I always wondered if a background in boring statistical analysis might ever come in handy. Well take a look at what I found amongst California's votes.

Tulare county use Diebold Opti-Scan equipment.
Tulare county gave 'obscure' candidates high percents of their state wide totals:
Palmieri - 995 out of 3,717 26.77%
Platform was 'don't vote for me or the recall'. Gay Rights activist who lives in LA.
Kunzman - 694 out of 2,133 32.54%
Lives north of Oakland and favored increased social programs. Said he would fire all school custodians to save money and have the kids empty the trash and clean the carpets.
Sprague - 546 out of 1,576 34.64%
Zero tolerance for discrimination. Lives near Sacramento
McClain - 46 out of 2,463 1.77%
Civil engineer, Berkley grad living in Bay Area

These were not local candidates. The 'local candidate effect' can be seen with Doctor Macaluso from Visalia in Tulare county. He got 7.2% of his state wide total vote from his home county.

As a percentage of the votes counted as of the time I ran this analysis, Tulare votes were 0.9% of the state total. For comparison, in the 2002 fall election, the county gave Bustamante 24,647 votes which has dropped to 15,487 even with an increase in votes cast from 61,884 to 68,891. Stats prove nothing but this strikes me as strong evidence for the need for an audit.

The county gave leaders the following percentages of their state wide totals:
Swartzenegger 1.028%
Bustamante 0.65%
McLintock 1.036%
Camejo 0.25%

These figures suggest a hypothetical scam in which the machines were used to skim Bustamante votes to 'fringe' candidates. That would leave the % for Swartzenegger close to that predicted by opinion polls and exit polls but decrease Bustamante's total. I am now running similar tests on every county where Diebold were doing their best to deliver the votes to George Bush - to paraphrase Wally O'Dell, their CEO.

Alameda's touch screens did something rather odd. A reasonable distribution of votes by candidate by county has a long thin tail, often ending with several candidates getting NO VOTES. The touch screens of Alameda seem to have managed to find a good number of votes for all sorts of people at the bottom edge of the ballot. Funny coincidence that Alameda has suddenly taken a dislike to Bustmante after giving him 62% of their vote for Lt. Governor last year, he is down to barely 50% of the total.

More later as I crunch the numbers. If you care to join me, go to the CA secretary of state's site for the raw data.

Faun

Faun, you are a genius. Those statistics make no sense whatsoever. I said it in an earlier post--something's rotten in the state of California. Keep at it--we're counting on you!

Get this OUT.

Very reminiscent of Volusia (?) County, where fringe candidates got unusually high numbers.


© samela 2003. Fair and balanced to the bitter end.

The great enemy of the truth is very often not the lie,
deliberate, contrived and dishonest, but the myth,
persistent, persuasive and unrealistic.
— John F. Kennedy

really are making polictics distasteful for
me. I don't think I'm alone.
They are going to drive everyone away from the polls
with their bs and the only ones
left voting will be their favorite....
angry white males.

Arghhh.

Where Gore's vote total "just went bacckwards" to the tune of NEGATIVE 16,022 VOTES!

Nothing to see here.

http://www.unc.edu/~pmeyer/usat29nov2000.html



"I miss America
And sometimes
She does too"
-D Byrne-

I hope you don't mind, but I fired those numbers off to Crossfire. Just for laughs. Fire a shot across the bow type of thing, just to see if anyone is paying attention. I don't expect any response, but at least we know they know. Or something like that.

-------------

NotBannedYet.US
Original cartoons, rants and liberal propaganda
"This time, the revolutions starts online!"

Maybe Cruz can sue for that stuff too.

One of the candidates that got a ballot box of extra votes is an attorney and I sent him a note to see what he thought. You may recall a book called "Jews for Buchannan" in which the most damning evidence of vote switching was that it would have required some Holocaust survivors to vote for Buchannan - a man who said that he felt the Holocaust was not as bad as history had portrayed it.

Even Pat Buchannan had the decency to say that he didn't think those votes were correctly cast for him. Perhaps if we could get the vote recipients to go public with similar statements, we could get some traction on the story. Back to number crunching. Can you guys send this info out to others? I have already passed it to Bart.

Faun

http://vote2003.ss.ca.gov/Returns/gov/00.htm#cty

I'll try to do some digging tonight. I hope I can understand what I'm seeing as well as you.

Thanks for reminding us that democracy is not a spectator sport.

"Do you promise to funk, the whole funk, nothin' but the funk?"
--Funkadelic

just now. The problem is: exactly *how* am I supposed to model this election with such a *heueueuge* list of candidates? I don't believe there is a lot of background literature on this. The tails do look very long and not noisy enough, almost like a continuum measurement. And the morte total votes cast the fewer 'zero' votes - as if they were trying to model a gamma-ish distribution or so.
X^(a-1) exp(-X), where 'a' is some fitted number < 1.
If it is the *same* a, we've got Charlie.

If you use different machines, you will 'lose' a different number of votes due to read errors but this should be random. Thus the machine type may affect the % of lost votes BUT it should not change the tail shape of the curve. If Diebold tails are different from other voting machine's tails, that would also tend to show vote tampering.

That is also why having only Diebold machines in Georgia made it hard to collect evidence that they had tampered with results.

Faun

results work well with (X^A)exp(-BX) for the expected number of votes for 'A' = 0.800±.08 and B = 1.275±.05. Both of those numbers are a little 'round'

Also, there seems to be less of a mess with the tail in Humboldt, but a whole lotta votes for weirdos for a county that presumably went big for Bustamante last time around.

the 'A' value is different, but the 'B' value is 1.279±.09 Hmmmmmm....
Modoc is another Diebold county, but not the same fit at all, hmmmm.

