(Huge) delegate vote anomaly in Alabama verified

[affa]

The overvote on the Ron Paul delegates is explained by a bunch of people voting in races. 10% would explain it fully, and even go over. Since the question we're trying to answer is "how can Ron Paul get so many delegate votes and so few real votes", we don't have to worry too much about the numbers not matching up.

No, that's not the entire question. Any answer must also be able to be applied to other candidates with similiar results. That is, you can't assume 10% just voted for Ron Paul delegates, even though his were halfway down the second row. You need to apply this 10% to, at the very least, all candidates on the first page. And when you do that, the 10% solution isn't as 'clean' as you want it to be.

If it were that simple 'oh, 10% of people just idiotically voted for Ron Paul', then we'd be able to explain absolutely -any- anomaly that way. OH! 57% of people just randomly voted for Ron Paul.

[parocks]

Whatever hypothesis you start with, don't you end up at some point checking to see if you can find numbers that make it work? With the "idiot voters" theory, 10% is a rough estimate (and there's more to the story that I'll get back to in a second) that makes the numbers come out about right. One theory about Alabama was that around some percentage of Ron Paul's vote got thrown out, but they didn't adjust the delegate race totals to be consistent; to make that theory work the percentage of discarded votes ends up being something like 66%. (But that doesn't fit the data nearly as well as the idiot voter theory.) Someone asked me if I could make the data work with the "flipping algorithm" theory, and I tried some numbers without much success, but if I'd come up with numbers that worked would the air they were pulled from be any less thin than for the number 10 in the idiots theory or the number 66 in the discarded votes theory?

Just saying that 10% voted in all the races doesn't tell the whole story, but it seems likely (although Liberty1789 claims to have data disproving this) that some unmotivated or uninformed voters would just vote for their candidate and ignore all the other races on the ballot. Having 10% dutiful idiots voting in every race would give Ron Paul a huge boost, but Santorum and the others a much smaller boost that might be offset by lazy or uninformed voters. Newt and Mitt did have some delegate races with an excess of votes, and especially for Newt ballot placement probably explains a lot of that. The point is that the "10% were idiots" theory makes the largest and most interesting anomaly go away, and other voting patterns that aren't unreasonable would easily get us the rest of the way. Combining Ockham's razor and the observation that lots of voters are stupid, I think this is the theory to beat.

"The point is that the "10% were idiots" theory makes the largest and most interesting anomaly go away, and other voting patterns that aren't unreasonable would easily get us the rest of the way. Combining Ockham's razor and the observation that lots of voters are stupid, I think this is the theory to beat."

Right. Exactly.

[affa]

Okay, the Alabama 2008 ballot looked just as ridiculous.
http://www.co.baldwin.al.us/uploads/...ballot_Rep.pdf

So, did we have the same problem that year?

can anyone find 2008 results?

[parocks]

No, that's not the entire question. Any answer must also be able to be applied to other candidates with similiar results. That is, you can't assume 10% just voted for Ron Paul delegates, even though his were halfway down the second row. You need to apply this 10% to, at the very least, all candidates on the first page. And when you do that, the 10% solution isn't as 'clean' as you want it to be.

If it were that simple 'oh, 10% of people just idiotically voted for Ron Paul', then we'd be able to explain absolutely -any- anomaly that way. OH! 57% of people just randomly voted for Ron Paul.

No, we didn't set out to answer the question about the other candidates.

We set out to answer the question about Ron Paul.

I'm in agreement with dsm here.

And don't forget, this matter was settled over 2 weeks ago. I initially thought it was fraud and I personally, looked at the data, trying to determine what happened. I figured out that the truth of the matter would likely be found in the very rare precincts where Ron Paul did well. I guessed where they might be - guessed near University of Alabama, was right - and noticed that Ron Paul's numbers looked normal in those cases. I'm not sure who came up with "stupid voters" first. But this has all been settled over 2 weeks ago.

[parocks]

I don't know about that assumption. I've clerked a lot of elections and seen all kinds of voters. With a ballot that complex, there will be more voter errors than with a simple ballot just by the ballot design. And there will always be a few voters who don't see well, are tired, etc. that will make mistakes. 10% spoiled ballots is awfully high.

If the party rules are that you can't "mix and match", why isn't every mixed ballot thrown out as invalid?

"If the party rules are that you can't "mix and match", why isn't every mixed ballot thrown out as invalid?"

That's what this is about.

