NSA Analyst Proves GOP Is Stealing Elections

By Denis G. Campbell and
Charley James
(c) 25 October 2012 

UPDATED: Why is Mitt Romney so confident?

In states where the winner will be decided by less than 10%, of the vote he already knows he will win. This is no tinfoil hat conspiracy. It’s a maths problem. And mathematics showed changes in actual raw voting data that had no statistical correlation other than programmable computer fraud. This computer fraud resulted in votes being flipped from Democrat to Republican in every federal, senatorial, congressional and gubernatorial election since 2008 (thus far) and in the 2012 primary contests from other Republicans to Mitt Romney.

This goes well beyond Romney’s investment control in voting machine maker Hart Intercivic and Diebold’s close ties to George W. Bush. Indeed all five voting machine companies have very strong GOP fundraising ties, yet executives (including the candidate’s son Tagg Romney) insist there is no conflict between massively supporting one party financially whilst controlling the machines that record and count the votes.

A retired NSA analyst has spent several sleepless nights applying a simple formula to past election results across Arizona. His results showed across-the-board systemic election fraud on a coordinated and massive scale. But the analysis indicated that this only happens in larger precincts because anomalies in small precincts can be more easily detected.

The bigger the precinct (x axis) the higher the number of votes for Romney and corresponding decrease for Santorum and Paul when all lines should look like the Gingrich line. (Source: MA Duniho)

“Easy to Cheat”
Retired NSA analyst Michael Duniho has worked for nearly seven years trying to understand voting anomalies in his home state of Arizona and Pima County. This publication has written extensively about apparent vote machine manipulation in a 2006 RTA Bond issue election that is still being fought in the courts. Said Duniho, “It is really easy to cheat using computers to count votes, because you can’t see what is going on in the machine.”

When Duniho applied a mathematical model to actual voting results in the largest voting precincts, he saw that only the large precincts suddenly trended towards Mitt Romney in the Arizona primary – and indeed all Republicans in every election since 2008 – by a factor of 8%-10%. The Republican candidate in every race saw an 8-10%. gain in his totals whilst the Democrat lost 8-10%. This is a swing of up to  20 point, enough to win an election unless a candidate was losing very badly.

Since sifting through and decoding massive amounts of data was his work for decades on behalf of the National Security Agency, he wanted to understand why this was ONLY happening in large precincts.

Nose Counting
The idea of examining large precinct results came via a link to a report written by Francois Choquette and James Johnson. Choquette became curious about South Carolina primary results in the February Republican contest. There a poll observer noted an unusually big gain of votes for Mitt Romney in larger precincts than in smaller ones. Choquette wanted to know why?

He examined and applied all of the normal statistical markers to see where a variance might occur: income level, population density, race, urban vs. rural, even party registration numbers. He found no correlation to explain why Romney votes trended upward while Paul and Santorum votes trended downward -yet only in large precincts.

Choquette then looked at all 50 states and found roughly a 10% switch in votes from Democrat to GOP. This was noted in every state except Utah, where the presumption was, as it was Mitt’s religious home state and very conservative, there was no chance of Romney losing and no variance was found.

Choquette even saw in Maricopa County, which is Phoenix and its suburbs, that in 2008 Romney used this technique against John McCain. But McCain beat him by too much for a 10% fraud gain to matter. McCain tried to do the same thing in the general election to President Obama but 9 million votes nationally were too many to make up.

Examining every county across America was too massive an undertaking for any one person so he included a simple set of instructions and encouraged others to do the same with raw vote totals in their county/state.

1. Download the text files of all raw actual vote results by precinct from the Secretary of State’s Office.

2. Arrange them in precinct order.

3. Put in all of the candidate totals for each precinct.

4. Sort the data by total vote smallest on the top.

Now here it gets a bit dense: He needed to add columns that show cumulative totals by candidate then compare them by candidate to establish trend lines.

That reveals trends should remain statistically constant throughout an election.

Stealing Votes
But as the spreadsheet shows, the larger the precinct, the numbers start to change dramatically.

“If percentages did not change from one precinct to the next, we would see a flat line, but what we are seeing is sloped lines downward for Democrats and upward for Republicans (or, in the case of the Presidential primary, upward for Romney and downward for his opponents), said Duniho.”

In every election contest, the trend lines dramatically crossed for no apparent reason. It was revealed that votes were being systemically bled off for Rick Santorum and Ron Paul and then being credited to Mitt Romney.

Once Duniho completed the spreadsheet, he pumped in actual vote totals from other Arizona election contests.

Chart showing Barber v Kelly special election to replace Gabby Giffords result in Pima County where the margin of victory was too large even with the supposed 'fix' in to overcome.

He looked at every 2010 race in Arizona from Governor Brewer to Senator McCain and Congresswoman Gabby Giffords. The trends lines all did the exact same thing. Someone had manipulated the election outcome, most likely one person inserting a programme inside the system’s central computer… that flipped votes.

The results were astounding.

They showed that Governor Brewer actually lost her election and Gabby Gifford’s razor thin less than 1% point re-election victory over Tea Party Conservative Jesse Kelly was closer to a 20 point victory for her.

Duniho added, “We need to have strong hand count audits to confirm the integrity of these elections. This means comparing hand counts with official reports of the election.”

