RPF members uncover possible evidence of election fraud in Romney's favor

[Mageant]

Mathematical evidence of election fraud in the recent Republican primaries has been discovered by people on ronpaulforums.com:

http://www.ronpaulforums.com/showthread.php?363915-We-NEED-more-hands-on-deck.-Significant-evidence-of-Algorythmic-vote-flipping.

https://docs.google.com/viewer?a=v&pid=explorer&chrome=true&srcid=0B_wWkfsJPShUMWQxMTc2NzgtM2MzYy00ZGJhLWI1MmYtMWU2ZGU1OWZkZjhk

The gist of the matter is the following. It seems as if votes are being flipped 1:1 in favor of Romney at the expense of another candidate who is almost always Ron Paul (Gingrich in some cases in South Carolina too).

The mathematical reasoning is as follows. If you order the vote results by precinct size then Paul always does better than Romney in smaller precincts and but this changes in a clean, linear fashion in favor of Romney as the precinct size goes up.

The proability of this happening by pure chance alone is astronomical (thus impossible).

Any imaginary "demographic factor" (because of the precinct size) that might explain this seems also unreasonable because it only affects one candidate positively and only one candidate negatively in a clean, linear fashion (ie without any ups or downs).

This also happens across different states. Rick Santorum, for example, is totally unaffected.

Mathematically, you would expect somethinkg like this to happen:


But look what really happens:

[bb113]

This is interesting. I'm not convinced yet though. How would this fraud be accomplished?

[Hawker]

Is Ron Paul supposed to have helped?  I noticed some stats that in 20 debates, Ron Paul has not criticised Romney once while he has attached every other candidate that was the "not Romney" flavour of the month.

[Maged]

Sorry, but the math is lying to you.

How's this for statistics? Out of all of these primaries, nobody has seen actual evidence of fraud. Moreover, Republicans are more likely to ensure that their vote was actually counted, so they'd pay attention to the counting and reporting. Knowing all this, what's the probability that all of this fraud is going on, yet nobody has any actual evidence? My guess: almost 0%. Hell, the general election would be easier to rig than the primary.

This is why people don't want to elect Ron Paul.

[Mageant]

This is interesting. I'm not convinced yet though. How would this fraud be accomplished?

By manipulating the software that tabulates the votes.

[Maged]

This is interesting. I'm not convinced yet though. How would this fraud be accomplished?

By manipulating the software that tabulates the votes.

Riddle me this: How could you do that without anyone detecting it? Remember, each precinct counted their votes by hand and reported them, meaning that each precinct has their own copy of the tally.

[Mageant]

This is interesting. I'm not convinced yet though. How would this fraud be accomplished?

By manipulating the software that tabulates the votes.

Riddle me this: How could you do that without anyone detecting it? Remember, each precinct counted their votes by hand and reported them, meaning that each precinct has their own copy of the tally.

Not all precincts count their votes by hand.
People have detected it.

[ineededausername]

whole thing got totally debunked on reddit...

[bb113]

whole thing got totally debunked on reddit...

Link?

[Maged]

whole thing got totally debunked on reddit...

Link?

http://www.reddit.com/r/AskReddit/comments/qb9ea/reddit_can_you_debunk_this_some_people_with/

Turns out that it's demographics that would cause this, as expected.

[grue]

This is interesting. I'm not convinced yet though. How would this fraud be accomplished?

By manipulating the software that tabulates the votes.

Riddle me this: How could you do that without anyone detecting it? Remember, each precinct counted their votes by hand and reported them, meaning that each precinct has their own copy of the tally.

Not all precincts count their votes by hand.
People have detected it.

tinfoil hat activate!

[bb113]

whole thing got totally debunked on reddit...

Link?

http://www.reddit.com/r/AskReddit/comments/qb9ea/reddit_can_you_debunk_this_some_people_with/

Turns out that it's demographics that would cause this, as expected.

I wouldn't say it has been "debunked" yet. As mentioned in the reddit thread, we require historical controls and current demographic info to draw any conclusions. It is interesting that the proportion of Romney voters seems to rise as a function of precinct size and the opposite for Ron Paul. I would say that on the face of it demographic differences are a more plausible explanation than election fraud, but this should be pretty easy to investigate.

I only looked at the Anderson County, SC data (discussed in the PDF) for myself. But from this I do see that Gingrich and Santorum proportions are not a function of precinct size, while Romney and Paul are nearly exactly opposite. This phenomenon is really interesting. How would we explain this? Is there a subpopulation of voters that vote Paul (if rural) or Romney (if urban) but are totally uninterested in the other two candidates? It just seems strange to me since Romney and Paul are at opposite ends of the spectrum as far as I am concerned. For example, I think it would make much more sense to go from Paul -> Gingrich. That is just me though.

