Jackley Breaks Krebs’s Random-Sample Table; Rule Fix Needed!

Marty Jackley broke the petition-sampling table, so Shantel Krebs had to order a new one!

Last Monday, our aspiringly gubernatorial Attorney General submitted his nominating petition. As a show of organizational force, Jackley submitted 8,264 signatures, 4.2 times the 1,955 Republicans need to run for statewide office. (Democrat Billie Sutton submitted 2,422 signatures, 3.4 times the 706 required of Democrats. Kristi Noem appears not to have tried making hay out of her signature count.)

As I reported last week, Jackley’s showing off created a problem for the Secretary of State’s office. For the first time, thanks to a new law passed in 2017, the Secretary of State certified statewide candidate petitions by using statistically rigorous random sampling rather than by counting and verifying all of signatures required. Statute directs that the Secretary sample “a number of signatures that is statistically correlative to not less than ninety-five percent level of confidence with a margin of error equal to not more than three and sixty-two one-hundredths percent” (SDCL 12-1-36; identical language is applied to ballot question petitions in SDCL 2-1-16). To avoid having to do math, Secretary of State Shantel Krebs recruited Northern State University statistician Sara Schmidt to run the numbers and make a table telling how many signatures to sample for various size petitions. The Board of Elections approved that table and wrote it into Administrative Rule 05:02:08:00.05.

Alas, the table only went from 250 signatures to 5,000, because, really, what knucklehead would submit more than 5,000 signatures when he only needs 1,955 (or, if he were an independent, 2,775)?

…best-laid plans…

But Secretary Krebs didn’t let that lack of foresight hold her down. She called statistician Schmidt and ordered a new table to cover sample sizes from 5,000 up to 20,000:

Attentive election nerds will notice that the “safe harbor” figures for signatures from 14,000 to 20,000 differ slightly from those listed in ARSD 05:02:08:00.05 for ballot question petitions. This difference likely arises because the existing ballot question signature table has increments of 1,000 while this ad hoc nominating petition table steps up by 100. Finer granularity means smaller safe-harbor round-ups.

But those increments expose a flaw in the wording of the existing rule. Read Part 8(b):

If the number of submitted signatures is between two numbers listed in the signatures submitted column, the Secretary of State shall round down to the lower number if the number is 500 or less and shall round up if the number is 501 and higher [ARSD 0:02:08:00.05, Part 8(b)].

This awkwardly written rule directs the Secretary to select a row on the table by rounding signatures to the nearest thousand. But both the existing nominating petition table and the above ad hoc table step up by 100. Rounding Jackley’s 8,264 to the nearest thousand would have meant sampling 762 signatures; rounding to the nearest hundred would have called for sampling 764. (My straight calculation from the formula, with no safe harbor, says the sample size should be 673.)

Now Marty’s primary challenger Kristi Noem could challenge Jackley’s petition (Krebs certified Jackley’s petition last week Monday, and statute gives five business days to challenge a nominating petition, so today’s the deadline for an in-office, non-court challenge) by saying that the Secretary certified the petition outside of existing rules. Such a challenge would likely fail: statute clearly tells the Secretary of State how to randomly sample any statewide petition; the failure of rule to translate that directive into practical spreadsheet guidance for every situation doesn’t prevent or invalidate the Secretary’s action in an extreme case.

But for the sake of completeness, the Board of Elections should add the above table to the rules. We should also rewrite Part 8(b) into four parts, the first to cover ballot question petitions, the second to cover nominating petitions, and the third and fourth to cover petitions exceeding the tabulated sizes:

(b) If the number of submitted signatures on a ballot question petition is between two numbers listed in the signatures submitted column on the statewide ballot question table, the Secretary of State shall round the number of submitted signatures to the nearest thousand.

(c) If the number of submitted signatures on a statewide candidate petition is between two numbers listed in the signatures submitted column on the statewide candidate table, the Secretary of State shall round the number of submitted signatures to the nearest hundred.

(d) If the number of submitted signatures on a statewide candidate petition exceeds the maximum number of signatures listed in the statewide candidate table, the Secretary of State shall apply the sampling table and rules for statewide ballot question petitions.

(e) If the number of submitted signatures on any statewide petition exceeds the maximum number of signatures listed in the statewide ballot question table, the Secretary apply the statistical formula specified in statute to determine the size of the random sample.

Amending the table and the rule discrepancy will go unnoticed by 99.99% of South Dakotans. It will have zero impact on who makes the ballot. But rules are rules, and they should be complete and correct. Bulk up that table and tighten those screws, in case Marty comes crashing in again with some extra-large petition!