Here are some of the statistical exhibits presented in court last week by consultant Clark Bensen --
King County Absentee Balance Sheet. He got to the 875 discrepancy the same way that I did, updated for subsequently discovered uncounted ballots.
Scattergram showing bias of unexplained surplus ballots in Gregoire precincts
Posted by Stefan Sharkansky at May 28, 2005 11:28 AM | Email ThisIf Fraudoire is able to pull this off, the DNC will KNOW that Hilly won't have any problems in 2008.
Bought and paid for!
Posted by: arky on May 28, 2005 12:25 PMI don't know if anyone else has made this observation, but it's entirely possible that this entire scope of statistical analysis used in support of an election challenge may, in the long run, may be used in later cases much as the way DNA evidence has gained in acceptance.
Even the scientific use of DNA is not always perfect. An uncle, for example, could have provided the material, rather than the subject in the dock. But still, there are some absolutes involved in both, and there are proving to be many ways to use stats in seeing how certain unknowns trend or to distinguish probabilities and conficence factors from noise.
Thanks again.
Posted by: scott158 on May 28, 2005 12:27 PMLogan's contribution was to make a bad situation worse in King County. In his changing of the method/process used to tally cast ballots. Instead of re-enforcing and re-training their people in the correct procedures needed to make the existing system faster and more accountable. You now have a process in transition. Without any additional training provide to the elections workers. Training that they would need to reduce error rates with the new method/process.
This is a classic case of impending disaster by lack of experience with a new method. A situation that could be (and I think was) exploited, if needed, by the Democrat's. Who can later say, "see it was just simple errors, no real conspiracy here." The real trick is to blame "the system", not "the individauls" who ran it. Which is exactly what the Democrats are doing. Remember by Logan's own admission, King Couny's error rate is THREE times, what it was in 2000.
Posted by: Mike P on May 28, 2005 12:50 PMAs for Nicole Way, she may be an unsung hero - which unearthed the pathway to fraud ! and put a big crimp in the Democrat machinery.. time will tell !
Posted by: KS on May 28, 2005 12:58 PMI didn't hear the GOP's lawyer rebut that assertion.
Was there any expert testimony provided by the petitioners to assist the judge in understanding that statistical evidence?
Posted by: Micajah on May 28, 2005 12:59 PMMy FOOT there isn't any! Dino was robbed and that's the truth. We won't stop tellingthe truth.
Posted by: Michele on May 28, 2005 01:23 PMIn other words, do they just go ahead and enter signature-less voter registrations into the system WHEN THEY'RE FROM SEATTLE, and then decide to 'follow the rules' when a signature-less registration comes in from, say, Sammamish and NOT enter it in??
Posted by: Michele on May 28, 2005 02:07 PMBTW I wonder what payoff CG had for Logan in he'd pulled it off? And that btw is why I think Sam Reed (in the wind) is so whimpy on all this!
Posted by: Victor on May 28, 2005 07:26 PMSeattle 36-1770
Seattle 43-1774
On the king county site, https://www.metrokc.gov/elections/pollingplace/address.aspx,
With just his (?her) last name, NO first name entered, King County elections site gave the correct polling site. I tried it with my own name, MISSING the FIRST NAME, and it failed.
Sharks copy (of KC RnE DB) for this record also indicates a blank first name.
Gregg
Posted by: gregg on May 29, 2005 02:16 AMMy guess (not knowing your last name) is that there are others in the system with the same last name, and the database engine would only respond to unique matches. The bigger question is is it legal to be registered with a single name? I don't know, but as long as 1. it is your legal name, and 2. it is unique, it shouldn't be a problem.
As I understand it, if you want to change your name to batman, it is legal. But it does create opportunities for monkey business.
Posted by: Dogbert on May 29, 2005 08:10 AMFuzzy logic, a multivalued (as opposed to binary) logic developed to deal with imprecise or vague data. Classical logic holds that everything can be expressed in binary terms: 0 or 1, black or white, yes or no; in terms of Boolean algebra, everything is in one set or another but not in both. Fuzzy logic allows for partial membership in a set, values between 0 and 1, shades of gray, and maybe—it introduces the concept of the “fuzzy set.” When the approximate reasoning of fuzzy logic is used with an expert system, logical inferences can be drawn from imprecise relationships. Fuzzy logic theory was developed by Lofti A. Zadeh at the Univ. of California in the mid 1960s. However, it was not applied commercially until 1987 when the Matsushita Industrial Electric Co. used it in a shower head that controlled water temperature. Fuzzy logic is now used to optimize automatically the wash cycle of a washing machine by sensing the load size, fabric mix, and quantity of detergent. Fuzzy logic has applications in the control of passenger elevators, airliners, household appliances, cameras, automobile subsystems, and smart weapons.
(Just because you don't understand or value a system, doesn't invalidate it.)
