 2000.

That was demonstratably too high by over 3%.
I will summarize:
50.45 mm voted for Bush in 2000.
1.75mm died.

That leaves 48.7mm maximum who could vote.
or
48.7/122.2 = 39.8% (assuming 100% voted in 2004)

That equates to 4 million phantom votes added to the Bush total.
I ask you, OTOH, where did they come from?

[OnTheOtherHand]

7. what, I'm Carnac the Magnificent, that I should know this?

Here's something I know (assuming the .pdf is accurate) -- the vote-in-2000 question wasn't on the New York state exit poll questionnaire. Actually, it doesn't seem to have been on any of the state questionnaires. It wasn't even on most of the national questionnaires, just version #3 of the 4.

The most likely answer, IMHO, is retrospective misreporting. In the 2002 General Social Survey, 50.6% of respondents who said they had voted in 2000 reported having voted for Bush, and 44.7% for Gore. http://webapp.icpsr.umich.edu/GSS/ , Analyze, Frequencies/crosstab, row variable "PRES00". I doubt that is _mostly_ participation bias, because the GSS folks work very, very hard -- much harder than the exit poll folks can -- to complete interviews.

[TruthIsAll]

15. So that is your most likely answer, huh? I won't let you off that easily.

This is not a question of sampling error.

At the 13047 timeline the number was 41% for Bush, 39% for Gore. That is also impossible, but within the MoE. And Kerry won that timeline by 51-48%.

Now, why was it ADJUSTED up to 43% in the 13660?
It was done deliberately.
Forget polling error.
43% is impossible.

So what should the number be?
WE know it can be no higher than 39.8%.
But that means that Gore goes up to 40%.
And that means Kerry won the election by MORE than 51-48%.

We have confirmation after confirmation after confirmation
after confirmation after confirmation after confirmation after confirmation that Bush lost the election - just like he lost it in 2000.

But you won't acknowledge it.
I appreciate your incisive analysis.

Why don't you apply your statistical prowess to this?

We are back to talking about constraints again.

Total 2004 vote (V) =
Total Bush 2000 voters still alive (B)
+ Total Gore 2000 voters still alive (G)
+ total Nader 2000 voters (N) sill alive
+ OTHER Voters who did NOT vote in 2000 (O).

V= B + G + N + O
V= 122.3
B= 48.7
G =49.2
N =2.7

O =122.3 - 100.6 = 21.7

The percentage of those who did not vote = 21.6/122.2 =17.6% (THE 13047 POLL HAS IT AT 17%, SO ITS CLOSE, WITHIN THE MOE))

Kerry got at least 57% of OTHER if you believe 13047 timeline, but just 54% if you believe that bogus 13660 final).
Kerry got at least 70% of Nader.

So where did Bush get his 12.5mm new votes from (50.5 to 63mm)?

CARE to explain it, OTOH?
DARE to explain it?

BTW, thanks for the fix to the OPTIMIZATION model.

Care to run any more scenarios? Maybe we can come up with just ONE plausible situation in which Bush wins - and rBr is justified.

But I doubt it.

[althecat]

8. Ok so lets compare this with the result in NY State

(1) The final weighted exit poll results say that Kerry won New York 66-33 among first time voters (10% of all voters), and 58-41 among the others.

Excuse me while I think out loud.

From:
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=372291&mesg_id=372364
(note these are preliminary numbers... before provisionals)

They show in NY

Bush received
390,371 more votes than in 2000 (an increase of 16%)

Kerry received
121,525 less votes than in 2000 (a decrease of 3%)

There were roughly 6.7 million voters in 2004
There were roughly 6.5 million voters in 2000

(= at first blush - 200,000 odd new voters at least)

We also know according to the exit poll that 10% of all voters were first time voters = 670,000 new voters...

Also according to the exit poll 2/3 of these 440,000 voted for Kerry. And 220,000 voted for Bush.

Therefore if the results are to be reconciled with the exit poll we need to believe.

561,000 voters who voted for Gore in 2000 did not vote for Kerry in 2004 (14% of Gore's vote)

...and...

Assuming everybody that voted for bush in 2000 voted for him again in 2004 (a courageous assumption) then 170,000 Gore voters voted for Bush.

And assuming that is correct then 400,000 of those who voted for Gore in 2000 didn't vote at all in 2004.

Does this make any sense?

[OnTheOtherHand]

10. I cannot tell you how depressing these numbers are

On 9/11, we lived about three miles downwind of the WTC. I never knew the air could turn that color. My brother-in-law is FDNY, but he wasn't sent there until the next day, and of course the day after that.