If anyone cares, here is the Anderson county data in scatter and line format. I think it is more intuitive to understand it this way rather than looking at cumulative votes. On the other hand, it is easier to detect the patterns if you transform the data to depict the fact that larger counties have more weight in affecting the final results (see final chart).

Source:
http://www.enr-scvotes.org/SC/Anderson/36835/66538/en/reports.html

Edit: I just realized forgot to keep the color code consistent... just check the legends for each.
Number of votes by precinct size:

Proportion of votes by precinct size:

Proportion of votes by cumulative total as if counted in order of smallest to largest precinct:

[Mageant]

Bingo. This is what we need to do to show the debunkers that urban=Romney/rural = Paul is false. I found this in only 20 minutes of looking. I'm sure we could do something similar to show that it's # of votes in a precinct that maters, not whether or not the precinct is urban or rural.

First I pulled up a road/city map of South Carolina here. Then I pulled up a precinct map here. If you look at the road/city map, you will see that the entire northern edge of Greenville County is rural. This means that Paul should win, right? Wrong. The three precincts that represent the northern rural area are Jennings Mill, Tigerville, and Gowensville. Here are the results: Jennings Mill cast 470 votes, with 61 to Paul, 167 to Romney; Tigersville cast 756 votes, with 109 to Paul, and 194 to Romney; Gowensville cast 558 votes, with 109 to Paul, 118 to Romney. Why did Romney win the majority of these rural votes? You tell me.

Now it gets better. If the debunkers are correct, then Romney should win the city of Greenville because "city folk love Romney!" It's not that simple. Yes, when you add together all 29 precincts that comprise the city of Greenville, Romney wins big time. However, if you use The Man's 277-vote threshold, here is what you come up with.

Precincts with less than 277 votes - Paul wins 10, Romney wins 3, 2 tie.
Precincts with more than 277 votes - Paul wins 0, Romney wins 14, 0 tie.

This is the came city! In the same urban setting Romney wins all the high-vote precincts, and Ron wins all the low-vote precincts!


Useful links.

Greenville precinct details.
Polling Locations for all precincts in South Carolina.
South Carolina results by county/precinct. (click on county on map, then "contest detail map," then "view all data."

http://www.ronpaulforums.com/showthread.php?363915-We-NEED-more-hands-on-deck.-Significant-evidence-of-Algorithmic-vote-flipping.&p=4233368&viewfull=1#post4233368

[Maged]

Bingo. This is what we need to do to show the debunkers that urban=Romney/rural = Paul is false. I found this in only 20 minutes of looking. I'm sure we could do something similar to show that it's # of votes in a precinct that maters, not whether or not the precinct is urban or rural.

First I pulled up a road/city map of South Carolina here. Then I pulled up a precinct map here. If you look at the road/city map, you will see that the entire northern edge of Greenville County is rural. This means that Paul should win, right? Wrong. The three precincts that represent the northern rural area are Jennings Mill, Tigerville, and Gowensville. Here are the results: Jennings Mill cast 470 votes, with 61 to Paul, 167 to Romney; Tigersville cast 756 votes, with 109 to Paul, and 194 to Romney; Gowensville cast 558 votes, with 109 to Paul, 118 to Romney. Why did Romney win the majority of these rural votes? You tell me.

Now it gets better. If the debunkers are correct, then Romney should win the city of Greenville because "city folk love Romney!" It's not that simple. Yes, when you add together all 29 precincts that comprise the city of Greenville, Romney wins big time. However, if you use The Man's 277-vote threshold, here is what you come up with.

Precincts with less than 277 votes - Paul wins 10, Romney wins 3, 2 tie.
Precincts with more than 277 votes - Paul wins 0, Romney wins 14, 0 tie.

This is the came city! In the same urban setting Romney wins all the high-vote precincts, and Ron wins all the low-vote precincts!


Useful links.

