by Carter Walker of Votebeat
Photo courtesy of Leise Hook / For Spotlight PA and Votebeat
This article is made possible through Spotlight PA’s collaboration with Votebeat, a nonpartisan news organization covering local election administration and voting. Sign up for Votebeat’s free newsletters here.
On Jan. 6, 2021, as former President Donald Trump rallied his supporters, he used a statistic that, though false, was making the rounds: “In Pennsylvania, you had 205,000 more votes than you had voters,” he screamed, throwing his arms wide open in front of thousands of angry followers. “This is a mathematical impossibility unless you want to say it’s a total fraud.”
The number appears to be the work of Heather Honey, a Pennsylvania-based “election integrity” investigator whose research has achieved a remarkable level of national salience among the far right, despite being replete with errors. The 205,000 figure, for example, is “false” according to the Department of Justice, and was based on incomplete data the state says can’t be used for this type of analysis. Honey herself has revised the discrepancy downward. While Honey’s current estimate is almost half of what it once was, it’s still inaccurate and the original number is also still routinely cited as fact.
“There were 202,377 MORE ‘votes’ cast, than actual REAL VOTERS THAT VOTED,” reads a November 2023 post on X from a popular rightwing account. It was reposted more than 13,000 times.
Honey has been among the most effective advocates for right-wing election talking points. Time after time, her research has fed into viral allegations about election integrity, fueling conservative pressure campaigns, forcing fact-checkers and public officials to attempt to piece together a more accurate picture and undermining confidence in long-trusted election practices. Often, her conclusions are misleading or based on incomplete information.
In the past year, working with a network of election integrity groups organized by conservative lawyer Cleta Mitchell, Honey has had perhaps the most success with her latest research: a 29-page report on the Electronic Registration Information Center, or ERIC, sent to Republican secretaries of state and legislators. Her information appears to have influenced the decision of several member states to withdraw from ERIC, an interstate program that election officials widely regard as the nation’s best tool to keep voter rolls up-to-date and free of bloat.
An analysis by Votebeat and Spotlight PA found the report’s conclusions are false, often based on out-of-context examples, and that her sweeping generalizations are frequently not backed by the data she presents. Presented with Votebeat’s and Spotlight PA’s findings — the product of redoing her many calculations and fact-checking the analysis she has offered to public officials — Honey defended her work.
Despite how widespread her research has become, Honey described herself in emails to Votebeat and Spotlight PA as “an ordinary citizen working hard to do what I can to restore confidence in elections.”
When asked for a final comment, she accused Votebeat of “slander” and personal attacks. “Who is pressuring you to write this hit piece? What is your goal?”
The way that Honey relies on real-but-incomplete data is a hallmark of those who spread misinformation, experts say.
“The fundamental misconception people have about misinformation is that misinformation is about the facts in front of you,” said Mike Caulfield, a research scientist at the University of Washington’s Center for an Informed Public who studies mis- and disinformation. Instead, Caulfield said, popular misinformation often relies on leaving out crucial details.
Calling it “misrepresented evidence,” Caulfield likened Honey’s report to a prosecutor accurately telling a jury that they discovered a murder weapon in a suspect’s possession but failing to mention that the fingerprints on it belonged to someone else.
“You can’t look at that and say, ‘The knife is real, we found the knife,’ but not look at it and mention you found fingerprints that would undermine the importance of the evidence,” he said.
Activated by 2020
As Honey tells her own origin story, she was standing in line at her polling place in Lebanon County, Pennsylvania, on Nov. 3, 2020, behind an older couple when her journey to election integrity investigator began.
“It really kind of struck me that, you know, this woman who just waited in line for an hour was told, ‘Oh, you have to cast a provisional ballot,’” Honey said on Mitchell’s “Who’s Counting” podcast in May 2022. She told Mitchell she wondered if the provisional would be rejected. “So it just got me a little bit worked up and I went home and I started doing, I mean, what I do.”
Honey has operated Haystack Investigations — a private investigations and supply-chain auditing consultancy — since 2017, her LinkedIn profile shows. Her website offers services ranging from supply chain audits to social media investigations.
