Regarding Math Degrees, Legitimacy and Transparency

[scottxyz]

Several people (most particularly, Euler) have been criticizing TruthIsAll for only providing the TYPE of math degrees he/she holds - not the institutions or the years.

This is barking up the wrong tree. I can see that from TruthIsAll's analysis that he/she is good at math. But this is still not the issue: You shouldn't have to be a math genius to believe that the election results are valid.

What we are dealing with here should not be terribly complicated math:

(1) Basically, it's ADDITION: Add up the votes.

(2) Aside from that, it's CHECKING YOUR WORK: See if the "official" vote totals agree with the exit polls.

(3) Maybe throw in a little PROBABILITY: How likely is it that the thousands of "errors" in the voting system ALL FAVORED BUSH.

You don't need a degree in higher mathematics to be able to answer these questions. Anyone can add (and so can Diebold ATMs - but for some reason, the voting machines made by Diebold and other vendors can't), anyone can check their work (but for some reason, our voting machine results don't check against our exit poll results), and anyone who's flipped a coin a few times knows what randomness looks like (but for some reason, our voting machines only make mistakes that favor Republicans).

You don't need a higher degree in mathematics to see any of this. The American voting system is dysfunctional: it can't add; the results don't agree with the exit polls (which was {one} reason the Ukraine's results got thrown out); and its errors always favor one party.

You don't need a degree in mathematics to see any of this. Nor should you - a properly designed voting system should be transparent enough that its legitimacy is apparent to ALL VOTERS, without needing explanations from mathematicians.

(Edited per suggestion from post 2 below.)

[DoYouEverWonder]

1. If you can balance your bank account

you should be able to add up a box of votes and get the same number twice. This is not rocket science.

[SicTransit]

4. Really? In Ohio on some ballots voters had more than 100 choices -

that is, there were more than 100 simultaneous races.

Let's see how hand-counting paper ballots would work. Here is a calculation. 100 things on the ballot. 1000 ballots in the precinct. If each choice takes 10 seconds to hand-count and record, and obviously that needs to be observed at least by 2 more people from the major parties (if not more) that would be 3*10*100*1000 seconds or 833 man-hours. Let's say there are 3 teams of counter+2observers per precinct. That would mean that it would take 92 hours to count all the votes. That is 12 working days - or 2.5 weeks of hugely repetitive monotonous labor for 9 people - and that is only in one precinct. If you pay these people $10/hr (which would be slave wages for this type of work) - that's $8.5K (observers would probably have to be paid by their respective parties). Multiply that by 11,000 precincts in the state of Ohio.

You can see that it is a bit more than "balancing your checkbook".

[DoYouEverWonder]

8. I said nothing about hand counting anything?

I am not opposed to using computers and tabulators. But I could write and Excel Spreadsheet in a couple of hours that could do the job. Please give me a break.

[SicTransit]

9. You could write an Excel spreadsheet that

reads in paper ballots, whether optical or punchcards? Funny - never seen an Excel spreadsheet with a card reader attached. Have you?

[DoYouEverWonder]

16. I meant to tabulate the votes

I would still use optical scanners to read the ballots.

They're cheap, easy and reliable and you still have a paper ballot.

However Diebold's election systems are written in MS Access, which is basically an overcomplicated crappy interface for integrating Excel spreadsheets. The big joke is they charge millions of dollars for their crap programs that are totally unreliable.

[SicTransit]

18. MS Access is unreliable?

Seriously?

Please show me how you can write a program in MS Access that "unreliably" adds up a column of numbers.

It is not the programs that are unreliable. It is the physical part of counting the ballots - the incorrectly filled in ovals, the overvotes, the undetached chads, etc.

[DoYouEverWonder]

29. Unreliable, in regards to security

MS Access and the Diebold election software is very easy to hack into.

Plus there are numerous reports of 'glitches' and program failures in many counties all over the country during this past election.

[mgr]

31. Problem wasn't with access but

the fact that it was on a Microsoft platform that if networked could be hacked from the outside.

The other concern was that the program itself could potentially add or subtract votes in a manner that voter intent was altered.

This was the basis of the blackbox voting issue that is still being vetted and addressed. I have not been keeping up for several weeks on this issue.

Mike

[scottxyz]

68. MS-Access is not secure, because it's not auditable

Counting a million physical pieces of paper in front of witnesses and TV cameras is almost impossible to rig - while counting invisible computer bits and bytes on memory cards and hard drives and modems can be rigged in a flash without anyone knowing it. Don't trust people like SicTransit who try to tell you that paper is less reliable than bytes and bits.

And Microsoft Access is not a serious database - it is only used for small applications where security is not an issue - never for large applications that need to be secure and auditable. Details below:

There are basically two main types of database software - "desktop" databases (such as Microsoft Access or "MS-Access", or Paradox) and "client-server" databases (such as Oracle, DB2, Microsoft SQL Server or "MS-SQL").

Client-server databases have important auditing and security features which desktop databases like Access lack - for example, "triggers", "transaction logs" and "mirroring". These client-server features allow you to independently track changes to the database, which is important for auditing and security purposes.

