April 25, 2014

avatar

Discerning Voter Intent in the Minnesota Recount

Minnesota election officials are hand-counting millions of ballots, as they perform a full recount in the ultra-close Senate race between Norm Coleman and Al Franken. Minnesota Public Radio offers a fascinating gallery of ballots that generated disputes about voter intent.

A good example is this one:

A scanning machine would see the Coleman and Franken bubbles both filled, and call this ballot an overvote. But this might be a Franken vote, if the voter filled in both slots by mistake, then wrote “No” next to Coleman’s name.

Other cases are more difficult, like this one:

Do we call this an overvote, because two bubbles are filled? Or do we give the vote to Coleman, because his bubble was filled in more completely?

Then there’s this ballot, which is destined to be famous if the recount descends into ligitation:

[Insert your own joke here.]

This one raises yet another issue:

Here the problem is the fingerprint on the ballot. Election laws prohibit voters from putting distinguishing marks on their ballots, and marked ballots are declared invalid, for good reason: uniquely marked ballots can be identified later, allowing a criminal to pay the voter for voting “correctly” or punish him for voting “incorrectly”. Is the fingerprint here an identifying mark? And if so, how can you reject this ballot and accept the distinctive “Lizard People” ballot?

Many e-voting experts advocate optical-scan voting. The ballots above illustrate one argument against opscan: filling in the ballot is a free-form activity that can create ambiguous or identifiable ballots. This creates a problem in super-close elections, because ambiguous ballots may make it impossible to agree on who should have won the election.

Wearing my pure-scientist hat (which I still own, though it sometimes gets dusty), this is unsurprising: an election is a measurement process, and all measurement processes have built-in errors that can make the result uncertain. This is easily dealt with, by saying something like this: Candidate A won by 73 votes, plus or minus a 95% confidence interval of 281 votes. Or perhaps this: Candidate A won with 57% probability. Problem solved!

In the real world, of course, we need to declare exactly one candidate to be the winner, and a lot can be at stake in the decision. If the evidence is truly ambiguous, somebody is going to end up feeling cheated, and the most we can hope for is a sense that the rules were properly followed in determining the outcome.

Still, we need to keep this in perspective. By all reports, the number of ambiguous ballots in Minnesota is miniscule, compared to the total number cast in Minnesota. Let’s hope that, even if some individual ballots don’t speak clearly, the ballots taken collectively leave no doubt as to the winner.

Comments

  1. Pete says:

    The smudge example got me to thinking; don’t we all leave fingerprints on the paper ballots we touch during the process of voting? Identifying a person due to inadvertent ink on their fingers would still leave quite a long way to go to identify the individual to deliver their reward; albeit one step closer, because the fingerprint is already visible.

    OK, time to get back to my CSI marathon.

    Thanks,

    Peter

    • Tel says:

      Unless wearing gloves you would certainly leave traces of DNA which are good identification. Finding those traces would be difficult, but hairy voters would be easier to track down. Mind you, every election official handling the ballot also leaves fingerprints, DNA, possible smudges, etc.

  2. James Grimmelmann says:

    Lizard People 2012!

  3. Mittch Golden says:

    Apropos your pure-scientist hat comment: I have long felt that the best way to resolve these situations is to recognize that there is a measurement error, and, in the cases of a close election, to use a randomizer to determine the outcome.

    In my view, it would work like this: Let’s suppose we have an election with two candidates, and consider the percentage of the ballots validly cast (i.e. so the total of A and B is 100%). If A gets more than 50.25% of the vote, he’s declared the winner; if he’s below 49.75%, he loses. Between the two values, we have a straight line of probability that he’s the winner. (I.e. p = (A-49.75)/(50.25-49.75) ). A random number between 0 and 1 is chosen, and that determines the winner.

    This approach removes the unrealistic assumption that we can know voters’ intent down to the very last voter, and would remove the necessity of battling over a few crazy ballots.

