If you read the headlines in the MSM on May 2, you might have thought Dino Rossi scored a major legal victory in his lawsuit seeking to overturn the November 2004 gubernatorial election. But after studying the transcript of Judge Bridges’ rulings, my initial impression hasn’t changed: it wasn’t a very good day for Dino. While the Judge hasn’t entirely shut the door on the election contest, he certainly clarified the enormous burden Rossi must overcome in order to prove his case. The Judge also seems quite conscious of the inevitable appeal to the state Supreme Court, and has carefully attempted to insulate his decisions from being overturned.
The first ruling of the day was the one everybody anticipated the most, and it’s the only one that didn’t go entirely against the Republicans. The Democrats had filed a motion to have evidence of a statistical analysis excluded from the proceedings, claiming that such proportioning would be inconsistent with the standard of proof required to invalidate an election. Judge Bridges denied the motion, concluding that neither the statutes nor the courts had established guidelines in this area, and thus he would consider such evidence, subject to a “Frye hearing.”
The AP immediately broadcast a headline that implied the court had accepted the Republicans “proportional deduction” methodology, ignoring Judge Bridges’ huge caveat:
However — and this is an important however. The denial of this motion should not be interpreted as a pretrial ruling adopting the statistical analysis methodology, so everyone understands that, and that’s the ruling of the Court.
As I had previously mentioned, my fly-on-the-wall reported that the first issue the Judge addressed in conference was scheduling the Frye hearing, at which the Democrats will argue that the GOP’s methodology is bad science, and thus inadmissible — an argument I don’t expect Judge Bridges’ to entirely accept. After all, we can all imagine a situation where some sort of statistical analysis makes perfect sense; for example, if Yakima County’s touch-screen voting machines had somehow randomly erased 10,000 votes, it would be absurd to argue that this error did not cost Rossi the election.
Thus I do not think the Judge will generally rule out the use of proportional deduction per se, but rather decide that the evidence is not “clear and convincing” in this particular case… a shrewd judicial move. If he were to reject proportional deduction as a point of law, it would leave his decision much more vulnerable to being overturned on appeal. But if he rejects its application as part of his evidentiary findings, the decision is virtually appeal-proof.
Clever.
If you are a Rossi supporter, that’s where the good news ends, such as it may be. The rest of the day’s rulings overwhelmingly favored the Democrats.
The next ruling concerned a motion by Democrats that the Republicans must actually prove a felon voted, by finding a signature in a poll book or on an absentee or provisional envelope, rather than simply relying on the voter crediting records. Judge Bridges ruled in favor of the Democrats, stating:
The process of crediting voters with having voted is a post-election administrative exercise that this Court determines does not bear upon the authenticity of election results….
Republicans shrugged off this ruling as little more than a procedural hassle, and indeed it’s likely the vast majority of felons alleged to have voted, probably did cast ballots. (How many were really illegal votes or not, is another question.)
But if Rossi had planned to argue the same case in a court of law as he has in the court of public opinion, then this ruling represents a huge defeat, for it entirely undermines what’s left of the “total mess” theory… that there were hundreds of unexplained “voterless ballots.” This charge was based on the discrepancy between the number of voters credited with voting and the number of ballots cast, but since the court has ruled that voter crediting has no bearing on the authenticity of election results, the so-called “voter credit discrepancy” is entirely meaningless. (And, I should mention, it also totally validates my refutation of the Snark’s OCD-like focus on this issue.)
In launching his contest, Rossi emphasized a number of allegations in support of his contention that the election was a “total mess,” including: that thousands of overseas military voters had been disenfranchised, that King County had illegally enhanced and duplicated tens of thousands of ballots, and that King County tallied hundreds more ballots than voters. It is telling that none of these allegations will be argued in court, their case almost entirely relying upon a statistical analysis of illegal votes by felons, and unverified provisionals.
“Total mess”… my ass.
After a recess the Court addressed two dueling, related motions, that due to some inexplicable bungling on the part of Rossi’s attorneys, resulted in the kind of ruling that legal malpractice cases are made of. The Republican’s motion argued that upon a prima facie showing that a voter is a felon that had not had his rights restored, the Democrats should bear the burden of proving that the vote was not invalid. Meanwhile, the Democrats’ motion asked the Court to exclude all evidence of illegal felon voters unless the R’s could prove six elements, that the voter was 1) convicted of felony, 2) as an adult, 3) had not received a deferred sentence 4) had not had their rights restored, 5) had cast a ballot, and 6) had marked the ballot to indicate a vote for governor.
But before issuing his rulings, Judge Bridges joined Jim West in sharing with us a glimpse of his bedroom fetishes.
As I was lying in bed last night, I had one of the fears that I think attorneys have had often, I’m sure, did I miss something. Am I going to get in court and realize that there is an issue that I just completely overlooked. Mr. Foreman started out his presentation a few minutes ago with the burden of proof argument, that is, is it by a preponderance of the evidence or is it clear, cogent and convincing evidence. And in actuality, I hadn’t anticipated specifically that that argument was before the Court, based on the written materials that the Court had been presented.
Ohmigod… did Rossi’s attorneys really raise the all important issue of burden of proof, without first briefing the Court? Uh-oh….
I’ll make a ruling. If counsel wish, however, to readdress the issue, I invite counsel to do that.
Translation of obtuse legalese: “You fucking morons.”
Up until this point, Judge Bridges rulings had been rather brief, but perhaps spurred on by the GOP’s amateur performance in arguing what, by any measure, was a truly stupid motion to begin with, he started lapsing into schoolmarm mode. In denying the Republican motion he ruled that “a felony conviction, coupled with the absence of a certificate of discharge… does not establish a prima facie case of illegal felon voting.” He then went on to lecture the GOP attorneys on some legal fundamentals, explaining why the burden of proof rests with the contesting party:
The reasons the burden of proof does not shift is grounded in both our case law as well as our statutes, and the Court, of course, as are counsel, we’re all mindful that the courts of this state presume the certified results of an election to be valid unless the contrary is clearly established. And unless an election is clearly invalid, when the people have spoken their verdict should not be disturbed by the courts.
Just to be sure the Republicans got the point, he then went on to cite statute that stated that registration is presumptive evidence of a person’s right to vote, and that when a voter’s right is challenged, the burden rests with the challenger, and must be proved by “clear and convincing evidence.” The same standard, the Judge ruled, should apply when election results are challenged under RCW 29A.68.020.
Inasmuch as voting is a constitutional right, no vote should be held illegal and discounted absent clear proof that the voter was legally disenfranchised.
Any questions? I didn’t think so.
Next the Judge denied the Democrats motion, but after doing so, issued guidelines for establishing that a felon vote was illegal, pretty much along the lines of what they had requested: that the individual was 1) convicted as an adult, 2) of a felony, not a misdemeanor, 3) was not given a deferred sentence, 4) did not have rights restored, 5) cast a ballot, and 6) marked the ballot for governor. Of course, number 6 is impossible, which the Judge duly notes, but says: try anyway.
And then came the killer. (Yes, it gets worse.)
With respect to and responding to Mr. Foreman as to simply what is the burden of proof, I’m going to say it’s clear and convincing.
At this point, while MSM headlines are still proclaiming a big Rossi victory, his attorneys are probably shitting in their pants. “Clear and convincing” is a very high standard, and it doesn’t just apply to illegal votes, but to the entire case. So for those of you still clinging to the fantasy that the legal definition of “appears” is “appears to Stefan,” it’s time to start moving towards the fifth stage of grief.
And it only gets worse, for now the Judge starts to get a little improvisational. Judge Bridges remarks that he’s notices a theme to the Republicans’ arguments — that an election may be invalidated where the number of illegal votes exceed the margin of victory, without proving which party was credited with illegal votes — and he wants to get this issue “out of the way.” Judge Bridges explains that,
Washington’s election contest statutes clearly require the contestant to show illegal votes or misconduct changed the election result based on RCW 29A.68.110 and .070.