Because MS-Access lacks these features, it is only suitable for use in small settings (such as a single office) where all users pretty much know and trust each other. For large-scale applications, such as banking or airline reservations, where there may be millions of transactions and where security and auditability are important, client-server databases are always used - never MS-Access.

It is very strange that Access is used for tallying votes. An Access program could quite easily "unreliably" add up a column of numbers - if it were written with oodles of mysterious, unnecessary add-ons and "back-doors" the way Diebold's vote-tallying software (GEMS) was, or if it were left on a computer attached to a dial-up connection, the way GEMS is. With GEMS, it has been demonstrated is possible to remotely and untraceably alter the vote tallies. I say "untraceably" because GEMS (written in Access), and the Windows operating system it runs on, have minimal, easily hackable auditing and security systems.

Diebold's GEMS, which is supposed to just "add up a column of numbers" in Access, is a complex morass of confusing code, including extra copies of the tables being added, and needless add-ins which complicate the program and could be used for tampering.

SicTransit claims that "the physical counting of ballots - the incorrectly filled-in ovals, the overvotes, the undetached chads" are what is "unreliable" - but do not be fooled by SicTransit's straw man here. The technology for counting pieces of paper is hundreds of years old, and its reliability is quite obvious to everyone - and being "obvious to everyone" should be an important feature of any voting system. It is quite easy to implement a system of counting paper ballots which is secure and reliable - and which is OBVIOUSLY secure and reliable to all participants: simply do things in the presence of additional witnesses and/or videotape cameras.

[pauldp]

10. Hey Dude

Banks have machines that count and sort and make microfiche copies of all types of different documents, checks etc., and they intelligently sort them at 30 documents PER SECOND. Diebold buys these fucking machines and they can't give us a paper trail? The microfiche can be randomly spot checked for accuracy. We could verifiably count multiple types of paper ballots securely in a matter of hours if we wanted to.

[SicTransit]

15. You could? Ok - why don't you start a

company that makes optican ballot readers/tabulators, show that they are more reliable than the existing ones and corner the market? In addition to saving democracy, you will make a bundle. Oh - make sure that you can reliably read filled in ovals, checkmarks, Xs and ballots where the voter both filled in the oval and wrote in the candidate, chads that were only detached at two corners, etc. etc. etc.

[pauldp]

20. Its already there.

The machines I was talking about CAN DO THAT. BankTech is the company that makes them. They use neural nets to intelligently sort. How do you think Banks function? By hand checking all of the different documents they receive?

[SicTransit]

25. No, it is not "already there" -

BankTech is not available to any BOE as an alternative to the current electoral machine technology. Why don't you start a company that would make BankTech-based election machines?

I see a lot of talk about unreliable machines and how they can be made better. I don't see anyone actually doing anything. Be the first one.

[pauldp]

48. Yes it is.

As I said Diebold has these machines.
I actually know activists working on this.
The problem is a thing called politics.




(new subject on edit)

[scottxyz]

69. SicTransit is lying - or, more charitably, uninformed

Banks routinely process millions of paper checks using machine technoogy. There is no need to invent these. What is needed is for these type of existing machines to be used in our voting systems, that is all.

SicTransit is lying - or, more charitably, uninformed.

Most of us already use Diebold ATMs - when was the last time a Diebold ATM lost or forgot a bank transaction you made?

What's going on here is that the reliability of ATM transactions is IMPORTANT - to you, to the bank. This is why Diebold ATMs work.

The UNRELIABILITY of vote-tallying must be important to someone - which is why Diebold vote-tallying machines DON'T work.

[Karenca]

72. very charitable of you...

and now,
sic is cold in his tomb.

[scottxyz]

73. I didn't think he'd last long

His lies were so transparent.

[davidgmills]

21. Two Ballots -- One National, one State -- Problem solved

Laws for both and timetables for both.

I thought I mentioned this to you before.

Perhaps it was someone else.

[SicTransit]

27. Once that is done, you can handcount national ballots and

machine-count the state ballots - so the problem will be solved for Presidential race and remain for all others. Another problem is separating National from State - officially, every race is a State race. There is no such thing as a National race. Even with President, you are not electing a President, you are electing a local slate of electors. But then, it's a formality.

That is one way for election reform - though partial. Unfortunately, if tomorrow any Senator stands and objects, that issue will be effectively poisoned for the next 4 years at least and nothing will be done about it.

[davidgmills]

62. Electors are a State Race

But the federal government has an interest in the outcome. The constitution gurantees the right to it. Congress has a right to regulate how it is done and it is time that it should.

As for the poison, your opinion -- not mine. Something that has never happened in a hundred years focuses lots of attention on the matter that it would not otherwise get.

This issue is a balance of power issue as well as an election issue. The question is whether Congress will abdicate its duty to insure the executive is fairly elected because obviously, the judical branch can't. It's a question not just of politics, but checks and balances.

[thanatonautos]

65. Don't agree with your calculations, but in any case Ohio took til Dec 13th

to fully count and certify its vote ... how many days
is that, now?

