Flipping the vote against Ron Paul in South Carolina?

[Liberty1789]

Excellent analysis Liberty1789. I want to voice some observations: There IS only one DEMOGRAPHIC that is related to Romney's vote receiving percentage in the 2012 GOP Presidential Preference Primary. That demographic is THE NUMBER OF VOTES AT EACH PRECINCT!

I am sure that you did not mean number of votes but % of total votes cast, right?

I have included the t-stat as per your kind request.

It measures a sort of standardized "wobbliness" of the slope. And here the numbers show that there no wobbliness whatsoever, which is amazing indeed. It completely violates poll science, a mathematical law of gravity!

To give an simple illustration. Take an urn with 10 red or black balls in it and your are trying to estimate the proportion of each. Normally, after having drawn 9 balls, you have a pretty good estimate of what the final proportion will be, right? Your interval of confidence is very narrow by now: 1 ball left to guess out of 10, so the final proportion can only move by 10% from the number after having drawn 9 balls. What Romney's numbers show is that your are no better guessing at 90% than at 50% what his final score will be. WHAT ??????????

 

[cjm]

All debunk attempts so far boil down to 'demographics' and all fall flat. The future report must spend an inordinate amount of time showing exactly why demographics can not explain these anomalies.

The original report had projections based on low turnout precincts. Those projections did not follow proper statistical methodology based on cluster sampling which depends on demographics. As long as projections are left out of a new report, I think everything else is fine with the caveat that I'm taking the high level statistics at face value and cannot confirm them myself. If the plot itself is non-random and algorithmically smooth, and proves vote tampering, let's not muddy the waters with projections.

 

[The Man]

OK- yes Liberty1789 you are right... again.

 

[affa]

is it worth sussing out an outline of the report here? That is, define sections, sequence, and scope of the report for peer review?

 

[bbwarfield]

is it worth sussing out an outline of the report here? That is, define sections, sequence, and scope of the report for peer review?

probably best to let the people who crunched the numbers do it over private messages and offer it on here later.... unfortuantely people like me would start complaining about them not following rhetorical style and other stuff ..... id probably have to resort to far side comics again

 

[affa]

demographics by county for South Carolina

Greenville breakdown
http://quickfacts.census.gov/qfd/states/45/45045.html

there is a tab in the upper left that lets you switch counties. you can also 'browse data sets' by county.

info on housing, education, income, race, more.

Liberty/The Man -- can you think of a way to chart votes by county against these demographics to prove demographics aren't at work? perhaps for each demographic, show all counties by candidate? not sure how. or does this only muddy the waters?
(note - i don't think demographics can ever explain what we're seeing... but perhaps we can use census data to prove it)

 

[bbwarfield]

demographics by county for South Carolina

Greenville breakdown
http://quickfacts.census.gov/qfd/states/45/45045.html

there is a tab in the upper left that lets you switch counties. you can also 'browse data sets' by county.

info on housing, education, income, race, more.

Liberty/The Man -- can you think of a way to chart votes by county against these demographics to prove demographics aren't at work? perhaps for each demographic, show all counties by candidate? not sure how. or does this only muddy the waters?
(note - i don't think demographics can ever explain what we're seeing... but perhaps we can use census data to prove it)

the amount of demographics would be astronomical.... the ethnic background seems to have no sway from what ive already looked up (and posted articles on)

wealth is the one thing that MAY... repeat MAY show some correlation.....


but the last chart liberty showed and the way it proves the unnaturalness of Romneys siphoning.... i think the demographics is become a moot point.... possibly look at three counties to see if the wealth demographic explained any of it...... but i think creating such a chart would be fruitless if a sample is done that proves the wealth demographic was not the issue

 

[bcreps85]

wealth is the one thing that MAY... repeat MAY show some correlation.....

One thing I read on Maine did suggest that many of Romney's votes were from the 5 richest counties, so you are correct that someone may find some correlation here, though I doubt it will be enough to actually disprove the original hypothesis. There just aren't that many "one percenters"...they can't make up a significant enough voting block to be the explanation for this, IMHO.