He cites the much debated Foulkes v Hayes case, where the court set aside an election based on evidence that ballots were fraudulently altered, without requiring proof that the result had changed. But, Judge Bridges notes, that Foulkes does not mention these specific statutes. Neither, he notes, does Hill v. Howell, where the court also suggested that “such a showing might not be required where fraud, intimidation or a fundamental disregard of the law had occurred.” Indeed, there’s only be one case where the courts did not require “proof of causation,” and that was Foulkes, a case that involved fraud.
But in our case here today, the petitioners have never alleged, to the Court’s knowledge, or even alluded to fraud or voter intimidation.
Hmmm. I’m guessing that may come as a shock to those of you who have gotten all your election contest coverage from talk radio and the right-wing blogs. So if that describes you, you may not want to read the following:
The rule urged by petitioners may be a wise one and a tempting choice for the Court. However, the Washington legislature has, by enacting RCW 29A.68.110 and .070, removed this choice from this Court’s discretion. The statutory command is clear and the Court should not invalidate the election upon proof the number of illegal votes exceeded the margin of victory.
The final issue of the day was a Republican motion to exclude all Democratic evidence of offsetting illegal votes and other irregularities, a gambit that perhaps was not quite the stupidest motion of the day, but certainly vied for the title of “Most Desperate and Futile.” Needless to say, the Judge denied the motion.
Conclusions
Rossi is toast.
It doesn’t take a genius to see which way Judge Bridges is leaning on this one, and “leaning” is a dramatic understatement… he’s virtually horizontal. Furthermore, he’s carefully constructing a decision that will be extremely difficult to appeal. Every motion to exclude evidence on a point of law has been denied; instead, the decision will be made based on his evidentiary findings.
Of course, none of this is a surprise… to me. I never thought Rossi had much of a case. There had been some thoughtful analysis coming from the other side, but most of it was based on a misinterpretation of the Foulkes decision, as I have previously pointed out here and here. And now even Rossi supporters are beginning to admit the inevitable.
Anyway, sorry to get so wordy. Now that all the evidence is in, and the standards of proof settled, I’ll follow up sometime soon with a brief overview.
Gosh Goldy–a Rossi supporter relying upon your legal expertise who be suicidal right now. Fortunately. while many of your observations may be correct…your conclusions are likely incorrect. The stand of proof issue is clearly a high barrier…however not impossible. EVERYONE agrees the burden should be high…but not impossible.
However, everyone agrees there have been numerous illegal votes cast. That precise number has yet to be determined. Once that number is clearly determined, the question shifts to “what does the Court do about it”? Goldy’s solution is NOTHING. I doubt that will happen Goldy….I have used the example many times of if a ballotbox were stuffed with illegal ballots, how would these illegal ballots be removed when they are already COMINGLED with legal ballots??? Clearly, there needs to be some form of relief. Proportional analysis is pretty much the only reasonable method once the Judge rules out “illegal votes exceeds margin of victory”.
Will Team Rossi show enough legal votes that when a reasonable proportional analysis method is applied, it will erase the margin of victory? I do not believe the Judge will strive for a one & only perfect methodology that is very complex. Precindt is probably reasonable and simple to understand and administrate.
What about the other issues raised in the R’s challenge??
Goldy can rest assured neither side will call him as an EXPERT WITNESS!
I am sure Goldy that the righties will immediately run to Lush Flimbaugh land and hope that God really is on their side after reading your piece on the decision in the Rossi case.
It looks like incompetence runs amok in GOP land. First, we get a guy running for Governor who doesn’t know the difference between a real estate agent and broker, then we get a party who hires Mr. Magoo for their attorney!
I love watching a GOP train wreck. I really do!
Goldy, all you said was we had bad lawyers and thus a bad case. All the facts laid out so far support our position that this election was intentionally bungled (by deliberately not checking the legitimacy of voters, the accuracy of the ballots vs reg. voters, etc, etc.
scott-
If Goldy is your guru of truth and legal analysis, you are in for a HUGE disappointment. Let’s put it this way scott, if you only have one LEGAL mind to choose from to keep you out of the BIG HOUSE…don’t pick Goldy!
The Accounting Grads at WSU got it right at their commencement..
when Gregoire spoke, the grads who spelled out A_C_C_O_U_N_T_I_N_G on the top their heads
shifted seats to spell out
C-O-U-N-T A-G-A-I-N
It was quite funny…but something the MSM predictably ignored.
GOLDY: Your points are right on. My take is that those that don’t want to accept the inevidable don’t want to see past the legal maze and haze created by the about-to-be losers. As I have always said, there will not be an overthrow of the past election no matter how much the Rossi Repers turn the pile of shit over to make it look more than it is, it still smalls and spells big-time loser for Rossi and camp.
I am rather upset that the state GOP hasn’t done anything to actually count the number of people who voted in King County in November 2004. This could easily be done (with the dedication of sufficient resources) by taking all the names from the poll books, the absentee envelopes, and the provisional envelopes, couting them, sorting them by precinct, and counting them again. I have guessed that this would cost $100,000 to $200,000, but that is chump change compared to the $4 million that the GOP has spent so far in this election contest.
It doesn’t seem like the GOP has done this, and they won’t be able to prove that more ballots were counted in King County than people casting them. There certainly are a lot of suspicions raised when the voter crediting records, adjusted downward by the irregular provisional voters (i.e. directly into the machine), and adjusted upward by things like no ballot in envelope and ballot not removed from envelope, appear to suggest at least 1200 more ballot counted than voters who cast them.
Obviously, the GOP could have spent $100,000 to $200,000 on this process, and not really found much. But this still would have indicated that King County’s records were seriously in error by failing to credit voters, especially if these were absentee voters. If the GOP didn’t prove their case, they would prove that King County Elections was screwed up either way.
As it stands, Vance will have hell to pay if this election contest is lost, and the GOP doesn’t present credible evidence to support the voterless ballot theory. There was so much said about this, so much circumstantial evidence (voter crediting records) which supports it, no actual evidence to contradict it, and such outrage if one assumes it is true (ballot stuffing!).
GOP supporters are (maybe WERE is the better word?) looking for Vance to present this theory at trial and hopefully prove it. Instead, it looks like Vance is punting the ball on this and giving up like a sissy. Four million bucks (and counting) of GOP money poured into this contest, and the voterless ballot theory is completely forfeited.
This election contest has diverted GOP members from calling Vance to account for some of his truly boneheaded actions. Such as naming lunatic Will Baker to run for State Auditor. Not only is this fellow completely crazy, he has a lengthy criminal record and absolutely none of the qualifications that an auditor should have.
This single dumbass act greatly undermined public confidence in Vance’s annointment of other statewide candidates (such as Governor and Senator) and easily cost Rossi far more than the 129 vote victory (?) margin. Obviously it didn’t help to have the party pick a candidate (Will Baker) with an obvious criminal record when Rossi’s association with criminals (such as his convicted felon boss and real estate mentor) in his personal career was a major concern of the voting public.
GOP supporters would be better served if Chris Vance got some sort of political appointment into a cushy government job, and someone more professional was hired to run the state party. Perhaps Vance could serve well working on the mayor’s staff in a larger city — Spokane comes to mind …
Mr. Cynical @ 1
”However, everyone agrees there have been numerous illegal votes cast…. Once that number is clearly determined, the question shifts to “what does the Court do about it”? Goldy’s solution is NOTHING. I doubt that will happen Goldy….I have used the example many times of if a ballotbox were stuffed with illegal ballots, how would these illegal ballots be removed when they are already COMINGLED with legal ballots??? Clearly, there needs to be some form of relief. Proportional analysis is pretty much the only reasonable method once the Judge rules out “illegal votes exceeds margin of victory”.”