[parocks]

2 things

1) Please tell me you are not a professional analyst. I hate to think how many astronauts would be dead by now if the guys at NASA said, "yeah, that looks about right; press the launch button"

2) this is more complex than other states we've seen but it does fit a certain pattern.

I fixed the problem. I did this 2 weeks ago. There is no reason (except to waste time) to figure out the best numbers. It isn't exactly 10%. And there is more than one variable.

But it's clear, because the RON PAUL numbers were so quickly cleaned up, that other numbers could be cleaned up.

This was done quickly - but most importantly - 2 weeks ago, right after the election.

Random voters randomly voting resulted in the outcomes we see. Nothing more than that.

This should go to hot topics.

[parocks]

That doesn't remotely explain what we're seeing.

You're telling me that voters just 'voting randomly' is supposed to explain this?

YA! You got it. A big chunk of the voters overvoting. Plus random. I'll give you the categories again.

X% - didn't vote for any delegates
X% - voted delegates / right candidate / partial
X% - voted delegates / right candidate / full
X% - voted delegates / right and wrong candidates / full
X% - voted delegates / right and wrong candidates / partial
X% - voted delegates / wrong candidate / partial
X% - voted delegates / wrong candidate / full

I also would like to add the "top to bottom overage factor". Because Gingrich was at the top - he would have a higher percentage of overvotes than Santorum. Paul and Romney would be in the middle.

With the top to bottom overage factor, you can normalize the various candidates.

So yeah, that should take care of it.

[BIG_J]

I find it odd that we are Re-hashing 2008 again. Put your energy into the delegate process or dedicate the "$5.50" an hour to a donation for Ron Paul or favorite liberty candidate.

Certain "Campaigns" Like to latch onto the successful liberty movement. One of them being the voter fraud camp. They just want people to sign up for "Their" cause. Kind of like all the emails from "Dick Army..." DOUCHE of the century...

[drummergirl]

I find it odd that we are Re-hashing 2008 again. Put your energy into the delegate process or dedicate the "$5.50" an hour to a donation for Ron Paul or favorite liberty candidate.

Well, frankly, the integrity of the voting process and the peaceful transfer of power this country has enjoyed for over 200 years is so important to me that the words to describe that level just fail me.

Let me paint a realistic worst case scenario for those of you who would like to sweep all the voter fraud under the rug.

By whatever means are necessary, Romney gets the nomination (worse things have happened; this in itself would not be the end of the world). The campaign is close going into November; Romney and Obama are neck and neck in several key swing states. Votes in those key states are "flipped" enough to give Romney the win on election night. Democrats start looking at the data, and begin complaining of vote fraud AFTER the election; investigative reporters start talking about how GOP insiders have known since early in the primary process about allegations of vote flipping, the history in Louisiana, etc. Pressure is put on electors to change their votes to Obama because of all the well documented fraud. Political protests begin. The electoral college votes for Romney and SCOTUS does not interfere. Will Obama hand over the reins if he KNOWS the election was a scam? Will there be riots? Could it get any uglier?

So you can say, "oh please bury in hot topics" or, "who cares what serious analysis says?", but the truth remains the truth. Pithy comments won't change that. I know you don't like doing the math. Einstein said, "If you think you've got math problems, you should see mine." Even he didn't like doing the math. It's important; it matters; so don't be sloppy.

[affa]

YA! You got it. A big chunk of the voters overvoting. Plus random. I'll give you the categories again.

X% - didn't vote for any delegates
X% - voted delegates / right candidate / partial
X% - voted delegates / right candidate / full
X% - voted delegates / right and wrong candidates / full
X% - voted delegates / right and wrong candidates / partial
X% - voted delegates / wrong candidate / partial
X% - voted delegates / wrong candidate / full

I also would like to add the "top to bottom overage factor". Because Gingrich was at the top - he would have a higher percentage of overvotes than Santorum. Paul and Romney would be in the middle.

With the top to bottom overage factor, you can normalize the various candidates.

So yeah, that should take care of it.

Except you're expecting everything to exist in a vacuum, and it doesn't. You can't seriously expect any significant number of people, say, voted for Gingrich, but skipped over Gingrich delegates, and voted for Paul and Romney delegates. Or, for example, a Romney voter just voted for Paul and Romney delegates. Or a Santorum voter only voted for Romney delegates.