    It would have been much more civilized (and scientifically valid) to resolve Bush-Gore in 2000 that way than the way it was.

  4. Lior says:

    There’s nothing wrong with optical-scan ballots. However, voters should have the option of filling them using machines (computers). A computer-filled ballot would not have overvotes, smudges, or marks. It would be easily readable by the optical scanner, but verifiable by the voter and available for recounts.

    In this setup you can ensure in hardware that the “cast vote” button also wipes the machine’s memory thus alleviating most of the privacy and security issues of ordinary voting machines.

    I’m sure Prof. Felten has good arguments why hand-filled optical-scan ballots are nevertheless superior, but I’m not convinced.

    • Anonymous says:

      Hand filled optical scan ballots are superior to machine marked ballots in one way: They cost less. Also, people can be voting by the dozen on any flat surface, rather than waiting in the rain for hours to get to a machine which cost 5K and flips the votes before the ballot is marked and printed. And on and on.

      Thank God Minnesota has 5% audits, otherwise the machines could not be trusted at all.

  5. I R A Darth Aggie says:

    If I understand the process, the first ballot with the “NO” mark next to Coleman should have been rejected by the Accuvote machine as an overvote. The ballot should have then been marked as SPOILED by a poll judge and the voter given a blank ballot to “try again”.

    So, why is this ballot even being considered?

    • Anonymous says:

      Different states have different rules and the auto-reject programming is optional. So at least where I live you are right that ballot would not even be considered. However the rules differ and not all optical scanners are programmed to give the overvote warning. I would imagine that whatever the law is in Minnesota they are following it.

      Having said that if you’ll recall Florida in 2000 the media loves the odd or extreme, the photogenic ballots. In my experience what the media sizes upon, particularly in the area of election issues has nothing to do with what is actually going on.

      I wouldn’t be surprised to hear that all of these ballots were rejected and that is why they were permitted to take pictures of them in the first place.

  6. Anonymous says:

    The rule needs to be simple: ballots are spoiled if they are not marked correctly. It doesn’t matter how they are marked incorrectly, or if someone could discern the intent of the voter with reasonable reliability. Only the ballots marked according to the rules are counted.

    This may seem strict, but it’s more important to have a clear, indisputable result, than to count every poorly marked ballot. The rules apply to everyone equally, so they are fair. In all liklihood, spoiled ballots are equally likely for all parties so there’s not much liklihood of them affecting the result.

  7. MinnesotaCentral says:

    First off, shouldn’t we be glad that we live in an open country where these types of ballots can be viewed online and discussed. When I looked there were over 90.000 voting in the poll … that’s great interest. And most of the poll results indicated a clear majority as to how the ballot should be treated … with probably less than 15% being “opinionated” party loyalists that were biased toward Coleman or Franken. The Canvassing Board should not have any trouble reaching a majority decision.

    What amazes me is that people did not advise the poll workers that they had a spoiled ballot and requested a new one. Why would they want to leave it to chance ?

    In the end, we should be glad that Minnesota has a system of paper ballots that can easily be reviewed.
    Sadly, we also know that whoever is determined to be the winner will serve knowing that 58% of the people thought that somebody else could do a better job.

  8. Beta says:

    “So… How does this work?”

    “Just take this ballot, fill in the little oval next to the candidate of your choice, and drop the ballot in the box there.”

    “What if I like both candidates?”

    “Then choose, and vote.”

    “But I can’t decide. How about if I fill in both ovals, but one darker…”

    “Fine, but it won’t be counted.”

    “WHAT? That’s not fair! You have to count my vote!”

    “Then you have to vote. If you don’t vote it doesn’t count. If you screw up a ballot a five-year-old could master, it doesn’t count. If you vote for the wrong person by mistake, that’s who gets your vote. If you make jokes and doodles, it might not get in the way you want it to. Want me to repeat the instructions?”