I agree with you that a “proportional vote reduction” method is pretty much the only reasonable method to use when illegal ballots commingle with legal ballots. However, valid use of the method requires that two conditions hold.
1. The illegal votes must have the same characteristics as the pool from which they are drawn. So, if the method is applied at the precinct level, the illegal votes being “proportionally reduced” must have the same characteristics as votes in the rest of the precinct.
2. Illegal votes must be identified in the same way and using the same degree of scrutiny for all units of analysis. In other words, each precinct in the state must have their voting records scrutinized to the same degree in order to yield unbiased results at the state level.
These two conditions are not ideals. They are requirements for a proportional analysis to be considered statistically valid. Condition one is the much-discussed question of whether ex-felons vote like others in their precinct. It is hard to imagine how to get around this problem given an absence of good information on how former felons vote. Of the two conditions, however, Judge Bridges may be convinced to overlook this one.
The second condition appears to be fatal for the Republican case. The Democrats should be able to show that Republicans cherry-picked illegal votes from Gregoire counties and precincts. The Democrats have found some 700+ illegal votes of their own by cherry-picking. They did not do this to help the Republicans in promoting a proportional vote reduction solution. Rather, they did this to show that condition two is not met. And, if the Democrats can show this then it is easy to show that the proportional reduction method is bad science.
The only way I envision getting around condition 2 is if the Judge decides that both sides have cherry-picked to roughly the same extend (but in opposite directions) and therefore their invalid voters should be pooled. Even though this solution fails the two conditions above, it has the psychological benefit that a remedy has been offered. In this case, it seems most unlikely that the Republicans can prevail. Even if one-half of the illegal votes on the Democrat’s list are thrown out and all from the Republican’s list are kept, the election still goes to Gregoire (I demonstrated this using a statewide analysis here, and my statistical intuition is that a precinct-wide analysis will shift the results in Rossi’s direction by well under 10 percentage points, which is not enough to overcome Gregoire’s lead).
Overall, I think it is extremely unlikely that the Republicans overcome all three of these hurdles to win their case.
”What about the other issues raised in the R’s challenge??”
What other issues, in particular, do you think are strong in the Republicans? From my reading, the proportional vote reduction is the only thing left in the case that has any possibility of showing that Gregoire lost the election. Judge Bridges seems to have ruled out the voter crediting discrepancy arguments.
@ 3 we have an early sighting of the inevitable metamorphosis of wingnut rhetoric that will occur over the next few weeks.
Yes, we’re moving from the euphoric, er, righteousness of January and February to the somber preparation for defeat later this month. Gone now are the daily bleatings of the self-evident correctness of the GOP challenge. No more celebrating that the case would be heard in friendly territory, by a tough and fair-minded judge who was, nevertheless, one of theirs. All of this has melted away as the GOP quietly dropped key elements of their case and began to face up to the lack of evidence for what remained.
So now we begin the move to the next stage: Our lawyers sucked! The judge was an earring-wearing librul! The Supreme Court is stacked against us!
Even this will be a mere transition to the final stage of perpetual begrudgement. Long after the final decision, when most of the voting public has lost interest, the true believers will still be telling each other their myths of eternal victimhood, keeping the faith alive. Fortunately for the rest of us, this last stage will take place largely out of public view save for the occassional cranky letter to the editor.
Richard @ 6: You’re correct in pointing out that the only way to make a case regarding a voter/ballot discrepancy is through analysis of polling place documents and absentee ballot envelopes. I think we’ve agreed on that since January when I tried pointing that out to Stefan Sharkansky.
You’ve pointed out that this might be costly, but it’s the only way to go for the GOP.
I’m not sure it would even be that costly — thanks to our good friend, Stefan. Most of his work is crap, based on specious methodology, but there is a small kernel of potential utility to his analysis. He’s already identified the handful of precincts that have the largest apparent discrepancy. The GOP could concentrate its attention on a careful examination of the records from those precincts — at far lower cost than what you propose.
I don’t think the court is going to be impressed with one- or two-ballot discrepancies randomly scattered over a number of precincts. The most likely explanation for those is careless inattention on the part of bored poll workers. However, the mysterious appearance of a dozen or more ballots at a single location or in a single absentee batch is harder to ignore. Sharkansky thinks this has happened in a number of precincts or batches, so why doesn’t the GOP take a look?
Now, I have no way of knowing, but my guess is they already have. It may be that on close inspection, a more mundane explanation becomes apparent — who knows? My bet is that the GOP has already found this to be true, or they are completely incompetent. Take your pick.
DJ @ 7
I agree with your first point, but your second point is, I believe, incorrect.
You say: “In other words, each precinct in the state must have their voting records scrutinized to the same degree in order to yield unbiased results at the state level.”
As a legal matter, I don’t believe this is correct. The GOP only has to show (by proportional analysis or actually showing how they voted) that the illegal/improper votes which they have identified changed the outcome of the election to make a prima facie case. It is them up to the Dems to rebut that showing by presenting their own offsetting illegal votes, or other off setting errors which deprived Gregoire of votes.
Proportional analysis is a (flawed) method to determine how to allocate the **identified** felon and improper provisional votes between the candidates. But there is no legal requirement that the challengers need to identify all the illegal votes in order to meet their burden of proof.
chew2@10
1) please describe how repubs will show where these votes changed the outcome. In detail please.
2) It is not up to the dems. to prove how felons did NOT change the outcome of a certified election. (see number 1)
Richard Pope @ 6
The bizarre thing about the Will Baker candidacy is that he got 1/3 of all votes, statewide. Which means that 1/3 of all voters (2/3 of Repub voters) were voting blind, on the basis of party alone, or else are wingnuts as extreme as the candidate.
As to the court case:
Ken Vogel writes three pieces on the felon vote and standards of evidence for today’s Tacoma News Tribune at http://www.thenewstribune.com/.....2823c.html .
“Comment by Mr. Cynical— 5/10/05 @ 4:49 am”
When do you sleep?
Re: Cyn@1 / Jon@13
Jon’s right Cyn. No wonder you’re so cranky.
Richard Pope @6,
I agree with you that it would be amazing if the Republicans did not at least sample some precincts and look for clear evidence that there were indeed ballotless voters. My understanding is that they did, though I have not been able to get anybody on their legal team to confirm this.
But as I have previously stated, the fact that the voter credit records show an apparent discrepancy distributed randomly across the entire county, makes it very difficult to conceive of a mechanism for executing such widespread ballot stuffing. And I continue to question what could possibly motivate election workers to risk career and/or liberty by fraudulently stuffing a few hundred ballots without any reasonable expectation that such small numbers would impact the result.
Without any supporting evidence, I can’t see how such election fraud theories make sense, unless you subscribe to Jim Miller’s thesis that Democrats are more likely to cheat than Republicans, and thus fundamentally corrupt.
Seems everyone’s forgotten that the innards from three “black box” voting machines were reported missing in WA state several months ago. No one as yet appears to know what’s happened to them. My “Spidey Conspiracy Sense” says the Democrats have them and there are some more nasty surprises awaiting our “too eager to yell fraud” WA state Republicans.
All the bluster aside, we will have to wait and see what happens in a few weeks. Until the verdict is in, all of it is idle speculation.
lucy@16,
please go on.
Chew2 @ 10
’I agree with your first point, but your second point is, I believe, incorrect. You say: “In other words, each precinct in the state must have their voting records scrutinized to the same degree in order to yield unbiased results at the state level.” As a legal matter, I don’t believe this is correct. The GOP only has to show (by proportional analysis or actually showing how they voted) that the illegal/improper votes which they have identified changed the outcome of the election to make a prima facie case.’