Using what she’s described as “open-source investigation” skills, Honey began researching Pennsylvania’s election laws and requesting data from the Pennsylvania Department of State in November 2020. Soon after, she decided to reach out to her state representative, Frank Ryan, a Republican, with her findings.
“I said, ‘Look, here are the things that I found,’” she told Mitchell. One of those things was “that there were more ballots reported than what we found in the voter files.”
Honey told Mitchell that Ryan gathered other representatives together to show them her work — that is, a calculation purporting to show a discrepancy between the number of ballots cast in the 2020 election and the number of people the Pennsylvania roll recorded as having voted.
Ryan did not respond to an interview request, though Mitchell’s version closely mirrors his public telling of events, and emails — on which Honey is included — show Ryan passed this information onto U.S. Rep. Scott Perry, a Republican who was in close contact with Trump at the time. According to the final report of the Jan. 6 committee, Trump would repeat the claim that Pennsylvania had more votes than voters numerous times, which was part of the Department of Justice indictment against him alleging election interference.
Honey told Votebeat and Spotlight PA that Ryan was receiving information from multiple sources at the time, but she did not directly dispute that Trump’s figures originated with her work.
Honey’s latest calculation of the discrepancy, revised after straggling counties uploaded fresher records, sits at around 121,000. Trump cited that number, and Honey’s work directly, in a document he released in January claiming fraud in swing states in 2020.
Still, election administrators in Pennsylvania say this number is based on a flawed analysis.
Voter roll data is constantly updated to account for moves, deaths, and other issues of eligibility, which stops in the weeks just before an election and resumes again immediately after, the Department of State and county election directors explained. The roll also does not contain information about some voters whose identity must legally be kept confidential, such as victims of abuse. In other words, the state’s voter roll at any one point in time doesn’t reflect a complete record of who voted in the last election. But Honey’s analysis treats the roll as if it does.
For example, if a voter in Philadelphia moved to New Jersey the day after the 2020 election, the Philadelphia Board of Elections would cancel their registration. And even though they legitimately cast a ballot in Philadelphia on Election Day, their voter history would not appear in the voter roll system after that because they have been removed from the rolls.
Given these limitations, the Department of State told Votebeat and Spotlight PA that Honey’s method of analysis is “not an accurate way to reconcile votes cast to the number of voters who participated in any individual election” and that the system “was not intended” for this purpose. The state pushed back on the figure at the time, but that didn’t stop it from spreading.
As Honey’s research gained traction, other election-integrity advocates began to take notice and solicit her for work. She was paid as a subcontractor in Arizona for Cyber Ninjas, the company hired by that state’s Senate to investigate the 2020 election in Maricopa County. That investigation ultimately did not prove any fraud. Documents from it — obtained by American Oversight and the Arizona Republic — show she was a “manager” and billed tens of thousands of dollars for her work, though it is unclear how much she was eventually paid.
Cyber Ninjas subsequently went out of business, but Honey’s work continued.
Days before the 2022 midterm elections, Honey released a report — through her election integrity research organization, Verity Vote — claiming nearly a quarter of a million ballots in Pennsylvania had been sent to “unverified” voters. Ryan again picked up the information in a letter to the Department of State. It was then echoed by Trump, and the conservative website the Gateway Pundit pointed to it as evidence that the election was fraudulent and should be decertified.
The report “flagrantly misrepresents” how the system works, the Department of State — which rebutted it after it gained traction online — said at the time. Voters were labeled as “NV” or “not verified” in the state’s voter management system, but as the Associated Press reported, the label is for internal workflow purposes and typically means the identification provided by the voter is still being verified by county workers so their ballots can be counted.
A more refined ERIC criticism
In 2022, as ERIC became the subject of right-wing fury, Honey began her own research.
The program was formed in 2012 as a project at the Pew Research Center before becoming an independent nonprofit. It continues to be governed by member states, and uses state voter rolls, death records, motor vehicle records, and other information to cross-check state voter rolls for accuracy. It also identifies voters who may be eligible to vote but are unregistered — referred to as EBUs — and requires states to reach out to those potential voters about registering.
