With a hand recount looming in our historically close gubernatorial election, there has been much debate over the relative accuracy of hand counts versus machine counts, and the error rate of vote counting technologies in general… most of it uninformed.
In my typically wonkish fashion I decided to dive into the most technical research I could find, and tediously share my gleanings with you. My primary source is the CalTech/MIT Voter Technology Project, and much of my data is drawn from the following reports:
Conflicting numbers regarding the error rate of voting machines have been tossed about in partisan blogs, the news media, and the John Carlson Show (which doesn’t really fall into either category.) On the one hand, the certification standard for voting machines in Washington state is indeed one failure in one million. But it is also true that the residual vote rate — the primary statistical measure of the performance and accuracy of voting technologies — is one to two votes in one hundred.
The residual vote rate is the difference between total ballots cast, and votes counted for a particular office, such as president or governor; these residual votes represent the “over” and “under” votes we hear about. And it is this measure that is significant in comparing competing technologies.
The fact that the counting machines themselves are virtually flawless is meaningless in the context of the larger discussion, because they are only flawless in counting flawlessly prepared ballots. For example, both optical scan and punch card machines will test to the same high certification standards, yet the average residual vote rate for punch card ballot systems is nearly double that for optical scan.
The large performance difference between the two systems is due, not to mechanical failure, but to a higher rate of “human error” by those voters using the punch card ballot system versus those using optical scan. This is a reflection of the way people relate to the particular technology… a concept that should be well familiar to user interface designers, or anyone who has ever manned the technical support line at a software company.
Indeed, Direct Recording Electronic devices (DREs) — including the touch-screen voting machines to which many counties are switching — have amongst the highest residual vote rates of all existing technologies, despite the fact they are programmed to make over-votes (voting for multiple candidates in the same race) impossible, and always tabulate ballots with 100% accuracy. (Or so Diebold tells us.) Meanwhile, hand-counted paper ballots, which provide no safeguard against over-votes, and risk introducing human error into the counting process, have one of the lowest residual vote rates.
Thus it is not the counting machine that introduces statistically significant error into the voting system, but rather the interface by which voters are asked to mark the ballot. It has become popular for bloggers and columnists to criticize voters for not following instructions, but when one voting technology produces error rates twice that of another, the technology deserves part of the blame. And clearly, voters in counties using punch card systems are being disenfranchised at rates nearly twice that of those in counties using optical scan… and at a rate well outside the margin of error in our gubernatorial election. (I’ll get to that in a moment.)
As to the relative accuracy of the various voting technologies, the CalTech/MIT studies found that voting systems fell into two clusters: paper ballots, lever machines and optical scanned ballots produced residual voting rates of one to two percent. Punch card and electronic voting methods produced rates of approximately three percent.
So, are hand counts accurate? According to the studies, hand counts are at least as accurate as lever machines and optical scanners, and significantly more accurate than punch card and electronic voting systems. In fact, when looking at counties that switched from one technology to another and comparing the resulting residual vote rates, the study found that while the overall results were consistent with other analyses…
Paper might even be an improvement over lever machines and scanners.
CalTech/MIT also explores a second measure of accuracy, tabulation validation rate — the agreement between initial counts and recounts of ballots in contested elections. This metric is less useful as a comparative tool, because it cannot be used to measure mechanical (lever) and electronic voting machines, as there are no ballots to recount. And the study in question only compared hand-counted paper ballots with optically scanned ballots.
The study found the tabulation invalidation rate was .83 percent for paper and .56 percent for optical scanning. Thus the discrepancy between the initial count and the recount was less for optical scan than for hand-counted paper.
To be honest, I’m not sure what, if anything, this says about the relative accuracy of hand recounting optical scan ballots, let alone punch cards. And here’s the part that might piss some people off… I’m not sure it even matters.
For what this study does tell us is that even the most accurate voting technology still is not accurate enough:
Considering these tabulation errors, how confident should we be in vote counts, and when should we have a recount? The tabulation invalidation rate was low, especially for optical scanning. However, it was not trivial. In a US House election with 250,000 votes, the invalidation rate of .005 for scanners amounts to 1250 votes. The tabulation errors may swing toward any of the contestants in a recount. Assuming a uniform distribution of tabulation errors, any race decided by less than .5 percent of the vote will have a non-trivial probability of being reversed in a recount.
A .5 percent invalidation rate in a gubernatorial election with over 2.8 million votes cast amounts to 14,000 erroneous votes! With only 42 votes separating the two candidates, no counting method can accurately tell us who really got the most votes.
Republicans scoff at Gregoire calling this election a tie, but statistically speaking, it is. This election is so far within the margin of error, that there is no practical way to accurately determine the winner.
Thus, the results of the third count — whether hand or machine — will be just as meaningless as the results of the first two.
Republicans argue that “winning” the first two counts gives Rossi legitimacy. It doesn’t. It’s like flipping a coin and having it land on heads two times in a row.
And Democrats argue that a hand recount will more accurately determine the winner. It can’t. This race is simply too close to call. We’ll never know who really got the most votes.
Fortunately, the law does prescribe an endgame. Gregoire will request a hand recount, and whoever “wins” that, will be governor.
And who do I think is gonna win? Flip a coin.
[Linking here from (un)Sound Politics? You can comment directly on Stefan’s critique, or join a discussion here: “Lies, damn lies, and statistics.”]
Let the rants commence, but please, read the CalTech/MIT studies before you start attacking them. I did my best to summarize the pertinent results, and if you disagree with my analysis, fine. But these are two of the finest technical schools in the world, and I’m not going to entertain charges that the studies are biased or shoddy unless you produce equally well researched studies that refute them.
