The Measure 67 Undervote

Cody Hoesly

I write this post out of curiosity.  According to the Oregonian's numbers as of now (over 97% reporting), Measure 66 received about 10,000 more Yes votes than did Measure 67.  Also, Measure 66 received about 7,500 fewer No votes than did Measure 67.  That means Measure 66 received approximately 2,500 more total votes than did Measure 67.  What explains this?

I assume most folks voted either Yes on both measures or No on both measures.  I know there was a significant movement urging a Yes vote on Measure 66 and a No vote on Measure 67, but I am not aware of any significant movement urging the opposite.  Taking the numbers at face value then, it would seem that 2,500 people voted Yes on Measure 66 without voting at all on Measure 67.  If you did that, or otherwise voted on only one of the measures, please comment below and say what your thinking was.

There are still some more votes to count, but presumably the ballots that have been tallied so far include both votes on each ballot.  Also, while it is possible that failed votes account for some disparity, it would seem equally likely that a bubble got improperly filled in for Measure 66 as for Measure 67, or that a scanner improperly read a vote for either measure.  But do any of you elections office gurus out there have any explanation?

I know I'm quibbling over something like .2% of the total vote, but that just happens to be the threshold for when an automatic recount kicks in, a la 2008's Measure 53 (for those who care/remember).  There'll be no recount here, but I'm curious whether this disparity results from voter choice, voter error, vote counting error, or some other cause.  Perhaps the answer could be useful in a more closely contested election.

  • Kienan Wear (unverified)
    (Show?)

    I think the most simplest explanation is that the ballots were double-sided, and some voters may have missed that.

  • Dumb Elections Decisions (unverified)
    (Show?)

    Just guessing, but double-sided ballots. With only two measures on the ballot, it makes no sense to put one on the front and one on the back, but that's exactly what some counties did. At least some voters presumably didn't realize there was a back side.

  • (Show?)

    The Multnomah County ballot wasn't double-sided. Which counties had double-sided ballots? If we can build a list, we can test that hypothesis.

  • John Silvertooth (unverified)
    (Show?)

    Kienan Wear: I think you will find that the ballots differ from county to county- our ballot in Wasco Co. were one sided.

    Cody Hoesly: That's exactly the rate at which No. 2 pencils break.

    McCarthy always used to say you could get at least 3% by people that mark the ballot wrong.

  • Kienan Wear (unverified)
    (Show?)

    The ballots were double-sided in Lane County.

  • Ricky (unverified)
    (Show?)

    @Cody

    Electronic vote counting machines have an error rate up to 5% on the first pass. Second pass of rejects cuts it down to less than 2%. Maybe some counties didn't run a second pass on undervotes, but if true that if ballots are different from county to county, that is kind of amazing and should be corrected to be uniform.

  • Christopher Anglin (unverified)
    (Show?)

    As soon as the ballots started coming in here (Lane County), announcements began to be made that voters needed to vote on BOTH sides of the ballot.

  • (Show?)

    Fascinating question. Thanks, Cody.

    You gotta look at the actual undervote, not the gap between 66 and 67 (though it doesn't invalidate your question at all). You can get the total ballots cast numbers at the Secretary of State site.

    Statewide undervote (technically, overvote+undervote) for M66 was 0.4%. Statewide undervote for M67 was 0.2% -- a gap of 0.2%.

    There was, interestingly, some tremendous variation between the counties.

    Yamhill had the biggest gap - with 349 fewer votes on 67 than 66, or about 1.1% of the 31,501 ballots. Malheur had 64 fewer votes on 67 than 66, or about 0.9% of the 6847 ballots. Lane had 991 fewer votes on 67 than 66, or about 0.8% of the 126,070 ballots.

    At the other end of the spectrum: Curry had 76 fewer votes on 66 than 67 - or 0.84% of the 9066 ballots.

    Squarely in the middle - with the least variation - were Jackson, Polk, and Tillamook counties. Over a combined 35,754 ballots, the gap between the undervotes was a total of four votes.

    Might be an interesting project to track down copies of all the ballots - and compare to these numbers. Did Yamhill, Malheur, and Lane all have double-sided ballots? Did Curry do something else different? Did Jackson, Polk, and Tillamook do something especially intuitive?

