TRUTHISALL: 1 in 76 BILLION Odds of 5.1% Discrepancy in Dem House Vote vs. Generic Poll

[mom cat]

The 116-poll House Generic Trend line projected a 56.4-41.6%
Democratic win, a 14.8% margin. The recorded vote was
51.3-46.4%, a 5.1% decline in Democratic vote share. The
Democratic margin declined from 10.6 million to 3.6 million
votes.

http://www.geocities.com/electionmodel/Election2006_16921_image001.png

116 Generic Poll Trendline		
Dem = 46.98+ .0419x
GOP = 38.06+ .0047x	

Substituting x=116 and allocating 60% (UVA) of the Undecided
vote to the Democrats: 
........Proj   UVA	Total
Dem	51.8	4.5	56.4
GOP	38.6	3.0	41.6

Assuming a 1.5% MoE, the probability that the Democratic vote
share would decline 5.1% from the Generic Poll (56.4%) to the
actual vote (51.3%) is 1 in 76 billion. The probability is
calculated using the Excel Normal Distribution function: 

Prob	= 1.310E-11	= NORMDIST(0.513,0.564,0.015/1.96,TRUE)
or 1 in 76,326,375,571

Calculate the odds for various MoE:
MoE.....Odds:1 in
1.00%	66,902,704,830,560,800,000
1.25%	1,501,199,875,790,170

1.50%	76,326,375,571

1.75%	181,561,494
2.00%	3,491,135
2.25%	227,190
2.50%	31,607
3.00%	2,333


U.S. House Vote
(in thousands)

	      Dem	Rep	Other	Dem%	Rep% Other% PMarg VMarg
Generic     41556	30946	1474	56.4%	42.0%	2.0%	14.4%	10610
Reported    37796	34194	1690	51.3%	46.4%	2.3%	4.9%	3602
Discrep    -3760	3248	216	-5.1%	4.4%	0.3%	-9.5%	-7008


Reported State Vote:
State Total Dem	Rep	Other	Dem%	Rep%	Other% PMarg VMarg

AL	579	224	352	3	38.7%	60.8%	0.5%	-22.1% -128
AK	202	81	115	6	40.1%	56.9%	3.0%	-16.8% -34
AZ	1127	479	576	72	42.5%	51.1%	6.4%	-8.6%	-97
AR	747	448	299	0	60.0%	40.0%	0.0%	19.9%	149
CA	6236	3549	2479	208	56.9%	39.8%	3.3%	17.2%	1070
.
CO	1371	728	572	71	53.1%	41.7%	5.2%	11.4%	156
CT	1079	652	421	6	60.4%	39.0%	0.6%	21.4%	231
DE	509	197	291	21	38.7%	57.2%	4.1%	-18.5% -94
FL	3727	1497	2162	68	40.2%	58.0%	1.8%	-17.8% -665
GA	1916	799	1117	0	41.7%	58.3%	0.0%	-16.6% -318
.
HI	338	220	118	0	65.1%	34.9%	0.0%	30.2%	102
ID	435	173	243	19	39.8%	55.9%	4.4%	-16.1% -70
IL	3127	1732	1381	14	55.4%	44.2%	0.4%	11.2%	351
IN	1646	803	821	22	48.8%	49.9%	1.3%	-1.1%	-18
IA	1028	490	520	18	47.7%	50.6%	1.8%	-2.9%	-30
.
KS	827	361	450	16	43.7%	54.4%	1.9%	-10.8% -89
KY	1244	596	609	39	47.9%	49.0%	3.1%	-1.0%	-13
LA	901	294	580	27	32.6%	64.4%	3.0%	-31.7% -286
ME	529	345	161	23	65.2%	30.4%	4.3%	34.8%	184
MD	1344	828	475	41	61.6%	35.3%	3.1%	26.3%	353
.
MA	1068	793	198	77	74.3%	18.5%	7.2%	55.7%	595
MI	3516	1793	1626	97	51.0%	46.2%	2.8%	4.7%	167
MN	2178	1154	925	99	53.0%	42.5%	4.5%	10.5%	229
MS	581	251	295	35	43.2%	50.8%	6.0%	-7.6%	-44
MO	2050	965	1031	54	47.1%	50.3%	2.6%	-3.2%	-66
.
MT	805	314	476	15	39.0%	59.1%	1.9%	-20.1% -162
NE	586	257	329	0	43.9%	56.1%	0.0%	-12.3% -72
NV	573	287	259	27	50.1%	45.2%	4.7%	4.9%	28
NH	402	209	189	4	52.0%	47.0%	1.0%	5.0%	20
NJ	1859	949	885	25	51.0%	47.6%	1.3%	3.4%	64
.
NM	545	304	241	0	55.8%	44.2%	0.0%	11.6%	63
NY	3561	2285	1268	8	64.2%	35.6%	0.2%	28.6%	1017
NC	1842	935	907	0	50.8%	49.2%	0.0%	1.5%	28
ND	433	284	149	0	65.6%	34.4%	0.0%	31.2%	135
OH	3763	1970	1785	8	52.4%	47.4%	0.2%	4.9%	185
.
OK	905	373	518	14	41.2%	57.2%	1.5%	-16.0% -145
OR	1264	713	523	28	56.4%	41.4%	2.2%	15.0%	190
PA	3815	2061	1705	49	54.0%	44.7%	1.3%	9.3%	356
RI	372	264	42	66	71.0%	11.3%	17.7%	59.7%	222
SC	1072	466	593	13	43.5%	55.3%	1.2%	-11.8% -127
.
SD	667	461	196	10	69.1%	29.4%	1.5%	39.7%	265
TN	1712	860	797	55	50.2%	46.6%	3.2%	3.7%	63
TX	3994	1783	2069	142	44.6%	51.8%	3.6%	-7.2%	-286
UT	549	234	283	32	42.6%	51.5%	5.8%	-8.9%	-49
VT	524	279	234	11	53.2%	44.7%	2.1%	8.6%	45
.
VA	2148	810	1220	118	37.7%	56.8%	5.5%	-19.1% -410
WA	1309	803	499	7	61.3%	38.1%	0.5%	23.2%	304
WV	446	258	188	0	57.8%	42.2%	0.0%	15.7%	70
WI	1852	1001	836	15	54.0%	45.1%	0.8%	8.9%	165
WY	377	184	186	7	48.8%	49.3%	1.9%	-0.5%	-2

.
.
.
.
.
.


HERE'S A COMPREHENSIVE ELECTION 2004 SITE: POLLING DATA,
ANALYSIS, DISCUSSION and...
THE EXCEL INTERACTIVE ELECTION MODEL
http://www.truthisall.net/

Downloads in a minute (4mb)
Easy to use (3 inputs)
Press F9 to run 200 simulations
Pre-election/exit polls
(51 State & 18 National)

A challenge to all those who still believe Bush won: Use the
National Exit Poll "How Voted in 2000" demographic
("NatExit" sheet) to come up with just ONE plausible
Bush win scenario. 

Note the feasibility constraint: The maximum ratio of Bush
2000 voters to the total 2004 vote is 39.8% (48.7mm/122.3mm)

Post the scenario on the Election Forum at
ProgressiveIndependent.com and/or Democratic Undergound.com 

http://www.geocities.com/electionmodel/InteractiveElectionSimulation_12255_image001.png

View the original 11/1/04 election model forecast of Kerry
winning 51.63-51.80% of the 2-party vote:
http://www.geocities.com/electionmodel/

[Cleita]

1. Sorry, I got lost in the figures.

Does this mean the vote was rigged in favor of Republicans again?

Anyway, thanks for posting.

[mom cat]

8. There is one chance in 76 billion that it was not rigged.

[Cleita]

18. Thanks mom cat.

That's what I wanted to know. So it appears our wins would have been much larger in a fair election.

[mom cat]

28. DATA TYPOS FIXED HERE ... (from TIA)

These typos were fixed:

1) The Democratic margin declined from 10.9 to 3.6 mm votes, a
7.3mm loss.

2) The Generic Poll projected Republican vote share was 41.6%,
not 42%. Therefore the Republican vote discrepancy was 4.6%,
not 4.4%.

         U.S. House Vote
         (in thousands)
         Dem    Rep  Other   Dem%  Rep% Other%  PMarg  VMarg
Generic  41556 30651 1474   56.4% 41.6%   2.0%  14.8%  10905
Reported 37796 34194 1690   51.3% 46.4%   2.3%   4.9%   3602
Discrep  -3760  3543  216   -5.1%  4.8%   0.3%  -9.9%  -7303

[mom cat]

45. Please note another eror in the text:

2) The Generic Poll projected Republican vote share was 41.6%, not 42%. Therefore the Republican vote discrepancy was 4.6%, not 4.4%.

This should read:
2) The Generic Poll projected Republican vote share was 41.6%, not 42%. Therefore the Republican vote discrepancy was 4.8%, not 4.4%.

[Donald Ian Rankin]

184. Be warned: she's very wrong indeed.

I've explained the maths in post 183; she's making a very fundamental mathematical error.

[Roland99]

56. Yikes!!!

[Donald Ian Rankin]

183. NO! NO! No, that's completely and utterly wrong, twice over!

Let U be the event "the election was unrigged"
X be the event "the polls gave the result they did"
R be the event "the results of the election were what they were".

The claim in the OP is "P( X | U & R) = 1/76,000,000,000

That's a *completely* different claim to P( U | X & R) = 1/76,000,000,000, which is what you're interpreting it as.

More specifically,

P(U | X&R) = P(U&X&R)/P(X&R)
= P(U&X&R)/P(U&R) x ( P(U&R) / P(X&R) )
= P(X | U&R) x P(U&R) / P(X&R)

So you're out by a factor of P(U&R) / P(X&R), which may well be *vast*.



Secondly, the correct interpretation of "the odds of there being this much difference between the polled result and the actual result as a result of random error", which is what the OP is claiming, is not "the elections were rigged", but "there was systematic bias between the result of the polls and the result of the election". There are lots of possible explanations for this - e.g. Republican voters were less likely to admit to being so than Democrats.

[bananarepublican]

220. Bush obviously prayed for a miracle! n/t

[iconoclastic cat]

2. Hey, how is TIA doing?

When will he be back up to posting speed?

[On the Road]

3. Oh, Dear

Still pushing that 76 Billion to 1 hogwash. It does incacluable harm to efforts to get clean elections.

What the 76000000000-to-1 odds would be correct only if the exit polls could be verified as a random selection of the ballots actually cast and counted. It bears no relation to a divergence in the exit polls.

[Name removed]

30. Deleted message

Message removed by moderator. Click here to review the message board rules.

[creeksneakers2]

77. T he generic poll is assumed to be wrong

because it doesn't ask which choice the voter will make in his own congressional race. It just asks which party the voter prefers nationally. Many people would want the Democrats to win nationally but vote for their own GOP Congressman because they think he's doing a good job. (and vise versa)

[mom cat]

127. TIA: Generic polls ask "would you vote for the Dem or Rep in your district"?

View the Generic Polls here:
http://www.pollingreport.com/2006a.htm

An Example:
FOX News/Opinion Dynamics Poll. Nov. 4-5, 2006.
N=900 likely voters nationwide. MoE ± 3.

"Thinking ahead to this November's elections, if the congressional election were held today, would you vote for the Democratic candidate in your district or the Republican candidate in your district?" If unsure: "Well, if you had to vote, which way would you lean?"

[smalll]

82. Ha! Ha, I say!

Oh, suddenly we care about PRE-ELECTION POLLS. What about all the PRE-ELECTION POLLS that, averaged out, told us BEFORE Election Day, 2004, that Bush would win by about 3%??? By which percentage Bush allegedly won? (Allegedly, according to the New York Times, CNN, the Secretaries of State of the 50 states, the BBC, C-SPAN, China Daily, etc. etc. - you know, the vast conspiracy.)

In November 2004, the Fraudsters would have nothing to do with pre-election polls! It is to laugh! :rofl: Laugh I say!

[snot]

84. Could you pls provide a citation or reference for your statement that

"all the PRE-ELECTION POLLS that, averaged out, told us BEFORE Election Day, 2004, that Bush would win by about 3%".

I ask because the polls I saw immediately before the election were neck-and-neck.

[smalll]

154. Take a look at the following link -

http://www.realclearpolitics.com/Presidential_04/chart3way.html

This is an average of major Presidential pre-election polls for 2004. As you can see, up through August, the advantage swung back and forth between Bush and Kerry. Bush then shot ahead at the beginning of September during the Republican National Convention, Kerry closed the gap somewhat at the beginning of October (after one of those debates), but by election day, Bush continued to hold a lead - NOT of 3% as I originally remembered, it's true, but a lead of 1.5% (49.0% over 47.5% for Kerry) which was pretty close to the 2.5% lead that Bush ended up with in the actual election.

I am a new member, but have been lurking for quite a while - since the 2004 election. In the Bush v. Kerry race aftermath, the fraudsters insisted that pre-election polls were worthless, and staked all their claims on the exit polls. Now suddenly, they want to rely on pre-election polls for 2006!

[mom cat]

129. TIA HAS THE LAST LAUGH

18 Final Pre-election Poll Summary: 
Kerry won 11, Bush 6, 1 tie 
Kerry won 5 of 9 Registered Voter (RV) Polls 
and 6 of 9 Likely Voter (LV) Polls 

Date   Poll   Sample Type   MoE  Kerry Bush
02-Nov Harris 1509    LV   2.52%   50   47
02-Nov Zogby  1200    LV   2.83%   50   47
01-Nov Marist 1166    LV   2.87%   50   49
01-Nov Econ   2903    RV   1.82%   50   47
01-Nov TIPP   1284    LV   2.73%   44   47

01-Nov CBS    1125    RV   2.92%   47   48
31-Oct FOX    1400    RV   2.62%   48   45
31-Oct DemC   1018    LV   3.07%   48   47
31-Oct Gallup 1866    RV   2.27%   48   46
31-Oct NBC    1014    LV   3.08%   47   48

31-Oct ABC    3511    RV   1.65%   47   48
30-Oct ARG    1258    LV   2.76%   49   48
30-Oct Pew    2408    RV   2.00%   46   45
29-Oct Nwk    1005    RV   3.09%   44   48
26-Oct ICR     817    RV   3.43%   48   48

24-Oct LAT    1698    RV   2.38%   48   47
21-Oct Time    803    LV   3.46%   46   51
20-Oct AP      976    LV   3.14%   49   46

[foo_bar]

143. that's called "cherry-picking"

i.e., omitting data that doesn't support the conclusion you desire.