As for local issues: they can be handled separately;
split off into separate ballots from the national and
statewide elections. Further, it is by no means true
that all Ohio counties had as many as 100 issues on the
ballot. Further, the time can be considerably reduced: your
factor of 3 assumes one must wait for all three observers
to agree on each and every vote. That's unlikely ... the
ballots can be passed from the counter to the observers, who
then verify, more or less at the same time and record their
objections separately, if any. That reduces the time
considerably. Ballots on which there are objections can
be separated and evaluated later. Most will have none
with a good ballot design. This will reduce the time
for counting considerably from 10 seconds. One person
calls the result, one records it. It will take no more
than 2-3 seconds on an undisputed ballot.

That's at least a factor of 10, so far.

Further, many countries manage both local and national
elections in this way, with no undue delay in presenting
the results. Canada manages general elections in this
way, and produces results within 24 hours.

This suggests many flaws exist in your reasoning.

Let's see how hand-counting paper ballots would work. Here is a calculation. 100 things on the ballot. 1000 ballots in the precinct. If each choice takes 10 seconds to hand-count and record, and obviously that needs to be observed at least by 2 more people from the major parties (if not more) that would be 3*10*100*1000 seconds or 833 man-hours. Let's say there are 3 teams of counter+2observers per precinct. That would mean that it would take 92 hours to count all the votes. That is 12 working days - or 2.5 weeks of hugely repetitive monotonous labor for 9 people - and that is only in one precinct. If you pay these people $10/hr (which would be slave wages for this type of work) - that's $8.5K (observers would probably have to be paid by their respective parties). Multiply that by 11,000 precincts in the state of Ohio.

Good try, but not very impressive. Where is your degree in math
from?

[SicTransit]

2. on exit polls and Ukraine -

no, exit polls were not "the whole reason the Ukraine's results got thrown out". Exit polls in Ukraine were a tiny corroborating corner in a mountain of hard proof of election fraud:

http://www.telegraph.co.uk/news/main.jhtml?xml=/news/2004/11/28/wukra28.xml

[scottxyz]

3. That's fair -- I will edit original post

Will change "the whole reason" to "one reason". Thank you.

[papau]

6. hard proof of election fraud in Ukraine is "only 30000 reports of problem

per GOP/Media in the US.

I love the words like HARD PROOF that have a different meaning for difference elections!

[SicTransit]

7. Find me something close to this in US elections:

"They started to beat voters and election officials, trying to push through towards the ballot boxes," he told The Telegraph.

"People's faces were cut from blows to the head. There was blood all over."

"Maya Syta, a journalist working at polling station 73 in a Kiev suburb, witnessed ballot papers destroyed with acid poured into a ballot box. "The officials were taking them out of the box and they couldn't understand why they were wet," she said.

"Then I saw they started to blacken and disintegrate as if they were burning. Two ballots were wrapped up into a tube with a yellow liquid inside. After a few moments they were completely eaten up."

"The most common trick was "carousel" voting, in which busloads of Yanukovich supporters simply drove from one polling station to another casting multiple false absentee ballots.

In another brazen fraud recorded by observers from the Organisation for Security and Co-operation in Europe, voters were given pens filled with ink that disappeared, leaving ballots unmarked and invalid."

[pauldp]

12. The Death of HR 2239

Murdered the fairness of our elections

[pauldp]

13. Old African American Seniors stood for hours in the freezing rain. n/t

[pauldp]

The Waren County "Terrorist Threat"

stinking bullshit.

[papau]

19. So we need physical violence before there is election fraud? - well

in Illinois I witness beating of blacks by GOP folks as they were going to the polls in prior elections.

In this election I have not looked for such reports in the media - but then back then the media had no such reports either.

Ballots being destroyed - yep saw that in the old days - and we have video tapes of folks throwing out ballots so Harris and crowd could not review.

Now "carousel" voting, in which busloads of Yanukovich supporters simply drove from one polling station to another casting multiple false absentee ballots, is replaced in the US with absentee votes included from people who did not vote absentee - we had the 2000 Florida problem - and now we have the problem of more absentee ballots being received than were requested (a 4 to 1 ratio giving the GOP a nice boost)

As to "voters were given pens filled with ink that disappeared, leaving ballots unmarked and invalid" we in the US use the invalid ballot trick - see the Illinois 1963 Supreme Count case on random pencil marks meaning a confused voter and ballot tossed - those tossed ballots were Dem ballots. This time round we have indications of sorting of ballots by who got the presidential vote, followed by pulling those "Kerry"cards and replacement by "Bush" cards, albeit we have no confessions yet. But then we have some good mathematicians in the US that will say that the above is the only way to explain certain reported vote.

Add in the default to Bush BBV fact - and the BBV tech's changing programs and recorded votes with no legal authority to do so - and I suspect Conyers speech will be interesting.

[SicTransit]

22. your post proves my point -

all you posted was conjecture. Not one piece of hard evidence. Unlike in Ukraine.

[papau]

26. Again you find math not hard evidence-I suggest avoiding a scientific

career -

indeed accounting and auditing and finance may be not to your liking.

Conjecture does not equate to the Harris/Allen/and all the other folks observations.

I wish the GOP had ethics - indeed almost as much as you refuse to believe that they have a major problem with ethics!

peace

:-)

[SicTransit]

30. Too late - I am the head of a software company with a degree

in Math and Computer Sciences.