Rather, you need to look at the ballot and figure out rational 'mistakes'.
Now, doing that, I can accept that a fraction of people:
1) Correctly voted
2) Voted 'down the line' for delegates for all candidates
3) Accidentally voted for a Gingrich delegate or two before realizing their mistake (due to his delegate's placement not only as first on list, but at top of second column)
4) Didn't vote for Santorum delegates because they didn't flip the ballot over.
5) Didn't vote for any delegates (note: Liberty has said this did not happen often; waiting on him for further info)
6) Some level of voter fatigue, however, Liberty has shown that voter fatigue did not seem to adversely affect Romney or Paul.

I can not, however, accept that a significant number (read: any) of Gingrich, Romney, or Santorum voters, say, voted for only Paul delegates. The ballot simply isn't set up that way to make that mistake.

You're asking me to accept that a sizable amount of people effectively drew happy faces on their ballots and that's not reasonable. People generally take voting seriously, and while mistakes can certainly be made, the mistakes will mirror the oddities of ballot design.

Also of note - previous elections and primaries used this same exact ballot design; still trying to see if we have any of the same errors, but given this is the first time it seems to have been brought, up, it seems like this might be the first time voters suddenly forgot how to vote.

I realize you love to keep saying 'you' resolved this already, and you very well may have to your own level of satisfaction, but if your 'level of satisfaction' is satisfied by 'random voting'... well, let's just say you didn't resolve anything to my level of satisfaction.

And confining 'mistakes' to those we can consider reasonable based on the ballot configuration, it's far harder to justify the numbers.

I mean, obviously, if you're willing to just go '10% extra voted for RP', 'X%' forgot to vote for Santorum, 'Y% voted for Gingrich by accident', and Romney voters do no wrong, you explain the problem away. But that's not a real explanation. It's just conjecture.

Again, I'm willing to accept a significant level of voter error due to the ballot.

But 'down the line voters', based on Ron Paul's excess votes, should have created over 30k 'bonus votes' for each of the other candidates. However, this is not what seems to have happened, which means it's not an open and shut case no matter how many times you insist it is.

[parocks]

Except you're expecting everything to exist in a vacuum, and it doesn't. You can't seriously expect any significant number of people, say, voted for Gingrich, then skipped over Gingrich delegates, voted for Paul delegates, then skipped over the rest. Or, for example, a Romney voter just voted for Paul and Romney delegates. Etc.

Now, I can accept that a fraction of people:
1) Correctly voted
2) Voted 'down the line' for all candidates
3) Accidentally voted for a Gingrich delegate or two before realizing their mistake (due to his delegate's placement not only as first on list, but at top of second column)
4) Didn't vote for Santorum delegates because they didn't flip the ballot over.
5) Didn't vote for any delegates (note: Liberty has said this did not happen often; waiting on him for further info)

But you're asking me to accept that a sizable amount of people effectively drew happy faces on their ballots and that's not reasonable.

I realize you love to keep saying 'you' resolved this already, and you very well may have to your own level of satisfaction, but if your 'level of satisfaction' is satisfied by 'random voting'... well, let's just say you didn't resolve anything to my level of satisfaction.

And confining 'mistakes' to those we can consider reasonable based on the ballot configuration, it's far harder to justify the numbers.

I mean, obviously, if you're willing to just go '10% extra voted for RP', 'X%' forgot to vote for Santorum, 'Y% voted for Gingrich by accident', and Romney voters do no wrong, you explain the problem away. But that's not a real explanation. It's just conjecture.

Again, I'm willing to accept a significant level of voter error due to the ballot.

But 'down the line voters', based on Ron Paul's excess votes, should have created over 30k 'bonus votes' for each of the other candidates. However, this is not what seems to have happened, which means it's not an open and shut case no matter how many times you insist it is.

We should take into consideration the fact that Gingrich was first, Paul was 2nd, Romney 3rd, Santorum 4th. That's the top to bottom factor I'm talking about.
Because Gingrich was at the top, he got more votes, Santorum got less, because he was at the bottom. So, you determine the factor, the multiplier, and multiply it by the delegate vote, or the guestimated vote.




Very many people did not vote right. And they didn't vote wrong in identical ways.

Many people seemed to pick races to vote in at random. Perhaps they scrolled down the list looking to vote for names of people they knew.

We don't really have to worry too much about that.

There was a lot of random people voting wrong in random ways.

If you can figure out the truth of that statement, the relevance of that statement, the importance of that statement, you'll start to understand how things work in the real world.

In the "flipping" threads, no one has really tried to understand that Romney does really really well in fast growing (therefore large) upscale suburbs of the major cities.