President George W. Bush holds a slight edge over Senator John Kerry in the final days of Campaign 2004. The Pew Research Center's final pre-election poll of 1,925 likely voters, conducted Oct. 27-30, finds Bush with a three-point edge (48% to 45% for Kerry); Ralph Nader draws 1%, and 6% are undecided.

http://people-press.org/reports/display.php3?ReportID=232

Of 10 major polls of "likely voters" taken in the week before the election, seven showed Bush leading:

Poll Bush Kerry
Marist 49% 50% (K+1)
Harris 49% 48% (B+1)
FOX News 46% 48% (K+2)
Reuters/Zogby 48% 47% (B+1)
USA TODAY/CNN/Gallup 49% 49% (K=B)
NBC/Wall Street Journal 48% 47% (B+1)
ABC/Washington Post 49% 48% (B+1)
CBS/New York Times 49% 46% (B+3)
Pew Research 51% 48% (B+3)
Newsweek 50% 44% (B+6)

http://www.usatoday.com/news/politicselections/vote2004/2004-11-03-polls-burn-pundits_x.htm

TIA lists the 11/02/04 Harris poll as Kerry 50/Bush 47, but Harris had it at 49B-48K-1Nader (right here) on the same day on the same sample, so it appears that TIA basically chose the polls that supported his thesis, and pretended the rest didn't exist. This is a recurring theme:

No, I leave them out because they are BIASED for Bush.

Why include them if they skew the averages against Kerry?

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=1998204#1998844

[mom cat]

161. TIA: Reply to foo-bar

foo, I see you're referring to RCP polls.
Are there any Registered Voter (RV) polls in that list?
All I see are Likely Voter (LV) polls.
Don't tell me your cherry-picking?
Actually you're not cherry-picking the polls, RCP dids by
providing an all-LV list.

But you are cherry-picking the RCP list of cherry-picked LV
polls.
Cherry-pick squared.

On the Other Hand, my list was taken straight from
http://pollingreport.com/

In 2004, RV polls made more sense, since newly registered
voters were mostly female, democratic and cell-phone users.
Not to easy to get a hold of those folks on a landline. In any
case my numbers include a mix of RV's and LV's.
The final Election Day Harris and Zogby polls both had Kerry a
50-47 winner. And you know what? They were LV polls.

The Final Pew RV poll had Kerry ahead by a point. 
And he came from 7 points behind the month before.
http://people-press.org/reports/images/232-1.gif
http://people-press.org/reports/display.php3?ReportID=232

Slight Bush Margin in Final Days of Campaign 
No Impact Detected from bin Laden Tape

Released: October 31, 2004
Summary of Findings

President George W. Bush holds a slight edge over Senator John
Kerry in the
final days of Campaign 2004. The Pew Research Center's final
pre-election
poll of 1,925 likely voters, conducted Oct. 27-30, finds Bush
with a
three-point edge (48% to 45% for Kerry); Ralph Nader draws 1%,
and 6% are
undecided. 

The poll finds indications that turnout will be significantly
higher than
in the two previous presidential elections, especially among
younger
people. Yet Bush gets the boost Republican candidates
typically receive
when the sample is narrowed from the base of 2,408 registered
voters to
those most likely to vote. (Among all registered voters, Kerry
and Bush are
in a virtual tie: 46% Kerry, 45% Bush). 

HERE'S THE TREND in the RV poll:
Kerry came from 7 points down to lead by a point.

Oct    3 19 30
Bush  48 45 45
Kerry 41 45 46

foo,foo, boo, boo

[foo_bar]

175. so he admits he's cherry-picking

But thinks I'm doing likewise by pointing it out. Almost a Rumsfeldian explanation for why he dropped data points that didn't fit the conclusion he wanted:

That's why I stick with Zogby and ARG, etc. I've stopped blindly incoporating the latest poll that Votemaster puts up - unless t makes sense. Call me a cherry-picker. I throw away the bad ones. That's why my numbers seem so far out there for Kerry.

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=104x4057474

Well, you said it.

[OnTheOtherHand]

177. thanks for the shout-out, guy

In 2004, RV polls made more sense, since newly registered voters were mostly female, democratic and cell-phone users. Not to easy to get a hold of those folks on a landline.

This has been explained to TIA so many times that I don't know what to say. A few points:

(1) The pollsters themselves provide likely voter estimates because they regard them as likely to be better, not worse -- so I'm not sure where TIA gets off overruling them. Did he read that in the Excel manual somewhere?

(2) Newly registered voters can show up as likely voters, so they don't offer a reason to prefer RV polls to LV polls.

(3) Being unable to reach people on a landline also wouldn't be a reason to prefer RV polls to LV polls (doh).

(4) An analysis by Scott Keeter at Pew showed that upweighting younger respondents largely compensated for landline bias in 2004. (Here is the abstract.) TIA has seen that link more than once, and he hasn't criticized the finding, so... well, cherrypicking doesn't begin to describe it.

[mom cat]

130. TIA: 2004 pre-election poll detail

2004 pre-election polls (RV and LV)
http://www.progressiveindependent.com/dc/dcboard.php?az=show_mesg&forum=120&topic_id=3192&mesg_id=3347

Polling Data Source:
http://www.pollingreport.com/wh04gen.htm
http://www.economist.com/media/pdf/YouGovS.pdf

[Cessna Invesco Palin]

112. I see his surrogates...

...have adopted his attitude. Lovely.

[OnTheOtherHand]

119. I assume that was just a direct quotation

I've inquired several times whether anyone is willing to defend TIA's arguments himself or herself. No takers so far.

[LeftCoast]

4. It's obvious some people have never taken Statistics

These posts are just embarrassing.

[Cleita]

6. No many of us haven't.

In my day math wasn't considered important for girls beyond arithmetic and some basic Algebra so you would be literate enough for an office job. So a little help for the proletariat would be appreciated.

[bleever]

7. TIA has three graduate degrees in mathematics. nt

[Zynx]

13. So he claims. This is the Internet.

[LeftCoast]

14. I keep hearing that

:eyes:

[survivor999]

31. And he uses Excel for everything...

Oh well...

[uppityperson]

95. oh dear. Next time will you be 1000?

[foo_bar]

47. "I'm not a statistician" -TIA

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x79760#86295

And it shows. Someone made up the "three graduate degrees" urban legend just now, as TIA himself claimed one of his mysterious degrees was a B.A. (see: http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x237348#23833 ). Regardless, none of the hypothetical degrees are in a statistical field, and TIA can't name a single college he attended, so it's "academic" as it were. Not that this is germane to his fantastical argument du jour, which is that random samples are a conspiracy theory.

[papau]

70. TIA's resume as to studies is decent - and I believe him to be an honest guy.

TIA says he studied:
Linear Programming.
Differential Calculus.
Integral Calculus.
Numerical Methods.
Mathematical Statistics.
Probability Theory.
Complex Variables.
Queuing Theory.
Partial Differential equations.
Ordinary Differential Equations.
Linear Algebra.
Matrix Analysis.
Non-linear Programming
Quadratic Programming

Degrees
BA Mathematics
MS Applied Mathematics
MS Operations Research

Work history is as an engineer - not a stat or math person - but the course list is standard for a BA in Math, and non-linear and quadratic programing sounds like senior year or first year grad school. But Matrix Analysis in addition to linear algebra is a bit curious - perhaps implying the linear algebra was very deep theory only, or the matrix analysis was a first year filler. The operations research semester yielded a fellow that can handle Excel - and stat based QC - and has more energy than I have.

As he says - he is not deep into stat - or for that matter polling and its social science/behavioral science angle.

But my contact with him indicates an honest fellow with a decent math background.

I heard he was ill.

Does anyone know how he is doing?

[nick303]

74. grepping around with his "areas of expertise" brings up the course list for Clemson?

seems kind of bizarre, and honestly, not that creative a lie.

[papau]

83. It's a standard course list for a math major - no big deal IMO n/t

n/t

[DemocratSinceBirth]

106. I'm A Billionaire

:-)

[goclark]

147. TIA is brilliant and I would trust his work over the NYT or

Gallup or any other PAID FOR BY BUSH mouth pieces.

TIA

:yourock:

[Joe Chi Minh]

153. They're losers, and they know it's all going to come out in the wash.

Sneer and scoff while you can, because you won't dare show your face on here when it does. They've probably already got secondary usernames up and running.

[foo_bar]

193. yeah, all the skeptics will be sorry when the TIA revolution comes!

It's funny that you mentioned secondary usernames because that wouldn't have occurred to me. It's almost as if you can inhabit the mind of a miscreant. Anyway, I hope this random stream of invective is going well for you.

[Joe Chi Minh]

197. Well, GV, I made two, what might be called, primary posts,

neither purporting to be the opinion of an expert on psephology, statistics or any othe field.

Unfortunately, it is not at all uncommon for people reputed to be experts, particularly in the scientific field, bearing sundry, often distinguished academic accreditations, to be intellectually dishonest. The result of this, of course, is that they involve themselves in arguments to their great discredit. Rightly or wrongly, that was the impression I gained of you and my fellow Edinburgher, whatsername, Elizabeth.

Even your username, OntheOtherHand, makes me suspicious, since it often happens that while often there is only truth and falsehood, Republican trolls often resort to affectations of a judiciously professional open-mindedness.

In the first of these posts, I said that the massive fraud would be exposed, (in this context evidently implicitly by statistics, as well as by all the machine malfunctions and voter suppression), and its deniers, exposed. I still believe this to be the case. How could anyone with an IQ above 40, not do so. The evidence is so overwhelming.

Having read Skinner's post, however, it occurs to me that TIA may indeed have based his case here on a false assumption. The fact is that I'm strongly inclined to believe that Skinner doesn't have a Republican axe to grind, and his explanation makes sense to a layman.

The second "primary" post was to take a secular fundamentalist to task for implicitly sneering at those who believe that everything that exists, bar God, himself, was created (by God).

You are a technical expert, if possessing a wayward heart, theoretically capable of positing rubbish at least as grotesque as any layman could. I am that layman, though positing rubbish is not a habit of mine. Happy now?

[OnTheOtherHand]

199. nice flame bait

OK, well, it's nice that you offer some semblance of thought, even if it took Skinner to evoke it.

[Joe Chi Minh]

200. You may be able to learn from that. "Vulgarisation", as the French put it,

has its merits. At least we, the non-technical people, can then learn.

[OnTheOtherHand]

201. shall we be candid?

If you actually wanted to learn about this topic, you readily could. As far as I can tell, you post on exit poll threads not to learn, not to engage the content, but to impugn the integrity of others, as in your penultimate post.

That is vulgar, and I learn from it. But I don't suppose that was your point.

[Joe Chi Minh]

204. Well, I certainly wouldn't expect to learn anything from you, GV.

Wasn't it you and Elizabeth who argued that the exit polls were useless as accurate indicators of the actual vote? So, the logical thing to do was to make them match the actual vote?

Even though Mitofski's exit polls had been used for precisely that purpose and with total success all over the world. I'm sure you will try to blind me with putative science, but you are evidently unable to explain what Skinner explained, and I suspect it's because your technical screeds claimed something quite different. Otherwise why would you be so dense as to disdain expaining it in the same simple terms that Skinner did? You can see, can't you, why you don't come across as being quite in full possession of the full deck.

[OnTheOtherHand]

205. do you actually read our posts?

Wasn't it you and Elizabeth who argued that the exit polls were useless as accurate indicators of the actual vote?

No, that isn't our position.

Honestly, I don't think you are inattentive enough actually to misunderstand our position so thoroughly, but I don't know.

[stepnw1f]

226. Absolutely They Won't

along with their own lack of credibility.

[RoyGBiv]

216. That's irrelevant ...

His being a mathematician may (or may not) mean the math has a higher possibility of being correct, but it says little to nothing of the statistical validity of the conclusion being offered. A+B may equal C, but if the actual equation should have been D+E=X, then the conclusion offered is not valid. Or to put it as Skinner put it, if the assumptions aren't valid, then neither are the conclusions.

I'm a little late to this party, but I am quite glad to have found that it took place. I have no personal issue with TIA or those who followed and have followed his work. I will offer, however, that when it was being presented in the month's leading up to the 2004 election, I was troubled by the methodology being used. I was further troubled by the fact that the methodology occasionally changed in subtle ways. That did not suggest to me dishonesty, per se, rather the possibility that TIA had become so captivated by the numbers themselves that he had not thought to ask himself whether the assumptions on which they were based meant anything.

Not being a statistician myself, but having done many studies using statistical breakdowns of voter behavior for historical research, something just didn't strike me as quite right about it all, so I forwarded several of the presentations to a friend who himself has a PhD in mathematics, is an occasional historian, taught for years, has written advanced mathematics textbooks, is very liberal in his politics, and currently works as an editor of a mathematics journal. I thought he might be able to explain what I was seeing in such a way so as either to remove my doubts or confirm them, but the answer he gave me was not at all what I expected.

I still have the e-mail, which was somewhat lengthy, but ended with this final conclusion.

"I am not qualified to judge."

"Why?" I asked him. He then explained to me the difference between doing math and doing statistics, using terms that meant more to PhDs than they do to me, but good enough terms I gathered the gist. The math in statistics is relatively easy to compute. The equations themselves are fairly well static and vetted. The numbers used in those equations, however, are the key to the validity of the conclusion, and the job of a statistician is in finding the right numbers to use. He did opine, tentatively, that a severe problem existed in TIA's method of combining and averaging and then comparing polls, which was a part of the methodology that initially had caused me pause as well. In the case under discussion here, the problem is glaring.

My friend then forwarded the presentation to a colleague whose life is statistical analysis. His conclusion? I'm paraphrasing here, but it boiled down to something similar to what Skinner said regarding this presentation, "The underlying assumptions are false. This is a case of a conclusion being sought and the methods and assumptions fixed to surround what is sought. This is the kind of thing that makes people distrust statistical analysis." And all this took place well before the election.

Since then I have paid little attention to any of this, except to offer my own little bit here, which I'm sure will draw the ire of many. The point for me is simply this. The worthy goal of exposing election fraud is not aided by bad statistical analysis and in fact works exactly in the opposite direction.

[ramblin_dave]

11. What about this analysis by the Election Defense Alliance?

It shows a significant deviation of election results versus exit poll data in favor of the Republicans...

Landslide Denied: Exit Polls vs. Vote Count 2006

There was an unprecedented level of concern approaching the 2006 Election (“E2006”) about the vulnerability of the vote counting process to manipulation. With e-voting having proliferated nationwide, and with incidents occurring with regularity through 2005 and 2006, the alarm spread from computer experts to the media and the public at large. It would be fair to say that America approached E2006 with held breath.