"math" is an all-inclusive word. 2+2=4 is concrete. Probabilities are less concrete. Statistics even less. Probabilities and statistics when the underlying data is spotty, involves people's opinions etc - even less so.

[tx_dem41]

32. SicTransit...has any of the analysis used non-whole numbers as

raw data? All I have seen is raw data that is rounded-off integers.

[SicTransit]

33. I am not sure what you mean by that -

the underlying raw data is usually whole numbers - since people are integers :)

If you mean the underlying percentages - that would not be "raw data", that would be at least one level above.

[tx_dem41]

34. Point taken....you are obviously a computer guy ...lol..

I mean the underlying percentages. This is involving exit poll analysis.

[papau]

46. A Great degree - Math and Computer Sciences - many schools have

had a problem with what to call such study.

In pre-history (mine) MIT used the usual EE degree, then my Kids saw pure Computer Sciences degrees - and Applied Math Deprees. Should be interesting what the grandkids find 5 years from now (God - I am getting old!)

In any case I agree with your point as to quality of data affecting any conclusion, but I know some of the folks doing the math - and I know of no data problems other just the roadblocks put in front of them by GOPer's trying to hide data.

As for heading a software company - that is a great job - my sister and her Husband owned/ran a company - an HR interactive training multi-media operation using the Plato lanuage -for many many years -

just remember to sell when a good offer comes buy - 'cause changing tech makes unique products no longer unique with unreal speed. Luckily the gig ran until he was of age for Soc Security - and he had kept some nice reserve funds on the books - but the world changed in 24 months for their little company and 50 workers went to zero very quickly.

[SicTransit]

49. Wow PLATO - I actually was on PLATO in the 80s at UIUC -

was a programmer there, surprised anyone remembers. Tell your sister that there is a Plato system running on a Mac somewhere that you can hook up to - go to www.cyber1.org and sign up. It is an exact emulation, and you can run avatar and empire, as well as other stuff.

[papau]

59. Cool - I will :-) By the way Paul invented the Plato lanuage - he will

be please to know that it lives!

[SicTransit]

61. Paul Tenczar is still around?

Yes, certainly send him the link to cyber1.org - and yes, he has a special place in every Platoite's heart.

[papau]

66. Good Grief - Paul is not "old" -- - spent Xmas with him and his family in

Florida at my place in Tampa.

So I'd say Paul Tenczar is still around!

:-)

[Karenca]

54. Oh..What a coincidence! So am I :))

I have my PhD. in Applied Math. :))
and, I teach this course at Columbia University.
and I'm not using my knowledge to twist the truth. :hi:

[papau]

60. Good to know - I miss NYC - I expect it is a great place to teach!

:-)

[euler]

45. This reminds me of another thing I want to rant about,

TIA and Stephen Freeman and a bunch of other writers have stated over and over that exit polls are inherently accurate. But to make their proof, they point to where - Germany, Utah, Europe ?

They state that exit polls are always or almost always accurate. This implies that the exit polls for past US presidential elections are accurate.

Ok, so you are righting a paper questioning the results in a US PRESIDENTIAL ELECTION, and you decide you need some examples that support your claim that exit polls are accurate, so where do you look for such examples ? How about the 1972, 1976, 1980, 1984, 1988, 1992, 1996 and 2000 US PRESIDENTIAL ELECTION exit polls ? Makes sense right ? All exit polls are accurate then these past US PRESIDENTIAL ELECTION exit polls are accurate. These are the ideal examples to use right ? Apparently not.

There is a good reason TIA and Freeman go to Germany to make their case. EXIT POLLS FOR PAST US PRESIDENTIAL ELECTIONS HAVE BEEN ANYTHING BUT ACCURATE. IN FACT, THEY HAVE NEARLY ALL SHOWN DEMOCRATIC BIAS. IN OTHER WORDS, AFTER RE_WEIGHTING WITH THE ACTUAL VOTE, A MAJORITY OF STATES SHOWED A SIGNIFICANT 'RED SHIFT.'

All this can be verified on Mystery Pollster.

[papau]

50. recent improved abilities by GOP in cheating is reflected in history

of exit polls.

But the better than regular polls and very accurate claim comes from 15000 folks being sampled rather than a few hundred, along with the removal of the poll problem of finding the "likely voter".

As to you EU question, the EU countries usually have 2 and sometimes 3 exit poll companies on an election day, and the EU media includes Union owned outlets that prevent corporate media story selection problems.

[SicTransit]

51. As you very well know, sample size does not really

matter if your sample is not representative. Since only some precincts are sampled, a representative sample is extremely hard to achieve, due to "clustering effect". That is why the data is massaged, adjusted, brought in line with actual results and cannot be used to prove fraud - as per Mitofsky, who is the father of exit polling.

[papau]

55. I assumption that the sample was not representive is not proven.

Granted any stratified approach has hurdles, but I do not see why we have become so bad at overcoming those hurdles.

I reject the shy GOP voter idea, and wait for some proof of non-representative precincts.