Well, I’ll start off with a rave. Having read those same studies, I can vouch for the accuracy of your summation; it’s certainly a positive contribution to the discourse. Nice job, Goldy.
Nice study, great amount of work on your part you are to be commended for your study…
This is an interesting read as well http://hl-mungo.dailykos.com/s...../03617/399
Goldy–You seem to have ignored my request for comment on the Dems action in Montana…trying to say voter intent not clear when name are clearly X’d or Squiggled out and 2 arrows connected. It shows how hypocritical the Dems are…clearly. You are supposedly the Progressive voice of reason.
What say you???
Very good article. A lot of work put in by you and a job well done.
Nancy Reagan used a medium and tarot cards to run the country. Ronnie Jr. lives in Seatle, has he been consulted?
Read the studies yesterday, and gee whiz, those guys sure spent a lot of time telling us that when humans are involved, it will be less than perfect. I should add it is even the collective us that builds the slightly imperfect machines, and that before the flawed ballots created by the not perfect voters are fed into them. If they were printed correctly to begin with, on the right weight paper, and correct ink.
I flipped the coin to play along, then I flipped it agin and again. Three times the same result. I may place a large bet on this toss.
Thanks, G.
“The residual vote rate is the difference between total ballots cast, and votes counted for a particular office, such as president or governor; these residual votes represent the “over” and “under” votes we hear about. And it is this measure that is significant in comparing competing technologies.”
Goldy- this is where I believe you start going wrong in your discussion. You make a valid point that people interact differently with different technologies, and that is largely due to familiarity (when else in life so we EVER fill out punchcards?)…
BUT, over and under votes are NOT errors, at least not at the counting end. People choose to not vote in a race, or to vote twice (they might not realize they are doing so- sloppiness, etc., but it is the human doing so). A higher “residual error rate” is NOT machine error, but a different level of human error in the process.
NOTHING that can be done at the point of tabulation can correct that human error in voting (or not voting).
Then we get to “The study found the tabulation invalidation rate was .83 percent for paper and .56 percent for optical scanning. Thus the discrepancy between the initial count and the recount was less for optical scan than for hand-counted paper.”
So in going to a hand recount, we go for a HIGHER tabulation error rate- a fifty percent higher rate (.56% vs .83%)- for the bulk of the votes that will be handcounted.
Which all comes down to- this has NEVER been about counting all the votes, vote accuracy, or any of the other bullshit advanced by Berendt or Gregoire. It is about making a high-stakes wager on a throw of the dice, and going to a less accurate method of tabulation.
It is her right- but that doesn’t make it right, and the fact that the Dems can’t be honest about their reasons is proof enough to me that they are at least able to be ashamed of what they are doing even as they proceed to do it.
Goldy- Leg in town, awfully busy- BUT another thought- maybe someone can look up an answer for all of us genuinely curious about things-
Is Yakima County the joker in the deck? Yakima had a perfect- absolutely perfect- recount. Not a vote different.
Of course, Yakima county just asked the electronic machine to spit out the results again (like the good ol’ lever machine days- a recount was opening the backs of the machines, jotting down the numbers, and adding them up again- no way to actually COUNT the votes).
Now, Yakima’s electronic machines will print a paper ballot for each ballot cast electronically, and those will be handcounted…
Will we see a difference?
Your data is superb. But I’d also like to throw in a “human factors” issue, that, in life, all of us always demands to recount something when there is any suspicion at all that the first or even second count might be in error.
Take a deck of cards. To check that it’s a full deck we always start with a basic count, face down, move then from our hand to the table and hope we get 52 as the number. If we don’t, we re-count them the same way. If it’s still, say, 51 (or 53), we typically then turn the deck face up, separate them by suits and count each suit, card by card, to see if each of them contain all 13 cards. If we find no missing cards, we know our initial count(s) were wrong and we have a full deck. If we find a missing (or extra) card in one of the suits we know our initial count(s) were right and it is not a “full deck.”
I know that even Republicans want to accurately count a deck of cards. Unfortunately, they never want to — proverbially — turn the deck face up and count accurately when it comes to votes in an election.
Jim… The studies address this issue. Yes, some of the residual vote is due to the intent of the voter. But since similar counties using different voting technologies in the same election will produce different residual voting rates, some of this residual is clearly due to the voting technology. As the study points out, there is a large difference (1 vote in 100) between the residual voting rates of those counties using paper, lever and optical scan, versus those using punch card or DRE.
How much of the residual is due to the technology? Well, by comparing the residual voting rate and the tabulation invalidation rate, the study suggests at least a third of the residual is due to tabulation error. For punch card and touch screen systems, that still means an error of 1 vote in 100 due to factors introduced by the technology.
The tabulation invalidation rate study does not provide useful data for comparing a hand recount of optical scan or punch card ballots versus the original machine count. It merely shows that there is greater variation between count and recount of hand-counted paper ballot systems than there is in optical scan ballot systems. Optical scan systems’ smaller discrepancy between the count and the recount only shows that it is more consistent, not necessarily that it is more accurate.
And if you insist on extrapolating the data to say that a hand count of optical scan would be less accurate than a machine count (and the study does not suggest this) then you would have to also conclude that a hand count of punch card ballots would be more accurate than the original machine count, due to the extraordinary high residual vote rate.
The only constructive conclusion the study draws from the tabulation validation data is that states should have a threshold of at least .5% for mandating automatic recounts.
I won’t attempt to get into the heads of Berendt or Gregoire, but I believe you are (perhaps intentionally) confusing the controversy over verifying provisional and absentee ballots, with that of a hand recount, since that rhetoric stems from the former.
But it doesn’t make it wrong either.