For many observers, the results on Election Day permitted a great sigh of relief—not because control of Congress shifted from Republicans to Democrats, but because it appeared that the public will had been translated more or less accurately into electoral results, not thwarted as some had feared. There was a relieved rush to conclude that the vote counting process had been fair and that the concerns of election integrity proponents had been overblown.

Unfortunately the evidence forces us to a very different and disturbing conclusion: there was gross vote count manipulation and it had a great impact on the results of E2006, significantly decreasing the magnitude of what would have been, accurately tabulated, a landslide of epic proportions. Because virtually all of this manipulation appears to have been computer-based, and therefore invisible to the legions of at-the-poll observers, the public was informed of “isolated incidents and glitches” but remains unaware of the far greater story: The electoral machinery and vote counting systems of the United States did not honestly and accurately translate the public will and certainly can not be counted on to do so in the future.

[mom cat]

20. Exactly. The analysis of fraud is coming in from several reliable sources.

[malaise]

69. I think we all expected fraud

but many of us also knew that the tsunami was too big for them to win. All election fraud from 2000 must now be made public. These criminals must be jailed.

[foo_bar]

48. this was debunked on the ER forum, but the dream lives on

This paper isn't peer reviewed, because it would disappear in a millisecond. Like Creationism, only people outside of their field can even pretend to make the argument:

Yet for the same question—“For whom did you vote in the 2004 presidential election?"—the final, adjusted exit poll showed a margin of 43% Kerry to 49% Bush. This 6% margin in favor of Bush was a dramatic distortion of the 2.8% margin actually recorded in E2004.
(...)In order to match the results of the official tally, the 2006 exit poll adjustment was so extensive that it finally depicted an electorate that voted for Bush over Kerry by a 6% margin in 2004: very clearly an undersampling of Democrats and an oversampling of Republicans.

http://electiondefensealliance.org/landslide_denied_exit_polls_vs_vote_count_2006

"Very clearly"? In 1964, exit poll respondents claim they elected JFK by a 27.5% margin (despite the 1960 election being decided by a fraction of a percent); the exit polls in 1972 had Nixon winning '68 by 18.9% not 0.7%, according to people's "recollection" of their prior vote (source). That this has happened every presidential election year since 1948 (with the smallest gain in retrospective approval being +0.8% for Carter in '80) throws a monkey wrench into the subsequent conclusion:

Conclusion

While the reported results of the 2006 election were certainly well-received by the Democratic party and were ballpark-consistent with public expectations, the unadjusted 2006 exit poll data indicates that what has been cast as a typical midterm setback for a president in his second term was something rather more remarkable ­ a landslide repudiation of historic proportions.

http://electiondefensealliance.org/landslide_denied_exit_polls_vs_vote_count_2006

I'm sure the effects of early voting and absentees will also be explored in the retraction to the retraction, beyond the little technicality about people misremembering voting for the victor of the previous election in every presidential-year exit poll from 1948-2000. I don't blame Simon for using the "ignorance is bliss" angle, after the success of Bev Harris et al. Which brings us back to:

"Experts" are commonly quoted. It is possible to find someone with scientific credentials who is against just about anything. Most "experts" who speak out against fluoridation, however, are not experts on the subject.

http://www.quackwatch.org/03HealthPromotion/fluoride.html

More to the point:

Apophenia is the experience of seeing patterns or connections in random or meaningless data. The term was coined in 1958 by Klaus Conrad, who defined it as the "unmotivated seeing of connections" accompanied by a "specific experience of an abnormal meaningfulness".

http://en.wikipedia.org/wiki/Apophenia

[truedelphi]

88. Two things for which I'll loan you enough rope to hang

Yourself -
>>the exit polls in 1972 had Nixon winning '68 by 18.9% not 0.7%,

Nixon won by a whole lot more than 0.7% - and he carried what, all but two states?

next your statement that: Most of the "experts" on fluoridation are not experts.

Study industry's control over every friggin' laboratory in this country - ya know what! The "independent " researchers are a dying breed. "Corporate" influenced science tells us that fluoride is safe - at least it tells us that here in the USA.

i lived for four months in Scandanavia - where flouride is seen to be for the pesky contaminant that it is. People are biking around in their eighties and nineties. The kids all had rosy cheeks. The experts had protected the population - because in so many places in Europe, the experts are not bought out.

Came back from Scandanavia with renewed energy. People would stop me on the street to ask me about my "glow" or what make-up I used. Constant compliments on my skin.

Within three months of drinking flouridated -chlorinated water, (as approved by US "experts) compliments were again a rare thing.

[foo_bar]

94. it's "fluoride" (and "fluoridation", "fluoridated")

It's a fairly elemental (forgive the pun) error for someone who claims to speak on the phenomenon.

i lived for four months in Scandanavia - where flouride is seen to be for the pesky contaminant that it is. People are biking around in their eighties and nineties. The kids all had rosy cheeks. The experts had protected the population - because in so many places in Europe, the experts are not bought out.

I thought that was Lake Wobegon, where the men are robust and the children are pink-cheeked. But seriously, personal anecdotes about your "glow" might be a little short of ironclad proof of TIA's competence or the correlation between fluoride and lifespan.

Nixon won by a whole lot more than 0.7% - and he carried what, all but two states?

[truedelphi]

152. Fluoride or flouride - It was also 11:15Pm when I posted

After a very long day

Prevention Magazine had an excellent article this summer on the risks of fluoride, the lack of real, independent research confirming its need as a water supplement and should be read by anyone seeking to de-bunk fluoride's debunkers. It probably does offer some benefit with cavity prevention - but if you brush three times a day with fluoride enhanced toothpaste-you'll get that benefit.

[Joe Chi Minh]

155. "Like Creationism, only people outside of their field can even

pretend to make the argument:"

An advanced education was never an obstacle to imbecility. Far from it. But I still find it difficult to believe anyone could be so dumb as you people.

Of course, creationism doesn't preclude the evolution that you nutjobs people are so fond of implying. In terms of empirical deduction, the origins of the universe are as much a closed book to you presumptuous nerds as to the rest of us; you're just too wilfully dumb to accept it.

The day you can get your mind around any of the paradoxes which ceaselessly proliferate at the outermost bounds of cosmology, let us all know, will you?

"Apophenia is the experience of seeing patterns or connections in random or meaningless data. The term was coined in 1958 by Klaus Conrad, who defined it as the "unmotivated seeing of connections" accompanied by a "specific experience of an abnormal meaningfulness".

Do you really expect people to be as gullible as you are - taken in as you clearly are by a hilariously imaginative polysyllabic Greek neologism?

During the slavery era in the US, the medical profession came up with a most impressive sounding latin neologism for a particular disease affecting only slaves... wait for it... a craving to escape from slavery. The sorry pathology afflicting runaway slaves.

"Scientific thinkers I have known." You'll just have to do better than that.

[foo_bar]

174. "you people"? minus Skinner I presume?

Do you really expect people to be as gullible as you are - taken in as you clearly are by a hilariously imaginative polysyllabic Greek neologism?

In statistics, apophenia is called a Type I error, seeing patterns where none, in fact, exist. It is highly probable that the apparent significance of many unusual experiences and phenomena are due to apophenia, e.g., ghosts and hauntings, EVP, numerology, the Bible code, anomalous cognition, ganzfeld "hits", most forms of divination, the prophecies of Nostradamus, remote viewing, and a host of other paranormal and supernatural experiences and phenomena.

http://skepdic.com/apophenia.html

"Scientific thinkers I have known." You'll just have to do better than that.

Did I write that? Did anyone?

In terms of empirical deduction, the origins of the universe are as much a closed book to you presumptuous nerds as to the rest of us; you're just too wilfully dumb to accept it.

Check out those DU rules. :thumbsup:

[Joe Chi Minh]

190. Cry baby. You can dish it out to believers, but you can't take it, when

they shoot you down in flames.

"Like Creationism, only people outside of their field can even pretend to make the argument".

But, without a rational argument to back it up, that snipe at Creationism is so fatuous as to beggar belief. If you secular fundamentalists could prove such a mysterious contention, that would be genuinely interesting. Truly, as GK Chesterton observed, people who don't believe in God, don't believe in nothing, they believe in everything.

In fact, since religious believers are quite well represented here on DU - not to speak of in the US and elsewhere in the world - it was insulting because the intention was clearly to offend. And the casual way in which it was expressed reflected a particularly tendentious quality of arrogance, by no means uncommon on these threads. But, heck, I'm not complaining. Just exposing the vacuousness of your mindset.

[foo_bar]

192. comparing bad science to bad science is offensive?

If you secular fundamentalists could prove such a mysterious contention, that would be genuinely interesting.

That's the first time I've read that phrase in an ostensibly left-wing context. I'm open to an argument for Creationism or the Infallibility of TIA, but blaming secularism for his shortcomings in the reality department has a certain historical resonance.

But, heck, I'm not complaining. Just exposing the vacuousness of your mindset.

So you mentioned, without any supporting evidence. I suppose you'll tell me one geneticist believes genes were created in Genesis, and therefore TIA doesn't have to take any of those stats textbooks literally. You're entitled to your beliefs, but your feelings about the Loch Ness monster are about as relevant to the scientific debate you can but allude to.

[Joe Chi Minh]

195. Funny. You didn't sound at all open to any argument for Creationism.

In fact, it was because you went so foolishly out of your way to disparage it, that I responded to your post.

But there is no empirical argument for Creationism (other than common-sense, which is not part of the reductionist paradigm essential, to some extent, to the pursuit of empirical knowledge), any more than there is one against it. That's where you seem to lacking in perception. I mean it's well known.

[foo_bar]

208. so the comparison is insulting to Creationism?

But there is no empirical argument for Creationism (other than common-sense...), any more than there is one against it.

You should try out for the debate team.

(...which is not part of the reductionist paradigm essential, to some extent, to the pursuit of empirical knowledge)

"But like most terms of abuse, 'reductionism' has no fixed meaning."
- Daniel Dennett

In related GD threads:
"Prepare To Believe!"-Creation Museum To Open

That's where you seem to lacking in perception.

That's where you seem to lacking an infinitive.

[Joe Chi Minh]

210. ""But like most terms of abuse, 'reductionism' has no fixed meaning."

There you go again. You're so gullible. Someone's half-baked opinion, and you swallow it whole! Like the mutt who coined that neologism you were so impressed by that you quoted it. Then wailed, it wasn't me who said it. Or did I scan your post too cursorily?


"That's where you seem to lacking in perception.

That's where you seem to lacking an infinitive."


Wow! What a debating point! A typographical error! And if you must be pedantic, don't you mean half an infinitive...?

[foo_bar]

211. "That's where you seem to lacking in perception."

There you go again.

JIM LEHRER: Was "There you go again" a line that just came to you spontaneously, or was it something that you had worked on?

PRESIDENT RONALD REAGAN: No, it just seemed to be the thing to say in what he was saying up there, because it was to me it felt kind of repetitious, something we had heard before.

http://www.pbs.org/newshour/debatingourdestiny/interviews/reagan.html (tip of the hat to anax)

Someone's half-baked opinion, and you swallow it whole!

In the social sciences this might be known as 'projection'.

[Joe Chi Minh]

213. More quotes? Tee, hee. Jim Lehrer's you latest guru, is he? I must say

though, I liked his jibe about Reagan working on it!

In your cloistered world, it is evidently news to you, but people were saying, "There you go again", long before Reagan. It's kind of normal conversational English usage, you know?

[foo_bar]

235. do secular fundamentalists have gurus?

The blue words with underlines are external references, not mantras per se. In faith-based statistics, this would constitute a trend:
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=380907&mesg_id=381274

In your cloistered world, it is evidently news to you, but people were saying, "There you go again", long before Reagan.

It's only the second time this cliché popularized by Reagan was rallied in TIA's defense, so I can't make too much of it. But there might be a shady portion of the Venn diagram when you begin blaming secularism for the OP's analytic demise:

For a generation, the notion of the secular; secularism, secular humanism, the secular left, and most recently (and oxymoronically) secular fundamentalism, and other variations, has become the bogeyman to be opposed. For this, we can thank the works of such religious right theorists as Frances Schaefer, R.J. Rushdoony, and Tim LaHaye,
(...)
Here at Talk to Action, I have written quite a bit about the misuse of the word secular, and its variants, but it is worth noting that among those of all points of view, few who use these terms actually bother to define them, and thus the meaning can depend on the biases of the reader, rather than clearly representing the intentions of the writer.

http://www.politicalcortex.com/story/2006/11/20/153119/94

That's a link to a left-leaning blog, but that shouldn't prejudice your reading of it. If I may summarize the remainder of your argument: "there is no empirical argument for Creationism", "other than common-sense".

"Has it come down to this? Laymen debating experts about technical matters in which they are totally ignorant?"
- "TruthIsAll"

[OnTheOtherHand]

194. yadda yadda yadda

Speaking of vacuousness, this is your second straight post that evinces no awareness of the actual topic of the thread.

Do you know anything about this subject?

[nashville_brook]

26. care to elaborate?

[Name removed]

38. Deleted message

Message removed by moderator. Click here to review the message board rules.

[LeftCoast]

41. Just curious...have you taken statistics?

And if you did, did you pass?

[Name removed]

42. Deleted message

Message removed by moderator. Click here to review the message board rules.

[LeftCoast]

44. I'll take that as a 'No'

:)

[Name removed]

49. Deleted message

Message removed by moderator. Click here to review the message board rules.

[LeftCoast]

50. If you'll re-read my initial post you may notice that my only point was

to say that many people posting on this thread had never taken statistics.

Thank you for helping to prove my point.

:thumbsup:

[Name removed]

52. Deleted message

Message removed by moderator. Click here to review the message board rules.

[LeftCoast]

53. It's not my job to teach you statistics

One wonders how you have managed to convince yourself of the accuracy of TIA's statistical gyrations when you lack even the most basic background in statistics.

[Name removed]

57. Deleted message

Message removed by moderator. Click here to review the message board rules.

[LeftCoast]

58. And I'm still waiting for ONE of you to tell me you took a stat course

Looks like we'll both be waiting a long time...

[Name removed]

67. Deleted message

Message removed by moderator. Click here to review the message board rules.

[LeftCoast]

75. And still you avoid the subject...

My point has been made.

[Name removed]

85. Deleted message

Message removed by moderator. Click here to review the message board rules.

[Skinner]

128. Bring it on, you say? Consider it brought.

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=364&topic_id=2775205&mesg_id=2781808

I suspect that after you read it, you will be apologizing for your repeated personal attacks in this sub-thread.

