Pennsylvania’s post-2022 election audit, explained

This text is made doable by Highlight PA’s collaboration with Votebeat, a nonpartisan information group overlaying native election administration and voting. This text is out there for reprint beneath the phrases of Votebeat’s republishing coverage.

An individual approaches the desk and picks up a 10-sided die. She rolls.

4.

The following curler in line takes her flip.

Seven.

The bizarre die, formed like an elongated diamond, could seem a curiosity besides to those that use it most: “board avid gamers and elections officers,” Jonathan Marks, Pennsylvania’s deputy secretary of state, mentioned.

However the rollers, workers of the Pennsylvania Division of State, will not be casting their cube to find out their subsequent transfer in Dungeons & Dragons. As a substitute, they’re producing a protracted, random quantity that may decide the course of Pennsylvania’s 2022 risk-limiting audit.

A risk-limiting audit is a kind of post-election overview designed to offer statistical confidence that an election end result was correct. This 12 months is the primary when all 67 counties are required to take part in considered one of these audits earlier than certifying their election outcomes.

The maths used to conduct the audit is out there to the general public, although practitioners agree it’s arduous to grasp. Elections officers who’ve used it mentioned the audit took time to grasp however they now trust in it, they usually hope it would give the general public extra confidence as effectively that election outcomes are correct.

What’s a risk-limiting audit?

As the results of a settlement in a lawsuit introduced by Jill Stein, the 2016 Inexperienced Social gathering presidential candidate, the state was required to implement risk-limited audits statewide by 2022. However regardless of the contentious origin, many directors will now inform you they’re a fan of the method.

“On the core, it’s a take a look at to verify that the margin of a race was not overstated,” Marks mentioned. “What we like about risk-limiting audits from an election administrator perspective is … primary it’s environment friendly, and one of many different advantages is it’s adaptable. And that’s actually the massive key for me.”

Beneath Pennsylvania legislation, counties have lengthy been required after every election to carry out an audit of two% of all ballots or 2,000 ballots, whichever is fewer.

However that model of audit has its limitations. For instance, an election administrator may select to solely look at ballots from one precinct. Whereas that overview may uncover an issue with that particular precinct or the ballots solid there, it gained’t reveal a lot concerning the election as an entire.

A risk-limiting audit, alternatively, is particularly designed to test the margin of a race to make sure that the reported winner truly gained.

It’s also “adaptable.” If a race has a better margin, the algorithm requires that extra ballots be checked to make sure the end result was right. If the margin is wider, fewer ballots must be checked.

The two% audit can be solely a neighborhood audit and never a statewide one, which means every particular person county is performing its personal audit that has no bearing on different counties, whereas the risk-limiting audit is a statewide endeavor that appears at contests as an entire.

“I believe county elections officers prefer it as a result of it solely requires them to do as a lot work as essential to verify the end result of a contest, whereas with a set proportion you’re doing the identical quantity of labor no matter the info about that election,” Marks mentioned. “In a single county [2%] could also be greater than sufficient, and in one other county, it will not be sufficient.”

One fan is Jeff Greenburg, Mercer County’s former longtime elections director and now senior adviser to The Voter Mission, a Pennsylvania nonprofit centered on voting entry. Greenburg ran one of many state’s first risk-limiting audits in 2019 as a part of a pilot program.

“[Getting] a 95% confidence stage would possibly imply you solely must depend a number of thousand ballots throughout the state,” Greenburg mentioned of how the adaptive portion works. “Logically, it doesn’t make sense. That’s one of many questions I keep in mind asking the state again in 2019. ‘How do I clarify to individuals if I depend 2,000 ballots versus 200, how do I clarify to them that 200 ballots is definitely giving me a greater quantity than the two,000?’ And that’s the place the mathematics is available in.”

After going by the method, Greenburg acknowledged its worth, and the Division of State working group he was part of really helpful that the Pennsylvania legislature make the risk-limiting audit statutorily required. The legislature has but to take up that advice.