And that is the point of my conclusion. The fact that Rossi “won” the first two counts does not suggest in any statistically meaningful way that more voters intended to cast their ballot for Rossi than for Gregoire. The number of random counting errors created by machines counting imperfectly marked ballots is so much larger than the margin of victory as to make it impossible to accurately determine the outcome, regardless of which technology is used.
Given these circumstances, neither side should feel ashamed of using existing law to their advantage. Whoever “wins” the final recount, well… that’s the luck of the draw.
Assuming the software is clean, no.
This is a pointless, sales-driven feature that any engineer would laugh at. A paper trail is useless if the voter can’t examine it to ensure that the machine registered what the screen said it did.
This, by the way, shows you the problem with using tabulation validation rates to compare the accuracy of different voting technologies, as lever machines and DREs will always produce a tabulation error rate of zero. Yet the relatively high residual error rate of DREs shows that there are interface-induced tabulation errors… we just can’t do a recount to find them.
Believe it or not, I was too busy doing real research to follow up on yet another one of your paranoid anecdotes.
I have no knowledge of what you allege is going on in Montana. But I would not support either party attempting to invalidate ballots with clearly discernible markings.
I shouldn’t think I have to make this statement, but…. I am not in favor of stealing elections.
Satisfied? (No, of course not. Because I’m clearly evil.)
One might suggest that if a hand count including voters “intent” if Chris were to win would either imply that votes were either scammed by the system used, or that Democrats are less inteligent than the rest of us and lack the brain power to vote correctly…
Thanks for the analytical effort.
Regarding Snohomish and Yakima counties, wouldn’t each have had a great many absentee and provisional ballots. Although a “hand” recount of the touchscreen, non-paper producing machines might seem useless, a hand recount of the absentees and provisionals makes sense. It is the existence of these ballots that makes Yakima county’s original recount look odd.
You’re absolutely right. I hadn’t considered the absentee and provisional ballots, which should in fact constitute a majority of ballots cast.
Goldy, this analysis is flawed.
Obviously there is the margin of error of a particular technology, and there’s the residual margin of error that you point out.
But, none of that really asks the important questions. For starters, what is the point of holding an election? Are we not trying to determine a winner? Now you can dismiss this as statistically insignificant using the residual blah x 3. But the reality is that a winner needs to be determined given the process as defined by law and not by a coin toss. Thus, the question that should be asked is “how can we lower the error rate of voting within the technologies that were used to hold this election?”
The best way to do this is to throw out the ballots that are not readable by machine. Once that happens, we arrive back at the 1 in 1,000,000 error rate of the machines, we get rid of the residual error of undervotes and overvotes which are simply improperly cast ballots, and we get rid of the bias of human divination of intent.
In short, if truly determining the winner with the least margin of error is the goal, then the best way to do it is to elinate the largest sources of error.
But let’s be frank, you nor Gregoire really want to eliminate illegitimate ballots because introducing a large margin of error is the only way to create a circumstance that “might” produce a Gregoire win. This is shameless partisan machination.
The most accurate results we have thus far are those produced by machine, and before divination bias. That shows a Rossi win by 261 votes.
Admit defeat and concede.
Thank you Jeff B sanity at last!
You fail to understand what causes a tabulation invalidation rate in excess of .5% in counting equipment that otherwise certifies at .0001%. These are not simply just over-votes and under-votes, there are also false-positives and false-negatives and other scanning anomalies. There are a large number of “readable” ballots that read differently each time they pass through the machine. If you did a hundred machine recounts of this election you likely would never get the same result twice, and if the margin is as close as the first two counts suggest, Rossi and Gregoire would each win roughly 50% of the time.
How do you throw out a ballot as “unreadable” when sometimes it registers a vote for Rossi and sometimes it doesn’t? What you are suggesting really amounts to cooking the books. If we followed your advice, we would certainly end up with a winner. But that final count would be no more representative of voter intent than what we are doing now.
The law defines guidelines for discerning voter intent. And the law defines that either candidate can request a hand recount, and that the winner of that final count will be governor.
You call it sanity to throw out the election laws? Actually it is anarachy. The ordered processes of the day are the State of Washington laws which cover these things. Sorry if the law gets in your way. Go to court and loose again.
A bit off the R’s image to think of them as outlaws, but here we are.
You guys act like sore loosers and your guy is ahead.
Goldy- you are still mixing the residual vote and tabulation error factors. The difference in residual vote is NOT because of increased or decreased machine error- it is the increased or descreased HUMAN error based on familiarity with the way to properly fill out a ballot.
That argues against punchcard technology, because most people NEVER fill out a punchcard, thus are MORE likely to make errors- undervotes and overvotes. But it does NOT indicate that any tabulation method is better or worse.
And in handcounting optical scan, you are handcounting a paper ballot. The tabulation error rate is higher from handcounting than optical scan machine counting.
The way to reduce human error is to take more time- but time is finite in this. Wait until the first group of pollworkers throw up their hands in frustration as they can’t get totals to agree while working under time pressure…
And as for Yakima- no argument we are not getting printed out verifiable ballots that we can be certain reflect the vote cast, but they SHOULD reflect the vote that the machine tabulated. I would argue any tabulation difference will reflect human error.
Re: Montana — The news reports that there is a tie after three recounts of the votes for Montana’s 12th District House seat: 1,559 votes each for the Democrat, Jeanne Windham and the Constitution Party candidate, Rick Jore. (The Republican candidate, Jack Cross, finished a distant third with 1,107.)
It’s all the more exciting because this race will determine whether Republicans or Democrats control the House next session (including committee appointments, chairmanships, etc.) — the Dems will control the Senate and the Governor’s chair, but the outgoing Governor (until Jan. 3) is a Republican. And if there’s a tie, the Governor gets to choose and appoint the winner.