    And can someone explain to all of us why the ballots would look different county to county?

    I've posted all the data into this public Google spreadsheet.

  • Kienan Wear (unverified)
    (Show?)

    That (double-sided ballot) would explain why their were so many undervotes in Lane County. I think the fact that different counties have different looking ballots is confusing for the system, as well voters. This is a great question to ask Kate Brown.

  • (Show?)

    from talking to many undecided voters (and listening to their concerns elsewhere), a fair number of people couldn't really decide if 67 was good or bad. 66 tended to be a bit easier to vote Yes on, but the (false) issues raised by the No campaign had an effect. i think that's why you had more vote on 66 than 67: some people never could decide what to do on the latter, so did not vote on it.

  • alcatross (unverified)
    (Show?)

    Kari Chisholm: And can someone explain to all of us why the ballots would look different county to county?

    Vote counting machines/apparatus may be physically different from county-to-county - if so, that could be a reason (e.g., some can handle double-sided ballots, some not - just like differences in regular copying/scanning machines...)

  • SeymourGlass (unverified)
    (Show?)

    Ricky: do you have some actual, verifiable sources to back up your claim of an error rate of "up to 5% on the first pass". And how does it drop to 2% on the second pass? In other words, explain your definition of "error". Thanks.

  • Ricky (unverified)
    (Show?)

    If you look at the specs on the machines, the error rate is 1 per 10,000 votes (this is also a federal requirement). That error is strictly a failure to read rate. In actual usage, the error rate is much higher, particularly when ballots are folded. The error rate also increases depending on the number of ovals per page. I have seen up to a 16% rate in various studies. A lot of counties have data, all over the net, from studies done. I could have misquoted the second pass rate, I will have to recheck some of the studies. Google "ballot counting error rate" and take your pick of the many studies out there.

  • SeymourGlass (unverified)
    (Show?)

    Ricky: so, how do YOU define "error" insofar as ballot counting.

  • (Show?)

    I am surprised to find the gap this close. In our predictions piece, I guessed that 66 would pass and 67 would fail. I was quite surprised to see how sticky they were. People really considered them a single issue, it appears.

    In addition to the double-sided thing, I could easily imagine someone voting on one and not the other.

  • Rangerhunter (unverified)
    (Show?)

    Each of the top four undervote counties listed by Kari was a county with a two-sided ballot.

  • marci (unverified)
    (Show?)

    Tillamook was a single sided ballot.

  • Edward I. O'Hannity (unverified)
    (Show?)

    Two-sided in Marion...

    ~EIO

  • (Show?)

    Fascinating. Thanks for the data, Kari.

    Does anyone know why an identical set of tax measures with no other candidate races would appear on a two sided ballot in some counties and on a one sided ballot in others?

    This is a recipe for problems. Anytime voters have to turn the ballot, undervotes increase. I can understand when there are a lot of races, but what was the rationale here? Different counting machines? Local choices? Something else?

  • 8secondblock (unverified)
    (Show?)

    I know my wife almost did not vote for 67 by accident. She was in a rush to fill out her ballot (7pm on election night)and just filled in the circle on 66. She thought they a yes vote would count for them both, since they have been linked together in advertising and whenever they are mentioned it is always 66&67. So if she missed this i am sure there were a few other people in the state that also made the same mistake

  • Kreditvergleich (unverified)
    (Show?)

    Sorry, what menas undervote?

  • Kathy Jackson (unverified)
    (Show?)
    <h2>Oregon counties have equipment from different vendors. Curry and Yamhill have new digital scanners from Hart InterCivic. Marion and Lane have scanners that use older infrared technology that is known to have difficulty detecting certain commonly used inks and therefore will show a higher error rate. The rest of Oregon's counties use equipment from ES&S, mostly the M-650 model. The undervote in different counties could be due to a combination of factors, including the humidity of the paper used in the ballots, the difference in the error rate for the technology used and problems with certain inks. Add in voter error and there are lots of reason why there might be different undervote rates in different Oregon counties.</h2>

connect with blueoregon