[Cameron27]

131. Thanks for stepping in Skinner. n/t

[Beetwasher]

135. Why?

I didn't say there weren't any valid criticisms, and I have no problems w/ valid criticisms, I have problems w/ people who merely snipe and then refuse to provide any. As long as you didn't do that, I have no problem w/ you or any criticism.

[Skinner]

138. OK.

Now excuse me while I go remove your personal attacks.

[Beetwasher]

139. Ok

Sorry 'bout the mess. I guess I got a pet peeve with people who merely want to snipe w/ nothing more constructive. The fact of the matter is I appreciate YOUR post that spells out clearly and responsibly what the problems are here, thanks.

[Cessna Invesco Palin]

171. Better watch it, Skinner.

You're turning into me. Before long you'll be ranting in the 9/11 forum. It's a slippery slope, man... a slippery slope.

[0rganism]

97. Okay, I passed a couple college statistics courses

Now, what was your critique of the TIA's statistical analysis? I'd like to read it. Since you've taken the time to offer a critique of the other posters in the thread, I'm hoping you have something equally pithy to say about the OP.

[ProfessorGAC]

142. Read Mine Below

The Professor

[Name removed]

66. Deleted message

Message removed by moderator. Click here to review the message board rules.

[DemocratSinceBirth]

107. Who's Chester?

eom

[stevietheman]

62. I took statistics in college.

What is your specific refutation of the analysis?

After you offer that, I will consider both sides and come up with my own opinion as to which side is more likely correct.

[Hav]

81. .

I had to take statistics classes, too. I passed but to be honest, I sucked. It just wasn't my field. I found it interesting but I didn't get a feeling for the matter.
I still have problems with some concepts even if I read the explanations again.

So could someone plese explain the major mistakes here? I'm just an objective observer here and I'm personally interested if someone is kind enough to explain what is wrong with the assumptions.
I guess the margin of error plays a role for some and I read an article explaining it and showing how many don't use it correctly.

And again, I don't take anyone's side. I'd just like to know.

[Name removed]

86. Deleted message

Message removed by moderator. Click here to review the message board rules.

[OnTheOtherHand]

109. the whole OP is a mistake

If you want to see some serious work on generic polls, you might head over to Charles Franklin's Political Arithmetik archive and read slowly and carefully.

A "trend line" that doesn't actually even have a time scale? the cosmic gong has rung.

[Beetwasher]

136. Then Provide Some

I have no problem w/ someone who wants to criticize, but to merely insult and snipe w/out providing anything constructive is bullshit.

[OnTheOtherHand]

173. I don't know who you are

but I've done my time presenting constructive criticisms of TIA, back when he was permitted to post here. Sorry to annoy you. (EDIT TO ADD: I'm not even snarking, just moving on. Time will tell whether we have anything to say to each other.)

[ProfessorGAC]

137. Here's Two Of Them, Beet

And, btw, i've taught more advanced statistics classes than TIA has even considered taking. (And you know that to be the truth.)

First, he uses an MoE of 1.5%. Nowhere does it show how that polling sample would be 25% less variable than a standard poll, which almost always round DOWN (due to sample size) to 2%. That would inflate the distribution of the final analysis by a factor of 4/3 which moves the tails farther from the mean. Since the distributions are not linear, at that end of the curve, it could move the probabilities out by a factor of more than 100.

Secondly, there is a broad unsupportable assumption that 60% of the undecideds would vote dem. Since the dems didn't get 60% of the total vote, that is an assumption for which there is ZERO basis in fact. Since the "undecideds" were a statistically significant portion of the sample, the entire rest of the analysis hinges COMPLETELY on that assumption.

Without acquiring the whole data set and doing a more supportable analysis, i would estimate the final conclusion to be off by a factor of AT LEAST 100,000.
The Professor

[Beetwasher]

140. THANK YOU!

You know, it just pisses me off that people come in here and just post these one line snipes and nothing else. If the work is shit, tell me HOW it's shit, otherwise I have to assume that the poster is just trolling.

[ProfessorGAC]

141. I've Had These Discussions Before With TIA

So, i'm not talking out of school. A lot of the methodology employed is questionable, but more importantly, the tendency to make unsupported assumptions is too common. Most stat analysis requires some base assumptions but there are fairly firm rules about that.

1) Have some supportive basis of those assumptions.
2) Demonstrate or illuminate that support fot the assumptions.

To assume something, and then use as support "because i think so" is a pretty weak analytical technique.

I do, however, agree with your earlier disagreement. If people don't believe the conclusions, fine. But, i would prefer that folks know WHY they disagree and not just because the number seems to big. That's the same reasoning that drives fundies to insist the world is only 6,000 years old. The idea that things go billions of years back just "seems too big a number".
The Professor

[mom cat]

156. TIA Replies to ProfessorGAC and Skinner

ProfessorGAC (1000+ posts) 
Wed Nov-22-06 09:33 AM
Response to Reply #86
137. Here's Two Of Them, Beet

And, btw, i've taught more advanced statistics classes than
TIA has even considered taking. (And you know that to be the
truth.)

First, he uses an MoE of 1.5%. Nowhere does it show how that
polling sample would be 25% less variable than a standard
poll, which almost always round DOWN (due to sample size) to
2%. That would inflate the distribution of the final analysis
by a factor of 4/3 which moves the tails farther from the
mean. Since the distributions are not linear, at that end of
the curve, it could move the probabilities out by a factor of
more than 100.

TIA:
The 1.5% MoE is justified. I take the FINAL COMBINED 10 polls
to calculate the MoE. There were 1000 sampled per poll. The 10
INDEPENDENT Final Generic Polls is equivalent to a 10,000
sample size. Using the formula to determine the MoE, assuming
a 56/44% voting split, P=.56, N=10000:

MoE = 0.97% =1.96*standard error = 1.96*SQRT((1-p)*p)/N)
So I'm being conservative when I use a 1.5% MoE

ProfessorGAC
Secondly, there is a broad unsupportable assumption that 60%
of the undecideds would vote dem. Since the dems didn't get
60% of the total vote, that is an assumption for which there
is ZERO basis in fact. Since the "undecideds" were a
statistically significant portion of the sample, the entire
rest of the analysis hinges COMPLETELY on that assumption. 

TIA
Both of your statements belie the facts. In a study of 155
elections, the challenger won the undecided vote in 82%, the
incumbent in 12%. In this election, there was a strong
incentive to kick the bums (The Republicans) out.

It's known as the incumbent rule:
http://www.pollingreport.com/incumbent.htm

This is what the most experienced pollster of all, Lou Harris,
had to say about the undecided vote in 2004:
http://www.harrisinteractive.com/harris_poll/index.asp?PID=515


As for your comment that my 60% UVA assumption skewed the
results:
Assuming a 60% UVA, the model projected a 14.8% Democratic
margin
.......Trend+ UVA = Projection
Dem 51.8+ 4.5 = 56.4
GOP 38.6+ 3.0 = 41.6

Assuming a 50% UVA, even though a split is unrealistic in
light of the above, the Democratic margin is reduced to 13.2%
- still a major landslide

.......Trend+ UVA = Projection
Dem 51.8+ 3.75 = 55.55
GOP 38.6+ 3.75 = 42.35

Now let’s recalculate the probability of the Democratic vote
discrepancy from the Generic poll trend line using the Excel
Normal Distribution function: 

Probability = 1.405E-08 =
NORMDIST(0.513,0.5555,0.015/1.96,TRUE) 
or 1 in 71,175,791


ProfessorGAC
Without acquiring the whole data set and doing a more
supportable analysis, I would estimate the final conclusion to
be off by a factor of AT LEAST 100,000.

TIA 
The odds went from 1 in 76 billion to 1 in 71 million. 
That's a factor of 1070.

Skinner should read Jonathan Simon and Bruce O'DEll at EDA:
http://electiondefensealliance.org/landslide_denied_exit_polls_vs_vote_count_2006

[rinsd]

157. In other words, he's full ot it.

[OnTheOtherHand]

178. yawn

Has TIA tried calculating the probability that the last ten polls are random samples from the same distribution? Is he familiar with "house effects"?

I have responded to TIA's undecided crap over, and over, and over. If anyone actually cares, let me know.

EDA's analysis is no better. We've discussed it here before. Again, if anyone actually cares, let me know. TIA pays no attention.

[foo_bar]

189. TIA should do his homework, any homework

TIA:
I take the FINAL COMBINED 10 polls to calculate the MoE. There were 1000 sampled per poll. The 10 INDEPENDENT Final Generic Polls is equivalent to a 10,000 sample size.

Meanwhile:

When combining polls from different survey organizations, house effects also are a problem. These effects represent the consequences of survey houses employing different methodologies, including survey design itself. Indeed, much of the observed difference across survey houses may reflect underlying differences in screening and weighting procedures. Results can differ across houses for other reasons, including data collection mode, interviewer training, procedures for coping with refusals, and the like (see Converse and Traugott, 1986; Lau, 1994; also see Crespi, 1988). Whatever the source, poll results can vary from day to day because polls reported on different days are conducted by different houses.

http://64.233.161.104/search?q=cache:VquKEP41EE8J:www.nuffield.ox.ac.uk/Politics/papers/2002/w27/wlezien.pdf+%22of+survey+houses+employing+different+methodologies%22&hl=en&gl=us&ct=clnk&cd=1&client=firefox-a

I have reservations about combining polls to reduce the margin of error because this requires an assumption that they are both random samples of the same population. This is a tenuous assumption for one poll let alone two, and often the discrepancies between polls tell us not about the population but about the respective sampling procedures of the pollsters.

http://www.biocrawler.com/encyclopedia/Talk:Margin_of_error

GAC writes:
Since the distributions are not linear, at that end of the curve, it could move the probabilities out by a factor of more than 100.

TIA responds:
Now let’s recalculate the probability of the Democratic vote discrepancy from the Generic poll trend line using the Excel Normal Distribution function:

Either TIA can't read, or can't do anything meaningful with the information.

TIA:
It's known as the incumbent rule:
http://www.pollingreport.com/incumbent.htm

The 155 polls we collected and analyzed were the final polls conducted in each particular race; most were completed within two weeks of election day. They cover both general and primary elections, and Democratic and Republican incumbents. They are predominantly from statewide races, with a few U.S. House, mayoral and countywide contests thrown in. Most are from the 1986 and 1988 elections, although a few stretch back to the 1970s.
(...)
Some examples of where more undecideds voted for incumbents or split evenly:

Last year in Minnesota, where Hubert Humphrey III challenged Sen. David Durenberger; and in Nebraska, where Bob Kerrey, the former governor, challenged David Karnes, who had been appointed to his Senate seat. In 1986 in Florida, when incumbent Sen. Paula Hawkins faced ex-Gov. Bob Graham. And in Chicago in 1979, where two-year incumbent Mayor Michael Bilandic split undecided voters with challenger Jane Byrne.

These examples and similar ones account for 17 of the 28 exceptions to the Incumbent Rule that we uncovered. In some of the remaining cases, the incumbent simply turned the race around in the final days. A good example of this is the 1982 Missouri Senate race pitting incumbent John Danforth against Harriet Woods. Other exceptions can be explained by sampling error.

http://www.pollingreport.com/incumbent.htm

Thus the "rule" is anything but. What's uncanny is TIA citing Nick Panagakis on this subject:

The conclusion that "exit polls differed significantly from recorded tallies" in the three states is incorrect.
(...)
Conclusion. All of the state estimates above are well within their error calculations below.
Ohio, n = 2020. Sqrt (.5 X .5) / Sqrt 2020 X 2.6 X 1.6 = +/- 4.6%.
Pennsylvania, n = 2107. Sqrt (.5 X .5) / Sqrt 2107 X 2.6 X 1.6 = +/- 4.5%.
Florida, n = 2862. Sqrt (.5 X .5) / Sqrt 2862 X 2.6 X 1.6 = +/- 3.8%.

http://www.mydd.com/story/2004/11/30/171641/64

This is what the most experienced pollster of all, Lou
Harris, had to say about the undecided vote in 2004:
http://www.harrisinteractive.com/harris_poll/index.asp?PID=515

This is the Harris poll showing Bush 49, Kerry 48, Nader 1 that TIA cherry-picked out. Not really a coincidence, it's at the point where TIA can't cite a single expert without bumping into his own rhetoric:

Skinner should read Jonathan Simon and Bruce O'DEll at EDA:
http://electiondefensealliance.org/landslide_denied_exit_polls_vs_vote_count_2006

Weird, TIA dissed Bruce O'Dell when their conclusions differed:

No class.

Your judgment in using that term is "fatally flawed".

I have read your paper.
And you don't put a dent in Ron's analysis.

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x374482#375181

More importantly, TIA can't respond to Skinner's point, because there's no defense:

The generic congressional ballot is not -- and was never intended to be -- an accurate prediction of how people will vote. The point of the generic congressional ballot is to get a general sense of the mood of the voters.
(...)
You can do the best, most accurate, most awesome mathematics in the history of the world, but if you start with completely false assumptions, your "analysis" is going to be worthless.

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=364x2775205#2781808

Indeed, comparing a "NIMBY" generic R vs. D to people's opinions of a named candidate is a ridiculous premise. But using a refuted paper about exit polls to defend the misapplication of a generic poll? Good riddance.