If you don't realize that different people do different things, and that not everybody behaves in ways that you think they do, you completely miss, or are blind to, what is actually happening.

In Alabama, many, many people thought that the directions said "do whatever you want". And they did.

[parocks]

Except you're expecting everything to exist in a vacuum, and it doesn't. You can't seriously expect any significant number of people, say, voted for Gingrich, but skipped over Gingrich delegates, and voted for Paul and Romney delegates. Or, for example, a Romney voter just voted for Paul and Romney delegates. Or a Santorum voter only voted for Romney delegates.

Rather, you need to look at the ballot and figure out rational 'mistakes'.
Now, doing that, I can accept that a fraction of people:
1) Correctly voted
2) Voted 'down the line' for delegates for all candidates
3) Accidentally voted for a Gingrich delegate or two before realizing their mistake (due to his delegate's placement not only as first on list, but at top of second column)
4) Didn't vote for Santorum delegates because they didn't flip the ballot over.
5) Didn't vote for any delegates (note: Liberty has said this did not happen often; waiting on him for further info)
6) Some level of voter fatigue, however, Liberty has shown that voter fatigue did not seem to adversely affect Romney or Paul.

I can not, however, accept that a significant number (read: any) of Gingrich, Romney, or Santorum voters, say, voted for only Paul delegates. The ballot simply isn't set up that way to make that mistake.

You're asking me to accept that a sizable amount of people effectively drew happy faces on their ballots and that's not reasonable. People generally take voting seriously, and while mistakes can certainly be made, the mistakes will mirror the oddities of ballot design.

Also of note - previous elections and primaries used this same exact ballot design; still trying to see if we have any of the same errors, but given this is the first time it seems to have been brought, up, it seems like this might be the first time voters suddenly forgot how to vote.

I realize you love to keep saying 'you' resolved this already, and you very well may have to your own level of satisfaction, but if your 'level of satisfaction' is satisfied by 'random voting'... well, let's just say you didn't resolve anything to my level of satisfaction.

And confining 'mistakes' to those we can consider reasonable based on the ballot configuration, it's far harder to justify the numbers.

I mean, obviously, if you're willing to just go '10% extra voted for RP', 'X%' forgot to vote for Santorum, 'Y% voted for Gingrich by accident', and Romney voters do no wrong, you explain the problem away. But that's not a real explanation. It's just conjecture.

Again, I'm willing to accept a significant level of voter error due to the ballot.

But 'down the line voters', based on Ron Paul's excess votes, should have created over 30k 'bonus votes' for each of the other candidates. However, this is not what seems to have happened, which means it's not an open and shut case no matter how many times you insist it is.

You have 5 choices there, and you're clearly missing one.

Did you notice that the first delegate election for a candidate has the most votes in it? And the the last delegate election for a candidate has the fewest votes in it?

How can you explain that?

Why, people got bored and stopped voting in those delegate elections.

There's #6 for you.

There are many many permutations. More than you can figure out.

[parocks]

No, it's quite easy.

Group 1) - started filling out all, and then, didn't realize their mistake, simply got bored. That's why Gingrich got more than Santorum. He was the top guy on the ballot. People started at the top. This was a big group.

Group 2) - started filling out the right candidates delegates. Got bored. Stopped. Another big group.

Group 3) - filled out all. another big group.

Group 4) - filled out none. another big group.

Group 5) - filled out all the delegates for the right candidate. another big group, but not as big as you think it is.

You seem to think that things that you don't understand = fraud. But the world is just more complicated than you feel like recognizing.

You know, there were 88 year old women there voting. Do you think they have the kind of eyesight you have? Do you realize that half the voters are of below average intelligence? Do you realize that some people are in a hurry, and see a bunch of ovals, and they figured out to pick one of 2, and scroll down that listreal quick, filling in whatever ovals they feel like.

There was a lot of all of that. And all of those people, doing all of those different things, gives us our result.

[parocks]

http://results.enr.clarityelections....n/reports.html

Affa - Let's look at some real data. U of A Student Rec

Let's look at this one precinct. In one Romney delegate race, there were 48 votes cast. In another Romney delegate race, there were 32 votes cast.

That's fully 1/3 of the total, 16, who voted in at least one Romney race, but not all.

33% voted in a way that you can't explain. Perhaps you should realize that you're imagination, your ability to figure out what people are doing, is limited.