AP reported Windham, who contends five ballots were illegally counted for Jore, obtained a court order that prohibits Secretary of State Bob Brown from certifying results of the recount and bars Gov. Judy Martz from appointing someone to the seat.
The Democrat claims five optical-scan ballots had the ovals filled in for both Jore and Cross; but apparently there is an X on or near Cross’s oval on each, which probably [to my mind, and the MT Board of Canvassers, incidentally all 3 Democrats] means the voter intended to vote for Jore. Probably — because it’s impossible to say without looking at the ballots. [I guess machines aren’t perfect vote counters after all…]
So it’s in court to determine whether the voters’ intent can be determined on those ballots, and meanwhile we can sit back and watch Jore and Windham fight to have certification happen before or after the Governor becomes a Democrat.
Great Falls Tribune article
Billings Gazette article
Leader Advertiser article with info about the recounts
Aw, Chuck, you have a follower. Or maybe it’s the other way around. How sweet. How sad. Jeff B. thinks we should “throw out the ballots that are not readable by machine. Once that happens, we arrive back at the 1 in 1,000,000 error rate of the machines [snip]”
Okay, enough already of the lunacy arguing that we must worship the numbers produced by the machines and ignore the statutory law of Washington that says votes matter as long as the voter’s intent is clear!
I guess votes don’t matter to you two if the paper isn’t perfectly flat or the ovals aren’t perfectly marked (with the “right” kind of pen) or the ballot was printed slightly out of alignment. Y’all are like the Soup Nazi for elections: No vote for you!
Also, this so-called “one-in-a-million error rate” is pure fantasy. Did you guys not read the post you’re responding to? The error rate you’re referring to is for reading perfectly prepared test ballots in lab conditions. Not the real world of elections. The problem isn’t that the machines are breaking or failing to do what they’re designed to do; it’s just that they’re not designed to read marginal ballots perfectly…because they can’t.
Chuck raised the example of a “hesitation mark” on an optical scan ballot last night — where you put your pencil down for a sec in a candidate’s oval, but decide not to fill it in. This is not a rare occurrence. But depending on how big the dot is, and how precisely the ballot is fed through the machine, and how much humidity is in the air, and myriad other factors, the scanner might read that dot as a vote (which would negate your vote if you had decided to vote for someone else). That’s just one example of how machines cause tabluation inaccuracies.
Goldy,
No, you are clearly failing to understand, or more accurately not wanting to understand.
You yourself state that the voting equipment has an extremly low failure rate of 1/1,000,000. So the equipment is not the source of error. The error is in the ballots. If a ballot sometimes registers for Rossi, and sometimes it does not, then the ballot must be thrown out. It is only when hand recounting techniques are introduced that the margin of error goes outside of the current difference in vote totals. (Out of 2.7 Million votes cast, if counted only by machine, and only those ballots that are machine readable, we would expect to see only +/- 6 votes. One vote either way for every million cast, according to the MIT vote study) So, the best way to get an accurate result is to simply throw out anything that the machines can’t count and live with the result.
Of course that would mean Rossie would win, because we would be throwing out all of the improperly cast provisionals, and other types of divined ballots that were not machine readable.
You are correct in that the law provies guidelines for discerning intent and for hand recounts, but the thesis of your argument is that it is not possible to be within the margin of error.
The truth is that it is possible to be within the margin of error, but to do so, we would have to take the least doctored reults. AKA the most objective results. AKA the machine results. The first machine count was incapable of inserting the kind of human bias we saw in the second count with King County adding bias. A third recount done by hand will insert even more error into the process. In short, by requesting a hand recount, Gregoire is actually making the results less accurate.
I also fail to see how throwing out a ballot that is not yielding a consistent result is cooking the books more than keeping a ballot that has been examined and intent divined? Is it not possible that in divining intent, humans could be wrong? Thus it would be far more fair to throw out ballots that are not absolutely deterministic than to keep them and have the possibility that the election was decided by human bias.
If our true goal is to find the most objective winner of the election, then it will not be served by a hand recount.
The democrats simply want to create a cirsumstance that might lead to a Gregoire win.
Jim King says: “Goldy- you are still mixing the residual vote and tabulation error factors. The difference in residual vote is NOT because of increased or decreased machine error- it is the increased or descreased HUMAN error based on familiarity with the way to properly fill out a ballot.”
It’s not just human error in not understanding how to mark ballots; machines not operating in lab conditions won’t do a perfect job reading real-world ballots. And even where there is human error or a slightly mis-marked ballot, the vote should count if it’s clear. The vote will count in Washington if it’s clear.
The pure tabulation error rate for optical scan machines, according to the Caltech/MIT VTP study, was 0.56%. And for initial hand counts, it was 0.83%. But the point is that the standard is a meticulous manual (read: visual) recount. The Using Recounts study points out (the obvious) that in hand recounts “greater effort is taken to arrive at the most accurate accounting of votes cast.” Stephen Ansolabehere, one of the study’s authors, reports in the New Scientist that “When votes are recounted, the error rate falls, often to as low as a few hundredths of a per cent.” That’s backed up by the 40 years of real-world experience of Bob Swartz, founder of punch-card company Cardamation, who told a New York Press reporter that “We read the cards through the machine twice, and if there are differences we look at the cards. If our goal is to get 100 percent accuracy, there’s no question that’s the way to achieve it.”
Yes it takes more time (really, it takes more care, which takes more time) to meticulously examine all the ballots and make sure the count is correct. But accuracy is the goal here, and we need to take the time to get it right when the totals are this close. And we will.