I was addressing the scientific validity of the method. Statistically, a proportional vote reduction method is valid only if equal scrutiny is applied. If this standard is not applied, then the final vote count will have “statistical bias” (and I am using the word “bias” in a technical sense). If the final vote count is biased, then it cannot be used as scientific evidence that that Gregoire owes her win to illegal votes. Again, this is an argument about scientific validity, but the court will most likely require scientific validity for any kind of statistical method it is willing to apply.
“It is them up to the Dems to rebut that showing by presenting their own offsetting illegal votes, or other off setting errors which deprived Gregoire of votes.”
I am arguing that it may well be sufficient for the Democrats to show that the results are statistically biased using the Republican list alone. I believe that is what the Democrats were trying to do in uncovering additional illegal votes. But, if Judge Bridges rejects the second condition (and I admit that just because a biased proportional vote reduction method is scientifically invalid does not guarantee the court will reject it), the Democrats will have their own list of invalid votes to throw into the mix. Hence, the third hurdle I mentioned.
”…But there is no legal requirement that the challengers need to identify all the illegal votes in order to meet their burden of proof. “
I did not claim that all illegal votes must be identified. I said that equal scrutiny must be applied to all precincts. For example (ignoring condition one) condition two would be satisfied, and the results scientifically valid, if each illegal vote was identified with a 50% probability across all precincts. That would yield approximately unbiased results.
pbj@17–
But without idle speculation, what would we do???
dj@7/Chew2@10–
My guess is chew2 is correct here. dj…I think you are making too big of a leap here. The Judge can only deal with the illegal ballots he can validate. He cannot speculate that there are others around the state (intuitively one could surmise that…but remember, Bridges want hard evidence of illegal votes to consider, period). I just can’t see him turning around and saying that because the Dems found some the R’s didn’t….there is no relief.
No, I believe Bridges ruling cuts 2 ways. High burden on R’s….but high burden on D’s. What can be validated as ILLEGAL to whatever final standard Bridges decides on will then have some reasonable proportional distribution method applied.
Again, I think the final method will be fairly simple and straightforward like by precindts.
I also think he will accept the R’s proposed methodology as reasonable. The Dems can come along and try to complicate it wo swing the result their way but…..
It all comes back to that darn word APPEARS!
Rujax206@14 and Jon@13–
Cranky???
I’M NOT CRANKY YOU LOWLIFE MF’S!!!!!!!
EAT SHIT AND DIE!!!!
Geez…I’m sick and tired of a bunch of SEATTLE PINHEADED LEFTIST ELITE A$$HOLES accusing me of being CRANKY!
Headless lucy@16
You mean the ones the aliens delivered at Roswell? You’re smoking dope again, thinking Mellon Scaife placed secret computer chips in all the voting machines to steal GOP victories. If he’s that good, why’d we lose (lost the 3rd count, that is)
Serious question?
If you libs were on the other side of this, what would an election recount effort/procedure be for?
Re-checking just the aritmetic
Re-checking, plus looking for stray uncounted ballots
Would you ever allow, or introduce question on the validity of the voters
Would you ever allow or question the above arithmetic, but looking at some audit of votes, ballots, et?
I’m sort of baiting you, but watching this unfold, cynic in me says, “ok, is there any recourse when the system is all screwed up, is the only time you challenge when you can prove outright fraudant ballots introduced (i guess innocent until clearly, factually proven guilty)?
righton @ 23: I don’t see why this is so hard to understand. You challenge when you can show that you lost the election due to fraud, a surplus of illegal votes credited to your opponent, or through gross negligence or illegal activity on the part of election officials. The last two options require that you can show that the cited activity was significant in that they caused you to lose an election you otherwise would have won. Mere showing of mistakes isn’t enough — they need to be shown to be the cause your guy’s defeat.
It’s all laid out in the RCW. Most of the GOP claims don’t even meet the minimum standard for the above; that’s why they’ve quietly dropped them. The remaining claim barely meets the threshold; that’s why the case is still alive — though barely.
We’ll see how it works out. My question for you: Will you accept the final result? I know I will.
dj @ 10
you say: I was addressing the scientific validity of the method. Statistically, a proportional vote reduction method is valid only if equal scrutiny is applied.
While that may be correct as a general statistical matter. That is not how proportional reduction is being used in this case. Once an illegal vote is identified, proportional reduction is being used to estimate for whom that illegal vote was cast. In that case, the sampling universe is simply the precinct in which the vote was cast. There is no sampling bias here, is there? I don’t believe this destroys it’s “scientific validity” at all.
(I agree that felons are not a random draw from the precinct, and that this should affect the “scientific validity” of proportionate reduction, but that is another issue.)
Proportional reduction has nothing to do with whether the identification of felon votes was complete or biased. No doubt they were conciously cherry picked by both sides. It is simply being used to apportion those illegal votes once they were identified.
I believe that proportional reduction is a suspect and too uncertain a method, and should be rejected by the court. But if it does adopt it, I don’t think that the fact that the illegal votes were cherry picked will prevent its use by the court.
As an aside, there were quite a number of cases cited by both sides which either permitted or rejected “proportional reduction”. None of those cases treated the method as a statistical or scientific issue. It was all common sense reasoning.
Scott
So if I could show 10,000 absentee ballots injected illegally into the system, I have no case unless I know who they all voted for?
righton: If you could show 10,000 absentee ballots injected illegally into the system, you’d have a great case, regardless of whether you could show who they voted for. See Foulkes v. Hayes.
Unfortunately, you can’t show that.
righton:
P.S. — Now that I’ve answered your question, how about answering mine?
Richard Pope @6
QUESTION.
I haven’t kept up with all of the allegations, but I recall that one of the things which disturbed me was that there were excess absentee votes shown for King county. It was also shown that absentee ballots are recorded and credited when they are received and before they are counted. Was there ever any explanation for this discrepancy by King County officials or how it could possibly occur? Can the Rossi folks show that these excess absentee votes are illegal or improper votes?
scott
Will I accept decision, appeal is legally binding, sure. Will I accept its morally/logically correct, I don’t think so. Since nobody in authority is honestly reviewing, all i can do is see the bad side of the election. Logan isn’t really self examining, Sims appted a token review committee, and the court process is about procedures and not really about the truth. I’m not a lawyer, so when I see Dems going back to houses after the election to get affirmation of their signature, and then poll workers taking absentee ballots home, and then some >10,000 vote delta between voters and votes, plus “benign neglect” of the dead/felons, plus KC seemingly far, far worse than other counties….I guess its down to, “have the AG or US Attorney investigate this sucker, otherwise to me Rossi won 2 recounts and then the KC machine swung into motion and robbed him of the election”
Scientific statistical analysis in this fine state of WA proves beyond doubt that those machines are rigged. The problem is not whether but how. And as for Mellon-Scaife, he does contribute a lot of money to the Evergreen Freedom Foundation—an anti-union/edcation front group for Wal-Mart, Mellon-Scaife and others of that ilk. Them’s the facts, Jack. So, I’ll just wait for the facts on the voting machines to surface and then gloat while you are once again “hoist by your own petard.”
Jerry Van Dyke
righton:
Some folks see dead people; you apparently see a “>10,000 vote delta between voters and votes.” It’s hard to say who’s more delusional.
If it exists, why doesn’t the GOP make an issue of the 10,000 vote delta in court? Seems to me, that kind of thing would help them out a lot.
BTW: Your response shows an interesting way of “accepting” the final decision. Sounds to me like you’re filling out your membership application for the Committee of the Perpetually Aggrieved. Have fun there; I hear it’s a tight-knit group.
Compare that with my response — I’ll accept the final decision unconditionally.
Mr. Cynical: Geez…I’m sick and tired of a bunch of SEATTLE PINHEADED LEFTIST ELITE A$$HOLES accusing me of being CRANKY!
Let’s break this down:
SEATTLE: Don’t live there.
PINHEADED: Well, I don’t think so; but you are entitled to your own opinion.