We ended up "upstate," not because we fled NYC (actually, I liked NYC more after 9/11), but because I lost my job and then got a job up here. We live in Ulster County now -- and just eyeballing these numbers, relatively tiny Ulster County may have given Kerry his biggest net gain over Gore (in absolute votes) in the state. Not that we had much competition. Tompkins, Columbia, that's about it.

I was startled when NYC elected Bloomberg that November. But I really never believed that NYC and Long Island would break Bush that way. Bush closed the gap by 90,000 votes in Brooklyn alone?

I can't tell you that I think Bush somehow jiggered the numbers in Brooklyn. I think he jiggered the minds. But I wouldn't mind being wrong, because I hate to believe that anyone in Brooklyn bought what Bush has been selling.

(EDIT: In fairness to Brooklyn, even assuming the official numbers are right, we're talking about 84% for Gore, 76% for Kerry. Not so shabby.)

[althecat]

13. Also odd is the difference between Manhattan & Brooklyn and Bronx

[autorank]

16. Your wish is my command...

"I hate to believe that anyone in Brooklyn bought what Bush has been selling." Not to worry, this is an impossibility. I've spent a lot of time in NYC and know a fair amount about Brooklyn (not through political activities). The hatred of * is visceral. Just think back to the Republican National Convention and all those Brooklynites walking across the bridge. There were signs hanging from apartments all over Brooklyn telling * to go home. This is one instance where simple rational thought plus experience negates any of the bull shit in the numbers. Commander CookooBananas is an unwelcome party in Brooklyn.

[althecat]

17. Ok then so what happened... * picked up heaps of votes in King County...

The official results make that very clear...

Should we be looking at the precinct level results here perhaps?

[OnTheOtherHand]

18. I scoped out the city BoE website

http://vote.nyc.ny.us/results.jsp

-- looks like they have them by assembly district and (confusingly under the "assembly district" link) by election district. A good start would probably be the Assembly Districts, especially if it's possible to get shapefiles that delineate the boundaries (so we can see if there were big changes). But even raw maps would help to interpret the numbers. How curious are you?

[althecat]

19. I don't really see it as curiousity. More responsibility....

Clearly if we don't do this nobody will. And we have nothing to lose. I have noticed that in dealing with the GA numbers some very odd things have started to pop up at a precinct level...

That said to a fairly large extent I am relying on your opinion that what we are seeing in Brooklyn is suspicious. I think it is suspicious on its face... but then I do not know the area well...

al

[TruthIsAll]

2. I was going to say we did not have NYS exit poll data...OTOH

we apparently do. But if its the 13660 final, its bogus. WE have proven that.

We need the 13047 or 11027.

[Bill Bored]

5. This has nothing to do with exit polls, but...

anyone in NY who was allowed to vote in a swing state, students, snow birds with FL residences, etc. would probably have done so in 2004. A vote for Pres in NY is a wasted vote! It's a shame, but that's the way it is with the electoral college and all these wussy swing states that just can't make up their minds which way the country should be going and screw everything up for the rest of us!

We know you can't hack >10,000 lever machines.

I don't think there were central tabulators in NY, and our BOEs are pretty much bi-partisan.

One other thing is that NY didn't count many of our provisional ballots. This could have affected the exit polls, but I'm not sure by how much.

[autorank]

6. These are good points, except the first one.

New Yorkers love their state and local politics. There were state and local election plus congressional races which can be a real brawl. People turn out for this and, despite claims to the contrary about those shy and "reluctant" Eastern voters (ROFL), they New Yorkers wouldn't miss a chance to vote for President.

I am interested in the opportunities for election fraud there with the thesis that CA and NY were used to push up *'s margin.

I take your other points as valid but if you find out to the contrary or someone else does, I'd be most interested.

BTW, when is there going to be a thread here "This is for autorank." Was it your O'Dell joke that got you that privilege? Just kidding. And I don't want one that says "autorank Cost the Democrats the 2004 Electing."

[althecat]

9. If CA's bush numbers were padded...

Then he must have shed a lot of vote in CA as his increase in vote there was the lowest of any state.

[LightningFlash]

11. And the only way to iron that one out is through CA state exit polls.

Examine and cross-examine the state to region polls for CA, and USCV can do just that with their next table.

[Wilms]

12. There are 3186 "Election Incidents" reported for New York State

https://voteprotect.org/index.php?display=EIRMapState&state=New%20York&cat=ALL&tab=ALL

[Bill Bored]

14. on 2nd thought, self delete

I'm getting too cynical to even express my cynicism at this point.