Ohio Precedent
This isn’t the first time Republicans have been charged with vote theft. It happened in the 2004 presidential election, in Ohio and Florida.

In Ohio, GOP consultant Michael Connell claimed that the vote count computer program he had created for the state had a trap door that shifted Democratic votes to the GOP.

He was subpoenaed as a witness in a lawsuit against then-Secretary of State Ken Blackwell, and lawyers for the plaintiff asked the Dept. of Justice to provide him with security because there were two threats made against Connell’s life by people associated with Karl Rove. But in Dec. 2008, before the trial began, Connell was killed in a plane crash outside Akron Ohio.

There were problems in Florida, as well.

A study by the Quantitative Methods Research Team at the University of California at Berkeley found that anomalies between Florida counties using touch-screen voting and those using other methods could not be explained statistically. Noting the higher-than-expected votes for Bush in three large Democratic counties, Miami-Dade, Broward and Palm Beach, Michael Hout, a Berkeley professor who did the study said there were strong suspicions of vote-rigging.

“No matter how many factors and variables we took into consideration, the significant correlation in the votes for President Bush and electronic voting cannot be explained,” Hout said. “The study shows that a county’s use of electronic voting resulted in a disproportionate increase in votes for President Bush. There is just a trivial probability of evidence like this appearing in a population where the true difference is zero—less than once in a thousand chances.”

Don’t Trust, Verify
Indeed the only way to 100% verify this election fraud would be through handcounts of ballots by precinct, matching those results to the reported totals. But as was mentioned earlier, a group in Pima County has been trying unsuccessfully to get access to ballots to conduct such a count for almost five years since anomalies first surfaced in voting machines in 2006.

Is there a judge in Arizona likely to suddenly reverse past trends and allow access to conduct such a handcount of ballots 12 days before a national election? And if not, why not? Maybe someone needs to commission the Anonymous hacker group to re-level the playing field because the courts are not going to do it.

The results of Duniho’s analysis can only happen if votes are being stolen, and the only way that’s possible is if the computerised machines are programmed to steal them. Welcome to Zimbabwe.

More than 100 million Americans will cast their ballots thinking their vote will be fairly counted. It should be. Yet the crooks know they can safely flip up to 10% of votes without consequence. Anything more than that is statistically suspect.

President Obama won by such a huge margin in 2008 that even with this anomaly built into the system, he cruised to victory. This year the election is much closer. Can American democracy afford yet another election crisis placing three of the four last national Presidential election results in question or worse: The outcome was stolen, the outcome a victim of election theft?

Don’t Take Our Word
Use the spreadsheet above to do the maths in your own state, county or precinct. The results are compelling. Then demand that the Justice Department stop this insane view that results need to be reported by 11 pm for the television networks. Demand hand ballot counts!

We use paper ballots in the UK and results do not even begin to trickle in until 3 am. The final outcome can take up to three days to finalise. But voters in Britain know the count is accurate because every ballot is transparently hand-counted. When I read this article that Serbia, Belarus and Kazakhstan were sending election monitors to watch the US Election?, I knew we’d jumped the shark.

We are already being victimized by vote fraud on a scale that, in another country, would lead to calls for international election monitors. It is time for Americans to stop being victims of ghosts in the machine.

UPDATE: There was an error in the italicised paragraph in the section titled ‘Nose Counting.’ It originally read ‘from GOP to Democrat’ and the sentence was awkwardly constructed. The correct wording is now there. We thank our readers and regret the original error. -Ed

(This story was simultaneously released on the magazine and The David Pakman Show of 25 October.)

Denis G. Campbell is the author of 6 books including ‘Billionaire Boys Election Freak Show,’ ‘The Vagina Wars’ & ‘Egypt Unsh@ckled.’ He is the editor of UK Progressive Magazine and provides commentary to the BBC, itv Al Jazeera English, CNN, MSNBC and others. His weekly ‘World View with Denis Campbell’ segment can be heard every Thursday on the globally syndicated The David Pakman Show. You can follow him on Twitter via @UKProgressive and on Facebook.

Charley James is a long-time independent journalist who covers social justice, politics and economic issues. He’s worked in print and broadcast media for national magazines, large newspapers and major market radio and television outlets. Follow Charley on Twitter @SuddenlyHomeles.

 

Denis G Campbell View more

Denis G Campbell
Denis G. Campbell is founder and editor of UK Progressive magazine and co-host of The Three Muckrakers podcast. He is the author of 7 books and provides Americas, EU and Middle Eastern commentary to the BBC, itv, Al Jazeera English, CNN, CRI, MSNBC and others. He is CEO of Monknash Media and a principal with B2E Consulting in London. You can follow him on Twitter @UKProgressive and on Facebook.

59 comments

  1. Could this be why the GOP has started to dispute polls showing Obama ahead. In creating doubt with polling data – they will be able to say see…” we told you your polls were wrong – Romney was ahead”. Add in all the smoke and mirrors about Romney’s momentum, the “malarkey” about the first debate still hurting Obama (but of course the 2nd and 3rd didn’t diminish Romney) and you can see the groundwork being laid for voting machine manipulation.

    The article didn’t mention it, but I wonder about the Scott Walker recall race + the election of Prosser, the SC judge in Wisconsin.

