Flipping the vote against Ron Paul in South Carolina?

[The Man]

It's normal- Mainly democrat and Afro-American.

I also made this observation way way back yonder but it didn't get any traction. Any thoughts? Anyone familiar with this precinct have any idea of what happened there?

EDIT: Could these be the low turnout precincts where votes were taken from Paul, and maybe others, and given to Romney?

 

[wetroof]

time for some New Hampshire data. second largest county: Rockingham ~60,000 votes. 2008 and 2012 compared. On x axis is cumulative vote ordered by precinct size ascending . On y axis is candidates share of cumulative vote by percent. this is the way many have been showing the data.
2008

bysize08 by wetroof1, on Flickr

2012

bysize12 by wetroof1, on Flickr

Romney does better the larger the precinct. I'm not drawing conclusions.

edit: 42 precincts in this county.

 

[Liberty1789]

Cool to see. No convincing evidence of vote flip in either.

Do you have a link for 2008 NH primary results?

 

[affa]

Romney does better the larger the precinct. I'm not drawing conclusions.

edit: 42 precincts in this county.

a couple notes:

a) "Romney does better the larger the precinct." could be better stated: Romney [remains the only candidate that] does better the larger the precinct.

b) 2008 could be corrupted/fraudulent data as well.

c) if Romney had flatlined around 33-34%, and McCain in the low 40s, the only candidate that could be easily siphoned from to give Romney the win would be... McCain. I'm not saying this is clear cut (See point e) but it's possible.

d) Ron Paul stays pretty constant with precinct size, as to be expected.

e) McCain's flip with Romney could be explained off easily since they seem to be the only two 'players' in this race, which makes it more likely we'd see 1:1, as if we were watching a 2 man race.

f) If 'bigger' meant more Republican, as some have suggested, then this shows us in NH at least (assuming uncorrupted results), Romney and McCain aren't viewed similarly at all and thus we shouldn't use 2008 data for McCain as a proxy for Romney.

But yea, overall, nothing conclusive.

 

[wetroof]

here is the data I used:
2012: http://www.sos.nh.gov/presprim2012/RepSummaryPres.htm
2008 http://www.sos.nh.gov/presprim2008/rpressum.htm
the links for individual county data are at the bottom of the page.

it takes a little bit of messing around to get it into excel.

I'm going to take look at the largest county and probably post a graph or two. the other counties are small and not worth the effort.

 

[bbwarfield]

there would be a difference between 2008 and 2012.... increase in electronic voting precincts. Keep this in mind.... New Hampshire has some electronic and some hand counted. Dennis Kuccinich tried to get a recount on this cause of obvious changes between the two in democratic primaries that year. The reason given? "Obama does bettre in Rural NH and Clinton does better in the cities"

 

[Mark37snj]

time for some New Hampshire data. second largest county: Rockingham ~60,000 votes. 2008 and 2012 compared. On x axis is cumulative vote ordered by precinct size ascending . On y axis is candidates share of cumulative vote by percent. this is the way many have been showing the data.
2008

bysize08 by wetroof1, on Flickr

2012

bysize12 by wetroof1, on Flickr

Romney does better the larger the precinct. I'm not drawing conclusions.

edit: 42 precincts in this county.

d) Ron Paul stays pretty constant with precinct size, as to be expected.

So it appears we have some Ron Paul data (2008) that may not have been altered. I think a good comparison to run would be Standard Deviations of those who do not have votes being siphoned off of them and comparing it to them when they do have votes siphoned off of them. You can create a comparison between individual canidates and as well as a cumulative group.

 

[The Man]

I found Rockingham to be untampered with. Romney legitimately does well.

time for some New Hampshire data. second largest county: Rockingham ~60,000 votes. 2008 and 2012 compared. On x axis is cumulative vote ordered by precinct size ascending . On y axis is candidates share of cumulative vote by percent. this is the way many have been showing the data.
2008

bysize08 by wetroof1, on Flickr

2012

bysize12 by wetroof1, on Flickr

Romney does better the larger the precinct. I'm not drawing conclusions.

edit: 42 precincts in this county.

 

[bbwarfield]

In December, 2005, for demonstration purposes, Mr. Harri Hursti, a recognized computer security expert, successfully modified the pre-election vote tally of individual candidates on the memory card of a Premier Accuvote® - OS optical-scan ballot-counting tabulator utilizing the 1.94w firmware, thereby changing the outcome of a Tallahassee, Leon County, Florida mock election. The change to the pre-election vote tally instructed the tabulator to reduce the beginning vote count of one candidate to a negative number and the beginning vote count of the opposing candidate to a positive number of equal value. This manipulation maintained an opening vote count of zero for that contest on the zero tape that must be produced at the beginning of each election. No evidence could be found of the manipulation.

This is from a commission by new hampshire secretary of state.... Pointing out some fraud can't be found easily.... Cause some people are smart enough to hide them

 

[bbwarfield]

http://www.sos.nh.gov/voting machines.htm

Here is the official list of who usese the voting tabulation machines in NH....Rochester town uses it in it's 6 precincts.... But I will try and determine the overall use in The county..... These are municipalities that use them... The remainder of a county would be hand counted

So in Cheshire county the town of Dublin for example would be hand counted but keene used electronic tabulation in 5 of its precincts.....