Jeff B. says, “If a ballot sometimes registers for Rossi, and sometimes it does not, then the ballot must be thrown out.”
No, Jeff, if a ballot sometimes registers for Rossi, and sometimes it does not, then the ballot must be looked at. I’d like to see a cite to Washington law saying that votes that an optical scan machine has trouble with are invalid ballots and should be discarded. Absent that premise, the rest of your argument falls. Oh, but you admit the law doesn’t say that. So you are basically saying that if we accept only ‘perfect’ ballots, then the machine totals will be authoritative. Well, sure. But that conclusion has no relevance to real elections.
I’m not suggesting that we throw out state law. I’m fully aware that our process allows for divination and hand recounts.
What I am saying though is that if the true intent is to get an accurate result, then it won’t be served by a hand recount.
In short, “every vote counts” is simply doublespeak for hoping for a less than accurate circumstance that might sway the election in favor of Gregoire.
This will be duly noted by all Washingtonions who prefer a more honest interpretation of their elections and it will destroy Gregoire’s political career.
And indeed, the polls at both KING5 and KOMO show that more people prefer that this end now with Gregoire’s concession.
David:
So you are basically saying that if we accept only ‘perfect’ ballots, then the machine totals will be authoritative. Well, sure. But that conclusion has no relevance to real elections.
It should. -Jeff B.
The true intent is to get an accurate result — not just a precise one. Throwing out any ballot that a machine has difficulty with will leave you with a set of perfectly machine-readable ballots that they can count precisely, with highly-reproducible results (your one-error-in-a-million stat), but the total will not be an accurate count of the votes cast in the election; just an accurate count of that subset. “Every vote counts” means just that — every vote for Rossi, every vote for Gregoire, every vote for Bennett. Magical ‘divination’ is not required; the manual recount process is done with existing standards and conducted under high scrutiny. The people who do the counting are dedicated and honorable citizens of this great state. (Republicans say they “trust the people” — perhaps it’s only another slogan.) The results of a careful recount are far more accurate than initial machine counts of all the ballots.
DIVINATION- I thought that was water seeking with willow switches. On he day I visited the counting center for King county, lo it seems years ago, and watched with a lot of other people, no problem at all, very welcoming in fact, saw no willow switches.
Did I miss something important, something left out of the statues?
Oh yes, lets have law by poll. Especially KOMO, maybe would not be too bad if lefty Schram is controlling the voice. But really, law by poll?
But if only way to get an accurate result that is within the margin of error given the current difference is to use a more precise method, then only that precision will yield a true result.
It is simply wrong that the results of a careful hand recount will be more accurate than a machine recount. Read the MIT study above.
I’m not saying that there won’t be dedicated and honest citizens doing the counting. What I am saying is that it is not possible to arrive at a total that is more accurate than what the machines can give us, given the known error rates for hand tabulation. And when you insert the bias of determining intent, the error rate increases.
It’s a question of objectivity. A machine is impartial to the result.
Ideally we would be both precise and accurate, but if we can’t have accuracy in an election, then what is the point of holding one? That’s why any attempt to introduce imprecise methods (every vote counts) is really an attempt to sway the accuracy (a circumstance for a Gregoire win.)
Jeff and Chuck are right on the money. In the end, the die-hard Dems will amount to be nothing more than sore losers, that simply couln’t fathom that it is possible for a Rep. to win in this state. But what the hell, let them try and drag this fucking thing well into the new year. All these idiots on the news ranting and holding signs saying “every vote counted” etc. are full of BS. Guess what, it turns out they WERE counted, and lil Crissy lost anyhow. In the end, that’s what it’s all about anyway-putting litter in it’s place. Thanks.
Jeff B. thinks that “it is not possible to arrive at a total that is more accurate than what the machines can give us, given the known error rates for hand tabulation.”
That’s not correct. See my previous post, which finally showed up (including hyperlinks slows things down). Machine counts are more accurate than quick hand tallies, but meticulous manual recounts are the most accurate method we have for tallying votes.
In the end, more accurately, if in the very end, if Christine Gregoire has more vote she will be the Governor No court will bail out the R’s, stakes too low for the US Supremes to vote their biased politics here, our egos to the contrary, Washington is no big deal on the eastern seaboard.
Law is clear in this state, a recount at this point is just following the very clear encoded civil law.
The sore loosers will be those who continue to whine and belly ache after the recount is complete. And in all fairness to Chrissy, she has no reputation about wimping out at any time in her career. Sorry, Chrissy is not Gary Locke. Yes, he got wimpy in his second term. Perhps that is part of the probleme here, a fighting feisty D party, folks got too used to Mr. Gary mumbling something about what could not happen. New era, stay tuned.
Still trying to
Chuck raised the example of a “hesitation mark” on an optical scan ballot last night – where you put your pencil down for a sec in a candidate’s oval, but decide not to fill it in. This is not a rare occurrence. But depending on how big the dot is, and how precisely the ballot is fed through the machine, and how much humidity is in the air, and myriad other factors, the scanner might read that dot as a vote>>>>>>>>>>>
But my point is depending on how the mark is a elections person might consider it my “intent” in a manual count and I would be considered having voted for someone I never did…all because of voter “intent”
David,
If you read the study closely, you will note that when the hand recounts were done, they were merely tabular hand recounts meant to identify marks that confused the machines. In those obvious cases, there is little difference between hand recounts and machine recounts, and indeed in instances where we are able to correct an obvious and fully deterministic flaw that throws off the machines, there will be no real problem with the hand recount.