    Every vote counts! If you are still on the fence, the Supreme Court is one of the least discussed, most important issues of this election. Please don’t let Romney/republicans be the one(s) to make the decision about who fills the next opening on the Supreme Court – PLEASE!!!

  2. Data Loving Independent
    /

    No, no, no. I am an Independent data analyst. I applied this technique for the Florida presidential race for 2000, 2004, and 2008, out of curiosity, and the trend is as one should expect – The Republican candidate leads the vote percentage (cumulative or not) in the smaller precincts by a huge margin (65% – 35%), and the gap closes once the more populated precincts are included in the tally since they are more Democratic. There is no evidence in the Florida General Election of this type of trend that shows a higher cumulative percent for the democrat in smaller precincts vs. the larger ones… This information is BOGUS, unfortunately. And, in general, if a trend is different in rural vs. urban areas, it should not be surprising, it is how it goes every time… Republican in rural, Democrat in urban. Plain and simple.

  3. Data Loving Independent, who’s to say your analysis method is not wrong. You know the first rule of statistics: numbers can be manipulated to reflect whatever you want. I’m more inclined to believe the article, and I’m a reformed Republican. I voted for Bush and after I saw how he stole the election I was done with the Republican Party.

  4. That’s a novel use of the word “proves”. I have to say that every mathematics professor I’ve ever known would chuckle at saying anything has been proved here.
    The article even admits that there is no proof.
    _Indeed the only way to 100% verify this election fraud would be through handcounts of ballots by precinct, matching those results to the reported totals. But as was mentioned earlier, a group in Pima County has been trying unsuccessfully to get access to ballots to conduct such a count for almost five years…_
    They can’t even verify the validity of this guy’s model in one county, but in the opening paragraphs make this statement, “This computer fraud resulted in votes being flipped from Democrat to Republican in every federal, senatorial, congressional and gubernatorial election since 2008…”
    Ridiculous.

  5. I have been a programmer for 30 years and don’t trust a computerized voting machine any farther than I could throw it. Ballots should be hand counted and the damned TV stations can just wait for an honest and accurate count.

  6. @Data Loving Independent

    You argue against these results based on differences of party affiliation yet clearly did not read this article nor the study it cites, which analyzed the results of only Republican primaries and their candidates. The result was that in every state but Utah, Mitt Romney won with vote shifts away from other candidates like Rick Sanitorum and Ron Paul that clearly followed linear trends upward as the precinct population sizes increased.

    Democrats, during their primaries in these same states, did not see this shift. Why is that? Do you have a demographic answer that is somehow strictly limited to Republicans who live in the exact same precincts?

  7. I don’t think so. Remember the extreme accuracy of Nate Silver’s predictions, especially taking into account that he was given their internal polling results. His numbers would’ve been very far off if there was a simple vote flip in 49/50 states.

    I don’t buy it. I don’t see any solid support for this.

  8. Nate: the fact that pollsters can extremely accurately predict the outcome of American elections is due to the fact that they based their predictions on “likely voter” models.

    So they find in one election that polls predicted 55% Democrat, 45% Republican, but due to vote flipping, the official result ends up being 45% Democrat, 55% Republican.

    The official explanation for that is then that Republicans are more likely to vote! So the fraud factor is then built into the likely voter model, which adjusts Democratic numbers downward and Republicans upward, and voila, the next time an election is stolen by the same margin, the polls predict the official outcome perfectly.

  9. While it’s possible that something like this is happening (though I’d want to see more evidence before believing Duniho’s claim that this “This computer fraud resulted in votes being flipped from Democrat to Republican in every federal, senatorial, congressional and gubernatorial election since 2008”), there are a few things in the article that I’d want more information about:

    (and to get this out of the way first, for those who will try to claim “Democrats do it just as much or more!”: I do know there are a few races occasionally where Democrats commit voter and election fraud, but the quantity appears to be vastly less than that committed by Republicans)

    • Which computers does Duniho claim are the ones that are being tampered with? The voting machines, or the machines that tabulate county-, state-, and nation-wide totals, or both? In what ratio? In other words, is the fraud perpetrated mostly through one or the other of these classes of machines? This might be important information if it indicates that one method is harder to tamper with than the other–if hacked voting machines are responsible for 80% of the vote flipping, and only 20% is due to the tabulating machines that count the larger vote, does this mean that doing away with voting machines might (at least for a couple elections) make it much harder for the Republicans to steal elections, if their most effective method for stealing votes is taken away from them? We know that there have long been methods for stealing elections that rely less on voting or counting machines (the Supreme Court, Republican operatives like Katherine Harris in government offices in various state, etc.), but would it throw a useful wrench into their works to do away with the use of voting and tabulating machines altogether, and instead to do all counts by hand (even though that too is subject to various methods of fraud, it might be significantly more secure)?

    • Are there already any large precincts where the votes are counted by hand (as part of the normal election process rather than during a court case after the election), either as the only method of adding up the votes, or as a doublecheck on voting machines in those precincts? If so, what do the results of those hand-counting precincts show?