 

[bbwarfield]

One thing I read on Maine did suggest that many of Romney's votes were from the 5 richest counties, so you are correct that someone may find some correlation here, though I doubt it will be enough to actually disprove the original hypothesis. There just aren't that many "one percenters"...they can't make up a significant enough voting block to be the explanation for this, IMHO.

1% probably not.... but there are alot of places with people making over 200,000 a year.... wich tend toward Romney according to available data. Theres an instance of one precinct wich is almost all 200,000 plus (kinda like a lake district) so while wealth could be a factor it wouldnt create these smooth lines and siphoning systems (In Greenville I have personal knowledge that most of the 1% went to Newt... the whose whys and hows of that are privledged information coming from my job)

im big on data sets being similar/well defined..... south carolina really has only two demographics at play (race/income) race is a dead end as less than 3% of minorities made up the vote in the primary.... wealth could be (but probably is not) a factor ....

So maybe some random sampling.... but no need for full scale search with the demographics in sc.... unless the sampling shows it has a chance of being a debunk

 

[affa]

So maybe some random sampling.... but no need for full scale search with the demographics in sc.... unless the sampling shows it has a chance of being a debunk

my point was more to 'pre-dunk potential debunks' than anything else. I have posted here before that I think demographics causing a 1:1 flip is astronomically unlikely in one precinct, let alone multiple. but if someone can come up with a way to quickly show that, say, race, education, and wealth (based on these numbers) do not explain it, we close that argument up forever. I do not know what charts would be needed to show this, however.

 

[bbwarfield]

http://www.cnn.com/election/2012/primaries/epolls/sc

this would prove race had nothing to do with it....

education a little

wealth a little

age is a good possibility i didnt think of......

 

[Danan]

http://www.cnn.com/election/2012/primaries/epolls/sc

this would prove race had nothing to do with it....

education a little

wealth a little

age is a good possibility i didnt think of......

Hm. Shouldn't the average age go down with increasing precinct size? Romney does better with older folks than with younger ones. And at least where I live younger people tend to move away from rural areas to the cities.

 

[cjm]

the amount of demographics would be astronomical.... the ethnic background seems to have no sway from what ive already looked up (and posted articles on)

wealth is the one thing that MAY... repeat MAY show some correlation.....


but the last chart liberty showed and the way it proves the unnaturalness of Romneys siphoning.... i think the demographics is become a moot point....

^this.

possibly look at three counties....

^not this

(note - i don't think demographics can ever explain what we're seeing...

affa, it may surprise you a bit given my posts on this thread (I don't think I've worded them as carefully as I could), but I agree with this statement 100%. I have already admitted that I was confused about the separate arguments for fraud (projections vs. smoothing), and since Liberty took the time to explain the results of his statistical analysis (thanks again, Liberty), I see that demographics is not an issue when we see an algorithmically smoothed curve -- a moot point, as bbwarfield says. I agree completely.

But, if someone issues a 300 page report on this statistical analysis without ever mentioning the word "demographics," but at the end says, "demographics can't explain this," then all will be for naught. This has nothing to do about being right or wrong, it's about argumentation, red herrings, and human nature. You've seen dozens of examples on this thread alone. That one statement that says, "demographics can't explain this" without an actual demographic analysis to back it up, opens the door for discussions on demographics. Personally, I strongly advise not going there.

Has anyone ever written an email at work or with some organization where a solidly crafted case was made for one item, and a few comments were added to editorialize? How many times do your adversaries respond to the solidly crafted case? Never. They pick one of the editorial comments and ignore the rest. Remember that we're not dealing with the scientific community here, this is politics. Whatever is put out needs to stay focused and be rock solid. I think the work will stand up to any demographic challenges, but until we're ready to perform that analysis, "we think" is all we can say. To which the establishment response will be, "They make a good case for suspicious activity, but never performed the demographic analysis to back up their claims." If demographics is a moot point, which I think it is, it's not worth mentioning at all. Heck, increased variety in the demographics should increase the randomness of the plots, right? But until we've done the work, "should increase" is all we can say and the detractors will run with it.