[foo_bar]

90. it's a variation on a theme

First off, it assumes that sociology follows a Gaussian ("normal") distribution, which was a good theory in the early 1900s:

These early sociologists were not concerned with theology, but they were seeking proof of the orderliness of society. Relying on the justifiably great prestige of Laplace and Gauss as mathematicians, they took the bell curve as proof of the existence of order in the seemingly chaotic social world. Unfortunately, the early social scientists often had a poor understanding of the fact that the mathematical formulas of Gauss and Laplace were based on assumptions not often met in the empirical world.

http://crab.rutgers.edu/~goertzel/normalcurve.htm

Bachelier's model of the stock market was, however, too simple and failed to capture many of the crucial aspects of price fluctuations - for example the possibility of crashes. Bachelier assumed that many of the fluctuations followed a Gaussian probability distribution, but crashes were absent from his model because the probability of extreme events is embarrassingly small in a Gaussian world (see left). Since then many scientists have tried to develop better models.

http://physicsweb.org/articles/world/12/1/7

Mandelbrot found that price changes in financial markets did not follow a Gaussian distribution, but rather other Lévy stable distributions, having theoretically infinite variance. He found for example that cotton prices followed a Lévy stable distribution with parameter α equal to 1.7, rather than 2 as in a Gaussian distribution. "Stable" distributions have the property that the sum of many instances of a random variable follows the same distribution but with a larger scale parameter.

http://en.wikipedia.org/wiki/Beno%C3%AEt_Mandelbrot

Like the dead microbiologist example, there's a host of hidden assumptions (in the latter, that you can predict the future by cherrypicking some "suspicious deaths" and formulating a bill-yuns to one probability from garbage-in-garbage-out data, and by throwing in people who weren't "world class microbiologists" to begin with): here he borrows the popular conception of margin of error, not its actual application in probability theory:

* The margin of error is a simple transformation of the number of respondents into an ambiguous term that is neither a "margin" nor the whole of "error".
* It is not a "margin" at all; the probability of the true percentage being outside the margin of error is low but nonzero.
* Many pollsters fail to account for the complexity of their sample design when calculating the margin of error, which usually makes their polls appear more accurate than they truly are.
* Perhaps most importantly, there are many different sources of error in polling, and variance due to sample size is not likely to be the only contribution. Other possible contributions to error include:
o Sampling bias, when the sample is not a representative sample from the population of interest. In particular, certain people may choose not to participate.
o The phrasing of the question may not be appropriate for the conclusions of the poll.
o Response error (Sudman & Bradburn, 1982)
+ Deliberate distortion (fear of consequences, social desirability, response acquiescence).
+ Misconstrual (not understanding the question).
+ Lack of knowledge (guessing to try to be helpful).

http://en.wikipedia.org/wiki/Margin_of_error

TIA's argument is that response error (and every other kind of non-sampling error) is a conspiracy, thus you can plug the results of polls into Excel and calculate the "billyuns and billyuns" chance of them diverging from the margin of error, in a perfect universe that only exists in the first week of your first stats course. To wit:

Battle ground state after battle ground state switched to the Republican column throughout the night violating the well established laws of mathematics concerning large sample polling.

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=364x2600324

Versus what they teach in stats class:

Non-sampling error cannot be controlled by making the sample bigger. Indeed, bigger samples are harder to manage. Increasing the size of the sample—which is beneficial from the perspective of sampling error—may be counter-productive from the perspective of non-sampling error. Non-sampling error itself can be broken down into three main categories: (i) selection bias, (ii) non-response bias, and (iii) response bias. We discuss these in turn.

http://www.stat.berkeley.edu/~census/sample.pdf

In TIA's belief system, people literally behave the same way as coin tosses (which are themselves not 100% free from bias, as it turns out). It's fine shorthand if you want a ballpark approximation, but to use it as proof of fraud (as TIA alleged NH, MA, NY, and VT were part of the 2004 conspiracy, since their exit polls were off by the most ("The Beast From The East"), despite being states with paper ballots and Dem control of the levers of power) is as unscientific as proving JFK was assassinated with Excel. In the real world:

Nothing in this paper should be taken as questioning the use of the normal distribution where it is appropriate (e.g., in estimating confidence intervals from random samples). To make this correct usage clear, it might be wise to revert to the earlier phrase, "normal curve of error." This would make it clear that the normal bell curve is "normal" only if we are dealing with random errors. Social life, however, is not a lottery, and there is no reason to expect sociological variables to be normally distributed.

http://crab.rutgers.edu/~goertzel/normalcurve.htm

In TIA land:

One can apply the Cum. Normal Distribution or Poission function (or both to confirm) to determine the probability of rare events occuring by chance only.. There are other models which may be applicable. I found these two perfect and easy to use. All you need is Excel; the functions are built-in.

I have also used this method to calculate the probability that
1- at least 15 JFK witnesses would meet unnatural deaths in the year following the assasination.
2- at least 16 world-class microbiologists would meet unnatural deaths in a 4 month period following 9/11.
3- The probability that at least a certain number of people would suffer from mad cows disease in a specific geographic area in a given year.

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=1777401#1777492

Here's an analogy to what you are doing: say I'm a broker; I look at hos Ford stock has done in the last couple weeks. Well, by golly it's up 25%!! So I tell my client, "Hey - buy Ford stock - it goes up 25% every two weeks!" and then the customer says to me, "That sounds really suspicious - are you sure? Do you have the numbers to show it?" and I say to them, "I have the numbers from this two week period. If you want better proof than that, you go look it up."

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=1005141#1005528

What is wrong with TIA's arguments. You could write a book, but for starters the guy MUST read up on the standard error associated with exit polls v. simple random samples. Actually, all he has to do is read MP. I've posted on the subject as well.

If you realize his error on this point alone, all the rest of his arguments don't need refuation - they refute themselves.

http://www.mysterypollster.com/main/2004/12/what_about_thos.html

[Hav]

125. .

Thank you very much, your effort is appreciated.

[RoyGBiv]

219. Brilliant post ...

Thanks for taking the time and effort you put into it.

[WI_DEM]

5. polls do have a plus and minus factor--it is not an exact science.

[mom cat]

9. All of that has been taken into consideration.

[LeftCoast]

If it had been we wouldn't keep having stupid threads like this.

[0rganism]

98. Is that why you troll these threads? Because of unaccounted MOEs?

Wow. That's pathetic.

You can do far better than that. At least say something about normal distributions and sampling bias.

[davekriss]

10. It's funny, however, that the plus/minus factor

what in TIA's three graduate degree language is Margin of Error (MOE), has trended in near-exclusive favor of the Republicans since 2000. Kinda of like rolling 5 hard-eights in a row in craps, or getting 11 reds in a row in roulette.

Speaking of 5 hard-eights in a row, I've actually done that. No, I didn't win much because I kept taking my winnings off the table with each roll. A student standing next to me let his winnings ride for 4 rolls, winning $5000. That same week I was at a table where one guy won $83000 and several won over $10000 -- not a good week for the casino! So it does happen, just not very likely.

[OnTheOtherHand]

61. no, that's not it

Most exit polls that I've looked at -- not just those since 2000 -- have overstated the Democratic vote share.

"Margin of error" just applies to random sampling error. There are other error sources as well. TIA claims to know this, but he acts as if it doesn't really matter. It does.

[davekriss]

103. That was my point

You make my case, thank you.

Let me put it another way: Exit polling, which prior to 2000 has accurately tracked the actual vote within MOE, has suddenly been rendered an unreliable tool because it now overstates the Democratic vote. Further, Exit Polling has tracked closely to general polling in the runups to election day. "Overstating the Democratic Vote" can mean one of two things: (1) the pollsters inexplicably forgot how to do their job -- this same job touted by the USG to claim election results invalid in other nations --, or (2) Democratic votes are stolen at the polls making it appear, since 2000, that exit polling is inflated toward the Democrats and therefore unreliable.

Did you ever think that what is unreliable is the vote counting on election day?

Almost all anamolies these past elections, 2000-2002(Georgia)-2004-2006, fall in favor of the Republicans. They gain by the unfortunate blip and snafu. Again, like rolling 5 hard-eights in a row in craps or 11 red numbers in a row in roulette. While certainly possible, not very likely.

What Rove failed to calculate was the ferocity of the anti-Bush vote. Even after slicing off in aggregate 5% to 8% of the Democratic vote they still lost the Senate and the House. Don't think they'll make the same error in 2008 (if there is an election then! -- Bush said recently that it's a wonderful thing that we can hold an election when the nation is at war, meaning in his head he thinks the opposite is possible, albeit less "wonderful").

[OnTheOtherHand]

108. no, your premise is incorrect

"Exit polling, which prior to 2000 has accurately tracked the actual vote within MOE"

That is incorrect. As I stated, "Most exit polls that I've looked at -- not just those since 2000 -- have overstated the Democratic vote share" (emphasis added). I have no idea why you would think that this makes your point, since it refutes your point. The exit poll discrepancy in 1992 was almost as large as the one in 2004, and probably larger than the one in 2006.

If people would stop repeating misinformation on this score, maybe we could focus on actual evidence.

"(1) the pollsters inexplicably forgot how to do their job -- this same job touted by the USG to claim election results invalid in other nations"

I haven't seen a case yet where the USG actually relied on exit polls to challenge election results.

"Did you ever think that what is unreliable is the vote counting on election day?"

Did you ever think that the vote counting and the exit polls could both be unreliable?

This is a travel day for my family, so we may have to resume the conversation another time.

[Cameron27]

132. Always happy when you show up OTOH...

I depend on your anaylsis, (and Febble's) to provide clarity in these murky situtations.

Have a great Thanksgiving!

[davekriss]

149. I view the response...

...as obfuscating, as my point is unrealized Democratic votes as measured by exit polls and general polling leading up to election day, amounts to vote theft. OTOH would claim it indicates either the unreliability of polling or the unreliability of polling and voting. I think any honest individual would agree that what we have is a major mess! No wonder centrist to right-wing politicians are conveniently elected year after year these days, even though it goes againt the self-interest of the vast majority of Americans.

[OnTheOtherHand]

172. (blush) have a great one too! n/t

[davekriss]

148. You still make my case

If polling over the last thirty days (including exit polling) of an election cycle show Democratic preference that is seldom realized in the voting booths, it should get our hackles up. We must get to root cause. We must answer "why?". Could be you're right, the polling companies aren't very good at what they do. On the other hand, Republican theft of elections may be rampant. And (I think this might be your position) it's possible that neither exit polling nor the actual election results are entirely reliable. This is not something to say, well, let's move on and forget about it.

If indeed exit polls have consistently shown a Democratic vote that exceeded the actual vote (which, by the way, I can't find evidence for on-line), and the exit polls track closely to the general polling leading up to election day, then I think we have a problem here, Houston: Entrenched vote theft of the well-known variety -- voter intimidation, scrubbed voter roles, strategic placement of voting equipment, anything and everything to reduce the probability that a Democrat casts a vote and that makes it easier for Republicans. Hey, to that old and well documented mix we've added easily manipulated, no paper trail electronic voting machines!

There's good reason to distrust Mitofsky, but here's what he has to say on the matter:

The mistakes made during the 2000 election were unusual. During the 10 years before that VNS and the poll before it made only one mistake from 1990 to 1998. Before that, when the broadcast networks made their own projections, there were similarly very few mistakes during the 1970s and 1980s. There were no mistakes during the limited coverage in 2002. There were no mistakes made during the 2004 Democratic presidential primaries.
-- Exit Polls From Edison Media Research and Mitofsky International (emphasis added)

Since exit polling has always been controlled by the major media, and the major media is in the business of corraling people to sell to corporations via the intermediary of advertising, they have a vested interest in pleasing these corporations. It's very possible that highly secretive (technique) exit polling is influenced by that interest.

However, why would agents of the major media generate bias for the Democrats? It makes no sense. Just pull up the demographics of any national election and one consistent observation can be made: The higher up the income ladder you go the more preference you find for Republicans. And it ain't the bottom demographics that occupy the board rooms of these firms! Further, board room influence has grown in this post-Fairness Doctrine/post-Rule of Sevens/post-Political Editorial Rule march to near-monopolization. If bias exists, it's the bias of the owning class.

I need a cite from you (I mean this in a friendly way) to believe there's more than the couple of errors Mitofsky refers to above (overstatements in favor of Democrats). Over all, exit polling has been fairly accurate. But to my point that you make my case for me (again): If exit polling consistently shows a preference for Democrats that is not realized in the voting booths, and if exit polling is consistent with general polling leading up to election day, then it is prima facie evidence for vote theft by the right and may (if your claim holds up) spill beyond the elections of 2000-2002-2004-2006.

[nashville_brook]

151. your points about BIAS are right on!

all your points are... but i especially like seeing someone address the giant elephant in the room; BIAS.

"exit polls favor democrats..." Orwell would LOVE this statement.

an exit poll is a check. when there is a discrepancy, it's dishonest in the extreme to blame the poll. the poll exists to make sure the vote is accurate.

one way to clear this up -- CHECK THE VOTE.

to blame the exit poll for anomalous data is like blaming your mammogram for being whacked when it shows you have cancer.

[OnTheOtherHand]

182. you have a lot to learn

"the poll exists to make sure the vote is accurate."

No, it doesn't. Exit polls in the United States were introduced to provide information on why people voted the way they did, and to get a head start on predicting outcomes -- not to verify the results.

I've read, over and over again, that U.S. exit polls were highly accurate until 2000. Well, there's no evidence for the claim. It used to rest on the incorrect assumption that the exit poll results were close to the official counts.

"to blame the exit poll for anomalous data is like blaming your mammogram for being whacked when it shows you have cancer."

OK, then, concluding fraud on the basis of an exit poll result is like getting a radical mastectomy on the basis of a mammogram result.

We've explained many times why the 2004 exit polls don't provide evidence of fraud. You haven't contributed much of substance to the discussion. But it could still happen.

[OnTheOtherHand]

185. well, if your point is

"it's possible that neither exit polling nor the actual election results are entirely reliable" -- sure, we agree.

"This is not something to say, well, let's move on and forget about it." I haven't. TruthIsAll has offered a seemingly endless succession of crap arguments (of course, he often repeats himself), and I and others have knocked them down. We try to learn more about our world by distinguishing between good evidence and bad evidence, good arguments and crap arguments. If that enterprise doesn't engage you, fine. But your comment about "obfuscating" in the other post seems rather bizarre.

"If indeed exit polls have consistently shown a Democratic vote that exceeded the actual vote (which, by the way, I can't find evidence for on-line)"

I already linked to some in this thread. Again: http://www.mysterypollster.com/main/2004/12/have_the_exit_p.html

This one is also fun: http://www.mysterypollster.com/main/2005/01/the_war_room.html

I could link to many other sources, but I'm not sure what you would regard as evidence. If you have an academic affiliation, you could actually download the datasets -- although who is to say that those shadowy media figures didn't fake them all? (The datasets don't interpret themselves. Ruy Teixeira pointed out soon after the '04 election that one of the 1988 exit polls, I think CBS, actually sampled more Dukakis voters than Bush voters -- but it appears that that is partly because CBS deliberately oversampled Democratic precincts.)

"and the exit polls track closely to the general polling leading up to election day"

Well, they don't. I suppose it depends on what you mean by "closely."

"then I think we have a problem here" -- of course we have a problem here. We don't need crap exit poll arguments to tell us we have a problem here, just as we didn't need dodgy documents to tell us that Shrub had some "issues" with his National Guard service.

"The mistakes made during the 2000 election were unusual."

As you probably recall, the "mistakes made during the 2000 election" were incorrect calls -- indeed, two incorrect calls of the same state in opposite directions. Neither of these calls was based on interview data alone. Mitofsky's record of (usually) avoiding incorrect calls has no bearing on bias in the underlying data. The estimation/projection system is designed to be robust in the face of data bias.

