Press "Enter" to skip to content

IM 23 Petition Error Rate Inflated by Secretary of State’s Counting Rules

Don’t knock my house down yet, bulldozer drivers! I spoke too soon about the quality of work done by Initiated Measure 23’s petition circulators.

Secretary of State Shantel Krebs calculated that the sponsors of the fair-share union dues petition, the International Union of Operating Engineers Local 49, had submitted 30,810 signatures and that only 14,861 were valid. That 51.77% error rate far exceeded anything we’ve seen this year or in recent petition years. However, when the heavy equipment operators submitted their petition, they said they had 20,000 signatures. How did the signature count balloon from 20,000 to 30,810?

It appears the problem is not the circulators but the administrative rules controlling how the Secretary of State counts petition signatures. Recall the signature-counting methodology Secretary Krebs explained last June when we submitted out referral petitions. At no point does the Secretary of State count every signature on a ballot question petition. Instead, the Secretary’s team follows the rules spelled out in South Dakota Administrative Rule 5:02:08:00.05 looks at the last filled-in signature line on each petition sheet: if that happens to be line #17, the Secretary’s team counts that sheet as having 17 signatures, even if some of the preceding lines on that sheet are blank or crossed out. If the Secretary’s 5% random sample calls up line #8 from that sheet, and that line happens to be blank, that counts as an invalid signature.

IM 23 Fair-Share Petition
There are two sides to every story of petition signature overcounts….

The fair-share union dues petition had signature lines on both sides of its sheets (see image at right), with signature lines #1–10 on the front and signature lines #11–25 on the back. Secretary Krebs tells Dakota Free Press that a majority of the IM23 petition sheets had blank fronts and signatures starting on line #11 on the backside. I’m just spitballing here, but if half of the sheets submitted by Local 49 (1,824, by Secretary Krebs’s count) had blank fronts, the Secretary would have included over 9,000 blank lines in her random sample. Including those obvious blanks in the random sample, blanks that the petition sponsors likely did not include in their signature count, inflates the error rate.

Speculating from the Secretary’s guess that a majority of the sheets have blank fronts, I estimate the real error rate for IM 23 circulators was more like 30%, not the 52% produced by Secretary Krebs’s rule-guided calculations.

But note this mathematical oddity: Secretary Krebs’s overcount doesn’t affect the calculated number of valid signatures. I can show you the algebra if you want, but trust me: if the 5% sample draws from the obvious blanks and actual signatures proportionately, the final number doesn’t change. So anti-labor Republicans, if you’re eyeing that thin 7% cushion for the IM 23 petition and thinking challenge, the overcount here does not appear to affect your calculus.

Secretary Krebs says some other petitions using sheets with signature lines on two sides exhibited the same phenomenon of blank fronts, but the highest overcounts among the petitions certified so far are 6.9% on Amendment T (the independent redistricting commission proposal) and 6.2% on Amendment U (the fake 18% rate cap). The Secretary’s overcount from the sponsors’ claimed signatures on IM 23 is 54.1%. Recognizing that including those obviously blank lines (a) is absurd and (b) produces deceptively inflated error rates, Secretary Krebs tells Dakota Free Press that she will ask the Board of Elections to amend SDAR 5:02:08:00.05 to allow a more statistically accurate method of sampling signatures on ballot measure petitions.


  1. grudznick 2016-01-19 18:03

    Do you think this means union members can’t read instructions or union members are trying to cheat on the signings?

    If this passes, I will be billing all of the heinous unions for the services I will be providing to them, whether they want or agree with them or not.

  2. caheidelberger Post author | 2016-01-19 18:53

    No, Grudz, the Secretary’s overcount does not indicate any failure to read instructions or intent to cheat by these circulators. Where do you get this stuff?

  3. leslie 2016-01-19 19:47

    Cory is “where do you get this stuff?” representative of 10%, 33%, 51%, 75% or 99% inaccuracy for Grudz… just spit balling?

  4. grudznick 2016-01-19 23:23

    I think it would be good to show us the Algebra, Mr. H. People like me could learn something. But goodness, if the union people immediately flip the page and start writing or had no consistent instructions to their paper hander-outers, what does that tell you?

    It tells me that unions are bad, and if this passes I will hold a class for them on paper-handing outing and then bill them for my services whether they attended or even wanted them. For I will have been providing a service to benefit current and future union members.

    May the public mock this law in the elections, and may the courts laugh it out of style.

  5. leslie 2016-01-20 00:22

    “Sumpthin fer nuthin grudz”

  6. mike from iowa 2016-01-20 05:30

    Algebra sounds Arabic,Grudz. What are you-a closet Muslim?

  7. caheidelberger Post author | 2016-01-20 09:47

    Algebra is one of my favorite Arabic pastimes. Grudz, here you go:

    signature validation algebra, CAH, 2016.01.19

    And you’re really stretching for an anti-union argument here, Grudz. Starting signatures on line #11 violates no rule and in no way fouls the petition. At worst, it wastes paper, but that marginal loss may be made up for in time-savings on the circulation trail.

  8. Troy 2016-01-20 11:57


    I don’t have time to check your envelope computation but I read through your explanation and did a cursory review of your algebra and both appear to support the results you posit. I love statistics used for any reason. Nice job.

    Other readers,

    I did not make the above statement because I think CH doesn’t know his Math but to provide a third party confirmation it isn’t voodoo math. Cory and I are used to non-math minds too often reacting to math with excessive skepticism not warranted by the statistics. :)

    If you are inclined to look for a weakness (I agree with CH being off on his estimate is likely to likely enhance the cushion vs. invalidate the petition because I think his estimate is likely liberal- which would get a more conservative result*) is CH made a estimate on the number of blank first pages. If that estimate (variable) is wrong, his result (not his algebra/computation) could be wrong.

    * enjoying the use of these terms in this context.

  9. caheidelberger Post author | 2016-01-21 11:43

    Thank you, Troy, for your love of math and wordplay.

Leave a Reply

Your email address will not be published.