LEFTIST: I’ve been accused of many things, but certainly not that.
ELITE: I could show you my 2004 return to disprove that one.
A$$HOLES: See pinheaded.
I guess I must be a moderate (although I consider myself conservative) if I get yelled at by both sides around here, so thanks!
P.S. Rossi (who I did vote for) is truly toast.
Hey–
I’ll accept Judge Bridges decision as final.
Is that what you are saying scottd???
Or are you reserving your right to support a Supreme Court challenge if the Dems lose???
Be very clear now….cuz it reads like “accept the final condition unconditionally” is a bit squirelly.
Chew2 @ 25
’While that may be correct as a general statistical matter. That is not how proportional reduction is being used in this case. Once an illegal vote is identified, proportional reduction is being used to estimate for whom that illegal vote was cast. In that case, the sampling universe is simply the precinct in which the vote was cast. There is no sampling bias here, is there? I don’t believe this destroys it’s “scientific validity” at all.’
If the goal were to determine an outcome of a precinct, then the sampling universe is the precinct and you would be correct. But, the goal in the election contest is to adjust a statewide election total. Therefore, all precincts must have the same degree of scouring for illegal votes in order to provide an unbiased statewide result. That is, a statewide adjustment would be biased by cherry-picking.
(I agree that felons are not a random draw from the precinct, and that this should affect the “scientific validity” of proportionate reduction, but that is another issue.)
I agree with you completely on this point.
”Proportional reduction has nothing to do with whether the identification of felon votes was complete or biased. No doubt they were consciously cherry picked by both sides. It is simply being used to apportion those illegal votes once they were identified.”
The issue is whether cherry picking can ever convince the Judge that Gregoire owes her win to the illegal votes. If the results are biased, then the case cannot be made (at least from a scientific standpoint, your judicial miles may vary). The courts are expected to use accepted science when they allow scientific methods into court.
Suppose a political scientist decided to publish a paper using this method and the Republican illegal vote list. The paper comes to the conclusion that Gregoire did not win the election, and might make a statement like, “We have shown that Rossi gained significantly more votes than did Gregoire at a significance level of 95%. The paper would have zero chance of making it through scientific peer review. The statistics would be deemed flawed because of the way the illegal vote data were collected. The author would be told that his methods do not and cannot show that Rossi acutally had more votes because the illegal votes were not sampled according to accpeted scientific standards.
Likewise, if the courts uphold the use of accepted science only, then the sampling method for finding the invalid votes will fail the validity test. (Note, the problem is not with the basic proportional method, the problem is that the illegal vote data cannot validly be used with the method to make inferences statewide).
”As an aside, there were quite a number of cases cited by both sides which either permitted or rejected “proportional reduction”. None of those cases treated the method as a statistical or scientific issue. It was all common sense reasoning.”
For this case, the Republicans have introduced proportional vote reduction as a statistical method (in particular, look up Katz’s two reports on the SoS web site), so this case may differ from others. This case is so close that some statistical reasoning is almost certainly needed. If one does an exact method using the Republican list only, Rossi only wins 96.1% of the time. That just barely squeaks by at the level scientists work at (usually 95%), but there is no guarantee that Judge Bridges will accept a 1 in 26 chance of erroneously turning over a valid win by Gregoire.
Richard Pope @ 6: Once again, excellent comments! We need more of that insightful analysis around here!
Mr Cynical @ 34
I know you didn’t ask me, but I’ll answer anyway. I will have to go with Don/Alan on this one. “I’ll accept as final a ruling by the highest court of jurisdiction. “
I am curious as to why you will accept as final Judge Bridges’ decision. I assume you accept as legal fact that an appeal is an option for the losing side.
Cynical: I wouldn’t want to leave you guessing.
I meant the final Supreme Court ruling. That’s the end of the line for our legal process.
P.S. : I don’t think the Dems are going to have any problem with Judge Bridges’ decision either — but I wouldn’t begrudge the GOP’s right to appeal.
dj–
I kinda have more than a strong inkling that Judge Bridges will basically put blinders on when it comes to his ruling….not worrying whether or not the election outcome is reversed, as hard as that may be able to believe.
Judge Bridges is akin to Tiger Woods putting for Birdie on the 18th Hole of the Masters when he is tied. Does Woods think…this is for the Masters? Sure. But he focuses on the task at hand…making the putt which means lining it up, judging the speed and break and making the stroke.
The Judge will focus on identifying illegal votes and a reasonable proportional distribution method. Then the math is done and it is what it is. If it’s less than 129…No Change.
If 129 or more…Change.
The math is merely a result of the underlying assumptions.
Folks like us can get all out of joint doing the math and trying to back into a formula that favors us or a method of identifying Illegal votes that favors our candidate. We all have our pre-determined outcome….Judge Bridges does not.
dj @ 35
We are talking past each other a little bit here. I am talking about what the law requires and you are talking about what scientific validity might require.
1) Let me try an example. The GOP identified 1000 felon voters, who they cherry picked to come from Gregoire precincts. The Dems idenfitied 500 felons from Rossi precincts. The GOP can attempt to prove their felons voted for Gregoire by proportional reduction based on precincts, or by asking them. If the GOP proposed to question them, the judge will not bar this evidence because the felon sample is biased and cherry picked. And if the GOP came up with enough votes by this method to change the outcome of the election, then they would meet their legal burden and the judge would rule in their favor, absent any contrary evidence by the Dems. The bias in picking the felons doesn’t prevent the GOP from meeting their burden of proof. It doesn’t matter that there were 500 other felon votes out there in Rossi precincts, unless the Dems introduced evidence to show how they voted. Well the same legal reasoning will apply if the GOP uses proportional reduction to prove how those felons voted. It will not be excluded because the sample of felons was biased or incomplete, although it may be rejected for other reasons.
2) You say that if Rossi claimed: “We have shown that Rossi gained significantly more votes than did Gregoire at a significance level of 95%. The paper would have zero chance of making it through scientific peer review.”
Maybe. But Rossi doesn’t need to show that, i.e. that globally he won or that Gregoire lost. Legally, he only needs to show that the illegal votes that he has identified would have changed the vote count sufficient to change the election result.
chew2-
I think you got it nailed.
PS.
I think the greatest flaw with proportional reduction is the assumption of a random draw from the precinct pool. But both you and Katz have to make that assumption to apply any statistical tests at all.
So having made that assumption the question is what can you say about the voting behavior of the group of identified illegal voters you have been given. You are attempting to make a statistical statement about the behavior of the identified sample you were given from the precincts they voted in. Does it matter how that group was selected, since you are only making an inference about THEIR behavior and no one elses.
Goldy @ 15, Chew2 @ 29
Comparing ballots counted to actual persons voting cannot be done unless the entire county is done.
First, there is no physical difference between poll and provisional ballots — same form.
Second, the absentee ballot form looks different, but has the exact same computer coding. The machine counts came up with poll, provisional and absentee totals by setting the machine to report under those categories for a given stack of ballots, not by actually reading the ballots themselves to determine category.
Third, absentee ballot cast numbers for a given precinct cannot be determined unless all accepted absentee envelopes from the whole county are gone through — the envelopes are not segregated by precinct.
Fourth, the same thing probably applies for accepted provisional ballot envelopes as well.
Fifth, not all absentee ballot envelopes contain a ballot from the same precinct as the voter signing the envelope. Shocking thought, but the truth. Some voters are registered twice, but voted only once. The ballot doesn’t necessarily match the envelope.
Without a full countywide effort, the best the GOP could do would be to count the number of signatures in a given precinct poll book and compare it with the number of ballots counted from that precinct on election day — i.e. the election day poll ballot tally from the machine, as opposed to the supposed poll ballot tally in the initial count, machine recount, or manual recount.