ERIC does not add or remove any voters itself. It only provides lists of voters to states, and states and counties then verify the accuracy of ERIC information before they remove voters from the rolls. Voters who register in response to the outreach must meet all voter registration requirements.
By June 2022, ERIC had grown to include 33 states and the District of Columbia. Prior to conservative attacks — including those fueled by Honey’s research — it drew bipartisan praise from officials for its ability to help clean voter rolls, including from some of those same officials who would take up Honey’s talking points.
In January 2022, an article from the right-wing outlet the Gateway Pundit accused ERIC of being “a left-wing voter registration drive disguised as voter roll clean up.” Louisiana soon after announced it was suspending its membership. Officials told Votebeat at the time the decision was unrelated to the coverage but that “numerous” experts on “election stuff” had advised the state to leave ERIC, office spokesperson John Tobler said.
Louisiana would be the only state to part with the program for several months after the Gateway Pundit’s story was published. In the interim, Honey released her report, offering apparent backing to the charges laid out by the site.
While it reached many of the same conclusions as the Gateway Pundit, Honey’s report offered a more professionalized critique of ERIC with historical research, original first-hand documentation, and data analysis. It appears to have contributed to Virginia, Texas, Missouri, and Louisiana withdrawing from the compact, as well as North Carolina’s decision to halt its process of joining.
Flaws and omissions in Honey’s ERIC report
In June 2022, Mitchell — a conservative attorney who represented Trump in Georgia in 2020 — hosted a conference in Washington, D.C. There, Honey gave a presentation on ERIC to several secretaries of state or their representatives, according to information obtained by Documented, a D.C.-based investigative news outlet, and provided to Votebeat and Spotlight PA.
Mitchell did not respond to an emailed request for comment.
At the same time, Honey released her ERIC report on her website.
The report repeatedly asserts misleading claims. For example, she writes, “After 10 years of ERIC, there is no evidence that it has led to an improvement in accuracy or clean voter rolls.” She attempts to prove that by comparing the number of voter roll removals in ERIC states versus non-ERIC states. That comparison, however, relies entirely on a single metric — the number of removals in one category released in one year.
She claims that her research — using data released biannually by the U.S. Election Assistance Commission (EAC) — shows that ERIC states removed proportionally fewer voters who had moved out of their voting districts than non-ERIC states had in the same time period.
Charles Stewart, a professor of political science at MIT who studies elections and works with EAC data, says the data shouldn’t be used in this way. It is often unreliable because it contains gaps and basic mathematical errors, and sometimes doesn’t match what states independently report. Stewart is a member of ERIC’s Research Advisory Board, which advises the organization on how it can measure its performance.
He also disagreed with the premise of Honey’s analysis, since ERIC is not responsible for removing any voter from the voter rolls and ERIC member states are “at the mercy” of local jurisdictions like counties that actually act on the data ERIC provides.
But the data can provide a rough measure of ERIC’s performance. Using data available to Honey at the time of her report, Votebeat and Spotlight PA repeated her analysis and found that the conclusion of her report is misleading. She appears to have chosen the singular data point that supported her claim, out of a series of voter-removal metrics in the same data set. And, despite the availability of multiple years of such data, Honey used only one — 2020, an outlier year. Taken together, Honey left out 11 of 12 relevant data points in her report.
Votebeat’s and Spotlight PA’s analysis, which used all 12 data points available at the time her report was published, suggests exactly the opposite: ERIC states remove a higher percentage of voters from the rolls than non-ERIC states. Data from a more recent year — 2022 — also supports that conclusion. (Read more about our analysis below the story.)
Asked to explain her selective use of data, Honey did not respond to some questions and offered answers to others that contradict the logic of her analysis. For example, asked to explain why she only used a single metric from one year, Honey said in an email that it would be inappropriate to analyze one of the other categories — removal of voters who died — in state-to-state comparisons, since some states, like Pennsylvania, do not use ERIC’s death data. But Honey’s analysis has the same flaw, as not all ERIC member states use ERIC’s moved-voter data uniformly, either.