.

This part of the state is made up of outliers, statistical extremes that don't show up most places. The multi-generational natives are right-wing to the core, and most who've moved here in the last 25 years just couldn't stand it in San Francisco 'cause it ain't liberal enough. We have equal numbers of 'Rush is Right' and 'Unrepentent Nader Voter' bumper stickers county-wide. What's more interesting is how the population is stratified ideologically. Arcata is extremely Green, while McKinleyville just to the north is known as Oklahoma By the Sea. Garberville in the south part of the county is a pot growing Mecca where the radio stations broadcast PSA's regarding the whereabouts of law enforcement officials at any given hour, yet communities all around them are populated almost exclusively by loggers and ranchers.

What I'm saying is that the incredibly broad distribution of votes here should be expected, given that progressives here are used to throwing them away. They know they can elect their own to local city councils in a few communities, and they know they don't stand a chance in any other election, so why bother. And the fact is, most Greens despise Davis even more than they did Clinton or Gore.

Besides, we use optical scan ballots in my precinct, and I'm pretty sure that's true for most precincts in all these northern counties.

All that said, as unusual as our voting patterns up here tend to be, perhaps that offered the perfect cover, making our tallies fertile ground for abuse. I just don't see how statistical analysis is going to show that.

But what do I know? I don't even get cable.

of the state. There was a reason why I chose your neck of the woods. How different is the rest of the state in terms of the tails? And it would appear -- not that much. {edit --ooops -- that is there IS less of a tail in Humboldt and you don't use Diebold, Modoc did use Diebold and has not much of a tail either, but maybe just too small to count Fresno did use Diebold, has tail and weird coefficient!}
And you have a Diebold/nonDiebold difference too. Now Modoc is probably too small for anyone to have bothered cooking, and they have a fairly short tail, more *typical* I would say.

"Do you promise to funk, the whole funk, nothin' but the funk?"
--Funkadelic

Word association almost caused me to type 'raw sewage' instead of 'raw data,' it might have been just as accurate....

Faun

Jump off point here:
http://www.ss.ca.gov/elections/elections.htm

Get 2002 and 2000 details from the links at the bottom of this page:
http://vote2003.ss.ca.gov/index.html

Get lists of machine types for voting here:

http://www.ss.ca.gov/elections/elections_vs.htm

I will post this information on a few other local boards that I post on. Lets hope we can do something before 04!!!

I posted this entire page on two other boards. I posted the whole schebang so that the back and forth questions could be seen too. Hope you didn't mind my using your replies, but this explained everything much better than I could have.

you too, secondharmonic. You guys are great!


I want my Crossfire at 7 pm!
Call CNN (404) 827-1700

It's easy to fool yourself looking for coincidences in data like this.

It would be counterproductive to claim something here unless there is good evidence.

>These were not local candidates. The 'local candidate effect'
>can be seen with Doctor Macaluso from Visalia in Tulare
>county. He got 7.2% of his state wide total vote from his home
>county.

These are very interesting data. The one thing I don't see anyone
looking at is position of the candidates' names on the ballot in
this thread, though.

You'd want to look at the physical layout of the ballot and see
where people were on it, top-to-bottom as well as adjacent to the
"big name" candidates to see if you thought Tulare (for instance)
was worth a recount.

>Alameda's touch screens did something rather odd. A
>reasonable distribution of votes by candidate by county has a
>long thin tail, often ending with several candidates getting
>NO VOTES. The touch screens of Alameda seem to have managed to
>find a good number of votes for all sorts of people at the
>bottom edge of the ballot.

Okay, but Alameda's a relatively big county. So where you'd see no votes in the tail in a smaller county, the same tail would be expected
to include a few votes in a larger county.

I think, but don't have time to look just now, that the ballots for each county are viewable in the order the candidates were listed over at that state website.

Here's my suggestion re: Tulare: Find someone there who's interested in the black box voting controversy. If the fringe candidates with the anomalous readings aren't up way high on the ballot or next to Bustamente or Ahnold, anyone registered to vote in Tulare can demand a recount of the county. They have to pay for it, and the payment is not refundable unless the recount changes the winner in the county.

So the payment would be final, almost certainly.

Tulare is relatively small, though; in LA, a recount costs 16k/day.

I wrote to BBV about your results, and pledged a contribution to a Tulare recount. If we can find someone there or another anomalous Diebold optical scan county, we could get a meaningful recount, since we do have an audit trail at least there.

I hope BBV will look into the numbers as well and decide if they're interested. If others are interested, think about how much you'd be willing to pony up to look into these results.

http://www.eff.org/Activism/E-voting/
You might want to contact some of the groups keeping an eye on e-voting on this site:
EFF "E-voting" Archive

Hey.. I don't know if this will bump up to the top.. but let me show it to you.

http://www.tularecoauditor.org/elections/recall/root1007/ballots/bt000001.pdf

now, lets snip the ballot to the relevant parts



All three fringe guys finished well for a reason. Some people just didn't know how to fill out a ballot.

The ballot I saw didn't have the vertical lines. Must've been sample only.

Now that you've posted the actual ballot, I realize this was just Deiboldt skimming votes off to fringe candidates and using "plausible deniability" to account for the anomalies.

Sorry, but this is an old Karl Rove trick. We all know he used to do the circuit giving lectures on tricky ballot construction.





© samela 2003. Fair and balanced to the bitter end.

The great enemy of the truth is very often not the lie,
deliberate, contrived and dishonest, but the myth,
persistent, persuasive and unrealistic.
— John F. Kennedy