If there are extra ballots in spite of all this, a likely source would be ballot duplication. It is possible that both the original and duplicate ballot got counted in some cases.
Hard to imagine deliberate fraud. If that was the case, why not mark all the extra ballots for Gregoire, and make sure she had a healthy victory margin?
Even when 336 mysterious extra ballots appeared in the machine recount, it was pretty obvious that these were distributed between Gregoire, Rossi, etc., rather than being cast all for Gregoire. Harder to say about the 59 mysterious extra ballots that appeared in the manual recount.
what a really great discussion between dj and chew2. Chew knows from posting at AlsoAlso that I’m interested in the statistical angle of things, but I have to say that I am leaning towards Dj’s point, even from a “common sense” perspective that the judge might seek to endorse.
Here’s why:
Chew finishes his last post with the point that selection bias is legally irrelevant to the task of apportioning the illegal votes that are known. It’s a very tempting argument, but I’m not buying. While the phrase “selection bias” is alien to common sense argument, its implications are ENTIRELY common sensical. What I mean by that, is that the judge knows good and well that Rossi’s people were attempting to find Gregoire votes to invalidate, not all illegal votes. If, in fact, it were possible to identify who voted on each specific ballot, I would agree that Bridges would reject the complaint of cherry picking. Why? IMO, because the factual legitimacy of EACH VOTE would be fairly established. Ballot A for Gregoire, Ballot B for Rossi, etc.
The problem is that Rossi does not have explicit information for each ballot. What he is therefore doing is making an IMPLICIT argument for each–that, all else equal, we can expect their votes to roughly follow the patterns of the votes already cast. In so doing, he has left the realm of the verifiable, and moved to the realm of the suggestible. The problem is no longer THESE illegal votes for Gregoire (or Rossi), but SOME illegal votes for Gregoire or Rossi–which we’ll need to apportion out in some way.
And that’s where he runs into trouble IMO, for if the outcome of the election becomes predicated on the imputed Rossi value of the TOTAL number of illegal votes in a given precinct, then the linchpin to Rossi’s case is not the percentage breakdown of the established vote–that doesn’t change–but the AMOUNT of the illegal votes, such that they may change the outcome.
Do you see what I’m saying here? An illegal vote with a known casting preference can be 100% apportioned–zero for one candidate, one for another. You cannot literally split a vote into tenths and hundredths of a vote, so by nature you are no longer considering the value of the individual vote, but the AGGREGATED value of all votes under consideration. And as such, the aggregated value is entirely dependent on the number of votes.
And here’s where common sense comes in: if the Democrats produce similar votes in sufficient numbers to suggest that the Rossi accounting of “illegal votes” has missed many in Rossi-dominated areas, then clearly the GOP’s assessment of the AGGREGATE illegal vote count is highly suspect. And if you can’t be sure about the aggregate, how can you assess the aggregate value? You can’t.
Let me put it one other way: if you know you have X number of illegal votes, and you know the preference of each one, you can superordinate those votes over a theoretical complaint of “surely there are others out there, perhaps voted for their candidate.” It may be the case, but here we have votes we KNOW were for this candidate, and we have no idea about those others. In this case, the found illegals are in practicum no more known than those we don’t know about. If we accept that we have to guess about the votes we’ve identified, on what basis is it not sensible to further guess about the votes we haven’t? If there are only a THEORETICAL total of bad votes for Gregoire, it makes no sense to dismiss the idea of other, “theoretical votes” to begin with.
Which, IMO, is why the RCW prescribes that illegal votes be enumerated as to preference in the first place–otherwise, we can sit and speculate about all the other illegal votes that we simply weren’t able to find. If, however, we allow ourselves to legally speculate about the character of the votes we have found (and that in a nutshell is what apportionment is), it makes no sense not to legally speculate about the total number of illegal votes plausibly out there. And when the Democrats show that indeed, a large number of illegal votes are in fact “out there,” it obliterates any confidence that the problem of illegal votes is being adequately recognized.
Hmm. I hope that is a comprehensible way to explain my position.
chew2: I also think you got it right. The GOP won’t have a legal problem with cherry-picking; their only requirement is to present evidence of sufficient illegal votes to give the election to Gregoire. If there are additional illegal votes that favored Rossi, it’s up to Dems to find them and present them. That’s what is actually happening.
The only problem with cherry-picking is that it allowed the GOP to delude itself into thinking it had a stronger case than it did. They seemed surprised by the notion that Democrats would be able to find their own set of felons in Rossi counties. Too bad for them.
However, I think the proportional analysis argument is going to face a tougher challenge — one that might even keep it from ever getting to open trial. The problem comes from the inference of the felon vote probability law from the samples of all precinct voters. So far, the analysis by dj and others has focused on statistical confidence in the predicted result, but that analysis assumes that the probability law is already known. Here’s an example:
If you already know that the probability law for some population is 60% Gregoire/40% Rossi, you could infer that a random sample of 1000 ballots from that population would include at least 600 votes for Gregoire. Your confidence level in that inference would be 50%. Or, you could infer at least 599 Gregoire votes with a higher confidence level, at least 598 with even higher confidence, and so on. The exact calculation comes from the binomial distribution. Our situation’s more complicated because we have to deal with votes for Bennett and undervotes, and we also have to deal with differences in probability by precinct, but that’s the general idea. (dj’s Monte Carlo analysis is an excellent way to deal with those complexities.)
dj’s analysis has centered on how many excess votes the GOP would need to present in order to reach some confidence level prescribed by the judge. In other words, how many illegal votes if the judge demands 67% confidence, or 95%, or whatever? Some folks seem to think anything higher than 50% is good enough; others think a higher standard will be needed.
That’s a reasonable discussion, but I don’t think confidence levels are the weakest point of the proportional analyis. The weakest point is in the inference of the probability law. By that I mean, concluding that there is a 60% chance a given felon would vote for CG because that was the probability estimated from the precinct population. The argument will be made that even within the precinct, there are identifiable subpopulations with voting patterns that significantly differ from the pattern of the whole. For example, it might be shown that women voted 70/30% for CG while men were evenly split. Other subpopulations can be identified by age, income, or level of identification. It will also be shown that the demographic makeup of the felon population does not match the makeup of the precinct voters.
All that means there is going to be a significant uncertainty in estimating the felon vote probability by looking at the voting pattern of the larger population — but even small changes in that estimate are going to have large effects on the number of illegal votes needed to reach any given confidence level. For example, if the probability law is 60% for CG, then the GOP would need to demonstrate 645 illegal votes to show that CG’s had at least a 129 vote advantage at 50% confidence. Or 700 illegal votes to reach 67% confidence, at 800 to reach 87% votes, and so on.
But all that was based on a guess that the felon probability law is the same as the probability law for each precinct, and from what we know about subpopulations, we already know there is likely to be an error in that guess. What happens if the felon probability of voting for Gregoire is only 2% less than the general population probability — maybe because the felon population is more male? Now you would need 806 illegal votes just to get to 50% confidence. 1000 illegal votes only gives you 84% confidence. That’s a big change.
The point here is that even a slight error in estimating the felon vote probability causes a dramatic change in the number of illegal votes needed to demonstrate the GOP’s hypothesis at any given confidence level. I don’t think the judge is going to rely on something so shaky to overturn an election.
Oops: I know it was a long post, but when I said “level of identification”, I meant “level of education”.
For reference, the five stages of grief as defined by Kubler-Ross are:
Denial
Anger
Bargaining
Depression
Acceptance
“DABDA”!
I think Cynical is mostly stuck on Denial, although he’s definitely taken forays into Anger, and he might have stuck a toe in Bargaining wrt to the accepting the Bridges court verdict as final.
Goldie, I have read the transcript of what Judge Bridges said, His calling the Democrats as sandbagging Civil court rule C26 doesn’t his lecturing that both sides need to prove their points doesn’t imply the slam dunk you are leading folks to believe. Yes, he is extablishing a burden on the Republicants to show that real illegal votes happened and that they were likely to change the results unless fraud is shown on the part of the winning political party.