This is a common fundamental misunderstanding: confusing accurate 'calls' with unbiased data.

Here's my most extensive analysis of the '04 exits. Let me know if you have any questions.

[davekriss]

196. Turkey calls

Thanks for your reply. Given the day, it's going to take a while before I can digest it. I will read your PDF. Given your erudition on the subject, is there a summary link somewhere that lists aggregate general polling the 30 days leading up to election day, aggregate exit polling, and actual voting for several elections prior to 2000? I have anecdotal cites (e.g. Georgia 2002, etc.).

My position is, that if general polling indicates x (e.g., again, Democratic preferences in Georgia 2002), exit polling is close to x, and actual voting turns out to be x - y (well beyond MOE) -- and this "minus y" factor is significantly more often in favor of Republicans rather than Democrats -- then we have prima facie evidence of voter suppression and count tampering.

This is not "exit poll fundamentalism", which I won't call a straw man until after reading your paper.

But I'm glad to see we agree that we have indeed a voting mess.

Now, off to carve the turkey...

:)

[OnTheOtherHand]

198. not really

You can try here for final pre-election presidential polls through 2000, but I don't know of any site with roundups for the last 30 days. Even in this document you can see how polls have proliferated in recent years, so it is hard to do comparisons across times. As for exit polls, the exit pollsters don't even attempt to project the popular vote, so it's a great big mess.

Here I demonstrate that the states where Bush did especially well compared to pre-election polls don't tend to be the states where he did especially well compared to exit polls. It strikes me as a fairly big problem. (We could conjecture that miscount was fairly uniform across the country, but that doesn't really fit very well either.)

Indeed, specific testable hypotheses aren't exit poll fundamentalism, a phrase that I define in the paper.

[davekriss]

236. OK, I read your paper

I have read your paper, Beyond Exit Poll Fundamentalism, as well as your "Surprise" page and an interview you gave "tng" at neuralgourmet. You do indeed call into sufficient question the easy belief that exit poll variance evinces election theft. However, I have some questions.

You lean heavily on the lack of correlation between swing and red shift. All else being equal, I would expect more votes for Bush in 2004 than he received in 2000 in precincts ("swing") where actual votes for Kerry fall significantly short of exit polls ("red shift"). However, that is not what you found (there is no correlation). I am unclear, though, on what you looked at.

I assume, when assessing swing, you refer to percentages and not simple vote count. So, if turnout increased, you expected new voters to vote along the same lines as exampled by 2000 voters (within your 10% tolerance)? Is that a risky assumption? If it is risky, does that weaken your conclusions in any way?

Also, shouldn't there be a "mal-pResidency discount" factored into the 2000 vote before conclusions are drawn about swing? Yes, yes, I know this is a "fraud of the gaps" argument, part of the fundamentalism you dispel so well in your paper, and not something that is easily discernible in the data nor readily testable. But, given all that had transpired up to Election Day 2004, maintaining a similar proportion to the 2000 vote in itself amounts to a fraudulently achieved "swing", no? (I speak mostly tongue in cheek.)

I may have evoked exit polling in my first post on this thread, but I never was an exit poll fundamentalist. Instead I rely more on the polling leading up to election day to tell me what to expect (mainly because of easy access). Unfortunately, in 2004, not enough of us were persuaded that Bush was (still is) the "worst President, ever". The actual vote may not have been out of line with pre-election polling, nationally or in Ohio, which means despite voter suppression (which I think did occur) and stealthy miscounts (which you call into question), a Bush victory was plausible. However, theft may have been achieved at the margins, a little bit here and a little bit there, Democracy threatened by a death of a thousand cuts.

[OnTheOtherHand]

238. thanks for the qq.

I'm about to go on the road, so my answers may be cursory -- feel free to 'try me again.'

It's not necessarily that the analysis 'expects' new voters to vote along similar lines if turnout increases. Indeed, a percentage comparison won't be sensitive to suppression of Dem votes in heavily Dem precincts, or padding of Rep votes in heavily Rep precincts. That also applies to exit poll "Within Precinct Error" analysis: (e.g.) massive vote suppression in Dem precincts wouldn't explain the observed WPEs.

I'm not sure I know what you mean by a "mal-pResidency disccount," but the analysis doesn't hinge on assuming that Bush should have done as well in 2004 as he did in 2000. It is possible for performance in one race to be a good "predictor" of performance in another even if the vote shares are very different. One of my responses to Kathy Dopp tries to spell this out in more detail, along with the possible caveats in the analysis. Certainly I don't think that a single correlational analysis demonstrates that 'there was no fraud.'

The evidence of voter suppression is solid, and there is some solid evidence of miscount too (if one considers the caterpillar crawl in Cuyahoga "miscount" -- it's certainly mis-something). I'm not persuaded that these factors altered the result in Ohio (partly because the pre-election polls there did give Bush the upper hand), but vote suppression and miscount are unacceptable whether they are decisive or not.

[davekriss]

239. mal-pResidency discount

What I meant (OTOH): If Bush received 100 votes in precinct x in 2000, is it reasonable that he can count on those same 100 votes in 2004? Could not events between 2000 and 2004 have measurably eroded support? Say, for illustration, going in to 2004 and without count tampering Bush could only count on 90 votes. Thus a "swing" back to 100 votes for Bush might be correlated with 10 votes of exit poll red shift, but you would have missed this in your analysis.

If some reasonable assumption could be made on "mal-pResidency discount" before applying your analysis, you might find a correlation that escapes you now. However, finding the basis for that "reasonable assumption" might prove difficult, given the slight Bush lead in pre-election voting. I therefore say this tongue-in-cheek as I know, if I say "surely such a discount exists", it examples the "fraud of the gaps" fundamentalism that you find so frustrating.

You conclude that suppression is clear, and perhaps a degree of miscount, but these were not sufficient to flip Ohio for Bush that otherwise would have gone to Kerry (based on the slight lead in pre-election polling Bush held going in to election day). I might agree but I'd need to know more. All you've achieved with me is unweighting of the exit poll argument. What about other states? Is it possible that other states were flipped sufficient to make Ohio decisive? Is anyone with your academic rigor looking at this now?

[OnTheOtherHand]

240. you're asking the right questions

It's certainly unlikely that Bush would perform identically in 2004 as in 2000, and his overall level might be higher in some places, lower in others. If we simply assumed that it was, say, five points lower across the board, that wouldn't affect the correlational analysis at all. In order to use 'real' change to conceal theft, the change needs to vary, and the theft has to be directed to 'cancel out' the variation in real change. I see no good reason to believe that that happened nationwide in 2004. Maybe if some of the folks who are convinced that it did happen would exert themselves to explain how.... (Bear in mind that I'm discussing a particular scenario of millions and millions of votes stolen nationwide, as supposedly proven by the exit polls. I didn't make it up; Steve Freeman wrote a book about it, sort of.)

By the way, it didn't shock political scientists in general that Bush would actually do better in the popular vote in 2004 than in 2000, because we talk a lot about incumbency advantage. Not that incumbency advantage guaranteed Bush's reelection, just that his vote share could have gone either way. Politically aware people tend to have a hard time accepting the existence of incumbency advantage; also, politically aware Democrats have a hard time accepting that Bush's approval rating in '04 could have been anywhere near 50%. I would be inclined to say that a "mal-pResidency discount" did exist but was offset by other factors.

I'm not trying to convince you about Ohio right now, and that's obviously important, because if Kerry wins Ohio, the rest doesn't matter. I'm shrill about the exit poll arguments because the most prominent ones are awful, and I'm shrill about a few of the Ohio arguments, but there are plenty of legitimate what-ifs. (And you mention other states in lieu of Ohio. New Mexico well could have been flipped. Beyond that, I don't know, you look around.) I would say that Mebane and Herron's analyses in the DNC Voting Rights Institute report are the fullest assessments of Ohio data right now. I want to put together an article that pulls together all available strands (like the cryptic information on the likely extent of vote suppression), but life keeps intervening. Some others also are working on aspects of Ohio. I would have liked to see what the RFK Jr. article would have looked like without all the overreach.

In my professional opinion at this time, the publicly available evidence doesn't add up to a Kerry win in Ohio. Richard Hayes Phillips thinks he has enough to tip the balance, and he may, although what I've seen of his work has been marred by faulty assumptions about voter behavior -- for instance, he seems to think that any instance of a Bush voter voting against the gay marriage ban constitutes prima facie evidence of fraud. But I won't declare myself unconvinced by evidence I haven't seen, any more than I will declare myself convinced by it. (Also, lest I confuse the issue, Phillips may have solid evidence of fraud whether or not he has solid evidence that the fraud altered the outcome.)

[Zynx]

12. Oh come off it already. This posts just cause loss of credibility.

[Name removed]

39. Deleted message

Message removed by moderator. Click here to review the message board rules.

[Cessna Invesco Palin]

117. He never had any credibility to begin with. n/t

[Last Stand]

15. That's why Chimp didn't fire Rummy before the election.

He figured that Diebold would provide enough of a bump to win the House and Senate. They misunderestimated how angry the masses are at his Accomplished Mission.

[LeftCoast]

16. Ah, so the fact that the election wasn't stolen PROVES it would have been.

Oy vey

[Last Stand]

19. what's your point? We're now talking numbers here.

I stated BEFORE the election that they would likely flip 2-3% of the vote, giving a 4-6% bump to Chimp's Chumps. Looks like that's exactly what happened, according to hard numbers (and not baseless paranoia.) The thing that they didn't count on is that a 4-6% swing wouldn't give them the House or Senate. Rummy would still be employed if they'd stolen enough votes.

[mom cat]

21. Thank you for that clarification.

[mom cat]

33. That is not what the poster said.

[understandinglife]

17. Recommended.

[autorank]

22. #5 here. This just keeps geting better.!!!!!!!!!!!! n/t

[mom cat]

24. Glad to see you. I thought you had gone on vacation.

Have a good trip if you are able to get away!

[robcon]

23. The problem isn't mathematical or statisticaL

It's that exit polls are not random statistical samples from the universe of voters, and this has been known for decades.

The "analysis" is garbage.

[survivor999]

29. The exit polls are known to have a Dem bias.

Not sure why.

[mom cat]

34. Actually, exit polls are quite reliable. The theory of the "Dem Bias"

in exit polls is a theory created by Repukes to "explain" why the exit polls favor the Dems more than the Diebold results do.

[survivor999]

37. Link?

[OnTheOtherHand]

51. riiiiiiiiight

Do you really think "Repukes" created a theory about "the Diebold results" back in 1992, when the presidential exit polls were almost as far off as in 2004 (and probably farther than the midterm polls in 2006)?

More to the point, can you support this notion with any evidence?

[crazylikafox]

35. Known to have a Dem bias by whom?

Can you give a source for that? Other than Fox?

[survivor999]

36. I heard in on NPR.

Probably because liberal leaning people may be more likely not to just walk away.

[OnTheOtherHand]

120. this roundup might help

http://www.mysterypollster.com/main/2004/12/have_the_exit_p.html

In short, Mitofsky and Lenski have reported Democratic overstatements to some degree in every election since 1990. Moreover, all of Lenski and Mitofsky's statements were on the record long before Election Day 2004.

(emphasis in original)

That isn't to say that the exit polls were accurate before 1990, just that Blumenthal couldn't find statements about their accuracy.

The big point that generally gets lost here is that the exit pollsters don't count on the interviews to be perfect, because they don't make quick calls in close races. So some degree of inaccuracy in exit polls isn't a big problem for exit pollsters -- only for exit poll inerrantists.

[mom cat]

40. From TIA: THESE ARE 116 PRE-ELECTION GENERIC POLLS, NOT EXIT POLLS

You either did NOT understand or did not READ the post.
So which is it, robcon?

There is NOT ONE REFERENCE IN THE POST TO AN EXIT POLL.
The analysis compares 116 PRE-ELECTION GENERIC POLLS, taken between Sept 2005 and Nov.6, 2006, to the ACTUAL VOTE COUNT.

DO YOU KNOW WHAT A GENERIC POLL IS?
Here they are- all 116 polls:
http://www.pollingreport.com/2006.htm

Do you know what a LINEAR TREND IS?
Here's the trend line graph- all 116 polls:


Either ALL 116 GENERIC POLLS were WRONG or the VOTE COUNT WAS RIGGED.
Which one was it?

[Ms. Toad]

59. Neither one is necessarily wrong.

Generic polls are useless to predict wins in individual district races unless (1) the districts have population distribution substantially similar to the polled population (gerrymandering by the party in power pretty much guarantees that this is not the case) or (2) the generic poll is specific to the district in which the contested race is being held (but then it wouldn't be very generic, would it).

The discrepancy between a generic poll and actual results will generally be sharpest when (as now) the minority party is rapidly gaining in popularity. The gerrymandering will still create wins for the controlling party or squeaker races in the many districts which are designed not to reflect the general population distribution so they will be safe seats for the controlling party.

Its fairly similar to a president winning the popular vote (equivalent of a generic poll) but still losing in the electoral college (because of safe states created, in this case by population distribution rather than deliberate manipulation of boundary lines).

[OnTheOtherHand]

65. well, the issue here isn't wins in individual districts

It's that the total vote (regardless of the number of wins) doesn't match some of the generic polls. I say "some" because the total vote is close to the Pew, Gallup, and ABC/WaPo results.

Lots of people were trying to figure out which generic polls were most likely to be accurate and why. I guess there's something to be said for refusing to choose. But claiming 1 in 76 billion odds is just ridiculous.

[Ms. Toad]

72. You're right - its not as bad as my initial quick look

I just get so frustrated at the number crunching based on misleading, unstated, or just plain wrong premises that I don't always read these posts as carefully as carefully as I should. Mostly I just refrain from posting since all it does is generate SHOUTING. I'd hoped the shouting would tone down after we actually won the elections, but it seems to have intensified - amazing how one can SHOUT without even being present.

[OnTheOtherHand]

110. yeah, it's all pretty remarkable

All this arguing with tombstones makes me feel like Ebenezer Scrooge. Lucky I will be traveling to see family.

I think it would be interesting to take all the generic poll results from the last week or so and figure the probability that they were all accurate within sampling error. It wouldn't be one in billions -- there aren't enough polls -- but it would be a stretch.

[smalll]

80. Ha! Ha! I say -

Oh, suddenly we care about PRE-ELECTION POLLS. What about all the PRE-ELECTION POLLS that, averaged out, told us BEFORE Election Day, 2004, that Bush would win by about 3%??? By which percentage Bush allegedly won? (Allegedly, according to the New York Times, CNN, the Secretaries of State of the 50 states, the BBC, C-SPAN, China Daily, etc. etc. - you know, the vast conspiracy.)

