TruthIsAll
04-03-2009, 09:13 PM
THE FINAL ROUND: TIA VS. THE NAYSAYERS
TIA:
In a recent Land Shark thread,
SHOULDA COULDA WOULDA: Why Election Fraud Naysayers R W-r-o-n-g---------
a PROMINENT naysayer came VERY CLOSE TO CONCEDING.
.....................................................................
"So I would argue that the Totality of the Evidence at present adds up to this:
1. The election was corrupt.
2. Democrats were the net losers from the corruption.
3. Voter (and vote) suppression remains a key problem, and may have cost Kerry Ohio.
4. Kerry probably lost the popular vote.
5. But we do not know for sure who actually won either the popular or the electoral vote, and this insupportable (sic.)"
.....................................................................
The naysayser confirms the corruption in points 1,2,3.
But still keeps from going all the way in points 4 and 5.
Even though all available evidence says otherwise.
Bush did NOT win the popular vote.
Kerry did. Easily.
Both the electoral AND popular vote.
Well, the naysayer is in Good Company.
Even Kerry won't admit that he won.
So let's summarize some important FACTS in this post in the hope that some of these holdout naysayers and democratic politicians get back to reality.
Deep inside, they know that's where they should be (Ms. Kerry and Ms. Edwards must be very frustrated).
But individual considerations keep them in public denial, while their heads are exploding as more and more information on the fraud surfaces every day.
Why don't they just read the GAO report?
Or scan the EIRS database?
Are they too busy trying to bring democracy to Iraq?
What about bringing democracy back here at home?
Don't we count?
Al Gore was ELECTED in 2000.
It was not just stolen from him.
It was stolen from US.
But at least Al fought.
John Kerry was ELECTED in 2004.
It was not just stolen from him.
It was stolen from US.
John, we hardly knew 'ye.
I want Al back where he belongs.
In OUR House.
TIME FOR SOME FACTS:
The naysayers continue their relentless attempts to mask the overwhelming evidence provided by confirmation of hundreds of pre-election and exit polls:
1) Pre-election state. Total sample: 50 polls* 600 = 30,000
2) Pre-election national. Total Sample: 18 polls *1500 = 27,000
3) Pre-election 48.5% Bush approval. Total: 11 polls*1000 = 11,000
4) 12:22am state exit polls: 73,600 respondents
5) 12:22am national exit poll: 13,047 respondents
And to this we must add have all the massive documented evidence of vote miscounts:
OH, FL, PA, NV, NM, VA, NC, MN, IA, WI, MO, NY…
Were ALL these pre and post-election polls BIASED?
They ALL confirm that Bush lost.
What is it about these polls that is so difficult to understand?
Where is the evidence that the exit polls were biased?
Was it shy Bush voters (rBr)?
Debunked in NEP by Mitofksky 43%Bush/37% Gore?
Was it early Kerry voters?
Debunked. View the time line Kerry led at 4pm, 7:33pm, 12:22am
Was it early women voters?
Debunked.
Female vote share
4pm: 58%, 7:33pm: 54%, 12:22am: 54%, 1:25pm 54%
Was it False Gore voter recall?
Ridiculous on its face.
Bush voters recall who they voted for...
Gore voters suffer from Alzheimer's?
Was it Bad weather early in the day keeping Bush voters home?
Right.
Breaking News! Republicans buy umbrellas at Walmart.
Was it inexperienced pollsters?
Mitofsky trained them. Who is better qualified?
Was it the exit poll "cluster effect"?
Do I hear 20%? 30%? 50%?
Ok, enter your "design effect" into the Interactive Election Model.
Let's see how many states will deviate beyond the MoE for Bush.
The model will calculate the probabilities.
Maybe not 1 in 19 trillion (16 exceeding MoE), but still astronomical.
And how does one explain 30% poll deviations in the Ohio 2005 election?
How much evidence is necessary to prove the DRE fix?
Were the pre-election polls biased, as well?
Naysayers can't blame it on "cluster effect".
Or bad weather.
Or shy Bush voters.
Or Gore voter Faulty Recall.
Or untrained pollsters.
These were PRE-ELECTION POLLS.
For naysayers to say that they wanted a Kerry win is a canard.
They claim to be Democrats or Indies searching for the truth.
To prevent fraud in the NEXT election.
As if THAT gives them credibility.
They want to have it both ways.
Deny that Kerry won and that the polls were right.
Yet at the same time claim that they wanted him to win.
Naysayer allegiance to Mitofsky is obvious.
They say the math is correct.
No argument there.
But they don't agree with the assumptions.
What assumptions?
That pre-election polls favored Kerry?
I can prove it. Go to pollingreport.com
That undecided voters went for Kerry by almost 2-1?
That new voters went to Kerry by 3-2?
That Nader voters went to Kerry by 4-1?
See the National Exit Poll time lines.
That the 43/37% Bush/Gore voter share of the 2004 vote was impossible?
It took a long time for the naysayers to agree.
After all, even they would not claim Bush voter immortality.
That the Final Exit Poll must be wrong?
Well, to match the vote, it applied fictitious weightings.
That's a no-brainer.
That all other Final demographics/vote shares must be wrong, as well?
Well, that's just simple logic.
If A = FALSE
and A = B
then B = FALSE
Do I hear heads exploding?
Or is it just another terror alert?
That Kerry's Gender share was manipulated?
It went from 54% FEMALE at 12:22am to 51% at 1:25pm.
Was it a massive sex change in 12 hours?
Christine Jorgensen never owned a computer back in 1952.
That the Party ID split was manipulated?
From 38 Dem/35 Rep to 37/37.
Was it Massive Fundie conversions in those 12 hours.
That the Census 2004 Vote Survey indicated 3.4mm votes were lost?
According to the Census, 125.7 million voted in 2004.
That's 3.4mm more than the recorded 122.3 million.
The Census Gender demographic MoE is 0.30%.
Should we believe the 122.3mm recorded vote?
The Final Exit Poll has a 1.0% MoE, according to Mitofsky.
That millions of votes are spoiled in every election?
Intentional or innocent spoilage?
Does it matter?
That the trend in Kerry/Bush response (alpha) disproves rBr?
The ratio declines from 1.50 in High Bush precincts to 1.0 in High Kerry.
USCV proved rBr was a myth using simulation.
I confirmed the USCV using the Exit Poll Response Optimizer.
That Bush job approval on election day 2004 was 48.5%?
That's an 11-poll average.
I can prove it.
You can look it up at pollingreport.com.
The combined MoE (11000 sample) is approx. 1.0%.
That there is no way Bush could overcome 48.5% approval?
Oh, well there is one.
He could steal it.
That the Ohio exit poll showed Kerry the 52-48 winner?
Of the 49 exit poll precincts:
36 deviated from Kerry to Bush,
10 from Bush to Kerry,
3 were unchanged.
That if Kerry won Ohio, he must have done better nationally?
Check the record books.
Ohio always LAGS the national Democratic vote.
Naysayers agree there was fraud in Ohio
But what about the other states?
NM, NV, FL, NC, NY, MN, etc..
That the 12:22am state and national exit polls did
confirm a Kerry victory ?
Well, forty-two of 50 states deviated to Bush.
That's 1 in 2 million odds.
That 50 state exit polls mirror 49 Ohio exit poll precincts?
Just a coincidence?
Move along. Nothing to see here.
Take a trip to Brazile, Donna.
That the 9% disparity between the voting shares of Florida optiscans and DREs is virtually impossible?
Dem/Rep registrations were essentially equal in Optiscan and DRE counties.
That sixteen of 50 states deviated beyond the MoE for Bush, none for Kerry?
The probability of that is 1 in 19 trillion.
That ALL 22 Eastern Time Zone states deviated from Kerry to Bush?
1 in 4 million.
That eighty-six of 88 documented touch screens switched Kerry votes to Bush?
See the EIRS database.
Do the math.
1 in 79,000,000,000,000,000,000,000.
THOSE ARE FACTS, NOT ASSUMPTIONS.
Take a look at the graph below.
It shows a time line of pre-election and post election polls.
Naysayers claim the "evidence" shows that Bush won the popular vote.
I ask, what evidence?
If Kerry won the popular vote, doesn't that mean the exit polls (state and national (12:22am) were therefore close to the truth?
Except, that is, for the 1:25pm Final National Poll
We know this one is pure, unadulterated BS. Why?
Look at the How Voted in 2000 demographic.
Focus on the 43%/37% Bush/Gore weights.
They are mathematically IMPOSSIBLE.
Here's the PROOF:
Bush got 50.45 million votes in 2000.
That's 41.25% of the 122.3mm who voted in 2004.
But 3.5% of them died, according to annual U.S. mortality rates (0.87%).
Therefore, AT MOST, 48.7mm of Bush 2000 voters came to the polls in 2004.
That's 39.8% of the 122.3mm total.
THE BOTTOM LINE:
Assuming REALISTIC, PLAUSIBLE, EQUAL weightings for Bush and Gore voters,
KERRY WINS EASILY, EVEN IF FINAL EXIT POLL PERCENTAGES,
WHICH WERE BIASED IN FAVOR OF BUSH, ARE USED.
PERIOD.
CASE CLOSED.
FINITO.
THE SMOKING GUN.
QUERE MAS?
Once again, I challenge the naysayers to a real-time debate using the Interactive Election model.
Let's begin where the DU "Game" thread abruptly ended:
http://www.democraticunderground.com/discuss/duboard.ph...
In the "Game", the naysayers claimed that Bush won 15% of Gore voters.
And that Kerry won only 52% of those who did not vote in 2000.
Can they ever come up with a plausible Bush win scenario?
I doubt it.
Note to Land Shark:
The TOTALITY of pre-election and exit poll data provide SOLID CIRCUMSTANTIAL EVIDENCE that the election was stolen.
The NUMBERS have been silently screaming for a year.
The fact is that AFTER A FULL YEAR there's still strong DOUBT as to who really won.
That should be sufficient incentive for the public to know that something must be done ASAP to restore our democracy.
____________________________________________________________________________________
FEBBLE CHIMES IN WITH AN EXPLANATION:
THE EXIT POLLS WERE “DOCTORED”, BUT THIS IS NOT UNUSUAL.
IN OTHER WORDS, IT’S OK TO MATCH THE FINAL EXIT POLL TO A VOTE – ANY VOTE.
EVEN IF IT’S A CORRUPTED, FRAUDULENT VOTE.
Febble (1000+ posts) Thu Dec-01-05 05:44 PM
Response to Reply #12
16. Well do check out
Late on election day, 2004, with Kerry ahead in the exit polls (scientific polls of voters leaving the voting booth), the corporate news monopolies doctored their exit poll data to fit the results coming from Diebold's and ES&S's secret vote tabulation software. This was not the normal adjustment of exit polls for election demographics. It was a very unusual--indeed, impossible--falsification of the numbers. It has been verified that the real exit polls showed a Kerry win by a 3% margin.
The exit polls were "doctored" not by the "corporate news media" but by those who actually conducted the "scientific polls of voters leaving the voting booth". However this was not "very unusual" - it is what the pollsters do in every election, and, ironically, is probably how they have achieved their reputation for extreme accuracy. Moreover, the "doctoring" was done "late on election day" in the sense that the process started as the first polls closed. This is because the exit poll estimates were as usual, based on three data sources: How are projections made?
Projections are based on models that use votes from three (3) different sources -- exit poll interviews with voters, vote returns as reported by election officials from the sample precincts, and tabulations of votes by county. The models make estimates from all these vote reports. The models also indicate the likely error in the estimates. The best model estimate may be used to make a projection if it passes a series of tests.http://www.exit-poll.net/faq.html#a10 As for your word "impossible" - the discrepancy between the unadjusted poll data and the vote count cannot have been due to chance. That is indeed "impossible". It may have been due to fraud in the vote-count. However, bias in the poll is also perfectly possible. Bias can occur when sampling is not random. It rarely is.
UK exit polls frequently show bias - and we know it is bias because we can rely on our count (fully hand-counted paper ballots). Sadly, you can not. This is a grievous problem. But having unreliable vote-counting methods does not make exit polls magically more reliable. "Impossible" exit polls are only too common, even in clean electoral systems. I am still traumatised by Neil Kinnock's defeat, in defiance of the exit polls, in 1992.(Edited to add: I do agree with you about Gore/Kerry! I was a Gore fan, and still haven't quite got over Florida 2000. That's why I'm here, actually, making a nuisance of myself....)
FEBBLE GETS IN AN EARLY JAB: TYPICAL VAGUE SUGGESTIONS ABOUT “BAD STATISTICAL INFERENCES” SHE CITES NO SPECIFICS, BUT SHE CLAIMS TO DISAGREE WITH ALL OF THEM…
Febble (1000+ posts) Fri Dec-02-05 07:56 AMResponse to Reply #37
42. Wow
Edited on Fri Dec-02-05 08:56 AM by Febble
well, as you've shown an interest in the recipient of the sarcastic jabs, here's the source (in context) of the quote:
http://www.democraticunderground.com/discuss/duboard.ph...
Yes, I admire TIA's enthusiasm too, but I do wish his assertions weren't billed as FACTS.Statistical inferences aren't facts at the best of times, and bad statistical inferences masquerading as facts are worse.But the dedication and diligence I can only admire. And the cause in which they have been exercised. to TIA. On edit: I should make clear, in case it isn't, that I disagree with virtually all TIA's inferences....
TIA MUST HAVE REALLY TOUCHED SOME NERVES WITH THIS ONE…
Febble (1000+ posts) Fri Dec-02-05 12:38 PMResponse to Reply #42
52. I should also make it clear
Edited on Fri Dec-02-05 01:19 PM by Febble
that I don't admire TIA's ad hominem attacks on people who are also diligently working to find out what went wrong with the 2004 election, whichever piece of ectoplasm he is embodied in right now.(On edit: by an ad hominem attack, I mean, for example a response that does not address the argument in question, but denigrates the perceived motivation of the person making the argument.)
OTOH ATTEMPTS TO DENIGRATE TIA’s NATIONAL PRE-ELECTION DATA SOURCE WITHOUT LISTING THE POLLS USED. THAT’S OK, WE’LL SURPRISE HIM SOON WITH A DETAILED LIST COPIED DIRECTLY FROM POLLINGREPORT.COM.
