(Huge) delegate vote anomaly in Alabama verified

[parocks]

The vote-for-all idiocy theory requires a lot of voters to keep voting wrong. In Tuscaloosa, parocks proposes a 6.4% number. Additionally he argues that some voters started to vote for all but stopped at some point in the process: those are the "partly wrong" votes cast. Their number should show gradual erosion as you go down the list of delegates on the ballot, but it is not the case. The model does not fit the data, as shown in the Tuscaloosa chart below: it requires a lot of "partly wrong" voters to get to Paul's delegates (hence the 0%), but then it requires a lot of them to suddenly disappear when you get to Romney (hence the -46%). Force-fitting in all its splendor.

The vote-for-all theory is clever, there is no denying that, but it does not work, there is no denying that either. And the chart above is further evidence that the anomaly is PAUL-SPECIFIC.

I'm not going to use the word idiot. Those aren't people who voted for ALL, those are people who voted in the wrong races.

These aren't exact numbers, these are plausible numbers, estimates, approximations. Certainly, the 540 and the 540 aren't going to be the same.

Something like this.

 

[dsw]

http://www.filedropper.com/alabamadelegateraces

Thanks! That matches my data, and I verified that only 39 precincts have zero total votes in the preference races. So I can't replicate the graph I wanted to look at more closely without some more information about why the datasets appear to be different.

 

[The Man]

I posted a variation of this earlier and wanted to expand. This is a graph of all candidates' "Votes minus Delegates" versus cumulative precincts vote totals ascending order. Its a great way to see ONLY the manipulated votes from the Alabama 2012 Primary.

 

[The Man]

Thanks! That matches my data, and I verified that only 39 precincts have zero total votes in the preference races. So I can't replicate the graph I wanted to look at more closely without some more information about why the datasets appear to be different.

Hmmmmmmmmmmmmmmmmm.........

 

[dsw]

Hmmmmmmmmmmmmmmmmm.........

Hmmm? The data Liberty just posted has 39 precincts with zero total votes. That agrees with the data I used. The data you posted in #549 here doesn't change in the x value for the first 54 rows, suggesting that your data has 54 precincts with zero total votes. That's not the only difference, it's just the first one, but later values don't line up because of it so it's hard to figure out exactly what else differs. As things stand I can't replicate the result -- which is not to say that this won't turn out to be a silly mistake on my part in the end, it's just where I got stuck.

 

[Liberty1789]

And, by the way, I was reminded of one of my favorite quotes when reading the heated exchanges of the last week:

"Truth springs from argument amongst friends." David Hume (genius)

Friends is the keyword here.

 

[The Man]

Hmmm? The data Liberty just posted has 39 precincts with zero total votes. That agrees with the data I used. The data you posted in #549 here doesn't change in the x value for the first 54 rows, suggesting that your data has 54 precincts with zero total votes. That's not the only difference, it's just the first one, but later values don't line up because of it so it's hard to figure out exactly what else differs. As things stand I can't replicate the result -- which is not to say that this won't turn out to be a silly mistake on my part in the end, it's just where I got stuck.

So post your graph then. Get on with it Let's see how different yours is.

"Rule#4. Use a straw man.
Find or create a seeming element of your opponent's argument which you can easily knock down to make yourself look good and the opponent to look bad. Either make up an issue you may safely imply exists based on your interpretation of the opponent/opponent arguments/situation, or select the weakest aspect of the weakest charges. Amplify their significance and destroy them in a way which appears to debunk all the charges, real and fabricated alike, while actually avoiding discussion of the real issues."

 

[dsw]

So post your graph then. Get on with it Let's see how different yours is.

"Rule#4. Use a straw man.
Find or create a seeming element of your opponent's argument which you can easily knock down to make yourself look good and the opponent to look bad. Either make up an issue you may safely imply exists based on your interpretation of the opponent/opponent arguments/situation, or select the weakest aspect of the weakest charges. Amplify their significance and destroy them in a way which appears to debunk all the charges, real and fabricated alike, while actually avoiding discussion of the real issues."

I posted it once, you objected that it didn't look the same, I posted my data, you posted your data, I went digging into the reasons that it didn't look quite like yours and ended up finding that your data does not seem to agree with Liberty's. Here it is again. The key difference is that that the graph I was trying to replicate had a sharp "knee" and this one doesn't.

And yet again: it could still be my mistake, but where things stand right now is that your data appears to indicate 54 precincts with zero total votes, and Liberty's dataset only has 39 (not that that's the only difference, but it's the first one) and it would sure be nice to start out by making sure that we're on the same page at least about what data we're using. If trying to replicate your results is what you consider a straw man, then fine, call it what you want.

