Results of the 2024 Kosciusko County primary post-election audit are in and they show that local voters can trust their votes are counted and counted accurately.
The audit “concludes with 99% confidence that the election systems functioned properly resulting in correctly tabulated and reported election outcomes,” according to the results provided to the county this week from Voting System Technical Oversight Program (VSTOP).
Kosciusko County Clerk for the Circuit and Superior Courts Ann Torpy said, “First I want to say that I always had full confidence in the voter system in our county.” She said the audit was worth the time of the county, state and VSTOP because it erases the fears of voters who may hear things about the election. She volunteered the county to participate in the audit because she wanted to show voters that they can trust the results.
On June 5, the VSTOP team completed a post-election audit in Kosciusko County for the 2024 primary election.
VSTOP was authorized by the Indiana Secretary of State to conduct post-election audits. At this time, Indiana does not have mandated post-election audits, and participation is voluntary by the county, and designated by the Secretary of State, according to a memo from VSTOP to the county election board. Torpy volunteered for Kosciusko County to take part, formally requesting designation on April 25 via a County Election Board resolution. Kosciusko County was granted designation and audit prep began.
An initial audit meeting was held to review the post-election audit process, and answer any questions county election officials had about the process. From there, the audit date, time, location and selection of audited contests were coordinated by the county clerk and VSTOP and were finalized.
Using the Stark Audit Tool, VSTOP proceeded to run ballot sample size estimates on the selected contests based on the initial election night reports found on the county website. While these may not have been the final vote totals (pre-certification), they work appropriately for estimating expected sample sizes and approximating how long it would take to complete the audit, the memo states.
Kosciusko County was advised by VSTOP to select only Republican contests for audit due to the overwhelming margin of Republican to Democratic ballots cast, and the inability to separate primary ballots by party when using direct-record electronic (DRE) voting systems. For example, auditing a Democratic race may require a sample of 25 ballots. Since the ballots cannot be separated, and Republican votes held nearly 90% of the vote share, it can be expected the random sample would pull many more Republican ballots. It could take exorbitant sample sizes to reach the 25 Democratic ballots that were needed for sample because of the significant disparities in party vote share. While no Democratic contests were selected for audit, the integrity of ballots cast and machine accuracy can be confirmed because the Republican ballots being sampled are from machines used by both parties on Election Day, the memo states.
In consultation with VSTOP and county election officials, it was determined the post-election audit will be inspecting election day ballots, so audit estimations and ballot manifests/inventories were generated using only election day ballots and reported election day totals.
It was also determined this audit would be completed with a 1% risk limit and 99% confidence level. This means there is 99% confidence the audit would catch an outcome error if there was one, and only 1% risk the audit would not catch any outcome errors. The only way a 100% confidence level can be achieved is through a contest’s full recount, and an audit is not a recount, the memo explains. The 1% risk limit represents the lowest possible risk, and the 99% level represents the highest confidence level that can be achieved in Indiana’s post-election audit process.
The ballot manifest/inventory was generated by Kosciusko County’s voting system vendor, MicroVote, and included all votes cast on election day that were recorded via Cast Vote Record (CVR) with Voter-Verified Paper Audit Trails (VVPAT). One voting system used on election day had a VVPAT error where votes cast were digitally recorded but did not print on the VVPAT. The issue was reported and addressed by MicroVote. Based on a review of the issue, VSTOP has full confidence that all votes cast were counted and this system anomaly did not impact vote totals, election outcomes or the integrity of the post- election audit.
A full statement from MicroVote President Mike Miller, confirmed by Kosciusko County, reads: “As we were removing VVPAT rolls in preparation for the audit one roll had not (no) votes and no election header on it. Both test print buttons on the unit itself worked correctly. We then tested the VVPAT with combinations of the same voting machine and red communication cable that was in the booth with it and a different voting machine and communication cable and all combinations of components worked correctly in each test.
“There was an entry in the audit logs on the election day voting machine that was in the booth with the VVPAT that confirmed the voting machine was programmed for use with VVPAT. At least one entry showed that a VVPAT paper ballot was voided. This is also a possible indication the voter saw a paper ballot. One explanation is the VVPAT unit, or the paper roll, was changed after the last ballot cast on the machine but Ann (Torpy) and her team were confident that this didn’t happen.
“Therefore, we have no explanation for the cause of the issue. If this were an equipment malfunction, we have not seen this type of malfunction before where a voting machine programmed for use with VVPAT was not printing to a VVPAT yet there were no error messages and voters were allowed to cast ballots. Also, this seems unlikely as the poll workers would have ignored the message on the screen that asks them to verify a VVPAT election header printed when starting the election. Poll workers would have also had to not realize ballot headers weren’t printing for each voter. Finally, all of the voters would have had to ignore the instructions on the screen to verify their paper ballot before casting their vote.”
With that being said, the total VVPAT count for election day ballots that were eligible for audit was 9,066. VSTOP then checked the manifest for functionality in the Stark Audit Tool and confirmed it functioned properly. All pre-audit prep was complete and the audit was ready to be conducted.
The VSTOP audit team, led by VSTOP Election System Audit Specialist Matt Housley, arrived in Kosciusko County at approximately 10 a.m. June 5 and began audit set-up. As the VSTOP Team set-up, Housley had a pre-audit meeting for all county election officials participating in the audit. After the pre-audit meeting was concluded and the audit set-up was complete. Kosciusko County’s public, post-election audit officially began at 11 a.m.
VSTOP conducted a ballot comparison audit using the Stark Audit Tool built for this type of post-election audit. This audit type is meant to inspect ballots for a match between the CVR and VVPAT. Every ballot’s CVRs should match its VVPAT counterpart, resulting in a 100% match rate. After the required number of ballots (or more) have been sampled, the VSTOP team enters the match information into the Stark Audit Tool and as long as there are zero non-matches, the audit will be successfully completed.
The audit began with an introductory presentation given by Housley, Dr. Chad Kinsella and Dr. Jay Bagga to all persons in attendance. This presentation provided general information about the post-election audit process and methodology used, specific information relating to Kosciusko County’s audit, and allowed for a question-and-answer session to address any questions or concerns. After this presentation concluded, execution of the audit began. A 20-digit seed number was generated to assist the Stark Audit Tool in selecting the random sample of ballots. This number was generated live-time through 20 rolls of a 10 sided-die by audit participants and members of the public. The live-time generation of the seed number assures the sample is truly random and no-one could know what ballots were going to be selected for sample prior to the audit beginning. The seed number also becomes a control for this audit, so should the audit ever need to be replicated, it can be done so in its entirety.
The seed number was entered into the Stark Audit Tool and a random sample of ballots was generated. VSTOP drew an initial sample size of 150 ballots. All three contests were reviewed using ballots from this initial sample.
Upon completion of all 2024 primary post-election audits, a full summary report of all post-election audit activities and their respective results will be prepared and submitted to the Indiana Secretary of State and all counties involved in a 2024 Primary Post-Election Audit.