OnTheOtherHand (889 posts) Fri Dec-02-05 04:37 PMResponse to Reply #57
58. "the final TIA poll" is contradicted by, well, the actual polls
http://pollingreport.com/2004.htm
(Bush leads in 5 of 8 projections, 10 of 14 trial heats)http://people-press.org/commentary/display.php3?Analysi... (slightly different list of trial heats, Bush again leads in 10 of 14)I could go on, but we have done this before.My vote is for the accuracy of the 12:22AM NEP and TIA. To say otherwise is to endorse the Diebold's of the world.An impressive new contender for the FEMME Trophy (Fallacy of the Excluded Middle: Most Egregious).
OTOH BRINGS UP THE POLLS- CHERRY-PICKING HIS (AFTER THE ELECTION) While TIA’s 18 NATIONAL POLLS WERE CHOSEN FOR THE MODEL BEFORE THE ELECTION
Note: I added this response after I had a chance to check out the poly sci election models that OTOH Referred to:
OTOH, since you are too lazy (or embarrassed) to show the poli sci model projections, I'll do it for you.
http://www.apsanet.org/content_13000.cfm
All the models presented here were not even close to the corrupted recorded vote, much less the TRUE vote.Except for Beck/Tien, that is.
Author Pick 2-pty Date Win Probability
Abramowitz Bush 53.7% 7/31/04 na
Campbell Bush 53.8% 9/06/04 97%
Wlezien/Ericson Bush 52.9% 7/27/04 75%
Holbrook Bush 54.5% 8/30/04 92%
Beck/Tien Kerry 50.1% 8/27/04 50%
Lockabie Bush 57.6% 5/21/04 92%
Norpoth Bush 54.7% 1/29/04 95%
First of all, look at the model dates.The latest was 9/6/04.
These guys should know that things change every day in a race.The electorate is a dynamic organism.Couldn't they have run the models on Nov.1, like I did?As I did every day for four months leading up to the election?Maybe next time they'll get it right, but I doubt it.They are by nature too conservative to change their approach.
Their combined Bush 2-party average forecast of 54% isn't even closeto the recorded vote, much less the true vote. The average is off by more than 5%, since Bush got about 48.5% of the two-party vote.The only one which was even close was Beck/Tien.They had it 49.9% for Bush, and were off by 2%.Not bad. The rest? Fuggedabout it.
One other thing.There is no way Lockabie's 57.6% Bush popular vote equates to anything less than a 100% EV and/or popular vote win probability.His 92% Bush win probability doesn't say too much for his model.At a 97.5% one-tail level of confidence, it implies an equivalent MoE of 7.6% based on the 57.6% projection. Not good, especially when winning 52% of the popular vote means a virtual 100% probability of winning the electoral vote. The same goes for the rest of the models. Only Beck-Tien's 50% probability made sense, since they projected a virtual dead heat.
Gee, OTOH, I sure teach you a lot, don't I? And you're the political science professor. Yet you say I'm irrelevant to your colleagues.Does MP also consider me irrelevant?He learned a little math from my postings also.What about Ruy Texeira?At least he questioned the Hispanic vote.Now, with all due respect to the poly scientists above:Have these guys ever used Monte Carlo simulation? Did they ever consider projecting individual states to,you know, calculate the Bush probability of winning the electoral vote? From what I could tell in my admittedly cursory review of the PDFs, they essentially all used the same factor analysis regression method, based on macro-economic/financial data. Polling was mentioned, butnot much. They should check out the Election Model for the next election (assuming it's relatively clean) and consider a Monte Carlo simulation.
IN MY OPINION, REGRESSION-BASED ELECTION FORECASTING MODELS ARE AN EXERCISE IN ACADEMIC ONE-UPSMANSHIP: "MY MODEL IS BETTER THAN YOURS".SORT OF A MASTURBATORY CIRCLE-JERK APPROACH.REGRESSION FACTOR-ANALYSIS IS NOT THE BEST WAY TO FORECAST ELECTIONS!ANALYZING SELECTED STATE AND NATIONAL POLLING RESULTS IS FAR SUPERIOR.IT MAY BE MUNDANE AND NOT AS SEXY AS FACTOR ANALYSIS, BUT IT'S MORE ACCURATE, AND THAT'S THE NAME OF THE GAME, ISN'T IT?
BUT... ALL FORECASTS ASSUMED AN HONEST ELECTION.I DID NOT SEE ANY FRAUD VARIABLES, OR SPOILED VOTE FACTORS, OR DISENFRANCHISEMENT PARAMETERS, OR VOTER INTIMIDATION FACTORED INTO THE MODELS. NOT A WORD ABOUT POTENTIAL FRAUD.JUST ECONOMIC AND/OR FINANCIAL FACTORS.WITHOUT A PAPER TRAIL, LOUSY FORECAST MODELS MAY CORRECTLYPROJECT THE "WINNER" IF HE STEALS THE ELECTION.SOMETHING LIKE THE FINAL 1:25PM NATIONAL EXIT POLL.
So what does it all prove? Which model was right?Well, if you believe the election was stolen, mine was.OTOH, do you believe the election was stolen?ONLY MY PRE-ELECTION STATE AND NATIONAL ELECTION MODEL MATCHED THE EXITS.BY THE WAY, OTOH.WHERE IS YOUR MODEL?Ah, what's the use?Talking to you is like talking to a stone.
OnTheOtherHand (889 posts) Fri Dec-02-05 06:51 PMResponse to Reply #58
63. ah, TIA seems to be channeling himself elsewhere on the Net
Edited on Fri Dec-02-05 07:12 PM by OnTheOtherHandbut
I have no idea what the DU protocol is for responding to someone who Isn't There. (EDIT: "but I have idea" could've been some weird meta pun, but wasn't)Anyway, the line seems to be: pay no attention to what anyone else -- including the pollsters themselves -- say their polls indicate. Sure, those darn MSM analysts at Pew said "Slight Bush Margin in Final Days of Campaign", but the illuminati know that it shoulda been Kerry +1. pollingreport.com, shmollingreport.com -- WeGo YouGov. All those poli sci models that called the election for Bush -- best not to mention those, much less refute them. The goofuses over at "Polly's Page", well, who wants to listen to parrots, anyway? Might as well argue with a stone.
FEBBLE DOESN’T QUITE UNDERSTAND THE ELECTION MODEL
Febble (1000+ posts) Fri Dec-02-05 05:41 PM
Response to Reply #59
60. Let me get this straight:
Edited on Fri Dec-02-05 05:54 PM by Febble your link to TIA's "final poll" says:Election Model Projections. If the election were held today, then based on recent state polling, the Electoral Vote Simulation model calculates that John Kerry has a 99.8% probability of winning an electoral vote majority by a 337-201 margin and 51.80% of the popular vote. Kerry won 4990 of 5000 Monte Carlo simulated election trials.
Based on the average of eighteen national polls, the National Vote Projection model calculates that Kerry has a 99.99% probability of winning a popular vote majority with 51.63% of the vote. So TIA was actually saying, on the eve of the election, that Kerry a 99.8 probability of winning the popular vote by nearly four percentage points?Well, I have to confess, I am gobsmacked, as we say around here.Every single pollster, including all those cited by TIA, was saying it was "too close to call" - and yet TIA thought that Kerry was a dead cert (99.8 is a dead cert in statistics) for a massive win of the popular vote?(On edit: well I suppose nearly 4% isn't really massive, but that makes giving it a probability of 99.8% even more, well... odd.)
OTOH ATTEMPTS TO EXPLAIN THE ELECTION MODEL TO FEBBLE WHILE DEMEANING IT AT THE SAME TIME…WHY….BECAUSE IT WAS CONFIRMED BY THE EXIT POLLS?
OnTheOtherHand (889 posts) Fri Dec-02-05 06:07 PM
Response to Reply #60
61. well, TIA said two things, both rather odd
TIA said that Kerry had a 99.99% (not 99.8%) probability of winning the popular vote. Let's not lowball things here, eh?He also said that If the election were held today, then based on recent state polling, the Electoral Vote Simulation model calculates that John Kerry has a 99.8% probability of winning an electoral vote majority by a 337-201 margin and 51.80% of the popular vote.But he cannot have meant quite that. Surely he didn't mean that Kerry had a 99.8% chance of winning 51.80% of the popular vote. (And 0.2% chance of winning 51.81%?!) Those figures (51.80% for the Electoral Vote Simulation, 51.63% for the National Vote Projection) are just his mean expected values, surely.Anyway, as I always say, if you disagree with TIA, you probably work for Diebold. (Actually, that is a freehand paraphrase, but real election reformers shouldn't get hung up on words.)
FEBBLE PICKS UP ON THE OTOH SARCASM…SHE DOESN’T GET THE KERRY WIN PROBABILTY Febble (1000+ posts) Fri Dec-02-05 06:43 PM
Response to Reply #61
62. Ah!
Riiiiight. So Kerry had a 99.8% (or maybe 99.99%)chance of winning, with an expected margin of a bit under 4%. And an EV margin of 136.You know what? I think TIA works for Mitofsky....
SO IT’S TIME FOR TIA TO GIVE FEBBLE A FREE EDUCATION ON THE ELECTION MODEL PROBABILITIES…AND TO CORRECT OTOH WITH PROOF THAT KERRY LED IN THE FINAL PRE-ELECTION NATIONAL POLLS autorank (1000+ posts) Fri Dec-02-05 08:02 PM
Response to Original message
65. Probability Questions & Poll Accuracy, the Muse Speaks
I’m getting new data from my channeling TIA function.
Voila! Someone said:"TIA said that Kerry had a 99.99% (not 99.8%)probability of winning the popular vote. Let's not low ball things here, eh?"Now it's my turn:Yes, I said it.Want to check the math?"If the election were held today, then based on recentstate polling, the Electoral Vote Simulation model calculates that John Kerry has a 99.8% probability of winning an electoral vote majority by a 337-201 margin and 51.80% of the popular vote".
Now for some pro bono public education.In the model, I calculated that Kerry had a 99.8% chance ofwinning the electoral vote (at least 270 votes).That's the result of winning 4990 of 5000 simulation trials.And 337 is the expected mean EV, assuming Kerry won 75% ofthe undecided vote.
The probabilities are:18 national polls- 99.9994% (to win the popular vote),
9 national independent polls: 99.9799% (to win the popular vote)
5000 trial EV simulation (51 states): 99.8% to get at least 270 EV
Want to check the math? The math is correct. Febble has always agreed in the past.
For 18 national polls (18000 total sample)MoE = 0.73%
Kerry 2-party projected vote (mean): 51.63%
Prob =normdist(.5163,.5,0.0073/1.96,true)
Prob= 99.9994%
For the 9 independent pollster subset (9000 sample)
MoE = 1.03%
Kerry 2-party projected vote (mean): 51.86%
Prob = normdist(.5186,.5,0.0103/1.96,true)Prob= 99.9799%
See ya guys...
Polling Date Accuracy Referenced FINAL NATIONAL PRE-ELECTION POLLS
18 Poll Summary:
Kerry won 9, Bush 8, 1 tie
Kerry won 5 of 9 Registered Voter (RV) Polls
and 4 of 9 Likely Voter (LV) Polls
Polling Data Source: http://www.pollingreport.com/wh04gen.htmhttp://www.economist.com/media/pdf/YouGovS.pdf
Total Poll Total Weighted Avg 67% 33% Sample Sample MoE KERRY BUSH KERRY BUSH
Date 26961 Group 0.60% 47.55 47.30 49.95 49.05
1-Nov Marist 1166 LV 2.87% 50 49 50.13 48.87
1-Nov Econ 2903 RV 1.82% 50 47 51.24 47.76
1-Nov TIPP 1284 LV 2.73% 44 47 50.35 48.65
1-Nov CBS 1125 RV 2.92% 47 48 48.35 50.65
1-Nov Harris 1509 LV 2.52% 48 49 51.31 47.69
31-Oct Zogby 1200 LV 2.83% 47 48 48.37 50.63
31-Oct FOX 1400 RV 2.62% 48 45 51.40 47.60
31-Oct DemCorp 1018 LV 3.07% 48 47 47.75 51.25
31-Oct Gallup 1866 RV 2.27% 48 46 50.51 48.49
31-Oct NBC 1014 LV 3.08% 47 48 46.01 52.99
31-Oct ABC 3511 RV 1.65% 47 48 48.65 50.35
30-Oct ARG 1258 LV 2.76% 49 48 49.05 49.95
30-Oct Pew 2408 RV 2.00% 46 45 50.55 48.45
29-Oct Nwk 1005 RV 3.09% 44 48 48.78 50.22
26-Oct ICR 817 RV 3.43% 48 48 53.16 45.84
24-Oct LAT 1698 RV 2.38% 48 47 51.26 47.74
21-Oct Time 803 LV 3.46% 46 51 47.38 51.62
20-Oct AP 976 LV 3.14% 49 46 53.16 45.84
OTOH IS FLUMMOXED; HE CAN NO LONGER CLAIM TIA POLL DATA IS WRONG, SO HE REVERSES COURSE. NOW HE ATTACKS THE METHODOLOGY OF CONSOLIDATING 18 INDEPENDENT NATIONAL POLLS…OTOH, DID YOU FORGET THE LAW OF LARGE NUMBERS?
OnTheOtherHand (889 posts) Fri Dec-02-05 08:25 PMResponse to Reply #65
66. whoa: I actually AM arguing with a stone... cool...
This is not a math question.In the real world, one can't get everyone to agree to work from TIA's list of Real poll results. See also my #63.Moreover, it is not S.O.P. in meta-analysis to just add a bunch of surveys together to make One Big Survey. And even if one did, it would be pretty wishful to assume that sampling error is the only error source in the model.But all this, and more, has been explained before.
FEBBLE CONFESSES HER FAILURE TO DO HER HOMEWORK ON THE ELECTION MODEL.SHE CLAIMS THE MATH MUST BE WRONG IF IT GIVES A 99% KERRY WIN PROBABILITY.SHE THEN PROCEEDS TO BLOW SMOKE WITH “META-ANALYSIS” JARGON.SHE PARROTS OTOH: “A GOOD META-ANALYSIS IS HARD TO DO”.FEBBLE, IT’S NOT THAT COMPLICATED.YOU ARE ABOUT TO BE VERY SURPRISED.TRUST ME, FEBBLE, THE MATH IS CORRECT.YOU WILL SOON FIND OUT WHY.
Febble (1000+ posts) Sat Dec-03-05 05:51 AMResponse to Reply #65
70. Qualification:
I confess I had not looked at this "final poll" before, but if "the math" gives a 99.9% (or even a 99.8%) probability of a Kerry win, then something about the math is wrong. I was optimistically putting it the odds at about 50:50 myself, and hoping that even if the popular vote was pretty evenly split, the EV would benefit Kerry. Which is sort of what happened (in the official vote anyway), in that a few thousand more (official) votes for Kerry in Ohio would have given him the presidency, but not the (official) popular vote.