Maybe I'm just paranoid or overly cautious. But my recommendation to the report generators is to stick with the non-randomness of the plots and focus on SC. If anyone is adamant about other states, make them separate reports so that they can each stand on their own.

I'm sure my posts on this thread have annoyed some of you from time to time, I appreciate you taking the time to read them. Honestly, I'm on your side in this.

 

[cjm]

Hm. Shouldn't the average age go down with increasing precinct size? Romney does better with older folks than with younger ones. And at least where I live younger people tend to move away from rural areas to the cities.

Depends on the area. Where I live, younger people are drawn to the cities, but those cities are heavily Democrat. Some precincts have turnouts of 10-15 people in the 2008 Republican primary, so they are much "smaller" than the "larger" (higher turnout) precincts on the outskirts of town that are more rural and more Republican. You can probably classify each case, but each case within a classification will still be fairly unique.

 

[affa]

But, if someone issues a 300 page report on this statistical analysis without ever mentioning the word "demographics," but at the end says, "demographics can't explain this," then all will be for naught. This has nothing to do about being right or wrong, it's about argumentation, red herrings, and human nature. You've seen dozens of examples on this thread alone. That one statement that says, "demographics can't explain this" without an actual demographic analysis to back it up, opens the door for discussions on demographics. Personally, I strongly advise not going there.

I do understand where you're coming from, but I think you're incorrect in thinking that if we discuss 'demographics' it opens the door to discussions. Why? I think that door is wide open to begin with. It's the 'go to' explanation for people who haven't took the time to read through this thread, and it will be the go to explanation for those who skim the report without understanding it.

That's why I think it needs to be addressed, and debunked as a potential explanation. I put up the census information in case someone can come up with a way to debunk it statistically, though I believe it can be logically debunked as well.

But I do not think simply not mentioning it will make the problem go away; in fact, I think it makes the report easier to dismiss (oh, they didn't take demographics into consideration)... when in fact, we did discuss demographics to death and the charts, once understood, dismiss demographics as a cause.

 

[bbwarfield]

This....



Im not suggesting including the three county random check in the report.... just for the sake of "due diligence" so we know we did it.... and to appease those who want it all checked... at least do some random sampling


okay... now this is gonna sound COMPLETELY ridiculous... but hear me out.

Many of you remember there was a "exit polling" companny ( named ccap or something) that came out with completely ridiculous numbers that were no where near the end total... the company had a good sounding system to come up with the numbers and it all looked scientific (except the numbers always being in youtube slide shows) now..... the MSM exit polls match the final numbers constantly.

Maybe this polling place actually was right? maybe the vote flipping actually is the reason they looked like pro-paul coocs? if we unflipped them and suddenly they are all within the margin of error of this group that tried to do an accurate number apart from media bias..... most places called it fraud..... but maybe unflipping the vote may prove them to be honest brokers that just have cruddy web design?

 

[cjm]

That's why I think it needs to be addressed, and debunked as a potential explanation. I put up the census information in case someone can come up with a way to debunk it statistically, though I believe it can be logically debunked as well.

Fair enough. Reasonable people can agree to disagree on matters of opinion -- and just to clarify, I mean the opinion whether to address this in a report or let it be.

 

[bbwarfield]

Fair enough. Reasonable people can agree to disagree on matters of opinion -- and just to clarify, I mean the opinion whether to address this in a report or let it be.

unfortunately for most people on this forum..... i am not "reasonable"

 

[cjm]

Many of you remember there was a "exit polling" companny ( named ccap or something)....

I remember them. I went to track down some of their information recently and the youtube clips that I found had been pulled for SC. YMMV.

 

[bbwarfield]

I remember them. I went to track down some of their information recently and the youtube clips that I found had been pulled for SC. YMMV.

your right.... BUT (and there always is one with me) in comes daily paul and they have it written down with some people ommenting on various aspects of them http://www.dailypaul.com/206718/early-exit-poll-has-ron-paul-in-second-with-26

paul in second at 26% actually sounds like the real number for south carolina