The final massage - when expressed in words - does not make logical sense as folks must change sex - and indeed their actual vote. And Hispanic Bush 44% is rejected by exit polls done by Hispanic groups that show no change from 2000 results.

Granted the final adjustment is SOP, but this time IMHO it is too much of a "required" adjustment

[scottxyz]

70. Disenfranchisement is disenfranchisement, SicTransit

Does SicTransit think disenfranchisement only counts when it involves clubbing someone over the head - not when it involves more subtle means?

In the Ukraine, you may get clubbed over the head when you try to vote.

In America, where we've perfected the art of deniability and high-tech Jim Crow, we are more subtle - officials may lose your voter registration card or illegally bounce you from the voter rolls, officials may make sure the machines "break", officials might even plant backdoors in the software.

Be patient, SicTransit. Someday the Ukraine may have a Karl Rove and a Diebold and Checkpoint of its own, and then they can steal elections and leave nary a trail, like they do in the US.

[understandinglife]

5. You also don't need to be an expert in category theory to define patterns.

[mgr]

11. Guys remember Dan Rather-gate

Something about the memo on Bush's guard duty being a fake, but that everyone was in agreement that he shirked. Look at the naysayer's concern this way: Any poorly thought through election fraud argument will be taken to refute the mountain.

The issue, I will say again, is that the analysis comparing exit polling and the actual results has been refuted. The there is specific evidence from the recount done in Washington, and we know that TIA was working with adjusted numbers to begin with. It has no legs, and it is harmful to our efforts. It is a fruitful hypothesis in that it led others to look at election irregularities, and for possible fraud. Think of it as Wittgenstein's laddar that you need to now kick away.

Mike

[understandinglife]

17. You are incorrect about WA State, and more, and bye (n/t)

[mgr]

23. Really? report this week from the recount

Basically stated that actual outcomes were below those reported by the exit polls. Are you saying that a hand recount was in error?, particularly when they were discovering absentee ballots favoring Kerry? Your tenacious hold onto a refuted position is astondingly close minded and emotional.

Give me a little more incisive analysis please? Or is it that all bloggers cannot construct an argument that requires more than two clauses?

Mike

[euler]

40. THIS is the whole point.

Until people understand this, people will never understand why MSM is ignoring us.

MSM knows that any argument resting on exit polls is simply not an argument at all. They hire experts too. In fact, I believe MSM has not ignored this stuff. They checked it out with experts, found out it has no legs, and rejected it. This is not the same thing as ignoring.

You can't listen to TIA or Stephen Freeman. Having a PhD in one field does not make you an expert in another. You can only listen to people who are exit poll experts. If they say this math is crap, it's crap. There is no higher authority to appeal to.

The other side will trot these experts out one after the other and they will destroy the exit poll thesis. Let's hope it doesn't get that far.

[papau]

52. Dan Rather-gate - there is no proof that memo was fake

Indeed the early "proofs" were disproved.

It was not MSWord, because impact differences could be seen by the Illinois lab that checked the memos.

Typewriters in the 70's did do proportional and various fonts available in the 70's fit the memo.

Indeed the best guess is that the simple Olypus typewriter the Sec had was used to type the memos.

The source - Bill - refused to admit he had pulled those files out of the trash when Bush friends were "cleaning" his file.

So Dan's problem was inability to give a source other than the silly dropped in lap story Bill used at the end.

As to the analysis comparing exit polling and the actual results has being refuted, I am seen that claim but it does not stand up. The fraud seems very real and proven - at least as to the math.

The specific evidence from the recount done in Washington is what? I had not heard of a Washington conclusion as to vote fraud.

TIA was working with the model in effect at 4pm numbers - what "adjusted numbers are you referring to? Indeed the 12:30 am numbers - prior to the forced fit to the recorded vote - confirm the fraud.

Has no legs is media talk for my boss will not air the story and I will not/can not push it.

But I like the "Wittgenstein's laddar that you need to now kick away" thought.

peace

:-)

[mgr]

57. Appreciate the response (long post)

Let me work this over a bit. I have the bad habit of recalling from memory, and not bookmarking. I have a lot of posts out there stating my position, and it has not evolved that much, just my arguments have.

1. The 4 pm numbers would correspond to the exit poll data released by Scoop earlier this week that they obtained from Edison/Mitofski. If you look at the table, the term 'weighted' is indicated. Statistically, weighted means sample numbers were readjusted to fit an expected pattern, so they are not 'raw'. I infer this to suggest that they reweighted the raw sample to expected frequencies, most likely census or previous precinct outcomes. What I suspect they did with the final, was to readjust by an extrapolation, probably changing the % republican turnout. (I am not about to suggest that exit polling is an exact science, but I can remember it as far back as Humphrey/Nixon, and suspect its design is based in a 1950-1960's demography)

2. A thread by a person working the Gregorie/Rossi hand recount that was an update specifically referred to the lower percentage of Gregorie votes than was anticipated by the exit polls. They did not elaborate, which to me spoke volumes. It is not the one that goes to great lengths discussing the black box problems in Snokomish Co(sp?). I wish I could produce it, because there is no reason in the world to take me at my word. Some one more savvy on this site may know it.