And more generally, when I say I agree with TIA's math, I certainly trust him to perform the computations he does correctly - the issue is usually whether we can legitimately infer from them what he infers from them. But his probability calcs are frequently invalidated by the assumptions that underlie them, and a probability calc based on erroneous assumptions is not a correct calculation).So what is wrong with this particular probability calc? Well for a start, as OTOH says, a good meta-analysis is a complex thing to do, and one thing it involves is using ALL data you can find (including unpublished data, if you can find that) and weighting it appropriately.
"Appropriately" is the hard part of course, and it is difficult to avoid the charge of "cherry-picking". There are, however, systematic ways of weighting studies according to statistical power and cleanliness of methodology. But the fact remains that meta-analyses in general (and meta-analyses of these pre-election polls in particular) are acutely sensitive to the studies you pick and the assumptions you make. In the case of the pre-election polls, some analysts called it for Bush (but with nothing like 99.9% confidence; others sat on the fence; and a few brave souls like Sam Wang (I think) called the EV for Kerry (with a wing and a prayer).
Yet TIA called it for Kerry with 99.9% confidence? From a eighteen polls in which (unweighted) the mean difference between Kerry and Bush was not (on my calcs) significantly different from zero?Look, I don't think that election was on the up and up, and I think the exit poll stuff was worth investigating. And apart from anything else, it also energised a lot of people (including me) into investigating stuff that brought the election result into serious doubt. But claiming that Kerry's probability of victory, in both the electoral college and popular vote, was 99.9% - well, as we say on our side of the pond, pull the other one, it's got bells on.Were any bookies offering 100:1 odds on a Bush win?
FEBBLE SCAMPERS BACK IN AFTER CHECKING PROFESSOR SAM WANG’ META-ANALYSIS FORECAST...BLIMEY, HE AGREES WITH TIA...!!!!!
Febble (1000+ posts) Sat Dec-03-05 01:52 PMResponse to Reply #70
76. Update
Just checked Sam Wang:
http://synapse.princeton.edu/~sam/pollcalc.html
Looks like he did indeed commit himself to a 98% probability of a Kerry EV win, bless him (he doesn't seem to give a probability for his Popular Vote estimate)And it's still on a wing and prayer, though. He modified his "gut estimate" to 6:1 after a Bayesian adjustment as to whether his assumptions were likely to be correct.Well, there you go. Sam Wang and TIA agreed, at least on the EV win.I'm afraid it doesn't convince me that either of them were correct.Actually, Sam Wang doesn't seem that convinced he was correct either:
http://synapse.princeton.edu/~sam/pollcalc_letters_afte...
but he has interesting things to say about why he might have been wrong:I think the purely statistical aspects of the analysis did extremely well. The electoral outcome looks like it will be close to the decided-voter outcome predicted by the polls. Victory margins are quantitatively close to the pre-election polls: out of 23 battlegrounds, the direction of the outcome was predicted in 22 (the exception was Wisconsin, where the polling margin was 0.4% for Bush and the actual margin was about 0.4% for Kerry). Quantitatively, 12 victory margins were within one standard error and 17 were within the 95% confidence interval. Not perfect, but not bad.
The most significant errors had to do the net effect of other factors not encompassed by polls. To make a final prediction, I used previous patterns of uncommitted voters breaking for the challenger as a guide, but this break either did not occur or was cancelled by other factors. My assumption of high turnout was flat out wrong! In the end, the likely-voter models of pollsters were not too far off.There has been talk of other factors, but a parsimonious explanation may be that the net effect of all other factors was zero. This isn't always true - in past years the outcome seems to have not matched final polls. There seems to be some mystery offset that varies a bit.
On the other hand, this year we had more data - maybe it's just a question of having enough data and the right answer falls out.One advantage of rigorous statistical modeling is that you can see a clear separation between factual information and assumptions of less certainty. In this case my baseline calculation was quite accurate, but the intangibles were wrong. As I said, in previous years at least one of the assumptions would have worked. What happened this year is a question for the political and policy people - in the end it goes to show that I am at my best with the numbers! (my bold)See also:http://election.princeton.edu /
OTOH IS QUICK TO EXTOL SAM WANG’S ANALYSIS. OTOH KNOWS THAT MR. WANG IS A PROFESSOR AT PRINCETON. HE’S NOT TIA. BETTER NOT TRY TO TRASH SAM.
OnTheOtherHand (889 posts) Sat Dec-03-05 03:09 PMResponse to Reply #76
77. yes, Wang gets big points for isolating crucial assumptions:
Note that all of these probabilites are conditional on the turnout and undecided voter assumptions being correct. The true probability is obtained by multiplying by a measure that is a function of whether my assumptions are accurate. The chance that I am wrong makes the true probability substantially lower than 100%! As Niels Bohr (and Yogi Berra) said, "Prediction is hard, especially of the future."http://election.princeton.edu Wang is confusing when he refers to his "assumption of high turnout" -- rather, his assumption was that high turnout would favor Kerry. Surely in some places it did (and conceivably fraud conceals some of its actual impact). But as we've discussed, it was never empirically all that obvious that high turnout 'had to' favor Kerry. TIA's parallel leap was to prefer Registered Voter results to Likely Voter results, although he uses some of each.
OTOH QUOTES TOE-SUCKER DICK MORRIS.HE SAYS THAT MAYBE THE EXIT POLLS WERE RIGGED.WELL, OTOH, TAKE A LOOK AT THE FINAL NATIONAL EXIT POLL.NOT THE PRELIMINARY 13047; THAT WAS FINE.LOOK AT THE FINAL 13660 – MATCHED TO THE VOTE.OTOH THEN PROCEEDS TO BASH THE U.S. CENSUS. WHY? THE SURVEY INDICATED THAT 3.5 MILLION VOTES WERE UNCOUNTED.
OnTheOtherHand (889 posts) Sat Dec-03-05 11:42 AM
Response to Reply #67
74. E/M, and the census
Some folks think that Mitofsky has been rigging exit polls for, umm, I guess decades. I guess for them, the disbanding of VNS is the least of the worries. (Dick Morris thought the exit polls were rigged to discourage Republicans; some folks have suggested instead that they were rigged to lull Democrats into complacency.) I honestly don't quite understand why anyone believes any of this. AFAICT, most public opinion researchers think that the exit polls were wrong, and a few think that the exit polls were basically right and the vote count was wrong; few see any need for the hypothesis that the exit polls (and/or pre-election polls) were rigged.
Anything is possible, I guess.I didn't notice the "Census 2004 Vote Survey" chestnut coming back. It's actually the Voting and Registration Supplement to the Current Population Survey (CPS). To quote a 1998 Census Bureau press release, "The CPS routinely overestimates voter turnout" (http://www.census.gov/Press-Release/cb98-146.html ), so most of us don't waste much time trying to reconcile the figures. You may well be right that presidential undervoting accounts for part of the discrepancy; I would have to look more closely.
FEBBLE PROVIDES A FULL-FRONTAL REBUTTAL TO THE ORIGINAL POST SHE RESPONDS WITH MOSTLY VAGUE STATEMENTS.SHE BRINGS NOT A SINGLE FACT TO THE TABLE. SHE CONSISTENTLY DENIGRATES ALL PRE-ELECTION AND FINAL POLLSSHE IS CAREFUL TO COUCH HER STATEMENTS, TO GIVER HER WIGGLE ROOM.. HER PRIMARY ARGUMENT IS THAT ALL POLLS ARE BIASED, THEREFORE ONE SHOULD NOT DRAW ANY INFERENCES FROM THEM. SHE REPEATS THE FALSE VOTER RECALL CANARD.BUT APPARENTLY ALZHEIMER’S ONLY AFFLICTED GORE VOTERS. THAT’S HER EXPLANATION FOR THE IMPOSSIBLE 43%/37% BUSH/GORE 2000 WEIGHTS She repeats her mantra – that polls are not random samples (WTF?):“1. Random sampling””2. Random sampling””3. Random sampling”
Febble (1000+ posts) Sat Dec-03-05 12:22 PMResponse to Original message
75. OK, these are my words:
....................................................................."So I would argue that the Totality of the Evidence at present adds up to this:1. The election was corrupt.2. Democrats were the net losers from the corruption.3. Voter (and vote) suppression remains a key problem, and may have cost Kerry Ohio.4. Kerry probably lost the popular vote.5. But we do not know for sure who actually won either the popular or the electoral vote, and this insupportable (sic.)".....................................................................
The naysayser confirms the corruption in points 1,2,3.But still keeps from going all the way in points 4 and 5.So, my response:Even though all available evidence says otherwise.Not in my view. So let's summarize some important FACTS ...Al Gore was ELECTED in 2000.It was not just stolen from him.It was stolen from US.But at least Al fought.Certainly Gore had more intended votes in Florida than Bush, and won the popular vote.John Kerry was ELECTED in 2004.It was not just stolen from him. It was stolen from US. John, we hardly knew 'ye. Not facts.
TIME FOR SOME FACTS:The naysayers continue their relentless attempts to mask the overwhelming evidence provided by confirmation of hundreds of pre=election and exit polls:
Not a fact
1) Pre-election state. Total sample: 50 polls* 600 = 30,000
2) Pre-election national. Total Sample: 18 polls *1500 = 27,000
3) Pre-election 48.5% Bush approval. Total: 11 polls*1000 = 11,000
4) 12:22am state exit polls: 73,600 respondents
5) 12:22am national exit poll: 13,047 respondents
These are facts in the sense that these polls were published. However the numbers are inferences from samples, and their margins of error assume random sampling. All the polls cited had high non-response rates, and cannot be assumed to consist of random samples, although one hopes they did.And to this we must add have all the massive documented evidence of vote miscounts:OH, FL, PA, NV, NM, VA, NC, MN, IA, WI, MO, NY…This is a qualitative statement, not quantitative. "Massive" needs to be supported by quantitative data before I will accept it as a "fact".
But yes, I agree there is good evidence of "irregularities", and probably in all these states. More than vote miscounts, however - some of the best documented evidence includes incidents where the voter was unable to vote at all.Were ALL these pre and post-election polls BIASED? Probably not.They ALL confirm that Bush lost.A pre election poll cannot "confirm" what happened in an election. And I was following the pre-election polls closely, and they certainly weren't telling me that "Bush lost". And yes, they may have been biased, some more than others, some in on direction some in another.
What is it about these polls that is so difficult to understand?
I dunno - what is it about "too close to call" that is so difficult to understand?Where is the evidence that the exit polls were biased? Was it shy Bush voters (rBr)?Debunked in NEP by Mitofksky 43%Bush/37% Gore?
I have no idea why TIA thinks the recalled vote data "debunks" non-response bias as an explanation for the discrepancy between poll and count. I certainly wouldn't describe the hypothesis as "shy Bush voters", but I find it perfectly plausible to postulate that slightly more of the "reluctant" voters (and we know there were a lot) were Bush voters than Kerry voters.
Of the "non-reluctant" voters, I also find it perfectly plausible that some Gore voters mistakenly recalled voting for Bush; OTOH cited an actual longitudinal study in which voters who voted for Gore later recalled voting for Bush. If similar proportions did so in the exit poll, it would account for the exit poll data Bush/Gore proportions.
Was it early Kerry voters? Debunked.
View the time line Kerry led at 4pm, 7:33pm, 12:22am
Was it early women voters?Debunked.
Female vote share
4pm: 58%, 7:33pm: 54%, 12:22am: 54%, 1:25pm 54%
Neither of these, as we both know. The timeline is irrelevant. The timeline simply reflects dynamic re-weighting processes. We know, in any case, from the E-M report that the problem was at the level of the precincts - a greater proportion of Kerry voters were sampled on the poll than were represented in the vote.
Was it False Gore voter recall? Ridiculous on its face. Bush voters recall who they voted for...Gore voters suffer from Alzheimer's?
I have to believe this assertion (not a fact) is willful denial of perfectly good evidence. Check back to the Game thread for OTOH's source. Inaccurate vote recall is a well-attested phenomenon, and often favours the incumbent.
Was it Bad weather early in the day keeping Bush voters home?
Right.Breaking News! Republicans buy umbrellas at Walmart.Well, they didn't have to queue as long in Ohio, so maybe. But the evidence suggests not.
Was it inexperienced pollsters? Mitofsky trained them. Who is better qualified?
This is my best guess. They had, I understand, a brief telephone training. Not enough for what is actually a remarkably difficult job.
Was it the exit poll "cluster effect"? Do I hear 20%? 30%? 50%?
TIA should stop using these scare quotes, and find out what a DESR actually is (I've attempted to explain, and even gave him a link).
Ok, enter your "design effect" into the Interactive Election Model.Let's see how many states will deviate beyond the MoE for Bush. The model will calculate the probabilities. Maybe not 1 in 19 trillion (16 exceeding MoE), but still astronomical.
No, because I do not share the assumption of the model that the only source of error in the polls is sampling error (DESR is also an estimate of sampling error).
And how does one explain 30% poll deviations in the Ohio 2005 election?
Well I think there are a number of possible explanations, but I thought we were talking about 2004? Possible there was fraud, although the finding by Klinkner that the discrepancy was not greater in DRE counties doesn't suggest DRE fraud particularly. But as I keep saying, until Ohio gets cleaned up, no-one will know the answer.
How much evidence is necessary to prove the DRE fix?
How much? Well, I'd like to see something more convincing that we've got. But I agree they are a terrible idea, on many, many, counts, and quite absurdly insecure, as the GAO report points out.
Were the pre-election polls biased, as well?
Oh, I thought we'd done this bit. Probably they were, but clearly not all in one direction. I think TIA eliminated the polls he thought were biased for Bush. I am sure they had non-sampling error, and tracking polls suggested some polls tended to track higher than others. This suggests that bias was a problem, and that certain pollsters tended to have a particular bias. This is highly likely, as bias is a function of methodology. One of the reasons I dispute TIA's lumping of polls together is that it ignores between-poll error (what is sometimes called "random effects" variance. It seems fairly clear that each poll had a characteristic "bias". It's the same in the UK. The one I like is ICM (for the Guardian) as it allows for "shy Tories" - it therefore tends to track Labour lower than the others - and generally gets closer to the result.
Naysayers can't blame it on "cluster effect".Or bad weather.Or shy Bush voters.Or Gore voter Faulty Recall.Or untrained pollsters.These were PRE-ELECTION POLLS.
This is pure rhetoric, not argument. See above.