    • Oregon uses voter-marked paper ballots nearly exclusively (correct me on the details)–no touchscreen voter machines (though they’re experimenting with using iPads–as if tablet computers are immune to hacking!)–and these paper ballots are tabulated using electronic scanners. But the article says the fraud, which it ties to electronic voting machines and electronic vote tabulators, was found to be occurring in all 50 states. Is the percentage of skew to Republicans in the final vote count, the same in Oregon as Duniho says happens in other states, even without Oregon using electronic voting machines? If so, this would implicate the scanners as being as hackable as the electronic voting machines the voters actually use to record each of their votes.

    • The article states: “Choquette even saw in Maricopa County, which is Phoenix and its suburbs, that in 2008 Romney used this technique against John McCain. But McCain beat him by too much for a 10% fraud gain to matter. McCain tried to do the same thing in the general election to President Obama but 9 million votes nationally were too many to make up.” Does this mean there’s a Republican vote-stealing system in place whose control is given to whichever Republican is the current front-runner, or the desired front-runner? And that its reins were given wholly to Romney during the primaries, but then to McCain when Romney lost the candidacy because his lead wasn’t big enough for the fraud to make up the difference?

  10. To me it looks like a statistical manifestation of a demographic phenomenon: denser populations correlate with moderation in political views as expressed in 2008 U.S. pres. election. More broadly, it’s an illustration of a phenomenon where larger sample sizes yield more accurate, or finer resolution of the actual total count. Because mathematically, at first, adding each district has huge effect on the cumulative tallies, but as more votes are added up, they barely sway the curves up or down. the slopes almost even out. This amateur statistician forgot to account for this natural phenomenon.

  11. I processed the data a bit more. There is another interesting trend.

    If you plot district size vs the vote counts for each candidate, the effect mentioned in the article can be seen. But then if you turn on trendlines, specifically the “power” trend line, look at the R^2 (correlation) values. All the candidates are between .90 and .92, except Romney, who is at .74.

    This means that whatever factor is being modeled by the power series, doesn’t apply to Romney for some reason.

  12. Most high level business persons that I have known consider ‘the rules’ to be something for others to follow but not they. I have no doubt that if there is a means to steal the election, they will try it. I have no problem being called paranoid on this issue, I am. All someone needs to do is prove me wrong. The thing is no one can prove this and that is exactly the problem.

  13. There has actually been much statistical analysis on this and the author is correct. See the work of Proffessor Stephen A. Freeman in 2004, for example. For a great article on the history of this over the last decade, READ THE ARTICLE BY VICTORIA COLLIER IN THE CURRENT ISSUE (NOV. 2012) OF HARPER’S MAGAZINE.

  14. I had a friend in the 1980s who was a computer programmer (like me) and part of his job was to provide legislators with information to make decisions, which can be kind of like the media who do not give you a full perspective on reality, maybe because they have no clue, like members of legislative bodies sometimes have no clue. It is rare to find legislators who are not self-serving and do not have an agenda. We may not always get what we want, but all we have is one vote, and at the very least these should be counted accurately by people with integrity. If machines are used as a tool to count votes, they must be error-free, and proper auditing measures must be used to ensure accuracy, for example the old textbook method of counting at least three times by three different entities, etc., so punched card receipts with dangling chads might be another tool to use. I hate being a witness to the destruction of the republic by those who would destroy democracy just because they can. If the U.S. has developed institutions that we all value, these must be cherished and passed on to the next generation, many of whom are immigrants, many of whom come from nations with histories of violent revolutions, from which we were born, but from which we have evolved. My values are truth, freedom, and justice but these values may not count anymore. I have seen very little evidence of equality in our history. It is little more than a word, but everyone’s vote should be equal.

  15. I utterly would not put this past the ‘Pubs, but I gotta ask: how is it that Nate Silver and Sam Wang have been able to *nail* their outcome predictions in recent elections, with predictions based on pre-election poll data? (Unless Karl Rove was also manipulating all the pollsters…)

    If the contention here was true, Nate and Sam’s predictions should have been off by 5 or 10%.

    What say you?

  16. CS-STAT MAJOR 30 years programmming
    /

    Ok I am a damn expert. I’ve been programming since 1981 and as soon as the collectors and tabulators (voting machines and the servers to you imps) were being sourced SOLEY from GOP I felt a rising sick feeling in my gut. Its not paranoia, they can AND HAVE rigged the elections. Its even worse than Denis points out. The vendors can make those machines do whatever they want. There’s is NO Independant Validation and Verification. To those of you who aren’t the paid posting shills: THE ONLY WAY OUT OF THIS MESS is to push for FEDERAL LAWS requiring a random 3.5 % (9% is better but the cost is geometric) HARDCOPY count to match results.

    To everybody in de-Nile .. go ahead and delude yourselves until those greedy b@$terds have you back in the factory working 7 days a week 12 hours a day at just enough pay to keep you handcuffed. Harkening back to their heyday before the coal miners and autoworkers forced them to cease and desist.

  17. Another Halocene Human
    /

    I’m confused by this article. Wouldn’t we expect larger precincts to trend Romney?

    There is a proven history of election fraud in Dade County so that’s nothing new, but I don’t think the conclusion up top on this article follows from the premise.

    Romney’s donors all come from highly populated cities so it’s hardly unexpected to find his voters there. JMHO.

    If you’re interested in election fraud, please check out BRADBlog. There is no better source on that issue on the internet. And I agree with with the other posters who mentioned hand counts of paper ballots. We need this in every state!

