Start Here
Donate

Case 10 — Cook County, 2016

10.1 — Introduction

On March 15, 2016, Illinois held its presidential primary. In Chicago — the heart of Cook County — 708,125 Democratic ballots were cast. Hillary Clinton won the city with 380,208 votes (53.69%) to Bernie Sanders’s 320,894 (45.32%), a margin of roughly 59,000 votes.

Three weeks later, credentialed election monitors appeared before the Chicago Board of Election Commissioners to describe what they had witnessed during the mandatory post-election audit. Their testimony, given under affidavit and recorded on the Board’s own video feed, described an audit process in which the official results were printed on the tally sheets before the hand count began, tallies were kept in pencil, and Board employees used multiple methods — erasure, Post-it notes, forced recounts — to make the hand count match the official numbers.

This case study is not about whether the 2016 Chicago Democratic primary was stolen. The evidence does not support that claim. It is about what happens when the process designed to verify an election is itself compromised — and what that failure reveals about the limits of post-election auditing as an integrity mechanism.

10.2 — Background: The 5% Audit

Illinois Election Code § 24C-15 requires a post-election audit of a random sample of voting equipment. The State Board of Elections selects 5% of precincts and 5% of early voting machines; the local election authority — in Chicago, the Board of Election Commissioners — conducts the audit by hand-tallying the permanent paper records and comparing the hand count to the machine-generated totals.

Chicago used Direct Recording Electronic Voting Systems (DREs) — touchscreen machines that produce a permanent paper record in the form of a long continuous roll. Each voter’s selections are printed sequentially on the roll as votes are cast throughout election day. This paper record is the independent evidence: if the electronic tabulation is accurate, the hand tally of the paper roll should match the machine’s reported totals.

The audit procedure is straightforward. One Board employee reads the votes aloud from the paper roll — “Clinton, Sanders, Clinton, Clinton, Sanders…” — while another employee tallies the votes by hand on a tally sheet. When the roll is finished, the hand tally is compared to the machine’s totals.

There is, however, a critical legal constraint. Section 22-9.1 of the Illinois Election Code provides that the results of the 5% audit “shall not be certified, used to amend or change the abstracts of the votes previously completed, used to deny the successful candidate… his certificate of nomination or election, nor used to change the previously declared result.” The audit exists as a check on the voting equipment, but its findings — whatever they may be — cannot alter the outcome of the election.

10.3 — What the Monitors Observed

Beginning March 23, 2016, credentialed election monitors from two organizations — Who’s Counting?-Chicago, led by Dr. Lora Chamberlain (a physician), and the Illinois Ballot Integrity Project (IBIP) — observed the 5% audit at a Board storage facility. Who’s Counting?-Chicago and IBIP were citizen-organized groups that formed in response to concerns about the administration of the 2016 Democratic primary; they were self-credentialed monitors, not court-appointed or government-appointed observers. Approximately twenty volunteers participated across four days of observation, March 23 through March 29.

What they described, independently and consistently, was a process structured not to verify the election results but to confirm them.

The answers on the test. Machine-generated final vote counts were printed on the tally sheets before the hand count began. The employees conducting the hand tally could see, as they worked, what the official result was supposed to be. As Dr. Chamberlain described it to the Board: the auditors had “the answers to the test” sitting in front of them. Attorney William Shipley, who observed on March 28, described the official count as “printed in bold on their tally sheets.”

Pencil tallies. The hand tallies were conducted in pencil — erasable. This is significant because it means the tally could be adjusted after the count was complete, and no trace of the original count would remain.

Multiple methods of forcing conformity. The monitors described at least four distinct methods by which the hand count was made to match the official results:

Erasure. Shipley testified that when the hand tally at his first table produced more votes for a candidate than the official result showed, employees erased tallies to bring the count down — and that he had photographic evidence of the erasures. He asked to see the final tally and was told that the current tally was “not the final version,” indicating an additional process would occur after the count was complete. That additional process, he testified, consisted of erasing tallies that exceeded the official vote count.