For naysayers to say that they wanted a Kerry win is a canard.They claim to be Democrats or Indies searching for the truth.To prevent fraud in the NEXT election.As if THAT gives them credibility.They want to have it both ways. Deny that Kerry won and that the polls were right.Yet at the same time claim that they wanted him to win. Naysayer allegiance to Mitofsky is obvious.
Again, this is pure rhetoric. I will leave well alone.
They say the math is correct.No argument there.But they don't agree with the assumptions.What assumptions?
1. Random sampling
2. Random sampling
3. Random sampling
That pre-election polls favored Kerry? I can prove it. Go to pollingreport.com.
See mine (and others', including OTOH's, responses on this thread.That undecided voters went for Kerry by almost 2-1?That new voters went to Kerry by 3-2?That Nader voters went to Kerry by 4-1?See the National Exit Poll time lines.Don't understand this. See my comment re time-line. We also can't distinguish "new voters" from "rare voters". New voters will tend to be young and might be expected to favour Kerry. Rare voters - who knows?
That the 43/37% Bush/Gore voter share of the 2004 vote was impossible?It took a long time for the naysayers to agree.After all, even they would not claim Bush voter immortality.
Oh dear, round, and round, and round we go. No, I agree, that it is unlikely that Bush voters are immortal. False recall seems much more likely as an explanation - we actually know this is a phenomenon. The only immortal I know personally is a Kerry voter (or I assume he is...)
That the Final Exit Poll must be wrong? Well, to match the vote, it applied fictitious weightings.That's a no-brainer.
That all other Final demographics/vote shares must be wrong, as well?Well, that's just simple logic.If A = FALSE and A = B then B = FALSE.
Do I hear heads exploding? Or is it just another terror alert?
That Kerry's Gender share was manipulated?
It went from 54% at 12:22am to 51% at 1:25pm.
Was it a massive sex change in 12 hours? Christine Jorgensen never owned a computer back in 1952.
That the Party ID split was manipulated? From 38 Dem/35 Rep to 37/37.
Was it Massive Fundie conversions in those 12 hours.
OK, this is getting silly. You seem to have forgotten that the projections were continuously reweighted (and that they actually had some glitch with the gender thing - it's in the E-M report, look it up).
That the Census 2004 Vote Survey was wrong?According to the Census, 125.7 million voted in 2004.That's 3.4mm more than the recorded 122.3 million.The Census Gender demographic MoE is 0.30%.Should we believe the 122.3mm recorded vote?
Well the Census people think that it overstates the # voted. But it seems clear that spoilage will contribute to the discrepancy.
The Final Exit Poll has a 1.0% MoE, according to Mitofsky. That millions of votes are spoiled in every election? Intentional or innocent spoilage? Does it matter?
Dunno about millions, but yes it matters. It matters desperately, whether it is intentional or structural discrimination. It is a serious Civil Rights issue, and virtually certainly resulted in the Wrong Man being inaugurated in January 2001.Yay, we agree on something!
That the trend in Kerry/Bush response (alpha) disproves rBr? The ratio declines from 1.50 in High Bush precincts to 1.0 in High Kerry.
Well, as the coiner of "alpha" (actually it was my husband's term, as he couldn't read my "x"s, but I suggested the measure, and gave it that name) yes, I vigorously dispute this. Mitofsky tested precisely this, and no, it does not decline. The line is statistically flat.USCV proved rBr was a myth using simulation.Correction: some members of USCV claimed their simulation indicated that rBr was unlikely. Another member claimed his simulation indicated it was perfectly possible. The first lot then accused the second lot of accusing them of claiming that they'd proved rBr was a myth. Then there was a big fight. I think the ones that claimed that alpha was higher in high Bush precincts got it wrong. I confirmed the USCV using the Exit Poll Response Optimizer. Well, I think "you" did too, for the same reasons as USCV did. I wrote an entire geeky paper about it. But more to the point, Mitofsky actually did the calculation, using the whole dataset, and it isn't.
That Bush job approval on election day 2004 was not 48.5%? That's an 11-poll average. I can prove it. You can look it up at pollingreport.com.The combined MoE (11000 sample) is approx. 1.0%.
Well, prove it then. But remember that MoE assumes random sampling.
That there is no way Bush could overcome 48.5% approval? Oh, well there is one. He could steal it.
Well, I seriously wondered if he could. My conclusion is probably didn't.
That the Ohio exit poll showed Kerry the 52-48 winner? Of the 49 exit poll precincts:36 deviated from Kerry to Bush,10 from Bush to Kerry,3 were unchanged.
Yup, something was biased. Was it the poll? Was it the count? That is the question.
That if Kerry won Ohio, he must have done better nationally? Check the record books. Ohio always LAGS the national Democratic vote. Naysayers agree there was fraud in Ohio.
Cite please. And remember, as they say in the investment prospectuses, past performance is no guide to future performance. Social science isn't physics. Also there is an IF in there.
But what about the other states? NM, NV, FL, NC, NY, MN, etc..
That the 12:22am state and national exit polls each confirming a Kerry victory is not believable? Well, forty-two of 50 states deviated to Bush.That's 1 in 2 million odds.
Oh boy, round and round and round and round. Yes, the poll, or the count, or both, seem to have been systematically. It certainly wasn't chance.
That 50 state exit polls mirror 49 Ohio exit poll precincts? Just a coincidence?
It certainly wasn't chance. It certainly wasn't chance. It certainly wasn't chance. It certainly wazzzzzzzzzz.......
Move along. Nothing to see here.Take a trip to Brazile, Donna
.....zzzzzzzz....
That the 9% disparity between the voting shares of Florida optiscans and DREs is virtually impossible? Dem/Rep registrations were essentially equal in Optiscan and DRE counties.
Ah! Now I did some of this stuff.... What do you mean? Do you mean Hout's finding that DRE's had greater discrepancies? Or Liddle/Dopp's/Mitteldorf's finding that Optiscans did? Actually I think both analyses are invalid, as there were huge demographic confounds - No DREs were used in very rural counties, and no optiscans in the very large urban counties. Florida smells, to me, but the machine thing is completely inconclusive.
That sixteen of 50 states deviated beyond the MoE for Bush, none for Kerry?The probability of that is 1 in 19 trillion
.....round and round and round and round......
That ALL 22 Eastern Time Zone states deviated from Kerry to Bush? 1 in 4 million.
Hey! I thought it was the swing states? Oh no, it was the Republican precincts - Oh no, it was the Democratic states....What is the point here exactly?
That eighty-six of 88 documented touch screens switched Kerry votes to Bush?See the EIRS database. Do the math.1 in 79,000,000,000,000,000,000,000.
The math here is truly bosh. Do we know how many machines these 88 incidents represent? Do we know whether the EIRS number was equally well publicised in Dem as in Rep areas? If not, then we can't do that math. I'll give you, it does look as though some machines may have defaulted to Bush. It requires serious investigation (and abolition of those dratted machines. It does not require idiotic probability calculations.
THOSE ARE FACTS, NOT ASSUMPTIONS.
Ha!
The bits that aren't pure rhetoric (including purely rhetorical questions) are not facts at all! They are inferences based on assumptions - every single one. You can't DO inferential statistics without making assumptions, and the biggest one is RANDOM SAMPLING. In not a single case cited can we assume anything like random sampling. What the probabilities therefore tell us is that the "facts" inferred ARE NOT DUE TO CHANCE. They do not tell us whether the non-chance cause was fraud or not. For answers to that question we have to do further investigation. I've done quite a bit of that myself, in terms of math, and many others have done quite a bit of on the ground investigation. And having looked at the accumulating evidence, some things are supported, some look shaky (like the exit poll story, which I now believe does NOT support the case for large-scale fraud).But to assert that these are FACTS is quite misleading, and to imply that the astronomical (im)probability estimates are prima facie evidence of fraud is to seriously misunderstand the nature of inferential statistics. And, frankly, to mislead.
Take a look at the graph below.It shows a time line of pre-election and post election polls.I'll pass on this, as the graph doesn't seem to have channelled properly.Naysayers claim the "evidence" shows that Bush won the popular vote.I ask, what evidence?Well, I say that the only real evidence that Bush won the popular vote was the exit poll discrepancy (I certainly don't count the pre-election polls), and that if this was due to fraud, then fraud should have been correlated with "swing" (change in Bush's vote since 2000). It wasn't, and I think this makes the fraud explanation difficult to sustain.
I certainly do not claim that the "evidence" shows that Bush won the popular vote. I merely say that I do not believe that the exit poll provide any evidence at all that he did not. And absent that evidence, I think it is unlikely that Kerry did.
If Kerry won the popular vote, doesn't that mean the exit polls (state and national (12:22am) were therefore close to the truth? Except, that is, for the 1:25pm Final National Poll. We know this one is pure, unadulterated BS. Why? Look at the How Voted in 2000 demographic.Focus on the 43%/37% Bush/Gore weights.They are mathematically IMPOSSIBLE.
Here's the PROOF:Bush got 50.45 million votes in 2000.That's 41.25% of the 122.3mm who voted in 2004.But 3.5% of them died, according to annual U.S. mortality rates (0.87%).Therefore, AT MOST, 48.7mm of Bush 2000 voters came to the polls in 2004.That's 39.8% of the 122.3mm total
.....round and round and round and round....
THE BOTTOM LINE: Assuming REALISTIC, PLAUSIBLE, EQUAL weightings for Bush and Gore voters,KERRY WINS EASILY, EVEN IF FINAL EXIT POLL PERCENTAGES,WHICH WERE BIASED IN FAVOR OF BUSH, ARE USED.PERIOD.CASE CLOSED.FINITO.
THE SMOKING GUN.
QUERE MAS?
As I said, if it makes you happy.
Once again, I challenge the naysayers to a real-time debate using the Interactive Election model.Let's begin where the DU "Game" thread abruptly ended:
http://www.democraticunderground.com/discuss/duboard.ph ...
In the "Game", the naysayers claimed that Bush won 15% of Gore voters. And that Kerry won only 52% of those who did not vote in 2000.Can they ever come up with a plausible Bush win scenario? I doubt it.
.....random sampling random sampling random sampling....
Note to Land Shark: The TOTALITY of pre-election and exit poll data provide SOLID CIRCUMSTANTIAL EVIDENCE that the election was stolen.The NUMBERS have been silently screaming for a year.The fact that ONE YEAR LATER, there's strong DOUBT about who really won, should be sufficient to convince the public that something must be done ASAP to restore our democracy.
And again, I whole-heartedly agree! The fact that this conversation is even happening is evidence of the degree to which your democracy has been sabotaged!And of course, I am encouraged, from a purely mathematical point of view, that you accept that there is DOUBT about who won (although I would much prefer it if Kerry was in the White House right now). Yes, indeed, there is doubt.Beyond that, we will have to agree to differ.Lizzie
TIA RESPONDS WITH FACTS, PLAUSIBLE UNDECIDED VOTER ASSUMPTIONS AND CALCULATIONS. FEBBLE HAD NO IDEA THAT PROFESSOR WANG’S META-ANALYSIS AGREES WITH TIA’S 98% KERRY ELECTORAL VOTE WIN PROBABILITY. OR THAT TIA ALSO CALCULATED KERRY’S POPULAR VOTE WIN PROBABILTY. TIA REMINDS FEBBLE THAT EDISON-MITOFSKY CLAIMED A 1.0% MOE FOR 13047 RANDOMLY-SELECTED RESPONDENTS IN THE NATIONAL EXIT POLL.
autorank (1000+ posts) Sun Dec-04-05 03:01 AMResponse to Reply #75
78. Well, OK these are the Truth's words...cheers Is All :toast:
Channeling to avoid "the Chunnel":
Febble, here is my response to your latest post.Do you want to post it on DU? Or do you want it channelled? Either way, it will soon appear on DU. Look at this as a heads-up. Before I begin, I have three questions: Of the 50 state pre-election polls, and of the 18 national pre-election polls, and of the 11 Bush approval polls, and of the 50 state exit polls, and of the first 3 national exit poll timelines...
1) Which were not random sample polls?
2) Which were not done by well-known pollsters?
3) In your experience, has there ever been a poll that youbelieved?
Just asking.
FEBBLE:I confess I had not looked at this "final poll"before, but if "the math" gives a 99.9% (or even a99.8%) probability of a Kerry win, then something about the math is wrong. I was optimistically putting it the odds atabout 50:50 myself, and hoping that even if the popular vote was pretty evenly split, the EV would benefit Kerry. Which is sort of what happened (in the official vote anyway), in that a few thousand more (official) votes for Kerry in Ohio would have given him the presidency, but not the (official) popularvote.
TIA:I thought that you would have some familiarity with the logic behind my election model by this late date. It is quite disappointing to see you criticize the analysis without having done your homework. It's obvious that you are not too familiar with U.S. elections; you fail to take the undecided vote into account. The undecideds invariably break for thechallenger - especially when the incumbent has a 48.5% job approval. It just makes logical, intuitive sense when you think about it. An undecided voter must be unhappy with an incumbent,otherwise why would he/she consider voting for the challenger? Makes sense, yes?
You fail to consider this in your election analysis- and the omission makes anything you say highly suspect from the get-go. Without considering undecides, your analysis is faith-based and unrealistic.The fact is, Kerry won the late undecided vote, according to all the polls. And the undecided allocation was a key driver in my state and national projections. Yet you never even consider this, other than to vaguely refer to faulty model"assumptions". You are never specific about them. Did you mean undecided vote assumptions? Read on.
FEBBLE:(And more generally, when I say I agree with TIA's math, I certainly trust him to perform the computations he does correctly - the issue is usually whether we can legitimately infer from them what he infers from them. But his probability calcs are frequently invalidated by the assumptions thatunderlie them, and a probability calc based on erroneous assumptions is not a correct calculation).
TIA:What you call “inferences” are in fact the result of valid assumptions. You may disagree with the assumptions, but they are based on historical precedent. You fail to recognize or appreciate the significance of the undecided vote factor in U.S. elections. The base case model assumption was that Kerry would win 75% of the undecided. But this was just amost-likely base case estimate. That's why I ran the analysis for a range of five undecided vote shares from 60% to an admittedly high 87%. Kerry wins all the scenarios.
If Kerry won just 60% of the undecided vote (and the exit polls say he did better than that) then based on the Monte Carlo simulation, the probability was 98.02% that he would win the electoral vote with a median 322 EV. Furthermore, based on 18 national polls, the probability was 97.55% that he would win the popular vote with an expected 50.73% of the two-party vote, again assuming that he won 60% of the undecideds. You really ought to take a trip to Monte Carlo sometime.