Honey did not address why she only analyzed the EAC’s data from one year, when four were available at the time she published her report.
Elsewhere in the report, Honey claims that the outreach to eligible but unregistered residents, or EBUs, that ERIC requires member states to conduct “results in significant swelling of voter rolls,” and that “EBU additions consistently exceed suggested removals by ten times.”
But available data doesn’t support Honey’s conclusions.
“The report is obsessed with EBUs,” Stewart said. “I just take, overall, the report as part of building the case that ERIC is this left-wing organization that is trying to get Democrats onto the voter rolls.”
To conclude additions exceed removals by “ten times,” Honey shows a graph of data from ERIC’s website. The wording of the title of the graph calls these “additions and removals instigated by ERIC participation,” as though all of the numbers displayed are actual voters removed from or added to rolls. But ERIC only supplies information about voters who may need to be removed or are eligible to register to vote — state and county election offices determine which are valid and make any changes themselves. In the case of eligible but unregistered voters, the state simply sends a mailer to the identified person informing them of their eligibility, and the voter must take action from there.
Honey acknowledges elsewhere in the report that removal and eligible-voter data ERIC provides to states are merely “suggest[ions].”
Her graph also leaves out two other categories of ERIC’s list maintenance recommendations to states. In reality, they’ve recommended roughly 41 million records be updated and identified roughly 57 million voters who may be eligible.
Votebeat and Spotlight PA performed a more accurate analysis of ERIC’s effect on voter roll additions, using data showing net changes to voter rolls, rather than the number of people who were sent mail.
The EAC’s data — which Honey uses elsewhere in her report, though not for this analysis — shows that there is typically less than 1 percentage point of difference between ERIC and non-ERIC states when it comes to registering new voters.
This figure aligns with academic research on the topic. A study of Pennsylvania’s ERIC mailings, published in 2020, found they resulted in only a 1 percentage point increase in registration.
Elsewhere, Honey’s report relies on real data to make inaccurate conclusions or omits relevant and readily available context.
For instance, she accurately states that Pennsylvania’s and Michigan’s secretaries of state received voter outreach grants from nonprofits dispersing money from Facebook founder Mark Zuckerberg and his wife, Priscilla Chan. Honey then incorrectly claims the grants were given to those states “in order to gain access to data needed to inflate the Democrat voter rolls and drive Democrat turnout.”
Honey did not respond to a request to provide evidence of that claim.
The portion of Pennsylvania voters registered as Democrats went down in 2020, as it has every year for more than a decade. Michigan voters do not register by party, so there are no “Democrat voter rolls” there. And a recent study from data scientists at the University of California Los Angeles found that the areas that received grants funded by Zuckerberg saw less than a 0.13 percentage point increase in voter turnout — ultimately significantly fewer votes than would be needed to swing the election even if they all had voted for Democratic candidate Joe Biden.
Honey also leaves out important context. For example, she quotes a 2021 Wisconsin Legislative Audit Bureau report which found many duplicate and erroneous voter registrations in the file ERIC sent to that state for voters who should potentially be removed.
Honey quotes this accurately but leaves out other parts of the report that reflect positively on ERIC, such as the state elections office saying ERIC does a better job “identifying individuals whose voter registration records may need to be inactivated or who may have more than one active voter registration record” than it does. Among its conclusions, the report says the state’s voter rolls would be more accurate if it used ERIC’s data more often.
Honey said the issues addressed by the Wisconsin agency were “very different things” than what she investigated in her report, but did not elaborate on why she didn’t include the positive aspects.
Honey’s claims gain traction
Despite the flaws in Honey’s report, it quickly gained traction, making its way across conservative media and into the hands of influential politicians and officials.
"Thank you for reading through this information!” a local Virginia GOP committeewoman wrote about Honey’s report in a March 2023 email to a state representative, expressing “hope” it would be shared with the Virginia state elections board. It ultimately was.