    “No, just promise me that if my intention is unclear you’ll dedicate plenty of time and effort to make sure you interpret my ballot correctly.”

    “No, I’ve got better things to do with my time. You’re supposed to dedicate a tiny amount of time and effort right now, and that’ll be enough to make your intention clear. Just fill in the–”

    “Yeah, yeah. Look, what if a lot of people mark their ballots unclearly? Will you at least be perfectly fair in the way you count them?”

    “If a lot of people screw up their ballots, or vote randomly, or stay home, then that’s what’ll happen. Just take care of your own ballot and let them take care of theirs. Fill in the little oval–”

    “Look, suppose I just put a little mark in both ovals like this. How do I know you won’t interpret that however you please? How do I know you won’t take all the borderline cases and count them for the candidate YOU prefer?”

    “All right, you got me. That’s exactly what I’m going to do. That’s why I’m here, to take all the spoiled votes and give them to the candidate you HATE. BWAH-HAH-HAH-HAH!!! And the only way you can stop me is by filling in the little oval next to the name of the candidate of your choice, and dropping the ballot in that box there. So you’re doomed! DOOOOOMED!”

    “I knew it! Well we’re going to win anyway, you hear me? And then Senator People is going to hear about this!”

  9. Anonymous says:

    I see your point about the ballots, then again I’ve always felt that was a potential. DREs without paper offer no system independent auditing. DREs with paper face problems with the security of the two but more importantly the basic problem that voters do not always read both. From a human factors standpoint they are quite bad. My favoritism for optical scanners hinges upon the fact that in this case people do examine the paper (because they are filling it out), they can obtain overvote/undervote warnings in advance, they can use ballot markers if they prefer, and the systems are cheaper.

    Additionally the auditing process with a optical scanner ballot is, in estimation, easier. The exiting register-receipt printers for DREs generate a long tape that is difficult to read (on the iVotronic it is simply a list of all actions with no summary of the vote) and prone to fading due to poor design. Moreover they must be read manually.

    With optical scanner ballots one can read them positionally, yes there are questions about the position but still it is easier to visually scan for a mark in bubble a or b than to read text. And one has the option of simply scanning them again.

  10. Anonymous says:

    I see your point about the ballots, then again I’ve always felt that was a potential. DREs without paper offer no system independent auditing. DREs with paper face problems with the security of the two but more importantly the basic problem that voters do not always read both. From a human factors standpoint they are quite bad. My favoritism for optical scanners hinges upon the fact that in this case people do examine the paper (because they are filling it out), they can obtain overvote/undervote warnings in advance, they can use ballot markers if they prefer, and the systems are cheaper.

    Additionally the auditing process with a optical scanner ballot is, in estimation, easier. The exiting register-receipt printers for DREs generate a long tape that is difficult to read (on the iVotronic it is simply a list of all actions with no summary of the vote) and prone to fading due to poor design. Moreover they must be read manually.

    With optical scanner ballots one can read them positionally, yes there are questions about the position but still it is easier to visually scan for a mark in bubble a or b than to read text. And one has the option of simply scanning them again.

  11. Anonymous says:

    Canadian ballots are not counted unless marked clearly and correctly. I think you can spoil a ballot and ask for a new one from the poll workers, but if you drop an improperly marked ballot in the box, you ended up not voting. Aside from the fact that a high number of spoiled ballots may suggest that the people are disgusted with both candidates, of course. :)

    P.S. what is the world coming to? It used to be debt consolidation and drugs, plus various shady deals and outright scams, but these days the number one spammer online seems to be Nike? For months, it’s been impossible to go anywhere near Usenet and not run into ads for Nike shoes, and now I see one right here on this blog.

    Yet another symptom of the economic meltdown?

  12. Steve R. says:

    With electronic voting you wouldn’t have this ambiguity!!! Of course no voting system is perfect.