 

[Mark37snj]

This is from a commission by new hampshire secretary of state.... Pointing out some fraud can't be found easily.... Cause some people are smart enough to hide them

With these electronic voting machines they can alter the results a hundred different ways on site and a hundred more different ways at the Central Tabulation Center. But the statistics of impossibility is the new radar you can't outrun.

EDIT: I'm adding this to my sig

 

[Liberty1789]

There is a gentle climb in Romney's score in Rockingham though, as innocuous as it looks, and I see it in the next large counties as well, which bothers me. So low intensity vote flipping could be there, but be visually discreet. I'll investigate.

Almost done with SC, results soon.

 

[Liberty1789]

Summary of story so far...

As ever X-axis cumulative votes cast sorted by ascending order of precinct vote tally, Y-axis candidate cumulative share of votes

Always flip to Romney

IA : from Paul, Perry, Bachmann. (Looks like a hinge at 80%. Romney gains accelerate to 15% of votes, from all allowing the victory).
NH: from Paul, Huntsmann
NV: from Paul. (Looks like a hinge at 80%. Romney goes flat, Paul bleeding as ever, now to Gingrich's benefit, securing his silver medal! Ain't that fun?)
ME: from Paul, Gingrich

Now the South Carolina treasure trove. Half a million voters, 47 counties. Perfect for forensic analysis.

Bear with me, there is a lot to say.

Santorum provides the perfect untouched dead flat benchmark, doesn't he?

Ok, so, the vote flip conveniently turned to the new threat: Newt...

 

[Liberty1789]

Let's start with a small hommage to The Man and his moment of genius. He has shown that in Anderson, SC and Greenville, SC, there is an apparent 1-to-1 vote flip between Romney and Paul. It is spectacular because the result is that Romney and Paul actual swap their final score!

Here are the 2 counties using my representation. It is totally mathematically equivalent to The Man's original work. The very same data from another angle.

On those chart, a 1-to-1 vote flip which ends up as a final score swap is very easy to detect. You want to see a cross between 2 candidates, and final candidates' scores at 100% equal to what they are on the left part of the chart. In Anderson, Romney scores at 100% is Paul's at 10%, and vice versa.

Notice as well that what The Man's plotted rightly as the fit-to-trend scores of Gingrich and Santorum become simply dead flat line on those charts. And boy, are they dead flat! "Natural", to be expected/predicted flatlines...

Now do those 1-to-1+final score swap crosses happen a lot? No. A quick perusal reveals another one in Greenwood and that's about it. Out of 47 counties. So flipping votes is the goal, complete score swap looks like a collateral, unintended coincidence.

 

[Mark37snj]

AWSOME work again as usual. I have one question, in Anderson SC after Paul and Romney swap positions, Paul continues to drop while Romney flatlines. Gingrich and Santorum are also flatlined. So where are the votes from Paul going? It's as if the algorithym, knowing what the final overall vote total is, and having locked in all the other canidates, has a known # of votes to allocate and then it figures out Paul's final slope, his end spot, so the overall vote totals add up. Romney gets flatlined, after he passes Paul, when there is only enough votes left to give Paul a negative slope so he cannot possibly pass Romney.

EDIT: If you were to look at this from the state level then once they have everything set at the state level the individual precincts are then fed this scenario/settings and they individually end up where they do. But if this State Level First scenario was the case then wouldn't there be a correlation between Ron Paul's final state level slope and his pricincts final slopes across all the affected pricincts, possibly given/correcting for vote %.

 

[Liberty1789]

Now the most important part of SC for me (except confirming that the vote eating algorithm would turn agressively to Gingrich) was: how populated will be the flatline counties. This is crucial.

What is the largest county with positively no symptoms of vote flipping?

[TABLE="class: grid, width: 450"]
[TR]
[TD]State[/TD]
[TD]County[/TD]
[TD]Votes Cast[/TD]
[TD]as % of state vote[/TD]
[/TR]
[TR]
[TD]Iowa[/TD]
[TD]Sioux[/TD]
[TD]2,072[/TD]
[TD]4%[/TD]
[/TR]
[TR]
[TD]New Hampshire[/TD]
[TD]Coos[/TD]
[TD]4,632[/TD]
[TD]2%[/TD]
[/TR]
[TR]
[TD]Nevada[/TD]
[TD]Douglas[/TD]
[TD]2,023[/TD]
[TD]7%[/TD]
[/TR]
[TR]
[TD]South Carolina[/TD]
[TD]Pickens[/TD]
[TD]17,787[/TD]
[TD]3%[/TD]
[/TR]
[/TABLE]

18,000 votes? No Romney climb???

Before SC, all counties with a high population exhibit a statistically significant linear Romney climb.

Why not Pickens and many other SC 5,000-10,000 vote counties?

Because the algorithm's trigger is not base on precinct population.
It is based on precinct vote tally as % of county vote tally.

If relevent PERCENTAGE WISE, then the county is flipped.

This puts an end to all demographic dependency and speculation. Demographics/socio-economic variables cannot be causative factors. The vote flip is fully explain with % of total votes cast and nothing else is necessary.

When I say fully, for our statistician friends, I mean that a simple:

Romney climb C= a x % of total votes cast + b

gives me R-squared of 98-99%% by the bucket.

It means that it explains 98-99% of the variance in Romney's score! That is why we feel entitle to say : crude algorithm!!

The switch on/off points are clearly visible on most charts.

Now as a reward for your kind patience with me, chart galore incoming!

 

[Liberty1789]

No vote flip

 

[Liberty1789]

 

[Liberty1789]

 

[Liberty1789]