  18. CS-STAT MAJOR 30 years programmming
    /

    Holo-man – no the large precincts follow a normal dist It is profoundly easier to bury enough outliers to skew a trend. Thats the stat side. If you dont get it … you weern’t paying attention in STAT 201

  19. Data Loving Independent
    /

    OK, I took the time to duplicate Duniho’s work for the last election he used as an example, CD 8 Special Election 2012. I used Pima County votes by precinct to duplicate the cumulative % plot, and it worked. I see the trends of Kelly picking up in cumulative fraction of votes when organized by precincts with smallest to largest number of votes cast. Then, to see if demographics in Pima could account for these trends, I performed the same type of data organization and plotting using the voter registration party affiliations, which they also provide. These voter registrations are provided for all precincts in Pima, whereas the precincts of Pima County that cast votes in CD 8 election are different, the demographics may be informative. In fact, when compared, the cumulative fraction of Democrat to Republican looks almost identical to the cumulative fraction of votes cast for them in the CD 8 Special Election in 2012…to within a percent. So, I think demographics can very easily explain the trends in that case. Of course there is no way to compare registrations to get an idea who would vote for which republican in a primary, I don’t find it surprising that more populated areas went more for Romney than Ron Paul…

    My first comment was based on my first test using 2000, 2004, and 2008 Presidential election results in Florida. In that case, the Democrat does better in populated areas. In Pima, it is opposite, perhaps because of the distribution of immigrants in the border county. The less populated areas may have a higher percentage of immigrants who favor the democratic candidate… Interesting but not to point that I suspect fraud.

    Having said all of that, I certainly find private ownership of voting machines with proprietary software problematic. It is important to have separate checks on these types of tally machines, such as exit polls and hand counts of a much more significant fraction than one or two or four percent.

  20. The best trick the devil ever played is to make people believe he didn’t exist . . . of course it’s possible and more likely probable . . . but as a nation it’s up to us to stand up and STOP it from happening.

  21. The HBO documentary “Hacking Democracy” pretty much proved the Diebolt Voting machines could be hacked. I have no doubt in my mind if it’s a programable machine it’s hackable, and will be hacked.

  22. The Koch Brothers, Rove, Grover, and many more will stop at nothing to get what they want. More money, more power, more, more, more. It’s disgusting.

    KARMA is the only thing that keeps me sane; they’ll get theirs eventually. If there is a God, that is.

  23. Statisticians are discounting your paper for the following reason.

    That’s Statistics 101 based only on the assumption that each precinct would have the same number of Romney marbles, which is a faulty assumption.

    From teraflop’s top comment in the thread I quoted:

    “This is only true under the assumption that the sequence of votes is IID[1] . That’s a very strong assumption which is almost certainly not true for electoral data. Just as a completely hypothetical example, suppose there’s a positive correlation between voting for Romney and working full-time. And suppose people who work full-time tend to get to the polls later, which means their votes are counted later. Wouldn’t this lead to exactly the effect you’re observing? The probabilities calculated using the hypergeometric distribution are meaningless unless there are no correlations of this type whatsoever, which is implausible. Not to mention that calculating p-values (which is essentially what these are) without correcting for multiple comparisons[2] is a really good way to get results that look more significant than they actually are.
    Laymen’s edit: You would need to eliminate any biases which could otherwise contribute to this correlation before claiming voter fraud. The effect could easily be coming from something else.”

    Do you have a rebuttal for this?

    This paper is very active on Reddit right now but the community is at a stand still.

    http://www.addictinginfo.org/2012/10/27/gop-rigging-elections-for-romney/

  24. I ran the analysis for the 2012 Minnesota GOP Primary, and it looks similar to the Arizona graph in this article: Gingrich is the only one with a flat line, while Romney’s trends up and Santorum’s and Paul’s trend down. I’d like to see more “control” graphs to show the contrast between R and D states before I believe this doesn’t just represent a correlation with voting tendencies. The original analysis has 2 such graphs: Utah 2012 (90% Romney) and Maryland 2012 (8 candidates). The original analysis has 14 R-state graphs showing the trend. To “prove” anything there need to be several D-state graphs that don’t show it. Where are they?

  25. If you think that these plutocates and the gop are willing to spend billions, suppress votes, lie to the public, buy voting machines and then let a few key strokes from a hack able voting machine stop them from buying this country…..think again.

  26. Not doubting that Romney is a creep, or that he may be trying to rig things. But this ain’t the proof. Yanggers, Holocene and Andrew have nailed the central problems:

    1)Population density –> relative liberalism

    So, the curves are as expected, given Romney was seen as the moderate of the bunch.

    2) Lots of statistics, but glaring lack of science

    If your point is that these (seemingly intuitive) curves are peculiar to Republicans & Romney, then prove it. A real study would show LOTS of graphs of non-curving voting results for Democratic elections (senate primaries would be good)and Republican primaries where Santorum or Gingrich won. I suspect that quite often we would see upward curves for more liberal/moderate candidates.

    The rest is smoke & mirrors.

    p.s. Gotta love CS-STAT’s retort that “large precincts follow a normal distribution”. Um, yes. So do small precincts. Then we measure the distance between the humps. That seems to have been the primary theme of STATS 101, as I recall.