Just because YOU CAN"T UNDERSTAND, doesn't mean there's fraud.

Oh, there probably is fraud. But you have to be pretty kinda smart to figure it out, and you can't even realize that this one isn't really all that tricky to figure out.

Lots of people voted in ways that you can't imagine, even when they're explained to you.

[parocks]

Here's some data from the Mary Phelps Center.

Ron Paul didn't do well here, less than 4%. That means there were overages in the delegate vote.

Let's see how many people, of the 1105, partially voted. The difference between the most and least popular delegate races for each candidate.
Or, the "range"

Total pres votes - 1105
Gingrich - range - 270 - 211 (59 people partial)
Paul - range 106 - 76 (30 people partial)
Romney - range 367 - 270 (97 people partial)
Santorum - range 238 - 207 (31 people partial)

If you add all those up you get 217 partials of 1105 total. That's around 20% right there.

Lots and lots of different behaviors.

Your theories do not account at all for this huge 20% variation.

In this particular precinct, around 9% partially voted for Romney delegates

97 of 1105

there were huge partials. It might be accurate to say that more people voted "wrong" (didn't complete the delegate votes for their candidate) than voted right.

You can see a big drop off from delegate election 1 to delegate election 2.

In Romneys case, more than half of his partial 53 of 97 was people who only voted in the first delegate race. So, right there, about 5% of the total voters voted in Romney's first delegate race, but not his second delegate race.

How do we explain that? 5% found the candidates name, looked at the one box, checked the name they wanted, and didn't go any further. Those might've been "people in a hurry". 5 whole percent, just Romney voters in a hurry, not voting for all the delegates.

See how we can solve these problems so quickly.

Santorum lost 11% of his from 1st to 2nd.
Ron Paul lost 6% of his from 1st to 2nd.
Gingrich lost 10% of his from 1st to 2nd.

About 10%, are busy.

24 + 6 + 53 + 27 = 110 / 1105

It's also the candidates. The choices on the ballots also matter some.
We can assume that there will be fewer and fewer votes, as people get bored.

This is not always the case. Look for races that are farther down the list that have more votes than races above.
Why would this be? More compelling candidates. Voters who are voting based on some sort of impulse created by the candidate or the candidates
name. Romney had a 315 vote delegate election after a 287. That's 28 impulsers, close to 3%.

Ron Paul had a 78 after a 76. 2 impulsers

Gingrich had a 237 after a 229 8 impulsers.

[Liberty1789]

Voters voted differently. Some voted down the line. Some voted not at all. Some voted randomly.

X% - didn't vote for any delegates
X% - voted delegates / right candidate / partial
X% - voted delegates / right candidate / full
X% - voted delegates / right and wrong candidates / full
X% - voted delegates / right and wrong candidates / partial
X% - voted delegates / wrong candidate / partial
X% - voted delegates / wrong candidate / full

voter behavior random, unpredictable.

Hmm.... You are not very familiar with statistics, parocks, are you? It's okay: few people are. Absolutely nothing to be embarrassed about, but my advice would be to adjust the obscurantist rhethoric accordingly.

You ignore 2 BIG statistical facts when you claim unpredictability: (1) random errors tend to cancel themselves out (2) bigger errors are much less likely than smaller ones. So, in theory, we should get the same numbers of votes in the presidential preference ballot and in the delegate race, but in reality we don't, as, among other factors, all the errors that you mention above get into play.

So instead of getting the 1st, theoretical, chart below, we get the 2nd, actual, one:



You see, there is a method to errors' madness.

A statistician will immediately note something critical: the average does not seem to have shifted and the most frequent observation, (called the mode) has not moved and the deviations remain nicely centered around it. 1 remains the most frequent data.

So what we need to focus on is rational, realistic factors, that could shift hugely the mode of 1 candidate whilst leaving the 3 other untouched, something truly unique to Paul, as evidenced in the chart below. And that is a tall order.



By the way, if 10% of the voters had voted for all delegates, the 4 bell curves above would have shifted to the left. Only 1 did. Mathematical proof that parocks' debunk is kaput.

[The Man]

Hey Liberty1789, can you detemine if a correlation exists between the ratio of votes for candidate/ 1st delegate votes to number of voters per precinct, in particular, for Romney, Gingrich, and Santorum? Specifically if you arrange the precincts in ascending order with respect to # of voters, how does this ratio look for the smallest precincts that represent the first 90k votes compared to the larger precincts?