Records obtained by Documented also show that Honey’s influence grew through the help of Mitchell. In June 2022, Mitchell organized the event for secretaries of state in Washington, D.C. Officials from Texas, Wyoming, Louisiana, and West Virginia were scheduled to attend, according to records obtained by Documented, which also show that Mitchell’s group paid for a Texas secretary of state official’s travel to the event. Records obtained by Votebeat and Spotlight PA show Missouri Secretary of State Jay Ashcroft was out of office that day for an unspecified “meeting in DC.”
Louisiana announced its official withdrawal from ERIC a month later.
As officials and activists took notice of the report, Honey’s work began to be cited by influential outlets and figures in conservative media, including The Federalist.
“Per government watchdog Verity Vote, ERIC doesn’t actually clean states’ voter rolls, but rather inflates them,” the publication wrote in March 2023, using the name of the entity through which Honey puts out her election-related research.
Her work also made appearances in a series of articles critiquing ERIC by Hayden Ludwig, director of policy research at Restoration of America — a conservative Christian nonprofit focused on the challenges presented by “the elite,” “Marxist neo-liberals” and “Communist China.”
And Judicial Watch, an influential conservative legal watchdog group, cited it in a white paper in March 2023. “States that do not participate in ERIC had a higher rate of identifying and removing from voter registration rolls individuals who relocated out of a jurisdiction than ERIC member states,” Judicial Watch wrote.
Policymakers took notice of the report as well.
In 2018, Missouri Secretary of State Jay Ashcroft, a Republican, said the state’s membership in ERIC “will help affirm voters are eligible and registered in the right location, identify potential duplicate registrations and identify unregistered voters so we can help them get registered.”
But last February, Honey gave Ashcroft a private presentation on her report, emails obtained by Documented and Votebeat/Spotlight PA show.
A month later, Ashcroft announced Missouri’s withdrawal, saying in part that it was because ERIC was “adding names to voter rolls” — language similar to the claims in Honey’s report. Missouri’s number of registered voters increased by less than 2% in the time it was part of ERIC. The state’s Republican state auditor recently criticized Ashcroft for the decision in a report, which cited the benefit of ERIC’s data in maintaining accurate voter rolls.
Ashcroft did not respond to multiple requests for comment.
Honey’s report was not the sole reason any state withdrew from ERIC. No state election official — even those who met with her or received her report and appear to have adopted some of her language — explicitly cited her research. But her report and advocacy appear to have influenced several states.
In January of last year, Devvie Duke, a member of the Texas GOP and an ERIC task force, said in a Zoom meeting that she had lined up ERIC training with Honey “so we can be informed and productive when we’re writing.” She paused. “Or, going to see our legislators in support of legislation getting rid of ERIC.”
State Sen. Bryan Hughes, who wrote the bill that pulled Texas out of ERIC, participated in that meeting, where he was introduced as “a regular” at the task force’s sessions.
Duke, who did not respond to a request for comment, is currently running for a seat in the Texas state House. Her campaign material claims that it was “through Devvie’s work that Texas terminated its membership” in ERIC, which she called a “liberal voter registration scheme.”
In Virginia, Susan Beals — a former Republican state Senate aide and now the commissioner for the state Department of Elections — also received the report last March from a member of the state House, according to an email obtained by Votebeat and Spotlight PA. A letter obtained by Virginia Public Media shows that when Beals withdrew from ERIC, some of her reasoning was worded similarly to the contents of the email. Beals did not respond to a request for an interview.
Internal communications from an election integrity network in North Carolina — obtained by Documented and shared with Votebeat and Spotlight PA — show that Honey met with a group of legislators about a bill to prevent the state from joining ERIC, and also met with at least some of members of the state Senate who would later vote for the measure.
And language mirroring Honey’s research has also appeared in the Capitol building of her home state, Pennsylvania, where state Sen. Cris Dush (R., Jefferson) said last March that non-ERIC states were doing a better job cleaning voter rolls than ERIC states, though he did not cite any evidence.