  27. Since this vote flipping has been done for past presidential elections, the voter models used by the polling companies are using models that incorporate the vote flipping without even knowing it. They have correlated polling data to the actual past result. So if the fraudsters repeat the fraud, the pollsters will be accurate – if the fraudsters don’t repeat the fraud then Obama will gain 5-10 points. The key is that the pollsters have the fraud “baked into the cake” so to speak.

  28. The authors correct their data to account for the fact that ratio of the number of republicans to democrats changes as the precinct size changes. From what I read this is a smaller effect compared to the vote flipping effect.

  29. “When I read this article that Serbia, Belarus and Kazakhstan were sending election monitors to watch the US Election?, I knew we’d jumped the shark.”

    oh come on now. This lends about as much credibility to your argument, as Michael Moore lauding the Cuban health care system in his film “Sicko”.

    Serbia… still has an axe to grind with the US re: the intervention in Kosovo.
    Belarus… a dictatorship which is in no position to criticize any country’s voting procedures.
    Kazakhstan… virtually a one-party state which conducts elections that are widely considered a sham.

  30. Scott, what exactly do you see in the report that indicates some correction for precinct size?

    I see some earnest assurances that they have, but as far as evidence, they only point to Figure 2 as their lone demonstration of the lack of impact of “pure demographics” on voter preference. As you can see, the chart a) only concerns income, b) is only for California, and C) is oddly described as “flat”.

    Am I missing something? The middle of page 22 promises the use of partial correlations to remove effects of population density on candidate choice, but we’re not shown any partial correlations anywhere.

    Some other questions and my guesses:

    Why use sampling in the first place when complete precinct-level data sets are readily available? (the better to confuse matters, introducing unnecessary levels of statistics).

    Why not just use straight-up correlation between precinct size and voter preference? (because that would highlight the central confound of the study — the “proof” of malfeasance is identical to the well-known positive relationship already existing between population density and ideology).

    Why graph cumulative votes across precinct sizes, rather than simple scatterplots of size by overall vote shares? (to give the illusion of chronological progression: true votes are coming in, THEN votes get flipped by the evil big precincts)

    If it were not so bizarrely convoluted and poorly written (seemingly unedited, really), this would make a good case study for how not to use statistics. My nasty tone is borne of frustration, but I’m actually curious if you’re seeing something that I’m not.

  31. Statistical analysis like this is too dense to ever provide any real motivation to change ballot-counting procedures, such as independent audits or using only open-source, verifiable code for electronic voting machines or not purchasing voting machines from companies with well-known partisan alignments. As we all know, there is an entire industry on one side of the political aisle which is quite skilled at obfuscating facts by spreading disinformation. It would be a trivial matter for them to come up with their own ‘experts’ to debunk this, they’ve been doing it with climate science for decades and that’s a matter that ought to be settled by now.

    If elections are being swung from one party or one candidate to the other, then someone wrote that code. In fact, this article alleges that not only are elections being thrown to one party, they are actually being thrown to a specific candidate, Mitt Romney. Considering how decentralized US elections are – they are run at the county level for the upcoming presidential election – this requires more than a few people to be involved.

    I have a very simple proposal to arrive at the truth about this: offer a sizable reward to the person who comes forward with definitive proof that they wrote the code that enabled someone(s) somewhere(s) to manipulate vote counting virtually nationwide at the county level – remember, we’re talking about over 3000 counties, and there are many different kinds of voting machines being used – in order to throw elections not only to a specific party but even to specific candidates.

    This is an interesting article, but it implies some sort of massive conspiracy. People are not very good at keeping secrets. The probability of a secret being kept secret, decreases inversely with the number of people who know the secret. If the authors are on to something, then surely somebody somewhere knows the truth and is willing to speak up about it for the right price.

  32. Upon further thought, I’m becoming rather skeptical.

    How do you think this works, what are the mechanics of it? I’ve worked the elections for the county. If you’re saying the vote flipping happens inside the voting machine… you’ve lost me right there. For this to work – for vote flipping to be done in only the larger precincts so that it is, as you say, harder to detect… somebody is really deciding which machines go to which precincts? We have hundreds and hundreds of precincts in our county. It’s always been my impression that the Optiscan machines are just sent out randomly, nobody keeps track of which ones go to which districts. There’s really somebody at the county who is deciding which machine to send to which precinct, who has that much time on his/her hands… and nobody knows they are doing this?

    Or are you saying that the vote flipping happens once the memory packs are delivered to the county and the vote tallies are downloaded to county computers at their headquarters?

  33. Did anyone see “Hacking Democracy” documentary? They investigated the past presidential election, I believe from just 2004 but might have almost mentioned 2000– and they had videotaped evidence of discarded ballot totals in Ohio in the trash that were supposed to be kept for a required amount of time for public records… and the counts on the paper receipts did not match the official vote tallies that were reported. You’d have to watch it to see. But they also bought used voting equipment, that was supposed to not be obtainable, very easily– and I believe some of the voting cards still had a simple executable program that changed a vote for D to a R. But the final vote tallies would sometimes be off, not matching the final vote- which should not have been the case! can’t remember the exact details from the movie.

    But I suggest anyone interested go watch “Hacking Democracy” that aired on HBO, and see for yourself how easy the machines can be to manipulate. They had a guy who could gain access to the vote machines in seconds. It would not be hard to change the final vote on the servers that were tasked with compiling the final vote tally. Thus the final numbers would not match the individual tallies.