[Liberty1789]

Hey Liberty1789, can you detemine if a correlation exists between the ratio of votes for candidate/ 1st delegate votes to number of voters per precinct

Interesting. Done it for Gingrich: R^2=0.01 and Paul: R^2=0.03 so no apparent correlation between relative delegate vote discrepancy and precinct size.

[dsw]

So what we need to focus on is rational, realistic factors, that could shift hugely the mode of 1 candidate whilst leaving the 3 other untouched, something truly unique to Paul, as evidenced in the chart below. And that is a tall order.

If 10% (or 5% or 8% or any significant percentage) of voters had voted for all delegates, that would disproportionately affect the mode for Paul, correct? Simply because Ron Paul got such a smaller percentage of the preference vote. A 10% idiot rate added to Ron Paul's 5% (assuming no overlap, obviously) triples the potential number of votes in his delegate races. For Santorum, it's a much smaller factor, especially if you assume that voting for Santorum and not understanding the directions are not independent variables.

So as a start, can we agree that this is a potential factor that we need to consider very carefully as possibly *part* of the explanation? It's the only thing I've seen suggested that could affect Paul's delegate race percentage disproportionately more than it would affect the other candidates.

By the way, if 10% of the voters had voted for all delegates, the 4 bell curves above would have shifted to the left. Only 1 did. Mathematical proof that parocks' debunk is kaput.

Now, I'm just a simple old retired engineer, and I don't have all the sophisticated statistical learning you've got. And thanks for reminding us that there's no shame in this!

But your conclusion is only a mathematical proof that parocks is valid only if you assume that there are no other factors involved besides that hypothetical 10% idiot rate. Can you show your work on that step of the proof, the part where you show that there are no other factors other than the hypothetical 10% idiot rate that could affect the other three candidates?

BTW, I read back over the recent messages here and didn't see an explanation for the claim that only 27 out of over 600,000 voters marked a vote in the presidential preference poll and didn't vote in any delegate races. Where did that data come from? It wouldn't refute it entirely but it would take some of the wind out of the sails of parocks's explanation.

[affa]

In the "flipping" threads, no one has really tried to understand that Romney does really really well in fast growing (therefore large) upscale suburbs of the major cities.

Because that's not what we're witnessing in that thread, at all.

And yes, we've spent tens of hours studying just that, and debunked it as an explanation.

[affa]

You have 5 choices there, and you're clearly missing one.

Did you notice that the first delegate election for a candidate has the most votes in it? And the the last delegate election for a candidate has the fewest votes in it?

How can you explain that?

Why, people got bored and stopped voting in those delegate elections.

There's #6 for you.

There are many many permutations. More than you can figure out.

Um. See #3, #4, and #6. I didn't 'miss' it, I just broke it up into groups. It looks like you responded before I added #6, but #3 and #4 are the lion's share of your suggested group.

1) Correctly voted
2) Voted 'down the line' for delegates for all candidates
3) Accidentally voted for a Gingrich delegate or two before realizing their mistake (due to his delegate's placement not only as first on list, but at top of second column)
4) Didn't vote for Santorum delegates because they didn't flip the ballot over.
5) Didn't vote for any delegates (note: Liberty has said this did not happen often; waiting on him for further info)
6) Some level of voter fatigue, however, Liberty has shown that voter fatigue did not seem to adversely affect Romney or Paul.

Your explanation DOES 'make sense'. Until you look at the numbers, at which point it does not hold up well. If you were right, Gingrich's numbers should have skyrocketed and Romney should be lower to account for RP's position. Romney, certainly, should have more voters.

[affa]

Has anyone found 2008 results yet? Same ballot configuration. Do we see the same oddity?

[dsw]

Has anyone found 2008 results yet? Same ballot configuration. Do we see the same oddity?

I could only find this:
http://www.sos.state.al.us/elections...nInfo2008.aspx
which doesn't have (as far as I could find) details about the delegate races.

[Liberty1789]

If 10% (or 5% or 8% or any significant percentage) of voters had voted for all delegates, that would disproportionately affect the mode for Paul, correct?

Correct.

For Santorum, it's a much smaller factor.

If you round up the 3 big guys at 200k votes each, you see that a 10% idiocy vote rate would shift the mode down by a good 20%. Did not happen for any of them.

So as a start, can we agree that this is a potential factor that we need to consider very carefully as possibly *part* of the explanation?

Agreed and thanks to parocks for the original good idea. But I do not see how it is compatible with unmoved mode, so I see it as fully debunked for now.