Solely 15 states at present carry out risk-limiting audits, and solely three of these statutorily mandate the method.

How does it work?

A random 20-digit quantity, referred to as a “seed,” is generated by rolling 20 separate ten-sided cube numbered zero to 9. This was carried out on a stay webcast by Division of State workers on Nov. 17.

At this level, uniquely numbered batches of ballots from all counties have been cataloged within the state’s risk-limiting audit software program. The seed is then entered into the risk-limiting audit algorithm to find out which ballots must be checked.

Mark Lindeman is coverage and technique director of Verified Voting, a nonpartisan elections know-how group that helps states together with Pennsylvania carry out risk-limiting audits. He mentioned whereas producing the seed quantity is totally random, as soon as that quantity is thought, anybody may double-check the work.

For the reason that algorithm is open-source, which means it’s out there to the general public, anybody may use that data and the seed to make sure that the batches the algorithm dictates be counted had been truly those counted. Lindeman mentioned this additionally prevents any administrator from realizing forward of time, and probably manipulating, which particular batch will probably be counted.

As soon as the batches are identified, counties start their a part of the method: a hand depend of the ballots in every designated batch to make sure the end result was correct. The variety of ballots in a batch varies by county, relying on how directors manage ballots.

“The audit math is tolerant of small discrepancies usually noticed in hand depend audits,” Lindeman mentioned, like if a poll is miscounted by a machine resulting from a stain on the paper. “Small errors within the machine depend would haven’t any bearing on the end result, however massive discrepancies that increase doubts concerning the end result may make it essential to conduct additional auditing or doubtlessly additional investigation, together with a full recount.”

This 12 months, the Division of State chosen the governor’s race for the audit. For the reason that margin of victory between Lawyer Basic Josh Shapiro and state Sen. Doug Mastriano was very vast — practically 15% — few batches needed to be checked to make sure accuracy.

Nonetheless, all counties should be ready to carry out the audit, since it’s unknown which batches will probably be chosen.

A spokesperson for the Pennsylvania Division of State mentioned that the seed dictated that 25 batches be chosen, which accounted for 10,209 ballots. These batches had been unfold throughout seven counties.

In Chester County, one of many state’s largest, elections administrator Karen Barsoum recruited six groups of three auditors every, with not less than one Democrat and one Republican on every staff.

Barsoum organizes her poll batches by precinct and vote kind (mail, in-person, and so on.) forward of the post-election interval to make sure a easy audit course of. The groups had seven completely different items of paper in entrance of them — every representing one of many selections within the gubernatorial race, together with write-in votes and situations the place voters chosen zero or multiple candidate.

“Initially, it’s unclear how lengthy it would final,” she mentioned of her system Friday morning, simply moments earlier than the audit was scheduled to begin. “It is determined by what number of batches the seed tells us to seize.”

For Chester County, it didn’t final very lengthy. No batches from the county had been chosen by the algorithm.

“We’re dissatisfied,” Barsoum mentioned with a considerably dejected however contently resigned smile. She had gave the impression to be desirous to implement her well-organized, color-coded system.

“We had been all able to go.”

The counties that had been chosen have till Friday to add their outcomes.

Marks, the deputy secretary of state, understands that the risk-limiting audit will take time for individuals to grasp, however he hopes it would present Pennsylivians with not less than some measure of better confidence within the course of.

“I’m not deluding myself into believing that doing [risk-limiting audits] is essentially going to persuade each skeptic,’’ he mentioned. “However I’m hopeful over time, as individuals see it work and folks see it in motion, that they’ll perceive it. That’s definitely my expertise.”

WHILE YOU’RE HERE… When you discovered one thing from this story, pay it ahead and turn out to be a member of Highlight PA so another person can sooner or later at spotlightpa.org/donate. Highlight PA is funded by foundations and readers such as you who’re dedicated to accountability journalism that will get outcomes.