 

[drummergirl]

I'm not going to use the word idiot. Those aren't people who voted for ALL, those are people who voted in the wrong races.

These aren't exact numbers, these are plausible numbers, estimates, approximations. Certainly, the 540 and the 540 aren't going to be the same.

Something like this.

Please parocks, your hypothesis has been discredited 6 different ways to Sunday. Nice idea, but the math does not back it up.

 

[The Man]

I posted it once, you objected that it didn't look the same, I posted my data, you posted your data, I went digging into the reasons that it didn't look quite like yours and ended up finding that your data does not seem to agree with Liberty's. Here it is again. The key difference is that that the graph I was trying to replicate had a sharp "knee" and this one doesn't.

And yet again: it could still be my mistake, but where things stand right now is that your data appears to indicate 54 precincts with zero total votes, and Liberty's dataset only has 39 (not that that's the only difference, but it's the first one) and it would sure be nice to start out by making sure that we're on the same page at least about what data we're using. If trying to replicate your results is what you consider a straw man, then fine, call it what you want.

Well try these other 12. You're exposed DSW.

 

[dsw]

Well try these other 12. You're exposed DSW.

Oh? My data agrees with what Liberty posted. Does yours?

(Each line = one delegate)

 

[RonRules]

Oh? My data agrees with what Liberty posted. Does yours?

(Each line = one delegate)

OK fine, after a week or so trying and wasting people's time you finally manage to reproduce someone's chart. Wouldn't it be more productive if you discovered new anomalies that indicate flipping? This is the purpose of this thread. Either find positive anomalies or go nag in the "no-fraud" thread. There's plenty of more stuff to find and prove in Alabama alone.

You tried to discredit my analysis in the other (main) thread by claiming that the Dane county WI Sheriff would not indicate flipping, without spending the time to try it. Instead I showed you that 7 of those ancillary races were favoring Republicans by flipping. That's now a major problem and embarrassment for the GOP.

Go grab fresh data analyze it and post positive results in this thread (for Alabama results) or the main thread.

BTW, has anybody contacted the Secretary of State yet?

 

[The Man]

Oh? My data agrees with what Liberty posted. Does yours?

(Each line = one delegate)

So you're saying that because YOUR graph isn't as sharp at the elbow that it somehow lessens the fact that Romney's candidate votes slope increases dramatically in relation to his delegate curve? Plllllleeeease.

 

[dsw]

You tried to discredit my analysis in the other (main) thread by claiming that the Dane county WI Sheriff would not indicate flipping, without spending the time to try it. Instead I showed you that 7 of those ancillary races were favoring Republicans by flipping. That's now a major problem and embarrassment for the GOP.

Without spending time to try it?? I'm the one who first posted the graph for the sheriff race, which didn't flatten out, and another old one that also didn't flatten out, and I could post a bunch more. Then I posted one from Oregon, a vote on a bond issue, and there are a bunch more in that county as well. Whether or not they indicate flipping is another question (I don't think they do) but if they do then I'm helping you prove that it's more widespread, and goes back more years, than anyone imagined.

I found that last one by looking for a large-area county with a good-sized college town surrounded by small towns and farmland. *Lots* of graphs that don't flatten out there, as I expected. And if it's mathematically necessary that a sample taken from the smallest precincts must be a representative sample of all the data, then that means that election fraud in that Oregon county has been rampant since the 90's, as far back as they've posted data.

And did you think I replicated The Man's graph? The point is that I have been unable to replicate his result, using Liberty's data, and finally boiled it down to showing that his data doesn't seem to be the same as Liberty's. It might still be my mistake, who knows? He doesn't seem to want to answer questions about that part. Hard to investigate a "knee" in the data if you can't replicate the knee.

But as to your real point: if only positive results (even bogus ones, as long as they're positive) are wanted here, then I'm probably in the wrong place. Am I getting the message clearly there?

 

[dsw]

So you're saying that because YOUR graph isn't as sharp at the elbow that it somehow lessens the fact that Romney's candidate votes slope increases dramatically in relation to his delegate curve? Plllllleeeease.

Well I thought the whole "knee" thing was the "just wow" factor. The straight lines superimposed on the graph showing *so clearly* the "crime" point, etc. I guess I just misunderstood.

 

[drummergirl]

I suspect dsw is asking questions other people have but they don't necessarily post. So, while I know it can be frustrating sometimes, I think it's good for those questions to get answers. Remember folks, we are on the same team (i.e. trying to get Ron Paul elected).

 

[affa]

Oh? My data agrees with what Liberty posted. Does yours?

