Some of the region's dopiest commentators are having a hard time grasping the concept of proportional analysis of invalid votes in the governor's race (e.g. Ken Schram, the P-I editorial board, the usual suspects at the lunatic fringe blogs).
But just because these dim bulbs can't shed any light on it doesn't mean that it's not a reasonable approach. The idea is simple, common sense and is used to resolve election irregularities all over the country.
Because of the inviolabilty of ballot secrecy, there's no way to know how any particular individual voted. The only legally valid information we have is (1) how the total number of ballots in a particular pool (precinct, county, legislative district, etc.) went to each candidate, (2) how many of the ballots in the pool are deemed invalid (whether because of a felon vote, or ballot stuffing, etc.). You then know the probability that an invalid vote was for a particular candidate. E.g. if a precinct with 100 ballots went 56 Gregoire, 39 Rossi and 5 for other candidates, you know that probability that an invalid vote went to Gregoire is 0.56, and for Rossi 0.39. If 3 of the votes in the precinct were ruled invalid, you would deduct 1.68 votes from Gregoire's total and 1.17 from Rossi's total. Repeat for every invalid vote and the district of finest granularity in which you know it was cast.
That is similar to, but more accurate than, what is used in, say, Wisconsin to adjust polling places in the case of ballot stuffing:
state law lays out steps to follow if there are more ballots counted than voters listed. First, poll workers are to look at each ballot in the machine and determine if any are blank, or lack the initial of the poll workers who gave it out.Of course, it's not a perfect solution, but it's not nearly as bad as deciding an election by hundreds of unexplainable mystery ballots that are known to election workers before certification but not to the public until after certification. Posted by Stefan Sharkansky at May 04, 2005 11:59 AM | Email ThisThese are to be removed.
If there is still a discrepancy, state law calls for workers to randomly remove ballots from the machine until the numbers match. These ballots are set aside and not counted unless a recount is requested.
Unfortunately, I've wasted some of my time trying to convince the lunatics, but you can't get these people to retreat from their position because they know the the slightest chink in their armor of denial is an opening for honest questioning of what happened in November 2004.
Rationally, it's pretty obvious that the court case is very valid and that there's going to be some harvesting of the liberal heads that are currently buried in the ground in the garden of eschew.
Posted by: Jeff B. on May 4, 2005 12:14 PMTo have pushed this matter to the point of "proportional deduction" shows the depths to which state politics has really sunk.All I could think was that he apparently does not have a grasp of what the law requires in the case of an election contest. I actually agree that this proportional analysis is a pretty ridiculous low to stoop to, but this is unfortunately the best thing that can be done given the requirements of the current laws.
These lunatics have convinced themselves that someone can stuff a ballotbox with illegal votes. Those clearly illegal votes are now comingled with legal votes. And there is absolutely no relief. The illegal votes must now be counted. Mind-boggling.
But once you realize what a lunatic Goldstein is, it's totally understandable. His LEFTIST excuse is him mom didn't breast-feed him long enough!
So Stefan...I guess it's wise we don't mention Goldstein's name too much. Although, I think if folks spent a minute digesting Goldstein's BS, they would see for themselves what a lunatic Goldstein is.
But, it's your blog Stefan. So I'll respect your decision not to specifically ID Goldstein as the lunatic blogger.
Posted by: Mr. Cynical on May 4, 2005 12:33 PMThey are comfortable with the level of fraud because it doesn't rise to the level that shames or damages them (yet). It's also acceptable because they are (currently) the beneficiaries of that fraud.
The lesson to be learned here is the illustration of the depths to which democrats will go in order to acquire and retain power.
All in all, it has been a fascinating glimpse into the naked application of power.
Looking forward to the 23rd!
Posted by: alphabet soup on May 4, 2005 12:36 PMIt’s obvious that the dems have absolutely no desire for an accurate reading of the law or the election, as those almost certainly would negate CG’s fraudulent win. The “never give an inch/house of cards built upon the sand” crowd has winning as the priority, not accuracy.
What is problematic are the “commentators/pundits” who enable the miscreants…Schram, et al…and those for whom “why all this fuss about a stupid election?” belies their shallowness of thought.
Obvious, but it bears repeating often enough that the clueless get it: this is about thousands of jobs and billions of dollars that come directly out of the wallets of regular people…not to mention major erosion of personal liberties.
I wouldn’t be the first person to draw out significant parallels between CG/WA state legislature of ‘05 and “1984”…and appropriately so. Or, alternately, the proverbial frog in the slowly simmering/boiling water.