  13. Hungerburg says:

    Wouldn’t it be enough to just have people look at those ballots, that the machine marked as invalid? does the machine produce false positives?

  14. Doug says:

    Wouldn’t it be reasonable to require an automatic runoff if the margin of error in an election is greater than the vote margin? IOW, to use your example, if Candidate A won by 73 votes, and the 95% confidence interval is ±281 votes, call it a statistical tie and hold a runoff.

  15. pfogg says:

    My father worked elections in the 60s and 70s (in a different state), and the rules for manual ballot counting would have explicitly rejected the first two cases. In fact, these were canonical examples of ‘what to reject’ given in the formal ballot-counting instructions (a partial mark counted as a full mark, and hand-written directions about which mark to count are ignored: double marking invalidates the vote for that specific contest, but the rest of the ballot counted). The need was for a set of easy-to-apply guidelines that would allow every ballot-counter to produce the same result as any other.

    I would argue that such simple, absolute rules are better for voters as well: each voter needs to be able to look at their ballot and be able to predict with great accuracy how it will be counted, so that they can tell whether an error (such as anyone can make) demands a new ballot, or if they can go ahead and submit it as it is.

    These rules also operate as a protection against vote-flipping — simultaneously subtracting a vote for an opponent and adding one for your own candidate by adding a note to one ballot. It seems unlikely that this was the motivation for the rules, though, since access to uncounted ballots would allow both the destruction and addition of ballots, which would accomplish the same thing.

  16. Carlos Gomez says:

    The use of optical scan from manually bubbled-in ballots is for the most part, a good solution. It provides a paper trail of actual ballots for recounts, yet still provides counting automation. In most cases, the margin of victory is large enough that ambiguous ballots do not affect the outcome. In the case where it is close enough to matter, the fact that there is a large portion of voters who didn’t cast a ballot means that however the determination on the recount, the person elected is close enough the choice of the people that it really doesn’t matter.

  17. PsiCop says:

    The hue and cry over “voter disenfranchisement” has been loud and vehement over the last 8 years (since the Florida recount debacle). The demand since then has been to ensure that “every vote is counted” and to discern “voter intent” under all circumstances, no matter how confusing a ballot might be. Any effort to discount any given ballot is interpreted as part of a plot by the Right to steal elections. This has effectively clouded the matter, making it a partisan one, when really, it should not be.

    Sorry, but I find it difficult to believe that the voter carries no responsibility for making his/her wishes clear, by following directions, not making stray marks, or adding gibberish to ballots. Mistakes happen, yes, but one can always request a new ballot if one has been botched somehow. It’s not that difficult. Really.

    Here in CT we have similar optical-scan ballots. On entering the polling place — before being handed a blank ballot to vote on — poll workers are available to answer questions. Where I voted, they had sample ballots and pens available, and there were more than enough people to explain the workings of what for many was a new balloting system (the 2004 election was decided on the old lever-machines; anyone who hadn’t voted since then would have been new to the optical ballots). There were accomodations for the blind, and telephone (audio-only) balloting for anyone who wished it. I find it impossible to see how any voter’s need was left unmet, or that any voter could possibly claim ignorance of how to vote in CT.

    The idea that any voter could claim — genuinely — not to understand the procedure or to be unclear on what constitutes a valid ballot, is laughable. Guidance and information were available in abundance, and could have been had merely by raising a hand.

    That said, I find it preposterous that anyone is actually wasting time trying to discern “voter intent” from ambiguous ballots which were not filled out according to the (simple!) directions given. The presumption is that voters are somehow not able to follow those rules … given how rudimenary they are, how much assistance is available for the asking, and that it’s possible to ask for a “do-over” ballot in case of a mistake, this is not a reasonable assumption to make.

    Unfortunately today’s political and ideological atmosphere make it next to impossible to suggest this overtly. Anyone who suggests disposing of a questionable ballot is often construed as being a partisan hack … but in fact this need not be the case.