Committee hears claims of inconsistent, missing and altered election data from Michigan Fair Elections Institute

At a House Committee on Election Integrity hearing, Braden Jacobasi and Tim Vetter of the Michigan Fair Elections Institute testified that multiple publicly available election data sets for Michigan contain inconsistencies, missing records and unexplained changes that they say undermine the ability to audit and validate election results.

The presenters told the committee they review local, county and statewide files (including the Qualified Voter File, or QVF) and found differences in precinct-level tabulator totals, voter lists and vote-history records. "There is no single source of truth," Jacobasi said, adding that files on the Secretary of State's website often combine or redact local precinct splits and that those changes can hide anomalies.

The issue matters because, the presenters argued, election systems must be auditable and allow independent researchers and officials to reconcile tabulator counts with voter lists. "Vote history records are like your bank's transaction records. According to state and federal law, they should never change, but they're changing all the time," Vetter said.

Jacobasi described four "pillars" of a world-class system: transparency and accessibility; standardization; validation/auditability; and accountability and security. He said data should be published in standardized, real-time, machine-readable formats, and that local tabulator files should preserve precinct "A/B" splits rather than being aggregated at the state level.

Using a sample from Livonia from the 2024 election, the presenters showed tabulator-level files they said differed from the Secretary of State's published precinct results. Jacobasi said the state version had combined some split precincts and had split candidate columns in ways that the local files did not, and that those transformations make it harder to detect or trace errors.

Vetter detailed six methods by which vote-history records appear to change after county certification: (1) records added after certification; (2) vote method switching (for example, from in-person to absentee) after certification; (3) records transferred from one voter ID to another; (4) vote-location jurisdiction fields changing so a vote appears to have been cast in a different county; (5) duplicate vote records appearing; and (6) vote-history records being removed. He said the statewide files do not consistently match local source data.

Presenters offered quantitative examples: a reported difference of roughly 1,500 votes between certified statewide totals and precinct-level totals for one election, an observed spike of roughly 20,000 duplicate votes on Oct. 28, 2024 (which they said the Secretary of State later removed and described as a "glitch"), and a previous primary in which they found about 6,000 duplicate votes. Vetter also stated that nearly 400,000 vote-history records from the 2020 election have been removed over time; he said some records were added after certification as well.

The presenters said they encountered delays and fees when requesting public files. Vetter said he was charged more than $600 and made to wait months for a FOIA response in one instance; Jacobasi said a 6-month delay occurred for a set of statewide files that the presenters expected to receive sooner.

Committee members asked the presenters about remedies and next steps. Representative Fox praised the testimony and asked about the magnitude of the error rate; he echoed the presenters' call for improved transparency. Representative Song said she trusts the election process and offered to work with the presenters offline to reconcile discrepancies in her special election; she noted some district and precinct boundary changes and asked how the work would be funded. Representative Wooden said he had checked some county clerk sites and county and state totals for certain special elections and found matching totals in those examples, and asked whether the presenters had referred evidence to law enforcement. The presenters said they have pursued civil litigation and made complaints to law enforcement and that some cases are pending.

The committee chair and other members said they wanted to explore solutions to reporting standards and data access. No formal committee action was taken on the presenters' recommendations during the hearing; members asked staff to review the testimony and continue oversight work.

Votes at the meeting included two procedural motions: Representative Fox moved to adopt the minutes from the May 20 meeting; the motion "prevails by unanimous consent" according to the transcript. Representative Fox also moved to excuse Representative Hoadley near adjournment; the committee approved the request and adjourned.

The presenters recommended: requiring local poll books and tabulator counts to be uploaded in standardized formats to the QVF at or before county certification; publishing machine-readable, unredacted versions of key files (with PII exclusions narrowly limited); providing upload receipts to local clerks; and commissioning an independent, broad technical audit to reconcile state and local files.

The committee hearing transcript records the presenters' claims, their examples and the questions from members. Committee members signaled interest in examining reporting standards and FOIA practices further; the hearing closed without substantive votes on policy changes.