"That cautious approach is heartening. Elections are to be decided by how many votes are cast for whom, not speculation on how many votes might have been cast for one or both candidates"
It is not "how many votes" it is how many legal votes. Votes with no voter are not legal. Provisional votes cast by those ineligible to vote are not legal. Votes castby people submitting multiple votes are not legal. Votes cast by felons who have not been reenfranchised are not legal. Illegal votes should not, in any way, determine the outcome of an election.
Because of our secret ballot laws, the next argument is specious:
"Guessing and speculation are hardly the preferred province of the courts. Demonstrable facts are supposed to hold sway under the law. Who voted illegally? For whom did they vote? Is the ballot no longer secret? Will felons be forced into self-incrimination to prove which candidate they voted for?"
It is never possible to determine, directly, for whom a vote is cast. The only avenue available to us is to determine for whom a vote was likely cast. This is the realm of statistics and probability. Because both these things require a knowledge of mathematics, they are probably beyond the grasp of the PU editorial board. Proportional analysis is one way of determining who a vote likely went to.
Finally, if the difference in vote totals between two candidates is less than the number of errors in the tabulation of the vote, then the election is clearly invalid, so the PU finally gets it right in their last sentence.
"Unless an election is clearly invalid, when the people have spoken, their verdict should not be disturbed by the courts."
Way to go, guys, your intellect is astonishing.
You are correct. It is flawed. I would prefer to toss out an election whenever the number of illegal votes and irreconcilable errors is greater than the margin of apparent victory. Unfortunately, that's not allowed in this case.
Posted by: Stefan Sharkansky on May 4, 2005 01:06 PMWhy? Because by adding illegal ballots to the pool, they raised the percentage with "D" on them, giving themselves an advantage when the proportional reduction is applied.
If by some magic method all those fake votes could just be removed from the pool, the drop in the "D" count would be even bigger than by the proportional method.
Posted by: Insufficiently Sensitive on May 4, 2005 01:08 PMInsufficiently Sensitive has nailed it. This analysis assumes that the illegally counted votes are random; this is a generous assumption (to the cheaters), and I don't buy it.
Posted by: Bostonian on May 4, 2005 01:30 PMGuess the republicans really like wasting money. Funny that they claim the government does that so much...looks like the government could take lessons from the republicans and really waste money!
Posted by: Peter Franklin on May 4, 2005 01:35 PMYou know no such thing! Only if the errors are random is it even a safe guess. If there is any bias to the error, it is a very poor, even harmful estimate.
As I pointed out in other comments, if you stuff ballots for Candidate A in Precincts that lean toward Candidate B, then you can make proportional "relief" work in your favor.
"Guessing and speculation are hardly the preferred province of the courts...." Absolutely right. So don't guess. What we know is that the number of error greatly exceeds the margin of victory. By any objective measure, that should be enough.
Posted by: StephenR from Houston on May 4, 2005 01:44 PMWhat is at stake here is suspicious and missing votes. More Ballots than voters in some precincts. Fewer ballots than voters in other precincts.
To presume they are "errors" is as wrong as to presume they are fraudulant.
Posted by: StephenR from Houston on May 4, 2005 01:57 PMThere is no problem with proportional analysis when you've got a "party that stuffed the ballot boxes" in the traditional sense. You know which way that criminal wanted the election to go, and you can assume (and a court would find) that the "stuffed" ballots were for their own candidate.
Here, you've got a situation where the allegation is that the system and/or inept and/or reckless elections officials improperly permitted felons, etc. to register to vote. These are at most a few thousand individuals, scattered throughout every corner of the state. No one really knows their voting patterns -- no party directs energy toward them. (For evidence of this, go no further than what Slade Gorton said immediately after the loss to Cantwell. He said that Libertarians -- who might've cost him the election by not voting R -- were too small and amorphous a group to figure out how to target in a campaign.)
Where you need proportional analysis, generally, is a situation like this. But it has to be more than just guesswork. I think none of us really knows the voting patterns of the particular group of felons who voted improperly in this election. They're smaller than the Libertarians, so nobody would bother studying them even IF they had the right to vote -- which, of course, is another obvious reason that no one would study them. Does anyone doubt, for example, that the voting patters of ex-felons in Montana might be entirely different than those of ex-felons in South Florida? And you could easily have voting patterns turn on the particulars of a given election. We know that, as a general rule, other things being equal women candidates do better among female voters, and somewhat less well among men. Can we say for sure that pattern wouldn't apply in this election, among a largely male felon population? I don't know.
And of course the same uncertainties apply to precinct or county by county analysis. Would we generally expect a poor felon living at the bottom of the socioeconomic scale to vote like the rest of the community? How about a felon in a heavily religious community?