    As a programmer, it is abhorrent that votes are not encrypted with the same technology that keeps your bank transactions secure, as well as keeping a duplicate paper ballot backup in the event of any irregularities. Any system of counting on computer can easily be broken, that’s why it is SO important to have verifiable a paper trails that can be referenced if need be. There needs to be a national systematized voting system implemented. Both in the types of machines they use, as well as the entire voting process– as well as which days are accessible to vote, and the hours available clearly defined for certain areas. Until we have adopted a more stringent national system, we will continue to be more susceptible to things like this occurring again in the future. Keep vigilant, and informed– this is the only solution to combat voter fraud. The same system the CIA uses to detect voter fraud abroad should also merit attention here at home for our elections.

  34. Ok, so if we believe that this theory is plausible, and that our elections are being manipulated, I have a solution to bring this into A National platform.
    The Democratic Party is not an assembly of idiots. We just tend to be less morally corrupt, and more empathic to a Humanitarian view of Mankind.
    To this end, I offer a challenge to you highly intelligent, and motivated Programmers and techno-wizards. If this can be done
    DO IT!
    Quit trying to prove that the Republicans are doing it, and instead prove that it can, and WILL be done. And that we can do it too.
    No one would work harder at reforming this perceived issue than a Republican who feels threatened. We need to get off our moral high ground, and stoop to the level of our enemy. If they know that they can buy or steal an election, they will.

  35. The program code would be simple. The following code would run just after the last vote was cast:
    If nbrBallots > 1 And 10000 And 100000 And < 1000000 Then
    nbr_to_flip = .1
    End If
    End If
    End If
    flip_count = nbr_to_flip * nbrBallots
    For vote_count = 1 To nbrBallots
    If dem_vote(vote_count) = True Then
    If vote_count < flip_count then
    repub_vote(vote_count) = True
    dem_vote(vote_count) = False
    End If
    End If
    Next

  36. Reposting because computer code got hammered during previous post:
    If nbrBallots > 1 And 10000 And 100000 And < 1000000 Then
    nbr_to_flip = .1
    End If
    End If
    End If
    flip_count = nbr_to_flip * nbrBallots
    For vote_count = 1 To nbrBallots
    If dem_vote(vote_count) = True Then
    If vote_count < flip_count then
    repub_vote(vote_count) = True
    dem_vote(vote_count) = False
    End If
    End If
    Next

  37. Guess I’ll try double spacing to preserve coding:

    If nbrBallots > 1 And 10000 And 100000 And < 1000000 Then

    nbr_to_flip = .1
    End If

    End If

    End If

    flip_count = nbr_to_flip * nbrBallots

    For vote_count = 1 To nbrBallots

    If dem_vote(vote_count) = True Then

    If vote_count < flip_count then

    repub_vote(vote_count) = True

    dem_vote(vote_count) = False

    End If

    End If

    Next

  38. I give up…my code gets hammered every time I submit. The first three “if” statements get dropped. The first sets the number to flip at .05 if the number of ballots is less than 10,000. The second “If” statement sets it to .075 if the ballots are between 10,000 and 100,000, and the last sets it to .1 if the ballot count is greater than 100,000. Then each ballot is examined in a loop, flipping the dem votes to repub until the number to flip is exceeded.

  39. Interesting discussion, however I propose that the factor causing the correlation is an effect caused by variations in voter turnout rates between different precincts. Since the analysis orders the precincts by the number of actual voters, (not the number of voters registered in the precinct), the trend line may be capturing voter motivation rather than fraud or any of the socioeconomic factors which were mentioned but discounted in the original analysis.

    Here’s a summary of the analysis which I performed:

    1. Choose an election for analysis. I chose to download the raw data for the Ohio 2006 governor election, Ted Strickland vs. Ken Blackwell. I completed the same analysis for the Ohio 2008 general presidential election, McCain vs. Obama, with similar results.

    2. Perform the original analysis described by the authors, the data shows the trend lines as described, indicating an alleged fraud.

    3. Calculate the voter turnout rate in each precinct (votes cast/registered voters).

    4. Calculate the ratio of votes in each precinct for Blackwell vs. Strickland (votes for Blackwell/votes for Strickland.)

    5. Generate a scatter plot for the data (turnout vs. candidate ratio.) Observe the wedge shape and the trend line which indicate a correlation between voter turnout and a vote for Ken Blackwell.

    As further confirmation of this analysis, note that While Blackwell lost the election overall by 60-37, Blackwell had more votes in 16 of the 25 precincts with the highest turnout ratios (excludes one precinct with only 2 voters.) It’s hard to believe that this anomaly was created solely by vote flipping, as fraud of this magnitude would be too obvious not to be detected.