(Each line = one delegate)

Why are you acting as if this chart does not show the same exact problem that The Man's charts do?
They clearly show that Romney Votes - Romney Delegate votes change in slope after a certain number of votes counted.

That is probably the clearest evidence of fraud I've seen so far, since in order for demographics to explain it you'd need the same people that start voting for Romney in 'big' precincts to also not vote for his delegates, which doesn't even begin to make sense.

Re: the 'knee'. C'mon. Your charts show the same thing as The Mans. This is a chart; the knee will be more dramatic based on scale of chart, and also is a bit more evident when you have the straight lines. However, yours show exactly the same problem. Why are you acting like they don't? Yours are just as 'wow' as The Man's. That your charts don't show the knee with quite as much 'wow' factor means absolutely nothing, since your charts still duplicate The Man's work and show the same evidence of fraud.

As I've said before, these charts are perhaps the best case for fraud I've seen yet, in combination with everything else we've seen, because we finally see Romney matched up against Romney (delegates) in the same place, same year, same election... and see him outpace his OWN delegates in the exact same way we've seen him outpace his opponents in other state's counties. Demographics can not explain this, no matter how much you torture the explanation. Fraud would explain it perfectly.

 

[affa]

But as to your real point: if only positive results (even bogus ones, as long as they're positive) are wanted here, then I'm probably in the wrong place. Am I getting the message clearly there?

Your charts ARE positive results. They replicate the problem The Man discovered.

As for you, you're welcome here, but you do have a history of driving conversation in circles, and focusing on random minutia while ignoring the elephant in the room.

Your charts are just as much evidence of fraud as The Man's. I'm more curious why you're acting like a trivial difference in the charts somehow negate's The Man's findings when your charts show the same problem.

 

[dsw]

Your charts are just as much evidence of fraud as The Man's. I'm more curious why you're acting like a trivial difference in the charts somehow negate's The Man's findings when your charts show the same problem.

A sharp knee, going from one linear slope to a different linear slope, would be a very interesting feature in a graph like this especially considering the smoothing effect, which will tend to turn correlations into nice gentle curves. (But not always. Don't forget VBC, where the "crime occurs here" point turned out to be where you hit the first of a large cluster of large contiguous precincts that also turned out to be the neighborhood with the largest concentration of homes listed for a million or more, and that one cluster of rich people skewing strongly pro-Mitt accounted for over 60% of the data to the right of the "crime" point, and that's without even considering a second geographic cluster near the first.)

To me the knee was *the* thing that made the result look like it might be worth replicating and investigating. Two linear regions connecting with a sharp knee would be a very interesting phenomenon, given how the data is being presented. But if it can't be replicated from Liberty's data then it suddenly gets a lot less interesting.

When I saw that it wasn't a sharp knee, here's what I did. The alleged crime point corresponds to a precinct size of around 724 IIRC. So we're looking for differences between the precincts with sizes below that and precincts with sizes above that. And since there are two numbers involved, Mitt's %vote and various ways of looking at his delegate scores, it also makes sense to separate those out and see how they're each behaving in the "under" and "over" subsets, so I did that. And I looked at the delegate numbers as a percentage of Mitt's %vote in order to remove the often visually misleading bias of graphing (x,y) values when there's a natural correlation between x and y, as there is here. But do you care? You see a graph that curves, make a simplistic argument against demographic explanations (assuming implicitly that there were not enough idiot voters to be part of that explanation, since to the extent there were idiot voters voting in every delegate race you would *expect* that as Mitt's %vote goes up the ratio of delegates to mitt would go down), and then you declare victory. Basically we're wasting each other's time here, so I'm going to help out by going into lurker mode. Forgive me if I can't resist kibitzing occasionally but I'll try to do it on the officially sanctioned skeptics thread.

 

[drummergirl]

A sharp knee, going from one linear slope to a different linear slope, would be a very interesting feature in a graph like this especially considering the smoothing effect, which will tend to turn correlations into nice gentle curves. (But not always. Don't forget VBC, where the "crime occurs here" point turned out to be where you hit the first of a large cluster of large contiguous precincts that also turned out to be the neighborhood with the largest concentration of homes listed for a million or more, and that one cluster of rich people skewing strongly pro-Mitt accounted for over 60% of the data to the right of the "crime" point, and that's without even considering a second geographic cluster near the first.)

To me the knee was *the* thing that made the result look like it might be worth replicating and investigating. Two linear regions connecting with a sharp knee would be a very interesting phenomenon, given how the data is being presented. But if it can't be replicated from Liberty's data then it suddenly gets a lot less interesting.

I'm not sure why you are getting bent out of shape here; your chart and The Man's are consistent with one another.