The problem comes in when there is an attempt to determine intent. Even given the guidelines, it is quite possible that those counting will incorrectly determine voter intent, thus introducing innacuracy into the process. And, that is what Gregorie is “counting” on. All it would take would be a few votes in each county to sway the election for Gregoire. This is where our laws are clearly outdated. These laws have been in effect since long before the accuracy of the machines that we have today.
Obviously we can’t change the laws now for this election, but there is no doubt that we would be better served by the impartiality of machines when it comes to the bias that can be inserted when “determining” voter intent.
Furthermore, the study also assumes meticulous hand recounting under very close scrutiny. But it is still much more likely that humans will make an error than it would be that machines would make an error in the purely tabular sense.
What’s perhaps most alarming, and what Gregoire is counting on the most is the punch card ballots that may lose chads after being handled several times. It’s quite possible that votes that were previously legitimate votes, lose a chad and then have to be thrown out. Again, all it would take would be a few of these votes in the punchcard counties, combines with other anamloies to sway the election.
All of these sources of error add up to more of a circumstantial win than a precise win, if indeed the election flips to Gregoire. Albeit a win under Washington law.
What’s obvious and so flawed here is that Gregoire’s only hope is a circumstantial win. It’s clear that she is doing whatever it takes to win, while sanctimoniously claiming that every vote must count and that it’s much better to do a statewide (more sources for error) count, etc.
There’s a real danger here for her political future either way.
And given the potentials for error in such a close race, the objectivity of the result will always be much more suspect after a hand recount where there is so much human handling of the ballots.
A convincing analysis, to me at least. Some of your observations also cast doubt on the rather facile argument that voters who are too dumb to follow voting instructions are to dumb to be voters at all. This policy would amount to using voting technologies as minimal intelligence tests, separating competent from incompetent voters. Only the “tests” would be more difficult in some locations than in others, and some more arbitrary as well.
Tom,
That is indeed what happens. Voters consistently have more problems with punch card ballots than they do with optical scan ballots.
Punch card vote machines are slightly harder to operate, so this does dorm a sort of low pass filter for voting.
However, all voting technologies are by design, very easy to use so anyone who claims that they did not understand how to vote, or were not able to vote is either lying or incapable of voting.
This is especially true given the enormous window of time to vote and the number of volunteers who are more than happy to assist voters who are having trouble with the process.
This is also why improperly cast ballots should simply be thrown out. Voter intent of those voters who take a lot of time to research the candidates and issues, carefully fill out their ballots, and make sure they are delivered to the proper precint is diluted by the “every vote counts” welfare of allowing provisionals without signatures, trying to determine intent of improperly filled out ballots, etc.
Whe hear the shrieks of “every vote counts” but what about the votes of everyone who took the time to do it right? Should the intent of these voters be swayed by those who make a mockery of this simple process?
Thanks for your analysis. I understand the statistics of this race much better now.
Jeff B. says that “If you read the study closely, you will note that when the hand recounts were done, they were merely tabular hand recounts meant to identify marks that confused the machines.” Eh? That’s not what the study says at all. See page 2, first paragraph, which states that “state or local election officials conduct an audit of votes cast” and “reexamine the ballots to determine voter intention.”
Jeff continues, “indeed in instances where we are able to correct an obvious and fully deterministic flaw that throws off the machines, there will be no real problem with the hand recount.” Yes. That’s right. The manual recount will correct errors in the machine count.
But, he says, “The problem comes in when there is an attempt to determine intent. Even given the guidelines, it is quite possible that those counting will incorrectly determine voter intent, thus introducing innacuracy into the process.” Also true — mistakes are possible. But given the level of scrutiny from all sides in a manual recount, the inaccuracy introduced is expected to be very small (see statistics cited earlier). And if people have trouble with a ballot, a machine certainly can’t be expected to get it “right”. Finally, if the voter’s intent can’t be ascertained clearly, the vote is not counted for either side.
And, not to be dissuaded, Jeff repeats his thesis from before: “Furthermore, the study also assumes meticulous hand recounting under very close scrutiny. But it is still much more likely that humans will make an error than it would be that machines would make an error in the purely tabular sense.” Only if you’re feeding the machines perfect ballots with perfect markings under perfect conditions, Jeff. Unfortunately not an assumption you can make for all the votes in real elections.
Jeff is also alarmed that “punch card ballots . . . may lose chads after being handled several times. It’s quite possible that votes that were previously legitimate votes, lose a chad and then have to be thrown out.” Punch cards are actually quite sturdy. In Washington, I believe that the standard for counting a punchcard vote is that at least two corners of the chad have to be punched out. And when you run a card through the machine, only chads that have been partially punched out like that are likely to fall off; unpunched chads don’t just fall out. It’s actually more likely that a real vote that wasn’t counted at first will later be seen. So worry not; they’ve been through machines all they’re going to be, and human eyeballs aren’t going to damage the cards.
The point I was trying to make earlier is it is my reasoned belief that machines are best at tabulation; unclear ballots require human determination, but then back to machine tabulation (oh, my GAWD- the King County process!!!!) unless you are talking about so few ballots that careful human, doubelechecked tabulation can be done in a timely manner.
Yes, the “unreadable by machine” require human intervention- this whole mess cries out for divine intervention. Goldy, I dun think that that thar Rapture came and left us all behind in recount hell!
Well, it is all irrelevant now- the lawyers are unleashed, every previous decision is to be second guessed, Gregoire has gone nuclear and Rossi will return fire. The politics of mutual assured destruction wins, we as a state lose. Worse than recount hell- maybe that was only limbo- we are in that hell that can only be created by the lawyers.
My conclusion, Jeff, was that whoever ultimately “wins”, Gregoire or Rossi, the win will be, in your words, circumstantial. There is no guarantee that a hand recount will be more or less accurate, but regardless, the margin of victory will be so far within the margin of error that the result will be statistically meaningless.