    In the original analysis, the number of votes in a precinct was used to order the data. Voter turnout clearly affects the number of actual voters in a particular precinct. Two precincts may have a similar number of registered voters, but a very different number of votes cast. Here is one example from the Strickland/Blackwell data set of 2 precincts with approximately the same number of registered voters, but very different turnout rates:

    Darke County – CV-Z 1045 votes cast/1240 registered voters – ratio of Blackwell votes to Strickland votes: 1.644

    Franklin County – Columbus-12-B 237 votes cast/1246 registered voters – ratio of Blackwell votes to Strickland votes: 0.1487

    Another interesting analysis to perform is as follows:

    1. Download the raw data for the Ohio 2012 primary election.

    2. Order the precincts by number of votes cast, then calculate the cumulative totals for the numbers of Democrat and Republican ballots completed. Since this is a primary, voters had to request either a Dem or a Rep ballot. This number is not actually a vote, but rather a count of the number of ballots requested. It seems that this number would be very difficult to falsify since all of the other totals on ballot must sum to the total number of ballots completed for the particular party.

    3. Perform the cumulative analysis as described in the original paper, graphing the Democrat and Republican ballot counts. Note that these graphs show the same type of trends that were noted in the original analysis, with the number of Republican ballots sloping sharply upward as the number of votes in the precinct increases.

    I suggest that the trend is caused not by vote fraudulent vote flipping, but rather by a hidden correlation in the data, such as the voter turnout metric proposed above.

    Please try this analysis on other data sets and let me know if you agree that this refutes the authors’ “proof” of fraud.

  40. @Data Loving Independent:

    I agree with you; my wrinkle in the conversation is that “no apparent” reason for a trend cannot be, and should not be contorted to say it is statistically significant evidence of a correlation between GOP votes and fraud using electronic voting machines.

    From a statistical standpoint, I don’t agree with using weighted variables for political party, as it is not a fixed characteristic. I’ve put an inquiry in with the author to gauge his rationale for using political party.

    @PT
    In fact, there could be an anomalous underlying demographic, which hasn’t been represented by this study, and that is geography, the Cartesian spatial autocorrelation of the linear trend and precincts. I’ve got a Master’s in Geospatial Science and I’ve seen first-hand some remarkable (not wholly ground breaking) but interesting correlations that are statistically significant for “no apparent reason” other than geography (at this point.)

    Just saying, I’m an independent, too; and a more robust spatial analysis should truly be done before alleging an entire political party is “stealing the votes.”

    The best example recently is Jake Welch and the employment figures (which turned out to be skewed, whether intentionally or not, doesn’t matter; it just matters that we took the time to check it out and let the people make the decision themselves.)

    For what it’s worth, until they can produce a spatial analysis that corroborates their “theory” of linear trends in Republican primaries, I’d read this with a scrupulous eye.

  41. @Data Loving Independent:

    Try plotting Dem vs. Rep affiliation for 2012 Ohio primary election: http://www.sos.state.oh.us/sos/elections/Research/electResultsMain/2012results/2012precincts.aspx

    The graph shows that there is a clear increase in % Rep affiliation as the number of voters in the precinct increases. Does the author claim that voter affiliation is also being flipped?

    @Holly:

    Precinct “size” is not distributed randomly because it depends on voter turnout in the precinct. In Ohio the law sets an upper limit on the number of registered voters per precinct at 1400. In more densely populated areas there are more precincts. Try looking at the precinct turnout and how that correlates with candidate selection.

  42. The statistical analysis highlighted in the article is weird and seems pretty unfounded in theory. Little shocked he didn’t just do a simple regression. There is in fact a slight correlation beween size of the voting population for romney and an inverse for Santorum and Paul. Gingrich was very stable. The r^2’s for each: Romney .23, Santorum: .07, Paul: .37, Gingrich: .003. Here’s the thing though – these are pretty weak r^2’s — while there is possibly a trend, it’s a weak one. A rough way to look at it is for 23% of the precincts, a upwards trend for Romney was observable and for Paul 37%. Not sure this is a clear signal of criminal activity though, and one would need to clearly prove other demographic factors are not correlated.

  43. To Commenter “RMW” and others.

    We have done the R^2 analysis early on and this analysis was also conclusive to our argument, not just for one state for for 49 states (Utah and Puerto Rico were flat)

    If you prefer to use R^2 that’s fine, but be sure to multiply the probabilities that Romney will do better than his opponents in ALL 49 states. You will also most likely come to the conclusion that this is an amazing anomaly.

    Regarding demographic factors, we believe that we have covered this subject extensively in the 50 pages of the combined papers. The most obvious way to see that demographics has a negligible role is to chart them directly as a function of cumulative precinct size. The results flat-line.

    As far as urban/rural population density, the statistics-oriented paper goes to great lengths to fully calculate the partial correlations required to factor out population density from the data and the anomaly favoring Romney remains.

    May I suggest that you read both papers and we believe your comments and all those above will be answered hopefully to your satisfaction.

    Of course, we welcome your analysis, with data and charts to challenge our arguments. Please note that the methodology described is to be used only for PRIMARY elections, where Republicans are challenging other Republicans or Dems vs. Dems. General elections are more complex and require voter registration party preference data for each precinct in order to make the same claims. We have done that for Ohio and it is shown in the paper.

    Please review both papers and most likely your questions will be answered:
    http://www.themoneyparty.org/main/wp-content/uploads/2012/10/2008_2012_ElectionsResultsAnomaliesAndAnalysis_V1.61.pdf
    http://www.themoneyparty.org/main/wp-content/uploads/2012/10/2008_2012_ElectionsResultsAnomaliesAndAnalysis_V1.61.pdf

    Francois Choquette

Leave a Comment