TIA:
In a recent Land Shark thread,
SHOULDA COULDA WOULDA: Why Election Fraud Naysayers R W-r-o-n-g---------
a PROMINENT naysayer came VERY CLOSE TO CONCEDING.
.....................................................................
"So I would argue that the Totality of the Evidence at present adds up to this:
1. The election was corrupt.
2. Democrats were the net losers from the corruption.
3. Voter (and vote) suppression remains a key problem, and may have cost Kerry Ohio.
4. Kerry probably lost the popular vote.
5. But we do not know for sure who actually won either the popular or the electoral vote, and this insupportable (sic.)"
.....................................................................
The naysayser confirms the corruption in points 1,2,3.
But still keeps from going all the way in points 4 and 5.
Even though all available evidence says otherwise.
Bush did NOT win the popular vote.
Kerry did. Easily.
Both the electoral AND popular vote.
Well, the naysayer is in Good Company.
Even Kerry won't admit that he won.
So let's summarize some important FACTS in this post in the hope that some of these holdout naysayers and democratic politicians get back to reality.
Deep inside, they know that's where they should be (Ms. Kerry and Ms. Edwards must be very frustrated).
But individual considerations keep them in public denial, while their heads are exploding as more and more information on the fraud surfaces every day.
Why don't they just read the GAO report?
Or scan the EIRS database?
Are they too busy trying to bring democracy to Iraq?
What about bringing democracy back here at home?
Don't we count?
Al Gore was ELECTED in 2000.
It was not just stolen from him.
It was stolen from US.
But at least Al fought.
John Kerry was ELECTED in 2004.
It was not just stolen from him.
It was stolen from US.
John, we hardly knew 'ye.
I want Al back where he belongs.
In OUR House.
TIME FOR SOME FACTS:
The naysayers continue their relentless attempts to mask the overwhelming evidence provided by confirmation of hundreds of pre-election and exit polls:
1) Pre-election state. Total sample: 50 polls* 600 = 30,000
2) Pre-election national. Total Sample: 18 polls *1500 = 27,000
3) Pre-election 48.5% Bush approval. Total: 11 polls*1000 = 11,000
4) 12:22am state exit polls: 73,600 respondents
5) 12:22am national exit poll: 13,047 respondents
And to this we must add have all the massive documented evidence of vote miscounts:
OH, FL, PA, NV, NM, VA, NC, MN, IA, WI, MO, NY…
Were ALL these pre and post-election polls BIASED?
They ALL confirm that Bush lost.
What is it about these polls that is so difficult to understand?
Where is the evidence that the exit polls were biased?
Was it shy Bush voters (rBr)?
Debunked in NEP by Mitofksky 43%Bush/37% Gore?
Was it early Kerry voters?
Debunked. View the time line Kerry led at 4pm, 7:33pm, 12:22am
Was it early women voters?
Debunked.
Female vote share
4pm: 58%, 7:33pm: 54%, 12:22am: 54%, 1:25pm 54%
Was it False Gore voter recall?
Ridiculous on its face.
Bush voters recall who they voted for...
Gore voters suffer from Alzheimer's?
Was it Bad weather early in the day keeping Bush voters home?
Right.
Breaking News! Republicans buy umbrellas at Walmart.
Was it inexperienced pollsters?
Mitofsky trained them. Who is better qualified?
Was it the exit poll "cluster effect"?
Do I hear 20%? 30%? 50%?
Ok, enter your "design effect" into the Interactive Election Model.
Let's see how many states will deviate beyond the MoE for Bush.
The model will calculate the probabilities.
Maybe not 1 in 19 trillion (16 exceeding MoE), but still astronomical.
And how does one explain 30% poll deviations in the Ohio 2005 election?
How much evidence is necessary to prove the DRE fix?
Were the pre-election polls biased, as well?
Naysayers can't blame it on "cluster effect".
Or bad weather.
Or shy Bush voters.
Or Gore voter Faulty Recall.
Or untrained pollsters.
These were PRE-ELECTION POLLS.
For naysayers to say that they wanted a Kerry win is a canard.
They claim to be Democrats or Indies searching for the truth.
To prevent fraud in the NEXT election.
As if THAT gives them credibility.
They want to have it both ways.
Deny that Kerry won and that the polls were right.
Yet at the same time claim that they wanted him to win.
Naysayer allegiance to Mitofsky is obvious.
They say the math is correct.
No argument there.
But they don't agree with the assumptions.
What assumptions?
That pre-election polls favored Kerry?
I can prove it. Go to pollingreport.com
That undecided voters went for Kerry by almost 2-1?
That new voters went to Kerry by 3-2?
That Nader voters went to Kerry by 4-1?
See the National Exit Poll time lines.
That the 43/37% Bush/Gore voter share of the 2004 vote was impossible?
It took a long time for the naysayers to agree.
After all, even they would not claim Bush voter immortality.
That the Final Exit Poll must be wrong?
Well, to match the vote, it applied fictitious weightings.
That's a no-brainer.
That all other Final demographics/vote shares must be wrong, as well?
Well, that's just simple logic.
If A = FALSE
and A = B
then B = FALSE
Do I hear heads exploding?
Or is it just another terror alert?
That Kerry's Gender share was manipulated?
It went from 54% FEMALE at 12:22am to 51% at 1:25pm.
Was it a massive sex change in 12 hours?
Christine Jorgensen never owned a computer back in 1952.
That the Party ID split was manipulated?
From 38 Dem/35 Rep to 37/37.
Was it Massive Fundie conversions in those 12 hours.
That the Census 2004 Vote Survey indicated 3.4mm votes were lost?
According to the Census, 125.7 million voted in 2004.
That's 3.4mm more than the recorded 122.3 million.
The Census Gender demographic MoE is 0.30%.
Should we believe the 122.3mm recorded vote?
The Final Exit Poll has a 1.0% MoE, according to Mitofsky.
That millions of votes are spoiled in every election?
Intentional or innocent spoilage?
Does it matter?
That the trend in Kerry/Bush response (alpha) disproves rBr?
The ratio declines from 1.50 in High Bush precincts to 1.0 in High Kerry.
USCV proved rBr was a myth using simulation.
I confirmed the USCV using the Exit Poll Response Optimizer.
That Bush job approval on election day 2004 was 48.5%?
That's an 11-poll average.
I can prove it.
You can look it up at pollingreport.com.
The combined MoE (11000 sample) is approx. 1.0%.
That there is no way Bush could overcome 48.5% approval?
Oh, well there is one.
He could steal it.
That the Ohio exit poll showed Kerry the 52-48 winner?
Of the 49 exit poll precincts:
36 deviated from Kerry to Bush,
10 from Bush to Kerry,
3 were unchanged.
That if Kerry won Ohio, he must have done better nationally?
Check the record books.
Ohio always LAGS the national Democratic vote.
Naysayers agree there was fraud in Ohio
But what about the other states?
NM, NV, FL, NC, NY, MN, etc..
That the 12:22am state and national exit polls did
confirm a Kerry victory ?
Well, forty-two of 50 states deviated to Bush.
That's 1 in 2 million odds.
That 50 state exit polls mirror 49 Ohio exit poll precincts?
Just a coincidence?
Move along. Nothing to see here.
Take a trip to Brazile, Donna.
That the 9% disparity between the voting shares of Florida optiscans and DREs is virtually impossible?
Dem/Rep registrations were essentially equal in Optiscan and DRE counties.
That sixteen of 50 states deviated beyond the MoE for Bush, none for Kerry?
The probability of that is 1 in 19 trillion.
That ALL 22 Eastern Time Zone states deviated from Kerry to Bush?
1 in 4 million.
That eighty-six of 88 documented touch screens switched Kerry votes to Bush?
See the EIRS database.
Do the math.
1 in 79,000,000,000,000,000,000,000.
THOSE ARE FACTS, NOT ASSUMPTIONS.
Take a look at the graph below.
It shows a time line of pre-election and post election polls.
Naysayers claim the "evidence" shows that Bush won the popular vote.
I ask, what evidence?
If Kerry won the popular vote, doesn't that mean the exit polls (state and national (12:22am) were therefore close to the truth?
Except, that is, for the 1:25pm Final National Poll
We know this one is pure, unadulterated BS. Why?
Look at the How Voted in 2000 demographic.
Focus on the 43%/37% Bush/Gore weights.
They are mathematically IMPOSSIBLE.
Here's the PROOF:
Bush got 50.45 million votes in 2000.
That's 41.25% of the 122.3mm who voted in 2004.
But 3.5% of them died, according to annual U.S. mortality rates (0.87%).
Therefore, AT MOST, 48.7mm of Bush 2000 voters came to the polls in 2004.
That's 39.8% of the 122.3mm total.
THE BOTTOM LINE:
Assuming REALISTIC, PLAUSIBLE, EQUAL weightings for Bush and Gore voters,
KERRY WINS EASILY, EVEN IF FINAL EXIT POLL PERCENTAGES,
WHICH WERE BIASED IN FAVOR OF BUSH, ARE USED.
PERIOD.
CASE CLOSED.
FINITO.
THE SMOKING GUN.
QUERE MAS?
Once again, I challenge the naysayers to a real-time debate using the Interactive Election model.
Let's begin where the DU "Game" thread abruptly ended:
http://www.democraticunderground.com/discuss/duboard.ph...
In the "Game", the naysayers claimed that Bush won 15% of Gore voters.
And that Kerry won only 52% of those who did not vote in 2000.
Can they ever come up with a plausible Bush win scenario?
I doubt it.
Note to Land Shark:
The TOTALITY of pre-election and exit poll data provide SOLID CIRCUMSTANTIAL EVIDENCE that the election was stolen.
The NUMBERS have been silently screaming for a year.
The fact is that AFTER A FULL YEAR there's still strong DOUBT as to who really won.
That should be sufficient incentive for the public to know that something must be done ASAP to restore our democracy.
____________________________________________________________________________________
FEBBLE CHIMES IN WITH AN EXPLANATION:
THE EXIT POLLS WERE “DOCTORED”, BUT THIS IS NOT UNUSUAL.
IN OTHER WORDS, IT’S OK TO MATCH THE FINAL EXIT POLL TO A VOTE – ANY VOTE.
EVEN IF IT’S A CORRUPTED, FRAUDULENT VOTE.
Febble (1000+ posts) Thu Dec-01-05 05:44 PM
Response to Reply #12
16. Well do check out
Late on election day, 2004, with Kerry ahead in the exit polls (scientific polls of voters leaving the voting booth), the corporate news monopolies doctored their exit poll data to fit the results coming from Diebold's and ES&S's secret vote tabulation software. This was not the normal adjustment of exit polls for election demographics. It was a very unusual--indeed, impossible--falsification of the numbers. It has been verified that the real exit polls showed a Kerry win by a 3% margin.
The exit polls were "doctored" not by the "corporate news media" but by those who actually conducted the "scientific polls of voters leaving the voting booth". However this was not "very unusual" - it is what the pollsters do in every election, and, ironically, is probably how they have achieved their reputation for extreme accuracy. Moreover, the "doctoring" was done "late on election day" in the sense that the process started as the first polls closed. This is because the exit poll estimates were as usual, based on three data sources: How are projections made?
Projections are based on models that use votes from three (3) different sources -- exit poll interviews with voters, vote returns as reported by election officials from the sample precincts, and tabulations of votes by county. The models make estimates from all these vote reports. The models also indicate the likely error in the estimates. The best model estimate may be used to make a projection if it passes a series of tests.http://www.exit-poll.net/faq.html#a10 As for your word "impossible" - the discrepancy between the unadjusted poll data and the vote count cannot have been due to chance. That is indeed "impossible". It may have been due to fraud in the vote-count. However, bias in the poll is also perfectly possible. Bias can occur when sampling is not random. It rarely is.
UK exit polls frequently show bias - and we know it is bias because we can rely on our count (fully hand-counted paper ballots). Sadly, you can not. This is a grievous problem. But having unreliable vote-counting methods does not make exit polls magically more reliable. "Impossible" exit polls are only too common, even in clean electoral systems. I am still traumatised by Neil Kinnock's defeat, in defiance of the exit polls, in 1992.(Edited to add: I do agree with you about Gore/Kerry! I was a Gore fan, and still haven't quite got over Florida 2000. That's why I'm here, actually, making a nuisance of myself....)
FEBBLE GETS IN AN EARLY JAB: TYPICAL VAGUE SUGGESTIONS ABOUT “BAD STATISTICAL INFERENCES” SHE CITES NO SPECIFICS, BUT SHE CLAIMS TO DISAGREE WITH ALL OF THEM…
Febble (1000+ posts) Fri Dec-02-05 07:56 AMResponse to Reply #37
42. Wow
Edited on Fri Dec-02-05 08:56 AM by Febble
well, as you've shown an interest in the recipient of the sarcastic jabs, here's the source (in context) of the quote:
http://www.democraticunderground.com/discuss/duboard.ph...
Yes, I admire TIA's enthusiasm too, but I do wish his assertions weren't billed as FACTS.Statistical inferences aren't facts at the best of times, and bad statistical inferences masquerading as facts are worse.But the dedication and diligence I can only admire. And the cause in which they have been exercised. to TIA. On edit: I should make clear, in case it isn't, that I disagree with virtually all TIA's inferences....
TIA MUST HAVE REALLY TOUCHED SOME NERVES WITH THIS ONE…
Febble (1000+ posts) Fri Dec-02-05 12:38 PMResponse to Reply #42
52. I should also make it clear
Edited on Fri Dec-02-05 01:19 PM by Febble
that I don't admire TIA's ad hominem attacks on people who are also diligently working to find out what went wrong with the 2004 election, whichever piece of ectoplasm he is embodied in right now.(On edit: by an ad hominem attack, I mean, for example a response that does not address the argument in question, but denigrates the perceived motivation of the person making the argument.)
OTOH ATTEMPTS TO DENIGRATE TIA’s NATIONAL PRE-ELECTION DATA SOURCE WITHOUT LISTING THE POLLS USED. THAT’S OK, WE’LL SURPRISE HIM SOON WITH A DETAILED LIST COPIED DIRECTLY FROM POLLINGREPORT.COM.
OnTheOtherHand (889 posts) Fri Dec-02-05 04:37 PMResponse to Reply #57
58. "the final TIA poll" is contradicted by, well, the actual polls
http://pollingreport.com/2004.htm
(Bush leads in 5 of 8 projections, 10 of 14 trial heats)http://people-press.org/commentary/display.php3?Analysi... (slightly different list of trial heats, Bush again leads in 10 of 14)I could go on, but we have done this before.My vote is for the accuracy of the 12:22AM NEP and TIA. To say otherwise is to endorse the Diebold's of the world.An impressive new contender for the FEMME Trophy (Fallacy of the Excluded Middle: Most Egregious).