3. Bill's problem, from my perspective, is that he is protecting the person who did pull the records from the circular file. Rather is getting too old to fight, that is probably why he retired. My point is that it will be costly for us to pull another 'imperfectly verified facts' campaign as with voting irregularities/fraud. It would be lethal for the cause. I have suspected that the IBM my mother worked on at home had the elevated th.

4. "As to the analysis comparing exit polling and the actual results has being refuted, I am seen that claim but it does not stand up. The fraud seems very real and proven - at least as to the math."

Whether the computations are accurate or not is beside the point, and a red herring. The question to be asked is whether the assumptions made are correct, that the analysis is appropriate, and whether any examples refute your analysis. The exit polling analysis makes the assumption that exit polling is accurate in characterizing the outcome of an election. Two factors would make this true, that verifying the election outcome is intrinsic to the design; and that the number of samples are sufficient to discriminate at the order of magnitude necessary to show it. In other words, one could have had the first condition met, but not anticipate an outcome of 70%, so you would not have sampled enough voters to discriminate with sufficient precision to show discrepancies at the level of 100,000's. I further think the first condition is not met with the Edison Mitofski exit polling, since they appear to smooth the data at least twice. I think they are being honest in stating that the purpose of exit polling is to evaluate the voting behavior of subgroups, such as african americans, hispanics, jews, union members, women; all of whom, you may note are supporters of the democratic party. In other words, by that admission, one can see how we could get a red shift problem if the exit polling is biased to sample democratic support. The fact that a table of the past 'raw' exit polling showed democratic support within the MOE, but not the republican share of the vote was not. What was addressed for the table was the spread was not very predictive, but when you looked at the percent of vote, with maybe the exception of Dukakis, the Dems were there within the MOE, but the republicans way off.

5. It is probably incorrect to refer to the irregularities outlined b/only/b by the exit polling analysis as fraud. These are probably the most innocent of variability. The exit polling conducted by Edison-Mitofski cannot discriminate the type of voter suppression and intimidation in Cleveland, since these people would not be at the polls; nor can it discriminate ballot stuffing at any precinct but possibly where the exit poll was taken; nor can it assess potential malfeasance with black box voting systems. Each is a discrete area of inquiry. Also, don't forget that you cannot exit poll absentee or early voters. I don't think we have an idea of what percentage of the electorate that was, I know that here in California, its about five million voters, that were never sampled.

6. The more appropriate analysis is probably comparing the population to the sample, determine if the sample falls within the population. Usually this is accomplished through a goodness of fit test. However, to do this now would probably be GIGO, since we do not have the raw exit poll samples, nor know how many votes (besides absentee and early voter ballots) there were on election day.

I appreciate the lack of the resorting to an ad hominen argument in your post, and always appreciate anyone who had to deal with Ludwig Wittgenstein.

Mike

[papau]

58. We are both working from memory - but I doubt we recall much that is

different.

1. The 4 pm numbers would correspond to the exit poll data released by Scoop earlier this week -

no not really - NBC via MSNBC's Slate released the 4pm numbers on the day of the election. They were not raw. The sample taken was evey 10th voter at sample precincts, which was then fitted (as in the weightings were applied) to the model that Edison/Mitofski thought was correct, yielding the "4 PM numbers". There were 2 pm numbers at Slate, and a 6 pm, and a 7:30. Scoop released the 12:30 am numbers early this week - still fitted to the orginal model.

What they did to get a "final" was to play with the weightings to get a best fit to the recorded vote. It is standard operating procedure and nothing evil in any way.

But I claim that the adjustments that result are unreasonable.

2. As to a lower percentage of Gregorie votes than was anticipated by the exit polls, the Wash Gov state poll was much smaller and therefore had a larger MOE - so a 1 to 3 % difference is to be expected - either way. The Low MOE expected that TIA speaks of concerns the various Stat rules for independent stats all having an error in the same direction and by so large an amount plus the pattern that is obvious in those errors (this is where math folks going for an MS will be runnibg cross-tabs for a few years!.

3. "Bill's problem, from my perspective, is that he is protecting the person who did pull the records from the circular file. Rather is getting too old to fight, that is probably why he retired. My point is that it will be costly for us to pull another 'imperfectly verified facts' campaign as with voting irregularities/fraud. It would be lethal for the cause. I have suspected that the IBM my mother worked on at home had the elevated th."

I agree as to everything except I think it will not be lethal for the cause - but that is just my guess. I note that being quiet about 2000 got us nowhere - and indeed we had a media cover-up of their own recount that showed Gore won!

4. "As to the analysis comparing exit polling and the actual results has being refuted, I am seen that claim but it does not stand up. The fraud seems very real and proven - at least as to the math."