That said, the law says that the hand recount is final, so that is how we are going to decide who is governor.
You have a lot of faith in people, David. I dare say it sounds like you dont even lock your house when you go to town because you are so trusting. The blind trust you display is amazing. But even as you try to justify things you seem to qualify elections people as seuth sayers able to determine the intent of someone that wasnt sure about what they were trying to do anyway. The Great Conyac of Carson fame couldnt do that with skill and niether can anyone else, besides he retired as should be any idea of a hand count. If we are going to hand count lets hire crusaders to do it so we can really feel like we are back in medieval times.
Sorry, Goldy- the count doesn’t appear to be what is going to decide things. The Gregoire camp has unleashed the lawyers, and is also advancing the novel claim that any dispute is to be settled by the Legislature, not the courts. Gee, if we lose in court, we’ll go to the Democratic legislature- if we win in court, well, that is good enough. After all, the GOP can’t appeal to the Democratic legislature. And thus do we vault over the edge of the abyss…
Well, Chuck, I don’t trust *you*.
By the way, it was “Carnac the Magnificent,” not ‘The Great Conyac’ (though there’s something to be said for great cognac).
I haven’t seen what you are referring to. I’ve heard there is a letter that went out from the Rossi campaign. I hope you are not predicting armageddon based on that.
Well, the proof will indeed be in the recount.
Anyone who is really serious about understanding this data should read the studies thoroughly and not David’s comments. The studies are mostly about residual votes which are either under or over votes. As the studies conclude, the residual percentage says a lot about which voter interface that voters are comfortable interacting with, but does not necessarily correlate that residual error rate to an outcome and as Goldy originally concludes, leaves the election within a margin of error, especially when there is human bias.
Furthermore, the studlies clearly conclude the opposite of what David says in response to my comments, which is that machine recounts are more accurate than hand recounts, quoting directly from Page 7.
“Stepping back from the details of the data, New Hampshire’s recounts speak directly to two important questions in election administration. Have we made progress? Do machine counts improve on hand counts? At least in the comparison of optical scanning and paper, the answer is yes.”
And, in the studies which were conducted in New Hampshire, almost all of the data comes from rather small local and state races. It’s obviously much easier to arrive at a definitive answer in a small race of a few thousand votes than it is to arrive at a much larger 3 million vote race as we have here in WA.
So the end result is that indeed as I originally indicated, machine counts are more reliable and that while many voters may indeed be confused by a particular voting technology that generates residual votes, that does not say anything about why those votes are residual and still leaves open the source of error of determining voter intent on residual ballots that are not clearly deterministic.
Goldy- the Rossi campaign released the letter that the Gregoire/Democratic Party (I’m not sure which of the two they are technically representing) lawyers sent to Sam Reed. And yes, it is that letter I am basing this on- the threats are ominous, the apparent strategy frightening. Just because Rossi’s folks got it and released it doesn’t mean it’s not from the other camp…
If you haven’t gotten it, I can e-mail it to you…
Jeff shows a talent for creative misinterpretation, stating that the VTP studies “clearly conclude the opposite of what David says in response to my comments, which is that machine recounts are more accurate than hand recounts,” and quoting the Using Recounts report at p.7, “Do machine counts improve on hand counts? At least in the comparison of optical scanning and paper, the answer is yes.”
Jeff, you are so half right. As far as the quote, I said as much before (see post #34 above): “Machine counts are more accurate than quick hand tallies,” — I’m representing exactly what the study concluded — but at the same time, I’m not ignoring the report’s reliance on “meticulous manual recounts [as] the most accurate method we have for tallying votes.” In fact, a meticulous manual recount is the standard that initial hand tallies and optical scan results are compared against. The experts say that it’s far more accurate in percentage terms than either a basic hand tally or a machine count; in fact, they say it’s the only way to get an accurate result: see post #26 above for citations.
I, too, encourage people to read the VTP reports thoroughly; I will stand by every comment I’ve made here.
No Jeff, you read more into the study’s conclusion than is intended. The conclusion is that in comparing hand recounts of hand counts to scanned recounts of scanned counts, there is less variation in optically scanned systems than in hand count systems.
The study clearly shows that when considering tabulation validation rates (and that metric only) optical scanning systems produce less error in the original count than hand count paper ballot systems. But you cannot extrapolate that the final result of a scanned recount is more accurate than the result of a hand recount.
The study does not evaluate hand recounts of optical ballots. The study does not evaluate hand or mechanical recounts of punch card ballots. And thus one cannot draw a conclusion from this study as to whether a hand recount of optical or punch card ballots would be more or less accurate than a machine recount.
Indeed, the only practical recommendation that his study makes is that the minimum threshold for automatically triggering a recount should be a 0.5% vote margin.
That said, I absolutely agree with you that people should read the studies for themselves. That’s why I provided the link.
Jim, please email it to me. I am far from the party operative that people like Cynicla-di-da comfort themselves in believing. The Democrats don’t send me anything except the occasional fundraising letter.
Goldy- This REPUBLICAN gets all the State Democratic Party e-mails- for some reason, Yahoo considers them spam and they go to my bulk file…
But the letter is on the way…
Remember those 400+ affadavits tearful Paul delivered to King County off the list of 900 rejected provisional ballots. Those affadavits, complete with tear-stains, were supposed to be set aside by King County election officials. Word has it a public records request was submitted for those 400+ affadavits. Seems simple, right??? WRONG!!! After waiting until the last minute, they were made available. Only 2 problems:
1) There were 225 of them
2) Only about 1/2 of those were on the original list of 900+
WHAT IN THE HELL IS GOING ON?????