But what is interesting is if the Dem’s have done the same level of documentation and homework that the Republicans & the Building industry have done on the felon and illegal vote question. Remember, he only needs to be convinced that something like 130 votes were improper and given to Chris G. to toss her to the curb. If the Dem’s have done their homework more carefully than the Republicans, Chris G might just remain Governor.
My feeling from reading things is that Bridges is trying to make this very difficult for both sides to establish evidence, but we are talking only about 129+ votes, so only a fraction of the data needs to pass a ridgid test. Likewise if the numbers are huge and the expert witnesses are convincing in their arguments this might not be so hard to prove for Rossi
The wingnuts don’t care what the facts are. They never do. The whole purpose of their huffing and puffing is to keep their right-wing sheep bleating.
Scottd @ 45
I (and dj) agree that it is incorrect to assume that an identified felon vote is a random draw from the precinct in which he/she voted. So one can attack proportional reduction as being scientifically invalid on that basis.
In principle one can attempt to estimate the felon vote based on their demographic characteristics, especially gender and race. This is called ecological inference (see King et.al.), but is subject to all sorts of pitfalls and criticisms. I asked dj about this and he was quite skeptical. While I have some statistical training, he is the expert so I will take his word for it. In any case, while the dems have raised this as an issue to invalidate proportional reduction, they themselves have not, as yet, proposed this finer level of estimation. So the judge is left with a bogus proportional reduction or nothing, a tough legal choice. Unfortunately I think he may opt to apply proportional reduction, although he may require some higher level of probability to be shown using the higher clear and convincing burden of proof as his justification. Indeed, I think the Dems would rather take their chances with proportional reduction, since it is simpler, and leave the sophisticated statistical critique as an appeal issue should they lose.
Things don’t appear to be going too well for our little chrissy… it’s the 2nd chart that is really a hoot… out of 50 governors chrissy is rated #47. Boy I bet she really thinks that 129 vote mandate is handy dandy!
So the judge is left with a bogus proportional reduction or nothing, a tough legal choice.
I dunno — that sounds kind of weak. You’re basically saying that the judge might accept a likely bogus method just because he doesn’t have anything better. I don’t think so. Bogus is bogus. You wouldn’t use chicken entrails just because nothing better was available :-)
It could just be that the GOP has an inherently weak case. It’s up to them to make the case. If the method they offer is fundamentally unsound, they don’t get to use it just because they have nothing better. Remember, the GOP is asking to overturn an election that is presumed conclusive until proven otherwise, and this judge has already indicated a tendency to set a high standard before granting this request.
The analysis I presented @ 45 showed that the proposed method is unstable because minor changes in the input assumptions lead to large changes in the output conclusion. In this case, the likely noise at the input is enough to flip the election back and forth, so the method is unreliable. The Democrats should hammer on that point.
Bob @48,
I think the point is — and I haven’t clearly enunciated this in a post yet — that even if he accepts proportional deduction in theory, and even if the final number of illegal votes suggests that Rossi won, he still may not find that to be clear and convincing, because of all the other variables involved. Proportional deduction is not absolute proof, and he’s going to require a very high degree of certainty. The Democrats understand this, and that is why their experts are going to hammer on the “ecological fallacy” theme.
That said, unless the Dems are complete and total fuckups regarding their offsetting votes, the R’s simply don’t have the data to prevail, even given their own methodology.
Oh… and Judge Bridges did not say the Dems were sandbagging… he was repeating the argument of the R attorney. He doesn’t lecture to the Dems at all on this subject.
I don’t care who is governor as long as they get rid of abortionists, deviants, drug pushers, pornographers, thieves, Social Service workers, vagrants, drunks, attorneys, bad surgeons, sleazy celebrities, and mean people.
What a fantastic discussion you all had today on this thread. Sorry I could not respond sooner—busy day today.
Mr. Cynical @ 39 (and Chew2 @ 40). I do, indeed, understand the points you are making, and you may well be correct that Judge Bridges will take the simplest possible analysis. The reason why I think a more sophisticated analysis will be used is that the Republican’s expert witnesses have introduced such methods!
Chew2 @ 40
”I am talking about what the law requires and you are talking about what scientific validity might require.”
Indeed, I am talking about scientific validity. But, that is because I am under the impression that courts using scientific methods require the methods to be accepted and valid science. (Lawyers, please feel free to correct me on this).
“ But Rossi doesn’t need to show that, i.e. that globally he won or that Gregoire lost. Legally, he only needs to show that the illegal votes that he has identified would have changed the vote count sufficient to change the election result.”
This is a point on which we disagree. I definitely believe that the Republicans must show that Rossi actually won the election, or in deference to Mr. Cynical, “Rossi appears to have won the election.” That is my reading of the RCW, but I am not a lawyer, so I might well be wrong on this one.
Chew2 @ 42
”I think the greatest flaw with proportional reduction is the assumption of a random draw from the precinct pool. But both you and Katz have to make that assumption to apply any statistical tests at all.”
I agree that this is a serious flaw (condition 1 in my original post on this thread). Scottd @ 45 agrees with this as well.
I am pretty certain that this is going to be a huge issue for this case. I know both of the D statistical expert witnesses personally (although I have not discussed the case with them yet). Professor Adolph, as a graduate student, worked for Professor Gary King at Harvard—the guru of ecological analysis in the field of political science. Interestingly, Professor Adolph was also a research assistant for Professor Katz, the Republican statistical expert! How weird is that? In any case, he has published papers on ecological inference. Professor Handcock is a first rate statistician, and I can think of only a few statisticians in the world who might be recommend over him. He has also published on ecological inference. So, my guess is that they will hit very hard on the “ex-felon non-representitiveness” issue. I am quite certain one or both will hit home on the second condition as well. That doesn’t mean that Judge Bridges will accept their arguments, though. The ability to communicate may be more important than the message (and statisticians are not known for their folksy explanations :-) )
”The point here is that even a slight error in estimating the felon vote probability causes a dramatic change in the number of illegal votes needed to demonstrate the GOP’s hypothesis at any given confidence level. I don’t think the judge is going to rely on something so shaky to overturn an election.”
I fully agree. Perhaps when (if) I get the precinct-level data, I’ll run the Monte Carlo analyses again using different assumptions about ex-felon voting preferences. That would nicely illustrate your point (small changes in input lead to large changes in outcome) with the actual data.
Chew2 @ 50
”This is called ecological inference (see King et.al.), but is subject to all sorts of pitfalls and criticisms. I asked dj about this and he was quite skeptical.”
Yes, although I was skeptical because there is no ground truth on how ex-felons actually vote. Simply matching by several demographic characteristics isn’t very sound with out at least a little ground truth.
you folk for got to take your med’s again today. we al still love you!
Ah goldy you should hope that this case goes on forever. the longer it goes on the more righties come to your site to read your legal meanderings about the Election Contest.
Me, I’ll just sit back and watch the governor do exactly what the majority of the state’s newspapers said she would do. Raise taxes and drive businesses away. She’s off to a great start so far. Can’t wait to see what she has up her sleeve next year.
Reggie… you know why you won’t find “legal meanderings about the election contest” over on (u)SP? Because they know they’ve lost.
Rossi got screwed again. The fact is, that whole election was ridiculous and no one won.
Why do Republicans whine so much? Between Rossi and that guy on the DownLow in Spokane, they’re really quite some crew aren’t they?
dj @ 55
Questions:
1. You say: “The reason why I think a more sophisticated analysis will be used is that the Republican’s expert witnesses have introduced such methods!”
What more sophisticated analysis was introduced by Katz or Gill in their papers?