OTOH BRINGS UP THE POLLS- CHERRY-PICKING HIS (AFTER THE ELECTION) While TIA’s 18 NATIONAL POLLS WERE CHOSEN FOR THE MODEL BEFORE THE ELECTION
Note: I added this response after I had a chance to check out the poly sci election models that OTOH Referred to:
OTOH, since you are too lazy (or embarrassed) to show the poli sci model projections, I'll do it for you.
http://www.apsanet.org/content_13000.cfm
All the models presented here were not even close to the corrupted recorded vote, much less the TRUE vote.Except for Beck/Tien, that is.
Author Pick 2-pty Date Win Probability
Abramowitz Bush 53.7% 7/31/04 na
Campbell Bush 53.8% 9/06/04 97%
Wlezien/Ericson Bush 52.9% 7/27/04 75%
Holbrook Bush 54.5% 8/30/04 92%
Beck/Tien Kerry 50.1% 8/27/04 50%
Lockabie Bush 57.6% 5/21/04 92%
Norpoth Bush 54.7% 1/29/04 95%
First of all, look at the model dates.The latest was 9/6/04.
These guys should know that things change every day in a race.The electorate is a dynamic organism.Couldn't they have run the models on Nov.1, like I did?As I did every day for four months leading up to the election?Maybe next time they'll get it right, but I doubt it.They are by nature too conservative to change their approach.
Their combined Bush 2-party average forecast of 54% isn't even closeto the recorded vote, much less the true vote. The average is off by more than 5%, since Bush got about 48.5% of the two-party vote.The only one which was even close was Beck/Tien.They had it 49.9% for Bush, and were off by 2%.Not bad. The rest? Fuggedabout it.
One other thing.There is no way Lockabie's 57.6% Bush popular vote equates to anything less than a 100% EV and/or popular vote win probability.His 92% Bush win probability doesn't say too much for his model.At a 97.5% one-tail level of confidence, it implies an equivalent MoE of 7.6% based on the 57.6% projection. Not good, especially when winning 52% of the popular vote means a virtual 100% probability of winning the electoral vote. The same goes for the rest of the models. Only Beck-Tien's 50% probability made sense, since they projected a virtual dead heat.
Gee, OTOH, I sure teach you a lot, don't I? And you're the political science professor. Yet you say I'm irrelevant to your colleagues.Does MP also consider me irrelevant?He learned a little math from my postings also.What about Ruy Texeira?At least he questioned the Hispanic vote.Now, with all due respect to the poly scientists above:Have these guys ever used Monte Carlo simulation? Did they ever consider projecting individual states to,you know, calculate the Bush probability of winning the electoral vote? From what I could tell in my admittedly cursory review of the PDFs, they essentially all used the same factor analysis regression method, based on macro-economic/financial data. Polling was mentioned, butnot much. They should check out the Election Model for the next election (assuming it's relatively clean) and consider a Monte Carlo simulation.
IN MY OPINION, REGRESSION-BASED ELECTION FORECASTING MODELS ARE AN EXERCISE IN ACADEMIC ONE-UPSMANSHIP: "MY MODEL IS BETTER THAN YOURS".SORT OF A MASTURBATORY CIRCLE-JERK APPROACH.REGRESSION FACTOR-ANALYSIS IS NOT THE BEST WAY TO FORECAST ELECTIONS!ANALYZING SELECTED STATE AND NATIONAL POLLING RESULTS IS FAR SUPERIOR.IT MAY BE MUNDANE AND NOT AS SEXY AS FACTOR ANALYSIS, BUT IT'S MORE ACCURATE, AND THAT'S THE NAME OF THE GAME, ISN'T IT?
BUT... ALL FORECASTS ASSUMED AN HONEST ELECTION.I DID NOT SEE ANY FRAUD VARIABLES, OR SPOILED VOTE FACTORS, OR DISENFRANCHISEMENT PARAMETERS, OR VOTER INTIMIDATION FACTORED INTO THE MODELS. NOT A WORD ABOUT POTENTIAL FRAUD.JUST ECONOMIC AND/OR FINANCIAL FACTORS.WITHOUT A PAPER TRAIL, LOUSY FORECAST MODELS MAY CORRECTLYPROJECT THE "WINNER" IF HE STEALS THE ELECTION.SOMETHING LIKE THE FINAL 1:25PM NATIONAL EXIT POLL.
So what does it all prove? Which model was right?Well, if you believe the election was stolen, mine was.OTOH, do you believe the election was stolen?ONLY MY PRE-ELECTION STATE AND NATIONAL ELECTION MODEL MATCHED THE EXITS.BY THE WAY, OTOH.WHERE IS YOUR MODEL?Ah, what's the use?Talking to you is like talking to a stone.
OnTheOtherHand (889 posts) Fri Dec-02-05 06:51 PMResponse to Reply #58
63. ah, TIA seems to be channeling himself elsewhere on the Net
Edited on Fri Dec-02-05 07:12 PM by OnTheOtherHandbut
I have no idea what the DU protocol is for responding to someone who Isn't There. (EDIT: "but I have idea" could've been some weird meta pun, but wasn't)Anyway, the line seems to be: pay no attention to what anyone else -- including the pollsters themselves -- say their polls indicate. Sure, those darn MSM analysts at Pew said "Slight Bush Margin in Final Days of Campaign", but the illuminati know that it shoulda been Kerry +1. pollingreport.com, shmollingreport.com -- WeGo YouGov. All those poli sci models that called the election for Bush -- best not to mention those, much less refute them. The goofuses over at "Polly's Page", well, who wants to listen to parrots, anyway? Might as well argue with a stone.
FEBBLE DOESN’T QUITE UNDERSTAND THE ELECTION MODEL
Febble (1000+ posts) Fri Dec-02-05 05:41 PM
Response to Reply #59
60. Let me get this straight:
Edited on Fri Dec-02-05 05:54 PM by Febble your link to TIA's "final poll" says:Election Model Projections. If the election were held today, then based on recent state polling, the Electoral Vote Simulation model calculates that John Kerry has a 99.8% probability of winning an electoral vote majority by a 337-201 margin and 51.80% of the popular vote. Kerry won 4990 of 5000 Monte Carlo simulated election trials.
Based on the average of eighteen national polls, the National Vote Projection model calculates that Kerry has a 99.99% probability of winning a popular vote majority with 51.63% of the vote. So TIA was actually saying, on the eve of the election, that Kerry a 99.8 probability of winning the popular vote by nearly four percentage points?Well, I have to confess, I am gobsmacked, as we say around here.Every single pollster, including all those cited by TIA, was saying it was "too close to call" - and yet TIA thought that Kerry was a dead cert (99.8 is a dead cert in statistics) for a massive win of the popular vote?(On edit: well I suppose nearly 4% isn't really massive, but that makes giving it a probability of 99.8% even more, well... odd.)
OTOH ATTEMPTS TO EXPLAIN THE ELECTION MODEL TO FEBBLE WHILE DEMEANING IT AT THE SAME TIME…WHY….BECAUSE IT WAS CONFIRMED BY THE EXIT POLLS?
OnTheOtherHand (889 posts) Fri Dec-02-05 06:07 PM
Response to Reply #60
61. well, TIA said two things, both rather odd
TIA said that Kerry had a 99.99% (not 99.8%) probability of winning the popular vote. Let's not lowball things here, eh?He also said that If the election were held today, then based on recent state polling, the Electoral Vote Simulation model calculates that John Kerry has a 99.8% probability of winning an electoral vote majority by a 337-201 margin and 51.80% of the popular vote.But he cannot have meant quite that. Surely he didn't mean that Kerry had a 99.8% chance of winning 51.80% of the popular vote. (And 0.2% chance of winning 51.81%?!) Those figures (51.80% for the Electoral Vote Simulation, 51.63% for the National Vote Projection) are just his mean expected values, surely.Anyway, as I always say, if you disagree with TIA, you probably work for Diebold. (Actually, that is a freehand paraphrase, but real election reformers shouldn't get hung up on words.)
FEBBLE PICKS UP ON THE OTOH SARCASM…SHE DOESN’T GET THE KERRY WIN PROBABILTY Febble (1000+ posts) Fri Dec-02-05 06:43 PM
Response to Reply #61
62. Ah!
Riiiiight. So Kerry had a 99.8% (or maybe 99.99%)chance of winning, with an expected margin of a bit under 4%. And an EV margin of 136.You know what? I think TIA works for Mitofsky....
SO IT’S TIME FOR TIA TO GIVE FEBBLE A FREE EDUCATION ON THE ELECTION MODEL PROBABILITIES…AND TO CORRECT OTOH WITH PROOF THAT KERRY LED IN THE FINAL PRE-ELECTION NATIONAL POLLS autorank (1000+ posts) Fri Dec-02-05 08:02 PM
Response to Original message
65. Probability Questions & Poll Accuracy, the Muse Speaks
I’m getting new data from my channeling TIA function.
Voila! Someone said:"TIA said that Kerry had a 99.99% (not 99.8%)probability of winning the popular vote. Let's not low ball things here, eh?"Now it's my turn:Yes, I said it.Want to check the math?"If the election were held today, then based on recentstate polling, the Electoral Vote Simulation model calculates that John Kerry has a 99.8% probability of winning an electoral vote majority by a 337-201 margin and 51.80% of the popular vote".
Now for some pro bono public education.In the model, I calculated that Kerry had a 99.8% chance ofwinning the electoral vote (at least 270 votes).That's the result of winning 4990 of 5000 simulation trials.And 337 is the expected mean EV, assuming Kerry won 75% ofthe undecided vote.
The probabilities are:18 national polls- 99.9994% (to win the popular vote),
9 national independent polls: 99.9799% (to win the popular vote)
5000 trial EV simulation (51 states): 99.8% to get at least 270 EV
Want to check the math? The math is correct. Febble has always agreed in the past.
For 18 national polls (18000 total sample)MoE = 0.73%
Kerry 2-party projected vote (mean): 51.63%
Prob =normdist(.5163,.5,0.0073/1.96,true)
Prob= 99.9994%
For the 9 independent pollster subset (9000 sample)
MoE = 1.03%
Kerry 2-party projected vote (mean): 51.86%
Prob = normdist(.5186,.5,0.0103/1.96,true)Prob= 99.9799%
See ya guys...
Polling Date Accuracy Referenced FINAL NATIONAL PRE-ELECTION POLLS
18 Poll Summary:
Kerry won 9, Bush 8, 1 tie
Kerry won 5 of 9 Registered Voter (RV) Polls
and 4 of 9 Likely Voter (LV) Polls
Polling Data Source: http://www.pollingreport.com/wh04gen.htmhttp://www.economist.com/media/pdf/YouGovS.pdf
Total Poll Total Weighted Avg 67% 33% Sample Sample MoE KERRY BUSH KERRY BUSH
Date 26961 Group 0.60% 47.55 47.30 49.95 49.05
1-Nov Marist 1166 LV 2.87% 50 49 50.13 48.87
1-Nov Econ 2903 RV 1.82% 50 47 51.24 47.76
1-Nov TIPP 1284 LV 2.73% 44 47 50.35 48.65
1-Nov CBS 1125 RV 2.92% 47 48 48.35 50.65
1-Nov Harris 1509 LV 2.52% 48 49 51.31 47.69
31-Oct Zogby 1200 LV 2.83% 47 48 48.37 50.63
31-Oct FOX 1400 RV 2.62% 48 45 51.40 47.60
31-Oct DemCorp 1018 LV 3.07% 48 47 47.75 51.25
31-Oct Gallup 1866 RV 2.27% 48 46 50.51 48.49
31-Oct NBC 1014 LV 3.08% 47 48 46.01 52.99
31-Oct ABC 3511 RV 1.65% 47 48 48.65 50.35
30-Oct ARG 1258 LV 2.76% 49 48 49.05 49.95
30-Oct Pew 2408 RV 2.00% 46 45 50.55 48.45
29-Oct Nwk 1005 RV 3.09% 44 48 48.78 50.22
26-Oct ICR 817 RV 3.43% 48 48 53.16 45.84
24-Oct LAT 1698 RV 2.38% 48 47 51.26 47.74
21-Oct Time 803 LV 3.46% 46 51 47.38 51.62
20-Oct AP 976 LV 3.14% 49 46 53.16 45.84
OTOH IS FLUMMOXED; HE CAN NO LONGER CLAIM TIA POLL DATA IS WRONG, SO HE REVERSES COURSE. NOW HE ATTACKS THE METHODOLOGY OF CONSOLIDATING 18 INDEPENDENT NATIONAL POLLS…OTOH, DID YOU FORGET THE LAW OF LARGE NUMBERS?
OnTheOtherHand (889 posts) Fri Dec-02-05 08:25 PMResponse to Reply #65
66. whoa: I actually AM arguing with a stone... cool...
This is not a math question.In the real world, one can't get everyone to agree to work from TIA's list of Real poll results. See also my #63.Moreover, it is not S.O.P. in meta-analysis to just add a bunch of surveys together to make One Big Survey. And even if one did, it would be pretty wishful to assume that sampling error is the only error source in the model.But all this, and more, has been explained before.
FEBBLE CONFESSES HER FAILURE TO DO HER HOMEWORK ON THE ELECTION MODEL.SHE CLAIMS THE MATH MUST BE WRONG IF IT GIVES A 99% KERRY WIN PROBABILITY.SHE THEN PROCEEDS TO BLOW SMOKE WITH “META-ANALYSIS” JARGON.SHE PARROTS OTOH: “A GOOD META-ANALYSIS IS HARD TO DO”.FEBBLE, IT’S NOT THAT COMPLICATED.YOU ARE ABOUT TO BE VERY SURPRISED.TRUST ME, FEBBLE, THE MATH IS CORRECT.YOU WILL SOON FIND OUT WHY.
Febble (1000+ posts) Sat Dec-03-05 05:51 AMResponse to Reply #65
70. Qualification:
I confess I had not looked at this "final poll" before, but if "the math" gives a 99.9% (or even a 99.8%) probability of a Kerry win, then something about the math is wrong. I was optimistically putting it the odds at about 50:50 myself, and hoping that even if the popular vote was pretty evenly split, the EV would benefit Kerry. Which is sort of what happened (in the official vote anyway), in that a few thousand more (official) votes for Kerry in Ohio would have given him the presidency, but not the (official) popular vote.