Whether the computations are accurate or not is beside the point, and a red herring. The question to be asked is whether the assumptions made are correct, that the analysis is appropriate, and whether any examples refute your analysis. The exit polling analysis makes the assumption that exit polling is accurate in characterizing the outcome of an election. Two factors would make this true, that verifying the election outcome is intrinsic to the design; and that the number of samples are sufficient to discriminate at the order of magnitude necessary to show it. In other words, one could have had the first condition met, but not anticipate an outcome of 70%, so you would not have sampled enough voters to discriminate with sufficient precision to show discrepancies at the level of 100,000's. I further think the first condition is not met with the Edison Mitofski exit polling, since they appear to smooth the data at least twice. I think they are being honest in stating that the purpose of exit polling is to evaluate the voting behavior of subgroups, such as african americans, hispanics, jews, union members, women; all of whom, you may note are supporters of the democratic party. In other words, by that admission, one can see how we could get a red shift problem if the exit polling is biased to sample democratic support. The fact that a table of the past 'raw' exit polling showed democratic support within the MOE, but not the republican share of the vote was not. What was addressed for the table was the spread was not very predictive, but when you looked at the percent of vote, with maybe the exception of Dukakis, the Dems were there within the MOE, but the republicans way off.

AND ONLY THE GOP BEING OFF DOES NOT IMPLY GOP VOTE PADDING VIA ABSENTEE AND OTHER FRAUD? This election added the default to Bush and the vote switch fraud to the 2000 game plan of vote suppression and absentee and fewer voting machines in Dem areas.

5. It is probably incorrect to refer to the irregularities outlined b/only/b by the exit polling analysis as fraud. These are probably the most innocent of variability. The exit polling conducted by Edison-Mitofski cannot discriminate the type of voter suppression and intimidation in Cleveland, since these people would not be at the polls; nor can it discriminate ballot stuffing at any precinct but possibly where the exit poll was taken; nor can it assess potential malfeasance with black box voting systems. Each is a discrete area of inquiry. Also, don't forget that you cannot exit poll absentee or early voters. I don't think we have an idea of what percentage of the electorate that was, I know that here in California, its about five million voters, that were never sampled.

Again this is where cross tabs work may confirm what we know by reports of witnesses - as in were those one time problems - or the result of a massive fraud.

6. The more appropriate analysis is probably comparing the population to the sample, determine if the sample falls within the population. Usually this is accomplished through a goodness of fit test. However, to do this now would probably be GIGO, since we do not have the raw exit poll samples, nor know how many votes (besides absentee and early voter ballots) there were on election day.

Knowing the the true population results is an almost impossible task ( unless one claims the reported vote is the population - and this was claimed in Florida in 2000 and shown to be false)- indeed books are written on the best math to use to get to the population numbers from a sample based on the type of population.

As to Ludwig Wittgenstein - there is an interesting PBS coming up tonight about lanuage - I hope not as deep :-)

But you have to love a fellow that spends the 2nd half of a career proving the first half not correct!

peace

:-)

[tx_dem41]

14. Has any of the exit poll analysis been done with anything but whole..

numbers (i.e. integers)? Thats all I've seen and that is sorely lacking for analysis where we are so close to 50-50 results.

[L. Coyote]

36. Numbers are rounded to the nearest percent in the posts I've seen

and then probabilities are extrapolated. You don't need to be a mathematician to see the flaws there. This is why Freeman reworks his study, to incorporate greater accuracy as the data has become available.

I compared Steve Freeman's numbers to the ones TIA posted. The deviation is near one percent on average per state.

Here are the Freeman numbers, in an Excel.

[TruthIsAll]

67. HERE ARE MY NUMBERS CONSIDERING 0.5% ROUNDING

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x243100#246102

[mgr]

39. Can't have a partial voter, now can we?

I was reflecting on a similar thing several days ago. I am not sure if this is what you are after. If you have more respondants in a particular category, and have to adjust that total to match, and your total respondants is not an even number, you have a fraction.

Of course you beg the question of whose exit poll analysis, the poll takers or TIA.

There appear to be several out there, but everyone seems to be focused on Edison/Mitofski, and TIAs little spreadsheet manipulation.

I am beginning to think that exit polls are iterative above the precinct level. What I think is that they adjust the raw numbers to what happened in the precinct the previous election. Then they compare the exit poll totals, say the presidential vote, and alter one of the multipliers to conform with the new results. Since all the categories are interdependant, that adjustment will alter all the measurements in the other categories, but not all will now add up to 100%.

Does this help or do you want to ask again with a little more clarity?

Mike

[tx_dem41]

41. I was referring to the percentages that were the results of the exit...

..polling (i.e. 49% vs. 49.3%).

[TruthIsAll]

56. As I have stated numerous times: I used the 2-party vote.

Simply:
Kerry%= Kerry/(Kerry+ Bush)
Bush% = Bush/(Kerry+Bush)

This way the total was 100%.

I did this to eliminate 3rd party votes (Nader et al). This had a NEGLIGIBLE effect on the percentages. The increments were added to Kerry and Bush proportionately.

The reason, as I have sated, was for data comparability, since I used 2-party analysis in the National and State polls in the Election Model. My final projection had Kerry winning 51.8%-48.2%, and ignored 3rd party votes.

Recall my comparison between the pre-election poll and the exit polls. There was a .12% deviation in the unweighted averages.
The numbers matched.

The pre-election numbers were 2-party in the Election Model, so I did the same with the exit polls.

My numbers were accurate.
There was no bias for Kerry or Bush.