In November 2004, the Fraudsters would have nothing to do with pre-election polls! It is to laugh! :rofl: Laugh I say!

[Joe Chi Minh]

209. The importance of your post, the point you make here, cannot be

overstated.

When they are shot down in flames, they murmur a few vapid blandishments and scurry on, as if it were a matter of no consequence that they'd had to concede on the main issue. They'll often bend over backwards to give an appearance of honest interest/enquiry/truth-seeking, yet they NEVER EVER lose sight of their objective, coming back again and again to dispute what they had hastily conceded as of not much consequence.

Winning their argument is everything. One bird thought that for me to apologise to another poster for getting something terribly wrong, as publicly as possible would be the ultimate humiliation. Instead of a blessing. The notion that some people are actually interested in the truth, and not driven by amour propre, is novel, and shocking to them.

I want to repeat your post here, and wish it was all in upper capitals:

"Oh, suddenly we care about PRE-ELECTION POLLS. What about all the PRE-ELECTION POLLS that, averaged out, told us BEFORE Election Day, 2004, that Bush would win by about 3%??? By which percentage Bush allegedly won? (Allegedly, according to the New York Times, CNN, the Secretaries of State of the 50 states, the BBC, C-SPAN, China Daily, etc. etc. - you know, the vast conspiracy.)"

[OnTheOtherHand]

223. what is this, self-parody?

If you care about the truth, start by considering the substance of your own posts.

"What about all the PRE-ELECTION POLLS that, averaged out, told us BEFORE Election Day, 2004, that Bush would win by about 3%???"

That's evidence against Bush having stolen the 2004 election. Doh.

In my opinion, it's actually overstated: the average poll margin was less than 3%. More like 1 1/2%. See e.g. http://pollingreport.com/2004.htm .

So explain it to me: how are you contributing to "honest interest/enquiry/truth-seeking"? Or even dishonest interest/enquiry/truth-seeking? Do you ever even look something up?

[Joe Chi Minh]

227. Just respond to smalll's post. How come the sudden interest

in PRE-ELECTION POLLS, on your part?

[OnTheOtherHand]

229. on MY part?

Um, guy, smalll was criticizing fraudsters, of which I am not one.

(However, smalll is apparently unaware that in TIAWorld (tm), the 2004 pre-election polls not only didn't show Bush ahead, but gave Kerry a decisive advantage.)

Care to try again?

[Zynx]

102. Individual house race polls showed results very similar to what happened.

The national generic vote is subject to really wild swings due to the likely voter model.

[Sancho]

68. You don't really know how well the polls generalize...

Most of us don't have access to all the poll data. We have parts of the polls without precinct ids, little information about the interviewers, and less about the details of the process. The sampling from exit polls may be representative, or they may not...and they may be perfectly representative for some states, races, or other units of analysis. They may also be useful to compare to other data outside the exit polls. Without precinct level data, it's had to say.

Unless E-M or others make the process more transparent, you can't say if the polls are representative and some of the elections are hacked, or if the polls are biased by sampling (reluctant voter) and process (gender/age of the interviewer) variables.

We don't know.

TIA's computations are based on normal distributions, and likely overstate the probabilities, (see Bonferronni), but the hypotheses are very interesting and may, in fact, be evidence that something is unexpected.

Personally, I think the defenders of the polls are too quick with excuses and pretty light on the data. On the other hand, now that election supervisors are posting more data at the precinct level and we can put it with annectdotal evidence for a given precinct, county or race...I suspect you'll see more EDA type reports. I'll looking at a few things now from the 2006 elections, and getting facinating trends....we'll see.

Meanwhile, TIA is interesting to read, and he certainly is creating some possible post hoc ideas that may be researchable by others.

PS. Yes, Virginia, I've studied statistics...

[Zynx]

101. Indeed. They are overwhelmingly concentrated in urban areas and

thus have always skewed disproportionately towards the Democrats in the early going in virtually every election.

[OnTheOtherHand]

116. I actually disagree with the first part

The exits aren't especially concentrated in urban areas. The exit pollsters aren't dumb. Their precinct samples are generally pretty good.

But their interviewers tend to get within-precinct samples that overstate the Democratic share. We're not sure why, but we know it's been happening for years. A lot of folks who like to trumpet how accurate exit polls prior to 2000 might realize, if they really stopped to think about it, that they don't know what the exit poll results were prior to 2000. Actually, a lot of them don't know what the exit poll results were in any year.

[Zynx]

159. I don't know about that. I did see a couple states that had some strange

geographical groupings. One thing that is for sure is that its harder to get a good sampling of a rural area than an urban one. The logistics involved are quite daunting. For example, in Wisconsin, southwestern and northwestern rural Wisconsin tends to vote Democratic whereas central and northeastern rural Wisconsin tends to vote more Republican. Logistically it would be hard to get a good rural sample though whereas getting good samples out of Milwaukee and Madison is fairly easy.

[OnTheOtherHand]

188. sorry, took me a while to find this again...

Even sampling an urban area might be trickier than one might think. The pollsters define "geostrata" in each state, intended to pick up on salient regularities such as the one you mention (I know nothing about WI, just taking your word for it!), and then use a sampling method designed to cover the partisanship spectrum more or less uniformly. Their models look at deviations from past performance. So if some region or 'type' of precinct is out of trend relative to the others, they may not immediately get a good measure of what is going on, but they at least have some early indication that something is.

On average in 2004, their precinct samples were reasonably accurate, but from state to state, there were pretty big discrepancies in both directions -- i.e., the vote counts in the sampled precincts (plus attempts to model absentee vote, etc.) didn't yield good predictions of the state totals. (Apparently the biggest discrepancy was in Kansas: paging Thomas Frank!)* Whether those discrepancies were driven by rural areas, urban areas, or neither in particular, I don't know. I just don't want people to think that there is some slam-dunk explanation, based on the exit pollsters stupidly undersampling rural areas and not compensating for it: that's not how the exit polls work.

* Here I'm drawing on pp. 29-30 of the evaluation report.

[EFerrari]

25. K&R

[earth mom]

27. Just as we figured, though we didn't use statistics...

My hubby and I figured that the rethuglicans had stolen some of the votes, only it wasn't enough of a margin for them to win because they actually lost to an overwhelming majority! :evilgrin:

[mom cat]

32. Bingo! They misunderestimated their opposition!

[foo_bar]

43. another great moment in TIA "history"

This analysis uses the same mathematics we used to compute the probability of 15 or more JFK witnesses turning up dead within a one year period. It's the SAME problem, but in this case the time interval is even shorter - 4 months, making it even more improbable.

We seek to calculate the probability of at least 15 microbiologists dying UNNATURAL deaths within a 4 month period. The deaths were a combination of homicides, suicides, accidents and undetermined origin.

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=1005141

In the words of DUer Rabrrrrrr:

You can play around with numbers all you want and come up with all osrts of charts and tables, but, as any decent scientist, mathematician, or business analyst will tell you, unless those numbers have meaning, it's just so much useless numerical masturbation.

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=1005141

TIA doesn't know a margin of error from a hole in the wall, so his arguments became increasingly shrill and dependent on religious enforcers, and the rest is history (not unlike the story arc of the Bush administration).

[mom cat]

54. You should know about shrill.

[foo_bar]

92. "I know what you are, but what am I?"

I think you reinforced my argument.

[DemocratSinceBirth]

63. Yes...

http://synapse.princeton.edu/~sam/pollcalc.html

[LeftCoast]

79. Sam Wang, assistant professor of molecular biology ?

Why is it that these 'experts' are never PhD's in statistics?

[OnTheOtherHand]

114. actually the PhDs in statistics aren't always a bargain either

It's good to have some background in survey research -- which many stats PhDs do, but many do not.

Wang actually did a pretty good job: as I remember it, he articulated his assumptions and said even in advance why they might be wrong. See also foo's post, with Wang's retrospective self-critique.

[DemocratSinceBirth]

121. I Posted Wang's Results Because He Had It "Right" And "Wrong"

When he added his own "assumptions" to the polls he messed up...

[OnTheOtherHand]

122. yup, and even his wrong was pretty right

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%!

http://synapse.princeton.edu/~sam/pollcalc.html#final2004 (emphasis and typo in original)

If only TIA could get that far, it might be the first step toward recovery.

As a bonus, Wang offered this savvy insight the day after the election: "Much of my mail today concerns exit polls, with calls for analysis to check for widespread fraud. Think about the lessons you have learned here. A more plausible possibility is that exit polls themselves are biased, for instance by the identity of the questioner or the temperament of the respondent."

I'm not disagreeing with you in any way, just taking an extra moment to celebrate Sam Wang, a mensch. (I've never met him, but the record is clear.)

[slackmaster]

180. Dr. Wang is an honest scientist

The one point that stuck out to me was the inherent futility of using that kind of analysis to predict the future. The graph of poll results with time ticks for events like debates brought it all home. We're not talking about something that follows a rational trajectory, that can be modeled and predicted like an inert mass in motion, the subject is human behavior; worse yet self-reported human behavior (adding another layer of uncertainty to the measurements).

It's a like trying to predict what the housing market will do next week based on recent price trends. You might get it right, but there is no surefire method of forecasting it accurately every time. No serious analyst will stick his neck out and say in detail what the market will do in the future. They'll tell you general trends but won't try to attach precise numbers to it.

[OnTheOtherHand]

186. yes, this reminds me of one of TIA's greatest hits

He fit an nth-degree polynomial (I don't even remember what "n" was -- 3? 4? 5?) to Bush's approval ratings, and projected that they were bound to go down. If he had fitted an "n+1"th-degree polynomial, he would have come to the opposite conclusion. Either way, the exercise didn't make much sense.

It's edifying to compare Charles Franklin's work. Franklin is much more rigorous and cautious in measuring and characterizing past trends -- and doesn't claim to foresee the future.

[foo_bar]

93. Wang subsequently admitted he was wrong

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!

http://synapse.princeton.edu/~sam/pollcalc_letters_aftermath.html

Your model does assume that the polls were random samples (Wang's did not, although my understanding is that he assumed that overall there was no net bias). This is the problem I have with it. I did understand your assumption about undecideds - that was not the problem. Although I think that you should also have put a probability value on your assumptions being correct, as Wang did. But your model allows us to put in our own assumptions, which is fair enough. But it still assumes that the only error in the polls was sampling error. It does not allow the possiblity for bias in the polls.

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=403521&mesg_id=403997

[Joe Chi Minh]

206. Yeah. Same with the work of the insurance company actuaries. Absolutely

spot on! (not)

Here's a little something that will turn you purple with rage:

http://www.guardian.co.uk/comment/story/0,,1599503,00.html

I don't know... Those religous enforcers, eh? Just don't appreciate secular fundies!

[livvy]

46. Definite K&R . It does just keep getting more interesting.

I think they did "misunderestimate" this time, and there wasn't enough time to re-adjust the software to make up for it.
That is a lot of "lost" votes.
One question: what does PMarg and VMarg stand for? Is it poll or probability margin, and vote margin, or what? Sorry to ask, but having never taken statistics, I figure it's the only way to understand the numbers better.

[kster]

73. Just a thought maybe the rigging was

done prior to the Foley story. :shrug:

[texpatriot2004]

55. Well then, odds are votes were stolen, again. nm

[bridgit]

60. k&r'd, hon...

:hi:

[chill_wind]

64. K&R'd

[Zynx]

71. This happens in virtually every election.

Republicans always do better than polling indicates in Congressional races. Why is it? Probably because Democrats are harder to motivate in midterm elections.

[ShaneGR]

76. TIA is like Fox News, only for the left....

Will say anything to make his "math" work.

[0rganism]

99. Interesting comparison -- one single "crank" with a theory vs. a major network

How many millions of viewers do you suppose TIA has?

How many billions of advertising dollars flowed through his bank accounts?

How many issues does he cover in his 24/7 newscasts? Does he do sports and entertainment as well as sketchy statistical analysis of elections?

If TIA is like Fox News for the left, then we really are fucked.

[Cessna Invesco Palin]

118. My guess?

He's still just trying to explain away the errors in his ridiculous predictions for 2004.

[Name removed]

78. Deleted message

Message removed by moderator. Click here to review the message board rules.

[truedelphi]

89. I'm pretty much for mental health

And it sure feels easier to obtain since Nov 7th

[helderheid]

87. WOW!!!!

[Swamp Rat]

91. k&r

[tritsofme]

96. Cook and Rothenberg predicted about a 30 seat pickup before the election

That's what happened.

There is a 99.995487548^998989% chance that I believe their judgement and examination of polls over banned DU sacred cow TIA.

[Zynx]

100. Exactly.

My own forecasts based on averages of polling and then some guess work on other races came out to a forecast of roughly 29-31 seats with my final projection being 233 Democrats when it was all said and done. It looks more likely to be 232, but still, it was in the correct range. The same goes for the Senate and governorships. If anything, a few races surprised the other way with us knocking off Jim Leach and Jeb Bradley of Iowa and New Hampshire respectively. Those really came almost out of nowhere. Granted, some races we did worse than expected, particularly in Ohio, but we did better than expected elsewhere. Some results were weird like knocking off Fitzpatrick, but not Gerlach and knocking off Simmons but not Shays. However, results in general were as expected.

The most remarkable thing about these elections was how right the polling was in most cases. It was very bad in Maryland where neither race was that close, but uniformly was pretty good.

[Cocoa]

104. I'm not impressed by the mumbo jumbo

I'm not an expert in stats, though I'm literate in math and statistics, with an undergraduate degree in math with a concentration in stats.

My educated though non-expert opinion is that this is garbage.

[DemocratSinceBirth]

105. Here's The Final Round Of Polls

They were all over the place:


http://www.pollingreport.com/2006.htm

[Cessna Invesco Palin]

111. Interesting.

Those are exactly the odds that he'll ever be allowed to post his vitriol on DU again.

[DemocratSinceBirth]

113. :-(

:-(

[OnTheOtherHand]

115. does getting other people to post it for him count?

Obviously the OP isn't vitriolic, but I've seen some others that have some 'tude.

[Perky]

123. Oh My God you folks are obsessed!!!!!

You actually are contending that there is a nationwide fraud involving what obviously must be at lesst 100,000 GOP operatives keeping the most extraordinary secret in US History...I suppose they have a secret handshake and decoder wings too...


Look I have no doubt some fraud occurred....It always does.