But your conclusion is only a mathematical proof that parocks is valid only if you assume that there are no other factors involved besides that hypothetical 10% idiot rate. Can you show your work on that step of the proof, the part where you show that there are no other factors other than the hypothetical 10% idiot rate that could affect the other three candidates?

Woah. Hold your horses. A 10% idiot rate fills the 50k gap for Paul. It opens up a 40k+ gap for Romney and one for Santorum, so you end up with more gaps to explain, not less, and no justification provided to date. The 10% idiot rate displaces the modes. Does not fit the data. So instead of explaining why 1 mode moved, you need to explain why 3 did not, alleging impeccable compensation from other factors, totally unidentified to date, not affecting Paul, but only the 3 others.

The data shows that the position on the ballot has an impact on the behavior of 10-20% of the voters. It helps Gingrich's delegate vote count, is neutral to Romney, is detrimental to Santorum. It is all logical. It fits impeccably the skew of the bell curves of the 3 candidates. Paul's characteristics should be between Gingrich and Romney, but they are in total violation of that.

If for any factor put forward that does not fit the data, one claims an unknown offsetting factor that brings it all back in line, then fine, I cannot disprove anything. Ockham shaking his head in his grave though. Lol.

BTW, I read back over the recent messages here and didn't see an explanation for the claim that only 27 out of over 600,000 voters marked a vote in the presidential preference poll.

Sorry, I have to go back, not sure what you mean (or what I meant) as I have only precinct data.

[dsw]

Your explanation DOES 'make sense'. Until you look at the numbers, at which point it does not hold up well.

What numbers are you looking at to reach that conclusion? Even if Liberty is right that only 27 out of over 600k voters marked none of the delegate races, what about the ones who marked just a few Newt races then quit? The difference between Newt's first delegate race and his third is about 30k votes. And when you get to his third delegate race we're down to a number very close to his preference vote. Some of those 30k who started out voting for Newt's delegates and then stopped may have gone on to find the section of races for their own candidate, some may have quit right there. And there are other effects to take into account too. What is it that you're finding compelling about the analysis so far?

[drummergirl]

Here's some data from the Mary Phelps Center.

Ron Paul didn't do well here, less than 4%. That means there were overages in the delegate vote.

Let's see how many people, of the 1105, partially voted. The difference between the most and least popular delegate races for each candidate.
Or, the "range"

Total pres votes - 1105
Gingrich - range - 270 - 211 (59 people partial)
Paul - range 106 - 76 (30 people partial)
Romney - range 367 - 270 (97 people partial)
Santorum - range 238 - 207 (31 people partial)

If you add all those up you get 217 partials of 1105 total. That's around 20% right there.

Lots and lots of different behaviors.

Your theories do not account at all for this huge 20% variation.

In this particular precinct, around 9% partially voted for Romney delegates

97 of 1105

there were huge partials. It might be accurate to say that more people voted "wrong" (didn't complete the delegate votes for their candidate) than voted right.

You can see a big drop off from delegate election 1 to delegate election 2.

In Romneys case, more than half of his partial 53 of 97 was people who only voted in the first delegate race. So, right there, about 5% of the total voters voted in Romney's first delegate race, but not his second delegate race.

How do we explain that? 5% found the candidates name, looked at the one box, checked the name they wanted, and didn't go any further. Those might've been "people in a hurry". 5 whole percent, just Romney voters in a hurry, not voting for all the delegates.

See how we can solve these problems so quickly.

Santorum lost 11% of his from 1st to 2nd.
Ron Paul lost 6% of his from 1st to 2nd.
Gingrich lost 10% of his from 1st to 2nd.

About 10%, are busy.

24 + 6 + 53 + 27 = 110 / 1105

It's also the candidates. The choices on the ballots also matter some.
We can assume that there will be fewer and fewer votes, as people get bored.

This is not always the case. Look for races that are farther down the list that have more votes than races above.
Why would this be? More compelling candidates. Voters who are voting based on some sort of impulse created by the candidate or the candidates
name. Romney had a 315 vote delegate election after a 287. That's 28 impulsers, close to 3%.

Ron Paul had a 78 after a 76. 2 impulsers

Gingrich had a 237 after a 229 8 impulsers.

I think it is far more likely that after starting voting for the delegates, that voters reread the directions, realized they did not need to vote in these races in order to vote for president (because it's always difficult to vote when you don't even know the names of the people on the ballot), and stopped voting for delegates.