This was a simple request…copy what weepy Paul delivered.
They were required to keep them aside.
Maybe King County attorneys misunderstood a clear-cut request??????
Never lose sight of these 400+ affadavits that allowed those related ballots to be counted!!!!!!!!!!!!!!!!!!!
Berendt is 100% responsible for them…he delivered them & it’s on tape. The question is how did 400+ dwindle to 225? And why are only 1/2 of the 225 on the provisional list??????
Well, Chuck, I don’t trust *you*.>>>>>>>
David, if your life is ever at stake and you need someone to trust, I hope you are lucky enough to find a person like me. You and I may disagree on EVERYTHING but I am honest to a fault and will defend your right to disagree all day long…dont let this get to you.
If every affadavit Berendt delivered is not accounted for and available, why bother with a recount???
Oh, I’m sure you’re an upright, trustworthy person, Chuck. It’s just your opinions I don’t trust very much. Sorry about that.
(If your life were ever at stake, and you needed someone to trust, you could count on me, too. But in that situation would you have a problem trusting a state [or King county] elections official? I imagine you wouldn’t. That’s a different kind of trust.)
What a drama! Gregoire gives in to republican demands….but will republicans accept the result or complain if gregoire ends up ahead? Too early to tell!
She wins, she wins, and all the R’s in the state can’t put Humpty D. Dino together again.
Either the errors are there in the counting and she wins, or Rossi is the G. elect.
Dems will have the $$$$$, full state recount, all counties and all precincts. Hard ball time is here. And it seems Chrissy is in charge. R’s will be soiling themselves in horror, angst and accusations. Remember Dino said it is up to her.
Sorry, I just wanted this recount all along.
If Rossi wins, wonder if he has enouh guts to put Gregoire in his cabinet? He said he plans to invite democrats, could be a master stroke to just offer. HUH.
Nope, sweet chrissy just commited political suicide, she is done put a fork in her!
Perhaps he will offer her a crossing guard job….
DUDES—
HELLO!!!! The provisional ballot affadavits!!! Remember??
Cynicla-di-da… it’s 2 o’clock in the fucking morning! Perhaps you’re so angry because you don’t get enough sleep.
No–had to take my wife to the emergency room. I’m not angry…take a look at SoundPolitics. One of your buddies “X” guesses no more than 4 of those affadavits turned in by “unstable Paul” were illegitimate. How did he come up with 4?? Don’t you think someone independent ought to get the same information the Dems got, plus copies of the affadavits and go out and confirm each one??? After all, if the bumbling Dems can find 400+…shouldn’t the stumbling R’s be able to also??? The problem is, the Public Records request appears to have yielded something entirely different. Look into it Goldy…ask for those affadavits yourself. You won’t because you can clearly smell the stench. Keep watching Goldy. This one ain’t goin’ away. Hey, I’ll bet the Dems made copies of each one too. Why don’t you call blubbering Paul and if he can shut the tears down for a couple minutes, maybe he can give you copies of what he turned in. That might help put this one to rest.
Cynical… I hope your wife is okay.
As to the affidavits, I’m all for all sides verifying the hell out of them. I’d rather have an honest election elect a Republican than a dishonest one elect a Democrat. I support working the vote count as aggressively as the law allows, but I don’t support cheating.
And here’s something to chew on for the “only count perfectly marked ballots from people with perfectly matching signatures” crowd. There is no question that in an election this large, a certain amount of fraudulent voting took place. But without a doubt, the number of legitimate voters whose legitimate votes were not counted was exponentially larger.
So hypothetically, if 4 of those 400 affidavits were fraudulent, should we have disenfranchised 396 legitimate voters to stop the 4 cheaters?
I as this because you have to make a cost-benefit analysis to determine at what point anti-fraud measures are no longer worth the cost in lost legitimate votes. The credit card industry makes a similar analysis in constructing anti-fraud measures, and their’s obviously favors the convenience of the consumer.
I had posted for a while regarding the innacuracy of Goldy’s analysis of hand counting in these comments.
I finally gave up when I realized that it’s useless to try and reason with a stubborn donkey.
For those who are interested in truly understanding the statistics and “Sound” reasoning behind why a third hand recount will be less accurate, swing over to soundpolitics.com.
If you just want to be filled with irrational Democrat/leftist ideas that allow you to continue your denial of the fact that even traditionally Blue states like WA are turning more to candidates like Rossi (you know the same ideas that were based on the statistically flawed polls that convinced you all that Kerry was winning until the results actually came in for Bush) then keep sniffing what is falling out of this HorsesAss.org.
However if you want to form a more objective opinion and actually understand why the Marxist ideas of the left are losing traction, go elsewhere, open your eyes, and learn.
Regards.
Jeff… I posted my analysis, cited my sources, and then encouraged people to read them and debate. In fact — while you may disagree with my conclusions — my analysis, in tone and content, was extremely non-partisan.
And even Stefan’s critique was fairly tempered and non-partisan… at least for him.
And yet you reply with your “Marxist” crap. You don’t want a debate. You want personal affirmation that you are right.
But in case I’m wrong, and you really do want a debate, let’s move the discussion over the CalTech/MIT studies to “Lies, damn lies, and statistics“, a thread I’ve opened for the purpose of discussing Stefan’s critique.
I am having trouble getting my mind around the posts that advocate throwing out votes that machines can’t read, equating expediency with accuracy. Somehow the poster has concluded that the machines should be arbiters of the process, rather than tools that the arbiters can use to speed up the process. The notion that such a count is more accurate when it excludes votes is ludicrous. The fact that the poster repeats it many, many times doesn’t make it any less so.