I think that the proportional analysis model may become mired in guesswork, and so found unacceptable by the court. If your election statute doesn't provide for a "too close to call" revote, and you have to prove that net illegal votes would've tipped the election, proportional analysis may be just too undertain to yield anything better than "on the one hand....on the other hand....."
But I'm not just here to spout uncertainties (important though that may be). I'm here with an idea. I think the kind of study that's needed to tell us what illegally-voting felons would likely have done in this state in this election is what *legally*-voting felons did. A lot of those have been identified. They could participate in a survey of how they voted without being criminally liable.
My personal belief is that proportional analysis based on geography alone is too uncertain to give any real indication of voting patterns in this election among ex-con illegal voters. I'm trying to be objective on this, even though I am a Dem. I'm open to being convinced, though -- we'll see what the "experts" say in court.
But I'd imagine fewer objections to tying felons' vote patterns to those of other felons. If the election were overturned on the basis of that kind of analysis, I'd have no problem at all.
Oh and for the record, if we had a statute that called for a new election based on illegal votes exceeding the margin of victory, we'd be warming up the voting machines right now and I'd completely support that. But we don't have that -- we have an election that was very close, where *maybe* (but who knows) things could've been different w/o these various illegal votes. My view is, unless you've got a right under the law to change it, those are the breaks. I am sure that Rossi would treat this exactly the same way if he were on the other side -- after all, remember his reaction to having a 42 vote lead?
Posted by: Bluebeard on May 4, 2005 01:58 PMTo them, rules are for fools.
Posted by: Danny on May 4, 2005 02:07 PMThere is indeed a problem with proportional analysis in a stuffing case.
At BEST (say a Precinct that leans 70/30 for Candidate A) proportional anaysis corrects 10 stuffed ballots for Candidate A into 4 net stuffed ballots.
At worst, 10 stuffed Ballots for Candidate A in a 70/30 precinct that goes for Candidate B, proportional analysis will "correct" those 10 stuffed ballots into a net 14 vote gain for Candidate A.
MAYBE proportional analysis is applicable for felon votes. Bluebeard's points here are correct. The felon's votes could correlate with lots of things, not just precinct. But his proposed study is ill advised on two points:
1. It is fundamentally wrong to violate the secrecy of the ballot in any state proceeding.
2. You would have to assume that felons that legally vote (have done the right thing to have their rights restored) have the same voting preferences as felons that voted illegally.
Posted by: StephenR from Houston on May 4, 2005 02:31 PMIf the number of phony ballots from 70%-Democrat areas exceeds fourfold the winning candidate's margin of victory, it becomes unreasonable to assert with any confidence whatsoever that the Democrat won legitimately. If the number of fishy ballots were smaller, it may be hard to know whether the election results should be voided (e.g. if the margin of victory is 1 and there are two fishy ballots statewide, should that justify throwing out the election?) but here it seems pretty clear that the results are not legitimate.
One thing I would like to see would be a rule that counties would be assessed the costs of a revote based upon the proportion of the total number of fishy ballots found therein. So if there are 1800 fishy ballots found in King County, and 200 found throughout the rest of the state, King County would foot 90% of the cost for a statewide revote.
Adding a rule like that would probably do a lot to clean up elections, since people would be likely to force out election officials whose incompetance hits them in the wallet.
Posted by: supercat on May 4, 2005 03:08 PMI wasn't clear. As I said in my post, in a traditional ballot-stuffing case, when the issue comes to light, you know which side the criminal was on. Then you don't do proportional analysis at all. You just assume that all of the ballots stuffed into the box were for the criminal's preferred candidate. No complicated mathematics. If you read my post, that's what I said. I said that there was no "problem" with proportional analysis in that case becase that type of analysis wouldn't be used.
I admit that the way I said it could've been clearer!
On your point about the survey idea not working because of the need to preserve secrecy of the ballot, I think it could. Responses could be kept confidential.....
Posted by: Bluebeard on May 4, 2005 04:19 PMThe second reason this theory is incorrect is that for it to have occurred, it would have needed to be organized in advance or on that day. The fact of the matter was that no one knew that the election would be as close as it turned out. Therefore, they wouldn't have tried such a shenanigan. No, the problems that occurred were due to the incompetence of election staff irrespective of any organized effort for one candidate or another.
Now, to the proportional analysis. The theory is invalid since we do not know how felons vote as a group. You are trying to tell me that male, low to middle income felons would vote for a female attorney general. What are you smoking?