Greenville precinct details.
Polling Locations for all precincts in South Carolina.
South Carolina results by county/precinct. (click on county on map, then "contest detail map," then "view all data."

http://www.ronpaulforums.com/showthread.php?363915-We-NEED-more-hands-on-deck.-Significant-evidence-of-Algorithmic-vote-flipping.&p=4233368&viewfull=1#post4233368

Because, clearly, the trend should hold in 100% of cases. This would be like me proving to these guys that there wasn't any fraud at all by providing a single precinct/city that doesn't appear to be manipulated according to their statistical methods. In both cases, the anecdote is just as valid. (as in, not at all)

[LoupGaroux]

For this to mean anything you would have to assume that all voters will behave statistically identically, regardless of precinct. Applying algorithmic logic to proclaim a fraud is using bad math to cook the books in your favor. A precinct has very little to validate it as a sociological analysis tool, other than the fact that it is a carefully constructed set of boundaries to favor one party over the other.

Let's look at Greenville, SC indeed. Your definition of urban/rural is skewed because of the influence of South Carolina law on annexing areas into urban units. What is being described as rural in this analysis is largely what most people would consider urban voters. The population is skewed dramatically towards a minority population, with almost 2.5X the percentage of African-Americans as the mean in the US, and a relatively economically disadvantaged community, with 16.1% of the population living below the poverty line.

That recipe will give you very squirelly numbers when you take a very small statistical sample, like 277 voters. They are probably not the majority party in the precinct, and in any primary, only the truly committed and faithful come out to vote- a strong Paul supporter will have an impact in a smaller precinct specifically because they stand out, and are going to certainly come out to vote. While the same number of Paul supporters in a larger precinct will be swallowed up the the volume of voters who feel compelled for other candidates. The volume of Paul supporters is not evenly distributed over every voting population.

So what they have uncovered is the reality of participatory statistics, and the facts of distributed populations. Fringe candidates will tend to have concentrated pockets of support, mainstream candidates will enjoy stronger support from the bulk of the voting population. Whatever pocket of support exists, the stronger their impact will be when the statistical sample gets smaller.

[Mageant]

So what they have uncovered is the reality of participatory statistics, and the facts of distributed populations. Fringe candidates will tend to have concentrated pockets of support, mainstream candidates will enjoy stronger support from the bulk of the voting population. Whatever pocket of support exists, the stronger their impact will be when the statistical sample gets smaller.

It doesn't make sense that the only correlation for these "pockets of support" seems to be precinct size, and furthermore only to the benefit of Romney as precinct size increases. The missing "randomness" is one the problems.

[LoupGaroux]

Precinct size is not a correlation, its the basis of the sample size. The distribution of Paul votes is not related to precinct size, but the voting preference of the voters in each precinct.

Analysis after the last presidential election showed that John McCain would have been the clear winner, by a dramatic margin if the result was based on precincts or counties won. It wasn't, it was based on electoral votes won. A fact that liberal strategists have learned from, and their conservative brethren are woefully behind on. Playing the game for votes is a surefire way to lose, play it to win by the numbers, and you have a shot. The Republicans fail to grasp that and have historically shot themselves in the foot time and time again by having knife fights in the primaries, and forcing lots of energy to be spent in campaigning for irrelevant votes, like the Iowa caucuses, New Hampshire and other insignificant states. It is all going to come down to buying enough votes in the swing states. And in that battle, regrettably, Paul has no shot.

[bb113]

Precinct size is not a correlation, its the basis of the sample size. The distribution of Paul votes is not related to precinct size, but the voting preference of the voters in each precinct.

Not saying there is any fraud being detected here (and I am someone who thinks the fix has been in with regards to Romney), but the proportion of paul voters definitely looks like it is a function of precinct size. Also it is obvious that the distribution of votes is related to the reference of the voters in each precinct (as long as the count is clean)... Please explain.

[LoupGaroux]

My point is that for this to be evidence of fraud, or result manipulation, one would have to accept that every single precinct will have exactly the same ratios of votes for the various candidates. If X got 13% in precinct #101 he also got exactly the same 13% in every other precinct in the country. In a smaller precinct, like our theoretical 277 vote precinct, two families coming out to vote for Candidate X will skew the numbers way beyond that norm.

The description of Greenville, SC is to identify that in that particular case the definition of "urban" is not an accurate reflection of the actual demographic because of a glitch in South Carolina annexation law. And looking at the demographics, there is a radical departure from the national averages in terms of racial make-up in the population. I am suggesting that you cannot make iron-clad rules, and allege improprieties because of those rules looking at an algorithmic set of rules. This is statistical analysis, and statistical realities apply, not simple mathematical constraints.

And if there was any cooking of the books, it was probably done by Democrats anyway, to try and get a perceptually wackier candidate than Romney for Emperor Barry the First to face in November. That way the toady media can have a field day tearing him apart and make their Anointed One President for another four years.

[bb113]

Oh yea, that Greenville "evidence" was dumb.