Posted by: starboardhelm on May 29, 2005 08:42 AMActually, the scattergram does show some pretty overwhelming circumstantial evidence that ballots were stuffed in pro-Gregoire precincts.
Draw a vertical line down through 40%, which is the percentage of votes that Rossi garnered in King County. This takes care of the 58-40% split - precincts to the right of the line were more 'pro-Rossi' than average, those to the left of the line were more 'pro-Gregoire' than average.
It shows that the precincts that lost ballots were evenly distributed (3 pro-Rossi, 3 pro-Gregoire, 1 on the line).
But look at the precincts that gained votes - 32 pro-Gregoire, 11 pro-Rossi. If a precinct gained votes it was 3 times more likely to be a pro-Gregoire precinct than pro-Rossi (again, taking into account the 58-40% split).
What are the odds that the precincts with ballots added are not evenly distributed, like those that lost ballots, but are 3 times more likely to be Gregoire precincts?
Since most people would not understand what the scattergram portrays, I would have preferred to have seen this compared to other scatter grams that showed what it should have looked like where no bias can be inferred.
I read in the paper this morning that this piece of evidence was disallowed by the judge. Is that true? If so, that's too bad, but this isn't the first problem I've seen in the Republican's case.
Posted by: Deadwood on May 29, 2005 09:08 AMIf Gregoire received the majority of votes in most King County precincts, and if more precincts were affected by a discrepancy involving more ballots than voters rather than more voters than ballots, wouldn't there be more affected precincts on Gregoire's side of the graph? She won more precincts. More precincts had an excess of ballots compared to voters. More + More = Gregoire's side of the graph looks as though she benefited disproportionately from stuffed ballots.
Drawing a line one place or another might be an acceptable way to adjust for the differences and see if there is an actual bias -- but where's the testimony of a true expert who could tell the judge what an adjusted graph would look like?
Posted by: Micajah on May 29, 2005 09:18 AMThey proved the illegal ballots and the willful neglect of elections officials. They presented good circumstantial evidence of fraud, but I'm not sure how well they explained it. And in the absence of cameras being allowed in polling places, and since we'll never catch Logan or Huennekens or whoever actually stuffing ballots, circumstantial is as good as it'll get.
But how well did they explain this evidence and tie it back to the neglect? We know there are extra illegal ballots, it was the neglect of the elections officials that allowed the extra ballots, and the extra ballots show patterns of fraud.
Posted by: Larry on May 29, 2005 09:26 AMThat is what I was saying on a different thread - This can all be summed up in a very simple question to be put to an expert witness statitician: What is the probability that the observed distrubution of overvotes can be attributed to random chance?
Even the donk statisticians wouldn't be able to wiggle out of that.
Posted by: Dogbert on May 29, 2005 09:36 AMIf not, you should be. Send me an e-mail when you get a chance - budaiguana at hot mail dot com. I'll explain to you why you should be based on some comments you made a few days ago (Andy, you and I were commenting about a similar topic).
Posted by: BananaLand (aka Iguana) on May 29, 2005 09:51 AMThat was May 25 at 4 hr 24 min on the recording.
So, we can look at the graph, but the court won't be looking at it. It is not an exhibit in the trial for any purpose whatsoever.
Sorry, but I must play devil's advocate and suggest that you are introducing a false dichotomy by implying that any such correlations must either be the result of chance or wilful misconduct. There is a third possibility, which is that there may be some common factors which affect both the percentage of votes Rossi receives in a precinct and the likelihood of ballots there being overcounted or undercounted.
Posted by: supercat on May 29, 2005 12:46 PMWhat, then, do you advocate as the possible common factors which caused there to be either too many ballots or too few, relative to the number of voters known to have cast ballots -- and to have caused the difference in distribution of those discrepancies?
Posted by: Micajah on May 29, 2005 02:16 PMThere are many things it theoretically could be. Note that I believe the simplest answer is in fact the dominant one, but there are probably many variables which are roughly correlated with precincts' political slants. These include per-capita income or intelligence of voters, per-capita spending on election personnel, local availability of good election workers, etc. Even if such factors should not account totally for the odd distributions of votes observed, they would tend to pull them far closer to a level that could result by chance.
Posted by: supercat on May 29, 2005 02:44 PMPolling place workers have nothing to do with that, since they don't credit voters with casting absentee ballots even when the ballots are properly cast at polling places (rather than being mailed) by putting them into sealed, signed envelopes and placing them in the side bin of the ballot box for later verification and crediting.
Spending on the processing of absentee ballots is the same, no matter which precinct they came from -- it's all the same people doing the processing of the absentee ballots.
If per capita income or intelligence of voters could make a difference, how did the rich/poor or smart/dull voters make their absentee ballots disappear?
If the discrepancy is the other way around, how did the rich/poor or smart/dull make more absentee ballots appear than the number of voters known to have cast absentee ballots?