[tx_dem41]

64. TIA.....I didn't imply that there was a bias..

so please don't think I am attacking anyone personally. You have done a yeoman's effort toward this. My question was based on PURE mathematics...not anything else.

Thanks for your efforts.

[euler]

42. There ya go.

No one has emphasised his point (or even brought it up that I've seen) but you are right.

[tx_dem41]

43. Why thank you "euler"...From your name I can assume you are..

a "math" guy. It makes a huge difference.

[euler]

47. I was reading Fermat's....

...Enigma, the chapter on Euler when I joined. I looked down at the book, saw the name, and typed it in, totally random. If I had joined sooner, my name might have been Pythagoras. Had I joined later, my name might have been Godel, or Wiles.

[deacon2]

24. Dead On

I don't recall Bucky Fuller having a degree... but he seemed to make many valuable contributions anyway. I take TruthIsAll at his/her word about the degrees and I, having read the analyses, couldn't care less about the sheepskins. What is postulated (massive irregularities tantamount to fraud) holds sufficient water and is certainly echoed by enough "outed" academics whether or not the conclusions are 100% accurate. And what is? The rest of this background noise is pure contention and I believe DISTRACTION from the core issue.

[tx_dem41]

28. Has any of the analysis been done using non-whole numbers? n/t

[mulethree]

37. all of it?

Lots of percentages, so fractions and ratios. Thats what? 5th, 6th grade math?

The probabilities and statistics, I didn't study till college sophomore year. Hopefully they teach this earlier now, at least the 'into' level needed for this stuff.

[tx_dem41]

38. Sigh....the percentages reported (from what I have seen) have ..

been whole numbers (e.g. 49% vs. 49.3%). This introduces a huge probem when using the numbers statistically.

[mgr]

44. Which percentage is expressed in whole numbers?

I'm beginning to see what your concern may be. The imprecision of measurement by the exit polls when you extrapolate to millions of voters will miss even if right on. If say the presidential outcome is 51.2 vs 48.8, if it will say it is 50/50. And what we are talking about is missing a margin of a million or more voters.

Do I have it?

Mike

[mulethree]

63. I think so

It's not an issue of whole numbers but of precision.

49% is actually 0.49, not a whole number, but could represent
0.48765 or 0.49432. So its 49% plus/minus 0.5%

So all numbers calculated from them would ideally include this 'margin of error' in addition to the survey margin of error and you couldn't consider anything 'strong' unless it exceeded both which might be 2% for some and 3+% for others depending on the survey margin.

[euler]

35. Your first sentence is simply false.

The thread I supposedly criticized TruthIsAll about his degrees is here:

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=237348&mesg_id=237348

I'll write a 100.00 check to anyone right now who can find a quote from me criticizing TruthIsAll about his education. Just search the thread for user name euler.

Your first sentence is wrong, and most of the rest of your sentences are misguided.

My problem, is not with the math, but with the very notion that math can be applied to the data we have to make any credible statement about the veracity of the 2004 presidential election.

TruthIsAll told us in at least 2 posts that he isn't an exit poll expert or an expert statistician. I take him at his word. So, since there are real exit poll experts available online, I thought that it would make more sense to actually get my information from them instead of TruthIsAll. I was somewhat surprised at the number of people who choose not to make this more rational choice. Then, I read in your post, "but for some reason, our voting machine results don't check against our exit poll results" and realized that perhaps I shouldn't be so surprised after all.

Ukraine and US exit polls are apples and oranges. Try Google.

Edit: Goggle - Google

[papau]

53. Do you think there are no statisticians on this board?

expert is a relative/opinion based term, but their are few folks here that can run with poll data and stats.

In private the real exit poll experts available may not be saying what you here being said to their corporate clients.

Ukraine and US exit polls are apples and oranges - the US ones are much better done!

[scottxyz]

71. Several people were jumping on TIA re math degrees

Here is what Euler originally said, which made me say that Euler was "criticizing" TruthIsAll for not revealing more info about his/her math degrees:

You do a lot of analysis here on DU and much of your work informs the work of others. You say you have 3 Math degrees, so I think is only fair that you disclose them given the importance of what you do here. For each, please provide the degree, year of degree and school name, or at least the degree and year.

For example, I have a BS Mathematics from the University of Texas. Thanks.

In all fairness, I should have also mentioned the others aside from Euler (LoZoccolo, foo_bar) on that thread

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x237348

who were hassling TIA much more severely over not providing enough info re his/her math degrees.

foo_bar: in other words, you have no math degree

LoZoccolo: And, he asked if we'd "studied" those things in the list and says he did, but doesn't mention if it was part of a degree program or not. So for all we know, there is no claim made to a math degree at all.

My point still is - you don't NEED a math degree to see that something is rotten in the state of Ohio - and you SHOULDN'T need a math degree to make sure your country's election totals are valid. Anyone who can add and who has flipped a coin a few times can see that TruthIsAll is onto something when he/she says that the numbers in Ohio don't check out, and that the glitches which all favor are probably not "glitches" after all.

The question is not "Where did TruthIsAll get his/her math degrees?" The question is: "Why are our voting machines so obviously screwed up?" Let's stay focused.