Your theory that so many voters are stupid seems unlikely (not to mention insulting to the people of Alabama). In order to vote, you have to 1) register, 2) know when the election is 3) care enough about the outcome to go out of your way for a half an hour 4) know where your polling place is 5) actually get yourself to the polling place on election day.

In my experience, the voters that actually turn out to vote in primaries are a more committed group than we see in the general election.

Your premise that 20% voted completely wrong (10% voting in all delegate races, 10% voting in none) seems way too high.

[Liberty1789]

BTW, I read back over the recent messages here and didn't see an explanation for the claim that only 27 out of over 600,000 voters marked a vote in the presidential preference poll and didn't vote in any delegate races. Where did that data come from? It wouldn't refute it entirely but it would take some of the wind out of the sails of parocks's explanation.

Ok I now see what you are referring to. This is what I wrote:

Out of 1,864 precincts, number of occurences with presidential preference vote and no delegate vote:

Gingrich	7
Paul	0
Romney	11
Santorum	9

I am talking precincts, not voters. "Out of 1,864 precincts". Sorry it was not clearer. I have counted, for each candidate separately, the number of precincts where you have at least 1 presidential preference vote and no delegate vote. I use it as a detector that it is not happening frequently. It would show in the low vote count precincts otherwise.

[dsw]

A 10% idiot rate fills the 50k gap for Paul. It opens up a 40k+ gap for Romney and one for Santorum, so you end up with more gaps to explain, not less, and no justification provided to date. The 10% idiot rate displaces the modes. Does not fit the data. So instead of explaining why 1 mode moved, you need to explain why 3 did not, alleging impeccable compensation from other factors, totally unidentified to date, not affecting Paul, but only the 3 others.

[...]

If for any factor put forward that does not fit the data, one claims an unknown offsetting factor that brings it all back in line, then fine, I cannot disprove anything. Ockham shaking his head in his grave though. Lol.

But we know that there are multiple effects at work here, so rejecting the hypothesis without taking those multiple effects into account isn't what Ockham's razor would tell you to do. I don't see why you think you can analyze that one effect in isolation. After all, it only represents (in the hypothesis) 10% of the vote. To take that in isolation, and treat the other 90% of the vote as if it were not skewed in any way, falls short of the "mathematical proof" you're claiming. If debunking were that easy there wouldn't be thousands of posts in the algorithm flipping threads.

You had a table that I (and others) had understood to be showing that only 27 people didn't mark any delegate races. I'm not sure what that meant now if you're saying it came from precinct data. If that was a misunderstanding of what you posted, and it's reasonable to assume that a significant percentage of voters would not be motivated enough to find the delegate races, then (assuming Ron Paul's supporter will tend to be better informed and more motivated) isn't that a valid compensating factor?


A 10% idiot rate fills a 50k gap for Paul and opens up a 40k gap for Santorum. But to put it another way, a 10% idiot rate erases a 200% surplus for Paul and replaces it with a 30% surplus for Santorum. But that's only a 30% surplus if we assume that 91% of Santorum's supporters found their way to the back of the ballot, to vote in the first of Santorum's delegate races. Are we to take it for granted that Santorum's supporters showed that level of diligence and motivation?

[The Man]

Someone please post the link to the delegates data in Alabama.

[dsw]

Ok I now see what you are referring to. This is what I wrote:

Out of 1,864 precincts, number of occurences with presidential preference vote and no delegate vote:

Gingrich	7
Paul	0
Romney	11
Santorum	9

I am talking precincts, not voters. "Out of 1,864 precincts". Sorry it was not clearer. I have counted, for each candidate separately, the number of precincts where you have at least 1 presidential preference vote and no delegate vote. I use it as a detector that it is not happening frequently. It would show in the low vote count precincts otherwise.

So in a small number of cases, there were delegate races with zero votes, corresponding to candidates with non-zero votes?

Sorry, I don't get it. Let's say 20% of Santorum voters didn't vote for Santorum delegates. A precinct that shows non-zero votes for Santorum in the preference race, but zero for Santorum in some delegate race, should be rare. If we looked at precincts with only five Santorum votes, for example, the odds of zero votes in a Santorum delegate race would be .2^5 or .003%. I don't know how large those 9 precincts with Santorum votes but no delegate votes were, but they wouldn't have to be very large in order for that to be an argument that a significant percentage of Santorum voters didn't vote in Santorum delegate races.

What am I misunderstanding?