Rossi claims that there are enough inaccurracies to invalidate the vote count. Fine. This means he did not win any of the counts. He can't claim CG didn't win the third count due to the inaccurracies and still claim he won the first two counts. Talk about hypocrisy. I am tired of people not looking at this in rationale minds. I think Ken Schramm is correct. Remember, he did vote for Rossi.
Posted by: tc on May 4, 2005 05:20 PMI beg to differ. There are over 5000 pieces of evidence of ballot stuffing. Every ballot without a voter is evidence. Circumstantial, but it is evidence. There are audit checks to make sure that the ballots are all correctly accounted for. When they cannot account for the ballots, stuffing is a legitimate reason.
Because stuffing is so hard to prove, you have to rely on the audit trail. What Shark found out is that audit trail is worthless.
5000 Ballots without Voters.
Another 1600+ Voters without Ballots.
The Audit Trail that KC elections BY LAW must keep is deamed worthless. Even the judge said the database can't be used to tell who voted.
These are facts, not statistical guesswork.
Its time to get back to looking at the forest.
Posted by: StephenR from Houston on May 4, 2005 07:04 PMWe have learned that KC could print ballots on demand.
We learned that KC change interpretations of voter intent during the manual recount.
We know that on 9 separate occations "lost" and "misplaced" ballots were "found" where and when they should not have been.
Some of the stuffing happened on election day because the poll books don't balance.
God, literally, only knows the stuffing that happened checking signatures with the absentees.
Remember, the manual recount was finished weeks after election day. They had plenty of time organize.
Far too much effort is going into felon voting. Maybe because the judge has blocked too many other avenues (why? I don't know.) and the lawyers think the felon vote will be enough.
Maybe they can win in court with the felon strategy. For the Reps to win big they have to play all the cards, if not in court, then in public.
Posted by: StephenR from Houston on May 4, 2005 07:26 PMI know...let's have a new election since the last one was so f***** up by King County. Yeah, that's the ticket.
Posted by: Danny on May 4, 2005 08:20 PMYes alphabet -
A power derived by fraud and corruption!
Your analysis is excellent. I have a few thoughts to add:
1) Your idea to poll felons who voted legally is fine, but conceivably we could poll felons who voted illegally instead. They would have to be asked who they favored rather than who they voted for, so they wouldn't worry about incriminating themselves.
2) If we polled felons who voted (legally or illegally), we would risk their gaming the system by saying the opposite of who they actually voted for, knowing that this would help their candidate. I'm not sure how many respondents would understand this and then actually do it, but it's a possibility.
3) Something like you suggested is indeed the best way to estimate the proper election result. That's how we'd solve a problem like this in the private sector.
However, courts tend to shy away from using inexact data (e.g., poll results) even if the effect is to correct a bigger error (illegal votes). An analogy is Bush v. Gore, where the Supreme Court was unwilling to accept minor discrepancies in how counties interpreted voters' intention, even though a recount would have compensated for much greater discrepancies in the voting technology used from county to county.
But one could argue that, even if the court shouldn't use polling data to decide an election, it's OK to use polling data to conclude that an election is tainted enough to require a re-vote.
4) I think both parties would be wise to conduct proportional analysis of the illegal votes as many ways as possible -- using precinct data; using existing poll data on vote distribution by geography, age, gender, income/education; and polling felons. Then they should tell the judge the results that are most favorable to their side. The judge would then have a choice to rule on the evidence presented or, conceivably, take the extraordinary step of ordering additional research (such as polling felons).
5) I think the court is between a rock and a hard place. It has correctly concluded that it is bound by law to accept proportional analysis. But that requires using either bad analysis (projecting based on precinct voting patterns only) or bad data (imprecise polls). What a dilemma.
Posted by: Bruce on May 4, 2005 10:49 PMThe RCW does not state that "an election challenge must be able to identify for which candidate specific illegal votes were cast". Rather, it states that the challenge must show who it "appears" they were cast for. Beats me what "appears" means, but it's a lower threshhold than certainty. And since the presumption is that a law is implementable, it must be interpreted as something that can be shown under our system of secret ballots. Which leads inevitably to statistical analysis...
Posted by: Bruce on May 4, 2005 11:10 PMSecond, if you followed the process for the recounts, you would understand that yes ballots were remarked, but the original was retained and again you had observers from each party overlooking the process. Additionally, questionable ballots were sent to the county canvassing board, which includes a Republican.
The reason for observers being in the process is to catch these types of errors. Are you stating that the observers did not do their job? If the observers aren't doing their job, then we will never have a fair and accurrate election no matter the rule changes. It wouldn't matter the technology or rules, if the observers are in on the scam, the scam will succeed. Please explain how the observers failed at their job. I have not heard any evidence of this.