Dush did not respond to a request for comment.
Friends in high places, and a new frontier
Honey has certainly been successful in getting her information in front of influential officials in a way her peers have not, forming connections with secretaries of state, their staffs, state lawmakers, U.S. Senate candidates, and members of Congress. Her influence, therefore, is likely to continue.
For example, she recently gave a talk in Pennsylvania to the Lycoming County Republican Committee, attended by U.S. Rep. Dan Mueser, who called it an “important presentation” on “securing and protecting the voting system.” She also appears to have provided draft language for a bill in June 2021 to state Reps. Seth Grove (R., York) and Russ Diamond (R., Lebanon), according to records obtained by American Oversight, a left-leaning watchdog group.
Honey also leads a local organization, PA Fair Elections, which hosts weekly Zoom discussions on election issues and organizes activists. Recent meetings have focused on how to advocate post-election hand counts to county commissioners, though some attract public officials, including at least two state representatives, a Commonwealth Court judge, and county commissioners.
Ahead of the 2024 election, her work is expanding beyond ERIC to include research and advocacy about ballots sent to and from voters living overseas, including military voters.
In addition to authoring a report on the subject, which she published in September 2022, she filed a complaint with the Pennsylvania Department of State, with the support of an attorney from the Thomas More Society. It alleged the department was violating federal law by not requiring identification for some overseas voters.
An attorney for Gov. Josh Shapiro’s office, who handled the complaint, dismissed it in November, offering that the complaint was more of an objection to federal law rather than an accusation that existing law was being violated. Honey is now appealing that decision to Commonwealth Court.
She’s also exporting this advocacy to states outside the commonwealth. In August, she gave a presentation on overseas voters to the Virginia coalition affiliated with Mitchell's election integrity network, according to records reviewed by Documented.
A now-deleted post from the North Carolina-based Asheville Tea Party, whom Honey worked with to contact legislators in that state, shows that she also gave a presentation in April on military and overseas voters at the request of Mitchell.
Meeting notes posted to the Asheville Tea Party’s website indicate Honey explained that her concern was that while people assume the ballots are from military voters, most of them come from overseas citizens, many of whom were likely unverified.
“These newer election officials just rubber-stamp them and don’t do anything to verify,” she said, according to meeting notes. “It’s the non-military that we need to really worry about.”
Honey’s work focused on influencing local election policy also continues.
At the first PA Fair Elections meeting of the year, Honey kicked off the meeting by talking about what changes members needed to push for in their counties, including pressuring election offices to conduct their required post-election audit — a recount of 2% of all ballots — by hand count. It is currently done with a mix of hand counting and machine tabulation.
“There are 104 days till the primary, so these things we are talking about, we are very much hoping to have them in place for the primary so we can work out any kinks,” she told the group of roughly four dozen. “If we can roll them out in the primary, we can hopefully make improvements to them by the general.”
Jason Armesto — a reporter for The Daily Progress, a newspaper in Virginia — contributed to this report.
Graphics and data analysis for this story were done using data from the U.S. Election Assistance Commission’s Election Administration and Voting Survey. Specifically, state-level data on the number of reported registrations, voting age population, overall removals, removals due to voter death, removals due to voter moving out of jurisdiction, and new valid registrations were extracted from the 2014, 2016, 2018, 2020, and 2022 datasets. EAVS data should be used and viewed with caution, as it is a voluntary survey and is often incomplete or inaccurate.
Data on overall removals, removals due to voter death, removals due to voter moving out of jurisdiction, and new valid registrations was then grouped by each state’s ERIC membership status for a given report period. Those figures were then divided by the aggregate reported registrations for the group of states. Reported registrations are the actual number of registered voters a given jurisdiction reports in response to the survey. The voting age population is an estimate of the number of persons eligible to vote, based on census data.
BEFORE YOU GO… If you learned something from this article, pay it forward and contribute to Spotlight PA at spotlightpa.org/donate. Spotlight PA is funded by foundations and readers like you who are committed to accountability journalism that gets results.