And more generally, when I say I agree with TIA's math, I certainly trust him to perform the computations he does correctly - the issue is usually whether we can legitimately infer from them what he infers from them. But his probability calcs are frequently invalidated by the assumptions that underlie them, and a probability calc based on erroneous assumptions is not a correct calculation).So what is wrong with this particular probability calc? Well for a start, as OTOH says, a good meta-analysis is a complex thing to do, and one thing it involves is using ALL data you can find (including unpublished data, if you can find that) and weighting it appropriately.
"Appropriately" is the hard part of course, and it is difficult to avoid the charge of "cherry-picking". There are, however, systematic ways of weighting studies according to statistical power and cleanliness of methodology. But the fact remains that meta-analyses in general (and meta-analyses of these pre-election polls in particular) are acutely sensitive to the studies you pick and the assumptions you make. In the case of the pre-election polls, some analysts called it for Bush (but with nothing like 99.9% confidence; others sat on the fence; and a few brave souls like Sam Wang (I think) called the EV for Kerry (with a wing and a prayer).
Yet TIA called it for Kerry with 99.9% confidence? From a eighteen polls in which (unweighted) the mean difference between Kerry and Bush was not (on my calcs) significantly different from zero?Look, I don't think that election was on the up and up, and I think the exit poll stuff was worth investigating. And apart from anything else, it also energised a lot of people (including me) into investigating stuff that brought the election result into serious doubt. But claiming that Kerry's probability of victory, in both the electoral college and popular vote, was 99.9% - well, as we say on our side of the pond, pull the other one, it's got bells on.Were any bookies offering 100:1 odds on a Bush win?
FEBBLE SCAMPERS BACK IN AFTER CHECKING PROFESSOR SAM WANG’ META-ANALYSIS FORECAST...BLIMEY, HE AGREES WITH TIA...!!!!!
Febble (1000+ posts) Sat Dec-03-05 01:52 PMResponse to Reply #70
76. Update
Just checked Sam Wang:
http://synapse.princeton.edu/~sam/pollcalc.html
Looks like he did indeed commit himself to a 98% probability of a Kerry EV win, bless him (he doesn't seem to give a probability for his Popular Vote estimate)And it's still on a wing and prayer, though. He modified his "gut estimate" to 6:1 after a Bayesian adjustment as to whether his assumptions were likely to be correct.Well, there you go. Sam Wang and TIA agreed, at least on the EV win.I'm afraid it doesn't convince me that either of them were correct.Actually, Sam Wang doesn't seem that convinced he was correct either:
http://synapse.princeton.edu/~sam/pollcalc_letters_afte...
but he has interesting things to say about why he might have been wrong:I think the purely statistical aspects of the analysis did extremely well. The electoral outcome looks like it will be close to the decided-voter outcome predicted by the polls. Victory margins are quantitatively close to the pre-election polls: out of 23 battlegrounds, the direction of the outcome was predicted in 22 (the exception was Wisconsin, where the polling margin was 0.4% for Bush and the actual margin was about 0.4% for Kerry). Quantitatively, 12 victory margins were within one standard error and 17 were within the 95% confidence interval. Not perfect, but not bad.
The most significant errors had to do the net effect of other factors not encompassed by polls. To make a final prediction, I used previous patterns of uncommitted voters breaking for the challenger as a guide, but this break either did not occur or was cancelled by other factors. My assumption of high turnout was flat out wrong! In the end, the likely-voter models of pollsters were not too far off.There has been talk of other factors, but a parsimonious explanation may be that the net effect of all other factors was zero. This isn't always true - in past years the outcome seems to have not matched final polls. There seems to be some mystery offset that varies a bit.
On the other hand, this year we had more data - maybe it's just a question of having enough data and the right answer falls out.One advantage of rigorous statistical modeling is that you can see a clear separation between factual information and assumptions of less certainty. In this case my baseline calculation was quite accurate, but the intangibles were wrong. As I said, in previous years at least one of the assumptions would have worked. What happened this year is a question for the political and policy people - in the end it goes to show that I am at my best with the numbers! (my bold)See also:http://election.princeton.edu /
OTOH IS QUICK TO EXTOL SAM WANG’S ANALYSIS. OTOH KNOWS THAT MR. WANG IS A PROFESSOR AT PRINCETON. HE’S NOT TIA. BETTER NOT TRY TO TRASH SAM.
OnTheOtherHand (889 posts) Sat Dec-03-05 03:09 PMResponse to Reply #76
77. yes, Wang gets big points for isolating crucial assumptions:
Note that all of these probabilites are conditional on the turnout and undecided voter assumptions being correct. The true probability is obtained by multiplying by a measure that is a function of whether my assumptions are accurate. The chance that I am wrong makes the true probability substantially lower than 100%! As Niels Bohr (and Yogi Berra) said, "Prediction is hard, especially of the future."http://election.princeton.edu Wang is confusing when he refers to his "assumption of high turnout" -- rather, his assumption was that high turnout would favor Kerry. Surely in some places it did (and conceivably fraud conceals some of its actual impact). But as we've discussed, it was never empirically all that obvious that high turnout 'had to' favor Kerry. TIA's parallel leap was to prefer Registered Voter results to Likely Voter results, although he uses some of each.
OTOH QUOTES TOE-SUCKER DICK MORRIS.HE SAYS THAT MAYBE THE EXIT POLLS WERE RIGGED.WELL, OTOH, TAKE A LOOK AT THE FINAL NATIONAL EXIT POLL.NOT THE PRELIMINARY 13047; THAT WAS FINE.LOOK AT THE FINAL 13660 – MATCHED TO THE VOTE.OTOH THEN PROCEEDS TO BASH THE U.S. CENSUS. WHY? THE SURVEY INDICATED THAT 3.5 MILLION VOTES WERE UNCOUNTED.
OnTheOtherHand (889 posts) Sat Dec-03-05 11:42 AM
Response to Reply #67
74. E/M, and the census
Some folks think that Mitofsky has been rigging exit polls for, umm, I guess decades. I guess for them, the disbanding of VNS is the least of the worries. (Dick Morris thought the exit polls were rigged to discourage Republicans; some folks have suggested instead that they were rigged to lull Democrats into complacency.) I honestly don't quite understand why anyone believes any of this. AFAICT, most public opinion researchers think that the exit polls were wrong, and a few think that the exit polls were basically right and the vote count was wrong; few see any need for the hypothesis that the exit polls (and/or pre-election polls) were rigged.
Anything is possible, I guess.I didn't notice the "Census 2004 Vote Survey" chestnut coming back. It's actually the Voting and Registration Supplement to the Current Population Survey (CPS). To quote a 1998 Census Bureau press release, "The CPS routinely overestimates voter turnout" (http://www.census.gov/Press-Release/cb98-146.html ), so most of us don't waste much time trying to reconcile the figures. You may well be right that presidential undervoting accounts for part of the discrepancy; I would have to look more closely.
FEBBLE PROVIDES A FULL-FRONTAL REBUTTAL TO THE ORIGINAL POST SHE RESPONDS WITH MOSTLY VAGUE STATEMENTS.SHE BRINGS NOT A SINGLE FACT TO THE TABLE. SHE CONSISTENTLY DENIGRATES ALL PRE-ELECTION AND FINAL POLLSSHE IS CAREFUL TO COUCH HER STATEMENTS, TO GIVER HER WIGGLE ROOM.. HER PRIMARY ARGUMENT IS THAT ALL POLLS ARE BIASED, THEREFORE ONE SHOULD NOT DRAW ANY INFERENCES FROM THEM. SHE REPEATS THE FALSE VOTER RECALL CANARD.BUT APPARENTLY ALZHEIMER’S ONLY AFFLICTED GORE VOTERS. THAT’S HER EXPLANATION FOR THE IMPOSSIBLE 43%/37% BUSH/GORE 2000 WEIGHTS She repeats her mantra – that polls are not random samples (WTF?):“1. Random sampling””2. Random sampling””3. Random sampling”
Febble (1000+ posts) Sat Dec-03-05 12:22 PMResponse to Original message
75. OK, these are my words:
....................................................................."So I would argue that the Totality of the Evidence at present adds up to this:1. The election was corrupt.2. Democrats were the net losers from the corruption.3. Voter (and vote) suppression remains a key problem, and may have cost Kerry Ohio.4. Kerry probably lost the popular vote.5. But we do not know for sure who actually won either the popular or the electoral vote, and this insupportable (sic.)".....................................................................
The naysayser confirms the corruption in points 1,2,3.But still keeps from going all the way in points 4 and 5.So, my response:Even though all available evidence says otherwise.Not in my view. So let's summarize some important FACTS ...Al Gore was ELECTED in 2000.It was not just stolen from him.It was stolen from US.But at least Al fought.Certainly Gore had more intended votes in Florida than Bush, and won the popular vote.John Kerry was ELECTED in 2004.It was not just stolen from him. It was stolen from US. John, we hardly knew 'ye. Not facts.
TIME FOR SOME FACTS:The naysayers continue their relentless attempts to mask the overwhelming evidence provided by confirmation of hundreds of pre=election and exit polls:
Not a fact
1) Pre-election state. Total sample: 50 polls* 600 = 30,000
2) Pre-election national. Total Sample: 18 polls *1500 = 27,000
3) Pre-election 48.5% Bush approval. Total: 11 polls*1000 = 11,000
4) 12:22am state exit polls: 73,600 respondents
5) 12:22am national exit poll: 13,047 respondents
These are facts in the sense that these polls were published. However the numbers are inferences from samples, and their margins of error assume random sampling. All the polls cited had high non-response rates, and cannot be assumed to consist of random samples, although one hopes they did.And to this we must add have all the massive documented evidence of vote miscounts:OH, FL, PA, NV, NM, VA, NC, MN, IA, WI, MO, NY…This is a qualitative statement, not quantitative. "Massive" needs to be supported by quantitative data before I will accept it as a "fact".
But yes, I agree there is good evidence of "irregularities", and probably in all these states. More than vote miscounts, however - some of the best documented evidence includes incidents where the voter was unable to vote at all.Were ALL these pre and post-election polls BIASED? Probably not.They ALL confirm that Bush lost.A pre election poll cannot "confirm" what happened in an election. And I was following the pre-election polls closely, and they certainly weren't telling me that "Bush lost". And yes, they may have been biased, some more than others, some in on direction some in another.
What is it about these polls that is so difficult to understand?
I dunno - what is it about "too close to call" that is so difficult to understand?Where is the evidence that the exit polls were biased? Was it shy Bush voters (rBr)?Debunked in NEP by Mitofksky 43%Bush/37% Gore?
I have no idea why TIA thinks the recalled vote data "debunks" non-response bias as an explanation for the discrepancy between poll and count. I certainly wouldn't describe the hypothesis as "shy Bush voters", but I find it perfectly plausible to postulate that slightly more of the "reluctant" voters (and we know there were a lot) were Bush voters than Kerry voters.
Of the "non-reluctant" voters, I also find it perfectly plausible that some Gore voters mistakenly recalled voting for Bush; OTOH cited an actual longitudinal study in which voters who voted for Gore later recalled voting for Bush. If similar proportions did so in the exit poll, it would account for the exit poll data Bush/Gore proportions.
Was it early Kerry voters? Debunked.
View the time line Kerry led at 4pm, 7:33pm, 12:22am
Was it early women voters?Debunked.
Female vote share
4pm: 58%, 7:33pm: 54%, 12:22am: 54%, 1:25pm 54%
Neither of these, as we both know. The timeline is irrelevant. The timeline simply reflects dynamic re-weighting processes. We know, in any case, from the E-M report that the problem was at the level of the precincts - a greater proportion of Kerry voters were sampled on the poll than were represented in the vote.
Was it False Gore voter recall? Ridiculous on its face. Bush voters recall who they voted for...Gore voters suffer from Alzheimer's?
I have to believe this assertion (not a fact) is willful denial of perfectly good evidence. Check back to the Game thread for OTOH's source. Inaccurate vote recall is a well-attested phenomenon, and often favours the incumbent.
Was it Bad weather early in the day keeping Bush voters home?
Right.Breaking News! Republicans buy umbrellas at Walmart.Well, they didn't have to queue as long in Ohio, so maybe. But the evidence suggests not.
Was it inexperienced pollsters? Mitofsky trained them. Who is better qualified?
This is my best guess. They had, I understand, a brief telephone training. Not enough for what is actually a remarkably difficult job.
Was it the exit poll "cluster effect"? Do I hear 20%? 30%? 50%?
TIA should stop using these scare quotes, and find out what a DESR actually is (I've attempted to explain, and even gave him a link).
Ok, enter your "design effect" into the Interactive Election Model.Let's see how many states will deviate beyond the MoE for Bush. The model will calculate the probabilities. Maybe not 1 in 19 trillion (16 exceeding MoE), but still astronomical.
No, because I do not share the assumption of the model that the only source of error in the polls is sampling error (DESR is also an estimate of sampling error).
And how does one explain 30% poll deviations in the Ohio 2005 election?
Well I think there are a number of possible explanations, but I thought we were talking about 2004? Possible there was fraud, although the finding by Klinkner that the discrepancy was not greater in DRE counties doesn't suggest DRE fraud particularly. But as I keep saying, until Ohio gets cleaned up, no-one will know the answer.
How much evidence is necessary to prove the DRE fix?
How much? Well, I'd like to see something more convincing that we've got. But I agree they are a terrible idea, on many, many, counts, and quite absurdly insecure, as the GAO report points out.
Were the pre-election polls biased, as well?
Oh, I thought we'd done this bit. Probably they were, but clearly not all in one direction. I think TIA eliminated the polls he thought were biased for Bush. I am sure they had non-sampling error, and tracking polls suggested some polls tended to track higher than others. This suggests that bias was a problem, and that certain pollsters tended to have a particular bias. This is highly likely, as bias is a function of methodology. One of the reasons I dispute TIA's lumping of polls together is that it ignores between-poll error (what is sometimes called "random effects" variance. It seems fairly clear that each poll had a characteristic "bias". It's the same in the UK. The one I like is ICM (for the Guardian) as it allows for "shy Tories" - it therefore tends to track Labour lower than the others - and generally gets closer to the result.
Naysayers can't blame it on "cluster effect".Or bad weather.Or shy Bush voters.Or Gore voter Faulty Recall.Or untrained pollsters.These were PRE-ELECTION POLLS.
This is pure rhetoric, not argument. See above.