I didn’t recall any. Katz posited both a legal voter propensity to vote for Gregoire, and an illegal voter propensity, but then said there wasn’t enough data to estimate the illegal voter propensity, so he said he would assume they were the same. I took this as a non-explicit nod towards ecological inference models. But he simply estimated the legal voter propensity based on the precinct proportion voting for Gregoire. The only statistical concept that Katz explicitly used was the confidence interval for his estimates.
2. You say: “there is no **ground truth*** on how ex-felons actually vote.”
My understanding is that there is never any actual voting data (ground truth), as opposed to opinion polling data, on how felons, black people, woman, or men actually vote. So estimates of how blacks, felons, and woman vote are made using ecological inference techniques on the raw voting data. E.g. in voting rights cases, experts estimate the proportion of blacks voting for a candidate from the raw precinct voting results and the racial composition of the precinct or precinct voters. They don’t actually know how black people vote, since there is no such data. Nevertheless, these studies are accepted by the courts in certain circumstances. The issue in voting rights cases is whether there races tend to vote as a block, so the precision required in the estimates is much much less than in the present case where we are trying to estimate the acutal number of votes.
chew @ 61
you say there isn’t any ground truth on population subcategories, which I think is a real stretch. If you mean a strict accounting of ballots and voter characteristics, of course the answer is no–ballots are secret. But volumes have been written validating the voting habits of blacks, women, etc. Exit polls, despite their knock in 2004, are highly accurate indicators of voting patterns–precisely why there was such buzz around their error last time; that doesn’t happen.
But there isn’t even any exit data for felons. You don’t ask someone at the polls, “How did you vote, and are you a convicted felon?” The imputation of votes by (eg) racial category makes reasonable scientific sense, because we have lots of indicators on how major racial category patterns break out. Not so for felons–we don’t have the first clue, often for the simple reason that in most cases we don’t even let them vote in the first place.
Speaking of experts, I should be asking my friend Dr. Keeter at Pew about this. He’s a well known voting behavior expert. If I can get him to discuss some things on record, I’ll bring the info back here.
TJ @ 61
While we have polling data on voter preferences based on race, gender, etc. none of that polling information is used when ecological inference estimates are made, only the raw voting data. And the polling data is not used to validate the ecological estimates as far as I can tell. In fact in some of the literature I’ve glanced at, opinion polling data regarding voting behavior is claimed to be unreliable as a guide to how persons actually voted. Such ecological estimates are published in peer reviewed journals, and in some instances accepted by the courts as evidence.
I grant you that there is no polling data regarding felon voter preferences. So the question is how do you estimate how they would tend to vote, or in this case how they actually voted. To what extent are ecological estimates feasible and acceptable, especially as compared with the proportional reduction alternative.
I wonder why someone doesn’t do an opinion poll from a random sample of the identified felons? Perhaps because they don’t know what answer they will get. -)
well, at the least we can say that insofar as there exist reasonable data on the voting habits of certain subpopulations, there do NOT exist any similar info on felons. And that’s the true point–any statistical analysis that purports to determine how felons might have voted in WA is working from almost zero data.
The Republican theme song should be Whine On, Harvest Moon
Chew2 and TJ
By “ground truth,” I was refering to any kind of information from Felons that would allow us to evaluate the question: “do felons vote like the people around them?”
This could be an exit poll that asks people if they have been ever been convicted of a felony (I don’t want the be the person doing this poll, however!).
Better yet, one could do a study like the one Chew2 describes, where ex-felons are surveyed and asked their voter preference. One could then use sex, ethnicity/race, education and other covariates and statistically infer the pattern of ex-felon voting in WA.
Chew2@61
You say: “The reason why I think a more sophisticated analysis will be used is that the Republican’s expert witnesses have introduced such methods!” What more sophisticated analysis was introduced by Katz or Gill in their papers?”
Professor Katz’s depositions introduced a statistical method for doing proportional ballot reduction. He makes the argument, using the precinct-level data with the revised Republican illegal vote list, that a Gregoire win is outside a 95% confidence interval. Specifically, he uses a binomial model for finding the expected number (and variance in number) of ballots to remove from each side, and then uses standard asymptotic theory to produce a 95% confidence interval.
That method is more sophisticated than the “simple” method of ignoring the variance in the outcome and going with the average outcome. In fact, a statistician would not think of the outcome of Katz’s analysis as a single number, rather a statistician views the final results as a whole distribution of outcomes. The Monte Carlo method produces exactly that—a distribution of final election outcomes. So do approximate methods like what Professor Katz used, but non-statisticians sometimes tend to take the average result as “the” answer. It is not. The average is one outcome with a particular probability.
The Republican theme song should be Whine On, Harvest Moon -Comment by C— 5/11/05 @ 10:17 am
Ho hum.
As opposed to the Democrat theme song?
PS, A great BIG thanks to Stefan Sharkansky at Sound Politics for his preview button where I regularly check my work.
dj @ 67
1. Thanks for the clarification regarding your reference to more “sophisticated methods”. That was my understanding as well rearding Katz’s study. He simply introduced the probalistic notion of a confidence interval, over a naive use of the proportions analysis.
What I was thinking about re “sophisticated” were more explicit ecological inference estimation techiques, like those espoused by King or Goodman.
2. As I stated previously, my understanding is that when ecological inferences regarding black voting blocs are done, there is no other demographic or polling data used and no assumptions regarding how black folks tend to vote.
“An ecological inference problem is one like this. In a certain election, we know how many votes were cast for the Democratic and the Republican candidates, and we know how many voters were men and how many were women. Can we say anything reasonable about whether men are more likely to vote Republican than are women? More generally and abstractly, given aggregates (number of Republican voters, number of men), we want to know things about individuals, or at least sub-aggregates (if not how Juanita Q. Public voted, then the proportion of women voting Republican). In particular, as King emphasizes in this book, in the USA questions about whether people belonging to different races have different political propensities are extremely important in implementing the Voting Rights Act and its sequelae, but, owing to the secret ballot, we have no direct knowledge of how people of different races vote.”
http://cscs.umich.edu/~crshali.....inference/
See e.g. King (an example using raw voting data only)
http://gking.harvard.edu/eicam.....0000000000
or King (range of problems studied)
http://gking.harvard.edu/eicam.....0000000000
Democrats
“The Party of No”
[Voice Over and Text on screen]
Wanna “no” something? Just ask a Democrat and see how well they “no” the presidents agenda.
[Cuts of Democrats saying “No”]
“Sad to say, the Democrats are becoming the party of no.” (Gloria Borger, Op-Ed, U.S. News & World Report, 2/28/05)
“So far this year the Democrats have been holding fast on a lot of issues, they have been really obstructionist and effectively obstructionist on the courts, on social security, not giving an inch.” (MSNBC’s “Hardball,” 5/2/05)
“Why should we put a plan out? Our plan is to stop him from – Stop him – He must be stopped.” (Congresswoman Nancy Pelosi (D-CA), Fox News 3/17/05)
[Cuts of Democrats saying “No”]
“I can’t believe the Democrats have painted themselves into this corner when it’s so transparently obvious that they are obstructionists.” (Hugh Hewitt, “Kudlow & Company,” 4/29/05)
“Voters are looking for reform, change and new ideas but Democrats seem stuck in concrete.” (Stan Greenberg & James Carville, The Washington Post, 3/8/05)
“It may take 60 days, it may take six months, it may take six years. But it doesn’t matter; We’re not going to back down.” (Sen. Harry Reid (D-NV) – Social Security Rally, 4/26/05)
[Cuts of Democrats saying “No”]
[Voice Over and Text on screen]
The Democrats think they “no” everything. Well… They do.
[Text: No to Qualified Judges. No to Preserving Social Security. No to the Agenda the American People Voted For.]
PS, A great BIG thanks to Stefan Sharkansky at Sound Politics for his preview button where I regularly check my work.