But the generic vote is a national sample that does not consider issues like gerrymandering and party strength in certain generic areas. As the seats become safer the total votes cast in a district decrease.

[DemocratSinceBirth]

124. Who Are You To Dispute 76,000,000,000 To 1 Odds?

-:)

[slackmaster]

181. The "Massive army of knowing conspirators" theory is just one version

There are also the "Lone corrupt software developer who controls everything", and the two-level "Dozen or so insiders and vast army of unwitting dupes" versions.

[Skinner]

126. This analysis is an embarrassment.

To those of you who keep demanding to see the problem, here it is:

The problem is not in the mathematics (although I have not checked the math, so it's possible that there are errors there, too). The problem is in the assumptions he used before he even started.

TIA assumes that the "generic poll" should match the recorded votes. This is, quite simply, WRONG WRONG WRONG WRONG WRONG.

Does everyone here know what the "generic poll" is? The generic poll (usually referred to as the "generic congressional ballot") asks respondents which political party they support in the upcoming congressional election -- but it does not provide any names of any candidates. The generic congressional ballot is not -- and was never intended to be -- an accurate prediction of how people will vote. The point of the generic congressional ballot is to get a general sense of the mood of the voters.

Think, people. THINK.

Most American voters are not political junkies or activists. Even during an election season they could not tell you for certain the name of their member of Congress. When they get a call asking them their preference between a generic Democratic candidate and a generic Republican candidate, they will simply respond based on party. A large proportion of voters would be unable to think of the name of the candidate that represents each party in their upcoming congressional election.

But what happens when those same voters enter the voting booth on election day? They are presented with a list of NAMES OF REAL PEOPLE, their candidates, listed by office and political party. And they remember what they like and dislike about all of these people. They might have expressed support for a particular party, and they might hold that generic opinion of that party, but they like their own member of Congress and don't particularly care what party he or she is in. Or they might remember the ads about how a particular candidate was a crook or sexual deviant, and now that they finally face the realization he or she is running in their congressional district, they cannot possibly vote for him or her regardless of what party he represents.

Bottom line:

You can do the best, most accurate, most awesome mathematics in the history of the world, but if you start with completely false assumptions, your "analysis" is going to be worthless.

ON EDIT: If you want to see how the outcome compares to the pre-election polls, you need to look at the pre-election polls that list candidates by name from each and every congressional district. Someone mentione up-thread that this is precisely what folks like Charlie Cook and Stu Rothenberg did before the election, and their predictions were quite accurate.

[ProfessorGAC]

134. Absolutely

I wrote my post (just below yours) just before i saw yours. I've made an attempt, without performing the complete reanalysis, to determine the impact of the inflation caused by those unsupportable assumptions.

My conclusions are the same as yours.
The Professor

[slackmaster]

162. Skinner, this is one of the best posts I have ever seen on DU

Thank you, and have a great Thanksgiving!

:toast:

[RoyGBiv]

217. I just want to say thank you ...

The mere ability to critique these conclusions has, in the past, been hampered by the almost taboo nature of offering any criticism of a pre-determined conclusion, to wit that fraud has taken place.

Before the 2004 election, even the mere suggestion that the methodology being used in cases like these was flawed drew me such a sound thrashing that I decided it simply wasn't worth the pain. And, yes, I know you weren't commenting on the larger question to which I just alluded, but having you voice added to this in this context at least allows the question to be asked.

[Bill McBlueState]

225. so here's my question

Why does someone who, for whatever reason, got banned from DU get to use his friends as mouthpieces to continue to embarrass the website?

[Mr_Spock]

237. It is a topic worth discussing

I am not for censorship - all ideas should be heard and discussed (within the rules), but repeatedly quoting a source that has shown little respect for the rules at DU (attacked those who would question the veracity of his conclusions) is possibly an issue for the administrators to consider. I also question the validity of the analysis which is believed without question by altogether too many folks. It is embarrassing if the assumptions made turn out to be invalid. It is quite possible to have ones conclusions eventually proved correct, but still be wrong in the assumptions/analysis that lead to those conclusions.

[ProfessorGAC]

133. Enormous And Unsupportable Assumptions Abound

TIA has made some egregious leaps in logic through a series of unsupportable assumptions. Since these assumptions all compound the distribution upon which the final conclusions are based, there is no telling how inflated these final probabilities might be. My GUESS, without doing a complete assumption free analysis, (only an educated guess), is that it's inflated by at least 10^5th, and that's probably underestimating the inflation impact of the assumptions.
The Professor

[Dick Diver]

145. You're correct, Professor. Those "leaps in logic" include

(1) The assumption that a "best fit" linear progression can predict predict outcomes within a complex social ecosystem. It would seem that both Malthus and the stock market would have dissuaded us of this long ago. This is not a purely qualitative objection, as this analysis forms the basis for TIA's base Democratic and Republican percentages, as well as the trend increase factors and presumably the UVA allocation.

(2) The use of a MoE of 1.5% when that MoE of the underlying data was much higher. In scanning the first ten or so generic polls referenced in the thread, I found MoEs ranging from +/- 3 to +/- 4. TIA's highest MoE as published here is 3. I would suggest a MoE of +/- 3.5 would be a much more reasonable assumption. It's incredible that one would attempt to base an analysis on a MoE that is significantly less than that of the base data.

If one looks at the impact of two simple changes of assumptions, allocating the UVA 50%/50% and using a MoE of 3.5, the impact on the result is significant.

Moreover, my analysis is simply from the mechanical side. Skinner's post noting the function and usage of generic polls is equally applicable from a different direction.

[ProfessorGAC]

146. You And Me, Dick. Let's Go To The Geek Club

Math geeks irritate me. And, i'm sure of it every morning when i look in the mirror.
The Professor

[Dick Diver]

150. Sorry, Professor. When I wrote my post

I hadn't seen your post #137, which says the same thing as my subsequent one, only in technically more accurate language (my BS is in Mathematics, but my MS is in Computer Sciences).

[natrat]

144. PELOSI GET RID OF THE MACMINES

[Name removed]

158. Deleted message

Message removed by moderator. 
[link:www.democraticunderground.com/forums/rules.html|Click
here] to review the message board rules.

[DemocratSinceBirth]

160. We Can Settle This Easily And Fairly....

Why doesn't TIA submit his work to a peer reviewed journal:


http://www.apsanet.org/


http://www.asanet.org/index.ww

That's what credible social scientists do...

[slackmaster]

165. YES!

:toast:

[DemocratSinceBirth]

176. Thank You

eom

[Gormy Cuss]

169. Or to the journals of AAPOR or the American Statistical Association, etc.

Excellent suggestion.

[rinsd]

163. The lunacy continues with the help of water carriers

Just go back to PI and be done with it.

[DemocratSinceBirth]

164. My Suggestion Is Eminently Reasonable...

.

[rinsd]

167. Yes, it was

I, on the other hand, am not being reasonable and I am well aware of that.

:toast:

Have a great Thanksgiving!

[OnTheOtherHand]

187. 'tis a consummation devoutly to be wish'd n/t

[Cessna Invesco Palin]

170. You know what?

This reads like The Naked Lunch if it were written by a masochist with a statistics fetish.

And speaking as a masochist with a statistics fetish, I find that highly offensive.

[savemefromdumbya]

166. Gore should still be President

[DaveinMD]

168. this is all bs

the generic vote doesn't take into account races where there was no challenger. Many more Dem seats went unchallenged. These seats are normally 80 percent and higher with a nominal challenger. And generic polls always close at the end as disgruntled partisans come home.

[CJCRANE]

179. This reminds me of Dan Rather's "Memogate".

Pointless arguments about technical details distracting from the issue.

[mom cat]

202. Precicely.

[Joe Chi Minh]

207. And they wonder why we pillory them!

[OnTheOtherHand]

221. well, I've always wondered

why is it a good thing to make bad arguments, and a bad thing to make good arguments?

why is it a good thing to undermine our credibility, and a bad thing to defend it?

[Joe Chi Minh]

228. I'm sure you have, GV.

However, policywise, it's not at all a good idea. And don't let anyone persuade you it is. I don't think you should even begin to entertain the notion that its a good thing to make bad arguments, and a bad thing to make good arguments. A good thing to undermine credibility and a bad thing to defend it. I feared you'd been lead astray by the neocons, and this just confirms it.

My advice to you, is, get as far away from the neocon ethos as you can.

[OnTheOtherHand]

230. I'm glad to see you say that

Many DUers seem to be under the impression that TIA should be supported no matter how awful his arguments are, because he 'raises important questions.'

I guess you and I agree that no matter what 'questions' he 'raises,' if his arguments are poor, they should be criticized. Excellent.

Unfortunately, you seem to have no aptitude for or interest in assessing the arguments.

[Kip Humphrey]

191. Hmm...a 5.1% decline in Democratic vote share. Reminds me of my pre-election prediction

that there would be a 5% vote flip to Republicans (10% shift). What Rove didn't anticipate was that Democrats would draw more than a 10% margin differential and consequently overwhelm the vote flip and win the Congress.

[bigriver]

203. Whoever TIA is, he's pushing garbage.

His statistical acumen doesn't even compare to your random poster at any baseball discussion board. The man's clueless about political polling.

As in all bad (and potentially fraudulent) studies, garbage data in, garbage data out.

[Cleita]

212. He accurately showed that both the 2000 and 2004 elections

were rigged. Other auditors not in anyway connected to him have proved him right months and even years after the fact. In other words, he has a track record.

The only garbage here is your post.

[bigriver]

214. And I should trust your words why?

"Other auditors not in anyway connected to him have proved him right months and even years after the fact."

Garbage confirming garbage is still garbage.

How come this fraud can't defend himself on this site and lets his groupies do the dirty work?

[T Town Jake]

215. The reason his sycophants have to post his bilge is outlined here:

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=391223&mesg_id=391545

The GREAT ONE doesn't play well with others, so his fan club has to do the actual posting for him here at DU....

[Cleita]

218. Curious.

I'm sort of wondering what your dog in this fight is?

You accuse him of a lot of garbage in your original post and yet you don't post any figures, charts, pie charts, anything to back up your accusations or anything to refute his figures?

[OnTheOtherHand]

222. no numbers or pie charts are necessary

Why don't you read the thread, note the substantive criticisms of TIA's arguments, and explain to us why they are all wrong? TIA was never able to do it.

If you think you have something to add to the discussion, by all means bring it on.

[Cleita]

224. Oh because some other mathematical geniuses have come

along with accusations, it makes them right and TIA wrong? What I am seeing is a lot of sniping and jealousy here. TIA took the trouble to shine a light on something we are all concerned about. Now, I'm not a math genius by any means, but I do understand that extrapolating statistics is open to margins of error.

Since I have been following TIA's work since the beginning, most of his statistical analysis have proven mostly right when various organizations, not connected to him, went in and recounted the Florida votes of 2000 and the Ohio votes and other malfaesance of 2004. So excuse me if I take the work of someone who has a track record over others who come along and say he's wrong but can't put up a study of their own to back up their allegations.

[OnTheOtherHand]

231. no, you are seeing a lot of ridicule

and it has very little to do with mathematics.

"...most of his statistical analysis have proven mostly right..."

No, they haven't. In fact, that conclusion could hardly survive a close reading of this comment thread. Sorry, but you have been misled.

If you aren't interested in considering the critical comments because they aren't "a study," then so be it. You'll find a bunch of my studies here, if you are curious.

[Dick Diver]

232. This post is incorrect on so many levels.

First, TIA is proposing a hypothesis that contradicts publicly-available results and data. Therefore, it is incumbent upon him to "put up a study of (his) own to back up (his) allegations." Others can simply point out the errors in his assumptions and reasoning (and many in this thread have) in order to invalidate his hypothesis. It is NOT incumbent upon others to "put up a study."

Second, I don't know that TIA ever did any significant statistical analysis of the 2000 election (except possibly in retrospect). I don't see any on his website. Certainly, I came to know his posts on this board as they related to the 2004 elections (since this board didn't even exist during the 2000 elections). You, however, seem to have been following his posts much longer than I, so if you could enlighten the rest of us, it would be much appreciated.

Third, it's a sad fact that no peer-reviewed journal will touch TIA's work, since it is essentially technically flawed and driven by partisanship (or maybe they're afraid the BFEE will assassinate them if they do :sarcasm: ). He is simply an amateur that professionals do not take seriously. Case in point, he happily calculated (and presumably acquiesed in making public) the odds that Lamont would get the exact number of votes in CT as Lieberman's 2000 opponent. This was based upon faulty information provided to him by the "journalist" autorank, who didn't bother to check readily-available primary sources. So regardless of the "garbage in" TIA, in his partisan zeal, was happy to supply "garbage out."

Fourth, in regards to "recounts" validating his projections, why don't you ask Ida Briggs about the recount in NH?

Fifth and finally, didn't you (and your calculator) recently suggest that if we set aside $300,000,000 dollars we could give every man, woman, and child in the country a cool million (or the interest on such)? Do you do TIA's calculations for him? Having looked at his results over the last couple of years, I wouldn't be surprised.

[Cleita]

233. Thanks for bringing up my calculator again.

How dare you!

I thanked you for the correction but that wasn't enough for you. I told you I had a problem with large figures like billions and trillions but you just had to make fun of me and drag me through that over and over again. I guess it really makes you feel like some bad ass big shot doesn't it? I don't care though because what I say isn't important.

But to trash someone through my post who has worked really hard over the years to try to shed some light on the really murky goings on with our elections is beyond shitty IMHO.

If any of you "brilliant" academics had anything important to say, why didn't you put up your own post? Then people could compare. Coming in and poopooing someone elses work while you don't put up your findings I find really, really deserving of everyone's contempt.

[Dick Diver]

234. Fine, then. Since you choose to participate in the conversation

so vehemently, please consider the following.

If we allow TIA's assumption that he can use generic polls as the basis for his analysis (an assumption that most real researchers would NOT allow)-- even given this -- his results are irrelevant if he increases his MoE to +/- 3.5%. This number would approximate the average of the polls he uses as the basis for his analysis, as their MoEs typically range from +/- 3% to +/- 4%.

It's TIA's contention, however, that he can use a MoE of 1.5% (or less) based upon the assumption that he can combine the sample sizes from 10 or more of these generic polls. Now keep in mind, these polls were conducted at different times, with different question groupings, by different organizations, with different party weightings, etc., etc., etc. Regardless, TIA states that by combining their sample sizes he gets a sample size of 10000+, thereby deriving a MoE of less than 1.5%.

What's your opinion of his hypothesis?

And btw, you didn't address the other four points of my original post.