For naysayers to say that they wanted a Kerry win is a canard.They claim to be Democrats or Indies searching for the truth.To prevent fraud in the NEXT election.As if THAT gives them credibility.They want to have it both ways. Deny that Kerry won and that the polls were right.Yet at the same time claim that they wanted him to win. Naysayer allegiance to Mitofsky is obvious.
Again, this is pure rhetoric. I will leave well alone.
They say the math is correct.No argument there.But they don't agree with the assumptions.What assumptions?
1. Random sampling
2. Random sampling
3. Random sampling
That pre-election polls favored Kerry? I can prove it. Go to pollingreport.com.
See mine (and others', including OTOH's, responses on this thread.That undecided voters went for Kerry by almost 2-1?That new voters went to Kerry by 3-2?That Nader voters went to Kerry by 4-1?See the National Exit Poll time lines.Don't understand this. See my comment re time-line. We also can't distinguish "new voters" from "rare voters". New voters will tend to be young and might be expected to favour Kerry. Rare voters - who knows?
That the 43/37% Bush/Gore voter share of the 2004 vote was impossible?It took a long time for the naysayers to agree.After all, even they would not claim Bush voter immortality.
Oh dear, round, and round, and round we go. No, I agree, that it is unlikely that Bush voters are immortal. False recall seems much more likely as an explanation - we actually know this is a phenomenon. The only immortal I know personally is a Kerry voter (or I assume he is...)
That the Final Exit Poll must be wrong? Well, to match the vote, it applied fictitious weightings.That's a no-brainer.
That all other Final demographics/vote shares must be wrong, as well?Well, that's just simple logic.If A = FALSE and A = B then B = FALSE.
Do I hear heads exploding? Or is it just another terror alert?
That Kerry's Gender share was manipulated?
It went from 54% at 12:22am to 51% at 1:25pm.
Was it a massive sex change in 12 hours? Christine Jorgensen never owned a computer back in 1952.
That the Party ID split was manipulated? From 38 Dem/35 Rep to 37/37.
Was it Massive Fundie conversions in those 12 hours.
OK, this is getting silly. You seem to have forgotten that the projections were continuously reweighted (and that they actually had some glitch with the gender thing - it's in the E-M report, look it up).
That the Census 2004 Vote Survey was wrong?According to the Census, 125.7 million voted in 2004.That's 3.4mm more than the recorded 122.3 million.The Census Gender demographic MoE is 0.30%.Should we believe the 122.3mm recorded vote?
Well the Census people think that it overstates the # voted. But it seems clear that spoilage will contribute to the discrepancy.
The Final Exit Poll has a 1.0% MoE, according to Mitofsky. That millions of votes are spoiled in every election? Intentional or innocent spoilage? Does it matter?
Dunno about millions, but yes it matters. It matters desperately, whether it is intentional or structural discrimination. It is a serious Civil Rights issue, and virtually certainly resulted in the Wrong Man being inaugurated in January 2001.Yay, we agree on something!
That the trend in Kerry/Bush response (alpha) disproves rBr? The ratio declines from 1.50 in High Bush precincts to 1.0 in High Kerry.
Well, as the coiner of "alpha" (actually it was my husband's term, as he couldn't read my "x"s, but I suggested the measure, and gave it that name) yes, I vigorously dispute this. Mitofsky tested precisely this, and no, it does not decline. The line is statistically flat.USCV proved rBr was a myth using simulation.Correction: some members of USCV claimed their simulation indicated that rBr was unlikely. Another member claimed his simulation indicated it was perfectly possible. The first lot then accused the second lot of accusing them of claiming that they'd proved rBr was a myth. Then there was a big fight. I think the ones that claimed that alpha was higher in high Bush precincts got it wrong. I confirmed the USCV using the Exit Poll Response Optimizer. Well, I think "you" did too, for the same reasons as USCV did. I wrote an entire geeky paper about it. But more to the point, Mitofsky actually did the calculation, using the whole dataset, and it isn't.
That Bush job approval on election day 2004 was not 48.5%? That's an 11-poll average. I can prove it. You can look it up at pollingreport.com.The combined MoE (11000 sample) is approx. 1.0%.
Well, prove it then. But remember that MoE assumes random sampling.
That there is no way Bush could overcome 48.5% approval? Oh, well there is one. He could steal it.
Well, I seriously wondered if he could. My conclusion is probably didn't.
That the Ohio exit poll showed Kerry the 52-48 winner? Of the 49 exit poll precincts:36 deviated from Kerry to Bush,10 from Bush to Kerry,3 were unchanged.
Yup, something was biased. Was it the poll? Was it the count? That is the question.
That if Kerry won Ohio, he must have done better nationally? Check the record books. Ohio always LAGS the national Democratic vote. Naysayers agree there was fraud in Ohio.
Cite please. And remember, as they say in the investment prospectuses, past performance is no guide to future performance. Social science isn't physics. Also there is an IF in there.
But what about the other states? NM, NV, FL, NC, NY, MN, etc..
That the 12:22am state and national exit polls each confirming a Kerry victory is not believable? Well, forty-two of 50 states deviated to Bush.That's 1 in 2 million odds.
Oh boy, round and round and round and round. Yes, the poll, or the count, or both, seem to have been systematically. It certainly wasn't chance.
That 50 state exit polls mirror 49 Ohio exit poll precincts? Just a coincidence?
It certainly wasn't chance. It certainly wasn't chance. It certainly wasn't chance. It certainly wazzzzzzzzzz.......
Move along. Nothing to see here.Take a trip to Brazile, Donna
.....zzzzzzzz....
That the 9% disparity between the voting shares of Florida optiscans and DREs is virtually impossible? Dem/Rep registrations were essentially equal in Optiscan and DRE counties.
Ah! Now I did some of this stuff.... What do you mean? Do you mean Hout's finding that DRE's had greater discrepancies? Or Liddle/Dopp's/Mitteldorf's finding that Optiscans did? Actually I think both analyses are invalid, as there were huge demographic confounds - No DREs were used in very rural counties, and no optiscans in the very large urban counties. Florida smells, to me, but the machine thing is completely inconclusive.
That sixteen of 50 states deviated beyond the MoE for Bush, none for Kerry?The probability of that is 1 in 19 trillion
.....round and round and round and round......
That ALL 22 Eastern Time Zone states deviated from Kerry to Bush? 1 in 4 million.
Hey! I thought it was the swing states? Oh no, it was the Republican precincts - Oh no, it was the Democratic states....What is the point here exactly?
That eighty-six of 88 documented touch screens switched Kerry votes to Bush?See the EIRS database. Do the math.1 in 79,000,000,000,000,000,000,000.
The math here is truly bosh. Do we know how many machines these 88 incidents represent? Do we know whether the EIRS number was equally well publicised in Dem as in Rep areas? If not, then we can't do that math. I'll give you, it does look as though some machines may have defaulted to Bush. It requires serious investigation (and abolition of those dratted machines. It does not require idiotic probability calculations.
THOSE ARE FACTS, NOT ASSUMPTIONS.
Ha!
The bits that aren't pure rhetoric (including purely rhetorical questions) are not facts at all! They are inferences based on assumptions - every single one. You can't DO inferential statistics without making assumptions, and the biggest one is RANDOM SAMPLING. In not a single case cited can we assume anything like random sampling. What the probabilities therefore tell us is that the "facts" inferred ARE NOT DUE TO CHANCE. They do not tell us whether the non-chance cause was fraud or not. For answers to that question we have to do further investigation. I've done quite a bit of that myself, in terms of math, and many others have done quite a bit of on the ground investigation. And having looked at the accumulating evidence, some things are supported, some look shaky (like the exit poll story, which I now believe does NOT support the case for large-scale fraud).But to assert that these are FACTS is quite misleading, and to imply that the astronomical (im)probability estimates are prima facie evidence of fraud is to seriously misunderstand the nature of inferential statistics. And, frankly, to mislead.
Take a look at the graph below.It shows a time line of pre-election and post election polls.I'll pass on this, as the graph doesn't seem to have channelled properly.Naysayers claim the "evidence" shows that Bush won the popular vote.I ask, what evidence?Well, I say that the only real evidence that Bush won the popular vote was the exit poll discrepancy (I certainly don't count the pre-election polls), and that if this was due to fraud, then fraud should have been correlated with "swing" (change in Bush's vote since 2000). It wasn't, and I think this makes the fraud explanation difficult to sustain.
I certainly do not claim that the "evidence" shows that Bush won the popular vote. I merely say that I do not believe that the exit poll provide any evidence at all that he did not. And absent that evidence, I think it is unlikely that Kerry did.
If Kerry won the popular vote, doesn't that mean the exit polls (state and national (12:22am) were therefore close to the truth? Except, that is, for the 1:25pm Final National Poll. We know this one is pure, unadulterated BS. Why? Look at the How Voted in 2000 demographic.Focus on the 43%/37% Bush/Gore weights.They are mathematically IMPOSSIBLE.
Here's the PROOF:Bush got 50.45 million votes in 2000.That's 41.25% of the 122.3mm who voted in 2004.But 3.5% of them died, according to annual U.S. mortality rates (0.87%).Therefore, AT MOST, 48.7mm of Bush 2000 voters came to the polls in 2004.That's 39.8% of the 122.3mm total
.....round and round and round and round....
THE BOTTOM LINE: Assuming REALISTIC, PLAUSIBLE, EQUAL weightings for Bush and Gore voters,KERRY WINS EASILY, EVEN IF FINAL EXIT POLL PERCENTAGES,WHICH WERE BIASED IN FAVOR OF BUSH, ARE USED.PERIOD.CASE CLOSED.FINITO.
THE SMOKING GUN.
QUERE MAS?
As I said, if it makes you happy.
Once again, I challenge the naysayers to a real-time debate using the Interactive Election model.Let's begin where the DU "Game" thread abruptly ended:
http://www.democraticunderground.com/discuss/duboard.ph ...
In the "Game", the naysayers claimed that Bush won 15% of Gore voters. And that Kerry won only 52% of those who did not vote in 2000.Can they ever come up with a plausible Bush win scenario? I doubt it.
.....random sampling random sampling random sampling....
Note to Land Shark: The TOTALITY of pre-election and exit poll data provide SOLID CIRCUMSTANTIAL EVIDENCE that the election was stolen.The NUMBERS have been silently screaming for a year.The fact that ONE YEAR LATER, there's strong DOUBT about who really won, should be sufficient to convince the public that something must be done ASAP to restore our democracy.
And again, I whole-heartedly agree! The fact that this conversation is even happening is evidence of the degree to which your democracy has been sabotaged!And of course, I am encouraged, from a purely mathematical point of view, that you accept that there is DOUBT about who won (although I would much prefer it if Kerry was in the White House right now). Yes, indeed, there is doubt.Beyond that, we will have to agree to differ.Lizzie
TIA RESPONDS WITH FACTS, PLAUSIBLE UNDECIDED VOTER ASSUMPTIONS AND CALCULATIONS. FEBBLE HAD NO IDEA THAT PROFESSOR WANG’S META-ANALYSIS AGREES WITH TIA’S 98% KERRY ELECTORAL VOTE WIN PROBABILITY. OR THAT TIA ALSO CALCULATED KERRY’S POPULAR VOTE WIN PROBABILTY. TIA REMINDS FEBBLE THAT EDISON-MITOFSKY CLAIMED A 1.0% MOE FOR 13047 RANDOMLY-SELECTED RESPONDENTS IN THE NATIONAL EXIT POLL.
autorank (1000+ posts) Sun Dec-04-05 03:01 AMResponse to Reply #75
78. Well, OK these are the Truth's words...cheers Is All :toast:
Channeling to avoid "the Chunnel":
Febble, here is my response to your latest post.Do you want to post it on DU? Or do you want it channelled? Either way, it will soon appear on DU. Look at this as a heads-up. Before I begin, I have three questions: Of the 50 state pre-election polls, and of the 18 national pre-election polls, and of the 11 Bush approval polls, and of the 50 state exit polls, and of the first 3 national exit poll timelines...
1) Which were not random sample polls?
2) Which were not done by well-known pollsters?
3) In your experience, has there ever been a poll that youbelieved?
Just asking.
FEBBLE:I confess I had not looked at this "final poll"before, but if "the math" gives a 99.9% (or even a99.8%) probability of a Kerry win, then something about the math is wrong. I was optimistically putting it the odds atabout 50:50 myself, and hoping that even if the popular vote was pretty evenly split, the EV would benefit Kerry. Which is sort of what happened (in the official vote anyway), in that a few thousand more (official) votes for Kerry in Ohio would have given him the presidency, but not the (official) popularvote.
TIA:I thought that you would have some familiarity with the logic behind my election model by this late date. It is quite disappointing to see you criticize the analysis without having done your homework. It's obvious that you are not too familiar with U.S. elections; you fail to take the undecided vote into account. The undecideds invariably break for thechallenger - especially when the incumbent has a 48.5% job approval. It just makes logical, intuitive sense when you think about it. An undecided voter must be unhappy with an incumbent,otherwise why would he/she consider voting for the challenger? Makes sense, yes?
You fail to consider this in your election analysis- and the omission makes anything you say highly suspect from the get-go. Without considering undecides, your analysis is faith-based and unrealistic.The fact is, Kerry won the late undecided vote, according to all the polls. And the undecided allocation was a key driver in my state and national projections. Yet you never even consider this, other than to vaguely refer to faulty model"assumptions". You are never specific about them. Did you mean undecided vote assumptions? Read on.
FEBBLE:(And more generally, when I say I agree with TIA's math, I certainly trust him to perform the computations he does correctly - the issue is usually whether we can legitimately infer from them what he infers from them. But his probability calcs are frequently invalidated by the assumptions thatunderlie them, and a probability calc based on erroneous assumptions is not a correct calculation).
TIA:What you call “inferences” are in fact the result of valid assumptions. You may disagree with the assumptions, but they are based on historical precedent. You fail to recognize or appreciate the significance of the undecided vote factor in U.S. elections. The base case model assumption was that Kerry would win 75% of the undecided. But this was just amost-likely base case estimate. That's why I ran the analysis for a range of five undecided vote shares from 60% to an admittedly high 87%. Kerry wins all the scenarios.
If Kerry won just 60% of the undecided vote (and the exit polls say he did better than that) then based on the Monte Carlo simulation, the probability was 98.02% that he would win the electoral vote with a median 322 EV. Furthermore, based on 18 national polls, the probability was 97.55% that he would win the popular vote with an expected 50.73% of the two-party vote, again assuming that he won 60% of the undecideds. You really ought to take a trip to Monte Carlo sometime.