Posted by: tc on May 5, 2005 05:58 AMTo become visible: a plane appearing in the sky.
To come into existence: New strains of viruses appear periodically.
To seem or look to be: appeared unhappy. See Synonyms at seem.
To seem likely: They will be late, as it appears.
To come before the public: has appeared in two plays; appears on the nightly news.
Law. To present oneself formally before a court as defendant, plaintiff, or counsel.
Let's say (for the sake of argument) that the court were to conclude that there's a 1% chance that illegal ballots, etc. threw the election to Gregoire, and a 99% chance that they did not. In the eyes of the "appears" advocates on this thread, that would mean that it "appears" that Rossi wuz robbed. But to say that ignores the fact that there's a much greater chance that it "appears" that he wasn't robbed.
In other words, the use of the word "appears" is really just a substitute for proof that it as least more likely than not that Rossi wuz robbed. Obviously, after extensive briefing by the parties (which may or may not have addressed this issue -- my guess is the Republicans wouldn't even try this argument as a "throw-away") Judge Bridges does not believe that he can invalidate the election on some unspecified "chance" that illegal votes swayed the outcome. To the contrary, he's talked repeatedly about the "high standard" to be met -- and from what he's quoted from prior court decisions, that seems to be the long-accepted rule.
To put it another way, using the example from one of the posts above, no one would say "It appears that I'm going to be late," if they thought it was actually more likely that they would NOT be late. To say "it appears that I MAY be late" is different -- but that's not what the law says. It doesn't say that you have to prove that it appears that the election "may" be invalid, it has to "appear" that it was actually invalid. That can't be the case if it "appears" to be even more likely that it was valid.
Posted by: Bluebeard on May 5, 2005 09:44 AMVery good point. First, only Shark amongst us has read the poll books to know what was objected to. Second, we have had reports on this blog of some objections from observers, but that is heresay. Third, object to what? Object that the person is not who they say they are when the voter does not have to offer proof (photo ID). Object to someone running a dozen ballots though the machine in the course of 2 minutes? Ok, I'll admit nothing that blatent happened. But ballots were run through the machines that should not have. Dispite the poll watcher, the number of error/mistakes/fraudulant votes coming from the polls is astonishing.
What about observers of the absentee counting and signature verification? None. (correct me if I am wrong). And how many ballots have been found in just the LAST MONTH still inside the second envelope? 95 at last count. Disproportionately from Rossi precincts. How convenient! Whatever observation there was was hardly enough.
if you followed the process for the recounts, you would understand that yes ballots were remarked, but the original was retained and again you had observers from each party overlooking the process.
It is you that have not been following the process. Some (tens of thousands!) ballots were "enhanced", no duplicate created, the original marks - if any - cannot be retrieved
During the recount process there were observers present, true. But, HOW FAR AWAY? There are reports in the January logs that some observers were kept 25 feet away from some of the tables. Heresay? From me it is. But it is verifiable.
Additionally, questionable ballots were sent to the county canvassing board, which includes a Republican.
... and TWO DEMOCRATS with majority rules. Really fair. Really objective. There should be records of those canvasing meetings where we can see and hear the objectins. (Are some of the audio tapes that some of the posters here are transcribing about these meetings?) You will notice that there is no photographic/digital scanning evidence of any of these questionable ballots. This the 21st century, people! There is affordable technology to digitally capture for later review all this stuff. It wasn't done. Could it be KC did not want it reviewable?
are you stating that the Democrats added felon votes after the fact.
No. I never said any such thing related to felon votes. My comment was in relation to general stuffing. You said they had to had it organized prior to or on election day. I replied that the manual recount process took weeks. Some stuffing at the polls, some stuffing/loosing in the absentee process, some stuffing in the enhancement, some stuffing in the voter intent canvas. Look at the table Sharkanski created on Dec. 24 for changes in the vote totals at each of the three recounts. There is only a 0.02% (1 in 5000!) chance that the vote changes during the manual recount in KC could happen by random chance. And that is according to the certified tallies in December.
Other than remarking of ballots, which hasn't been raised by the Rossi campaign, there haven't been issued raised on the recounts. The Rossi campaign has not fought the inclusion of the KC ballots that were not fully canvassed, nor any other issue specific to the third recount. Why?
Check your facts. The Rossi campaign screamed bloody murder at several places along the judicial trail. I don't have the time to research them all, but I've been an avid read of this blog since election day. They did not win, but they raised the issue. I hope they have not given up on these points.
Posted by: StephenR from Houston on May 5, 2005 11:12 AM