Post-it notes. At his second table, Shipley testified that employees used a different strategy. Instead of erasing excess votes, they recorded votes beyond the official count on Post-it notes separate from the official tally sheet, creating a split record: the official tally matched the official result, while the actual count — to the extent it exceeded the target — was captured on disposable notes. Shipley described a subsequent “concordance process” to reconcile the two.

Stopping the count. Monitor Michelle Suzanne Gale, a political scientist with a degree from Richmond University in London, testified that she observed employees counting hash marks on the tally sheet to the official number and then stopping — tallying additional votes read aloud by the reader as votes for a different candidate, or simply ceasing to tally once the target was reached.

Forced recounts and replacement of counters. At his third table, Shipley testified that when the hand count did not match the official result, the employees were told to recount. When they recounted and still did not match, they were told to count a third time. When they insisted their count was correct, other individuals were brought in to conduct their own recount. The original counters, he testified, “were made to feel as if they were doing something wrong, as if this was somehow a mistake on their part.”

Obstruction of monitors. Multiple witnesses described active efforts to prevent monitors from observing the process. Gale testified that Board employees placed ballot boxes between monitors and the tally sheets, refused to move them when asked, and turned tally sheets away from observers. Chamberlain reported that General Counsel James Scanlon had directed employees to obstruct the monitors’ view. The 7th Circuit opinion, accepting the plaintiffs’ allegations as true, states that Board employees “actively attempted to obscure Plaintiffs’ view of the tally sheets.”

A specific example. Dr. Chamberlain provided one particularly detailed example. On one early voting machine, she testified, she watched the entire hand tally of the paper roll. At the end of the count, the hand tally did not match the official result. According to her testimony, the auditors erased 21 Bernie Sanders votes and added 49 Hillary Clinton votes — a swing of approximately 70 votes on a single machine — to make the hand count match the official numbers. She described this as forcing “the hand count of the audit of this toilet paper roll to the recommended Chicago Board of Elections official results.”

10.4 — The April 5, 2016 Board Meeting

What happened at the Board meeting on April 5 is itself part of the evidentiary record, because the meeting was recorded on the Board’s own video feed and posted to the official Chicago Elections YouTube channel. The video record is also discussed in a recap posted to a political forum.

The meeting began with the canvas of returns and proclamation of results — the formal certification of the March 15 primary. Chairwoman Marisel Hernandez opened the meeting, introduced Commissioners Jonathan T. Swain and William J. Cressy, and immediately moved to the agenda item: consideration of the tabulated statement of returns.

Dr. Chamberlain, present in the room, attempted to object: “Can we object?”

The response: “No, not at this time, Dr. Chamberlain. Thank you though.”

A motion to approve and adopt was made, seconded, and passed. The certification meeting was over. It had lasted less than two minutes.

A separate special board meeting was then convened, in which public comment was permitted. Six witnesses — Chamberlain, Gale, K. Rubenstein, Rebecca Kerlin, Shipley, and others — gave affidavit-based testimony describing what they had observed during the audit. Their testimony occupied over an hour. The Commissioners listened. The Board’s General Counsel responded. But the election results had already been certified before a single word of testimony was heard.

The sequence matters. The Board did not refuse to hear the testimony. It heard it — after the results were certified. The monitors’ concerns, however well-documented, arrived at a meeting that had already concluded its official business.

10.5 — The Witnesses

The credibility of the testimony rests in part on the qualifications of the witnesses and the specificity of their observations.

Dr. Lora Chamberlain is a family physician who led Who’s Counting?-Chicago. She coordinated approximately twenty volunteer monitors across four days. Her testimony was the first and most sweeping, establishing the overall pattern: official results on the tally sheets, pencil tallies, erasures, obstruction. She provided the most specific single example (the 21/49 vote swing) and submitted affidavits. She was careful to caveat her own conclusions: “It appears to us that the inaccuracies were mostly in favor of Hillary,” she told CounterPunch after the meeting, but added: “We are not going to say that in a court of law because we didn’t have every table covered every day” and “there was a lot of blocking behavior.”

Michelle Suzanne Gale of the Illinois Ballot Integrity Project observed on March 23, 25, and 28. A political scientist, she provided testimony about the counting methodology: employees counting to the official number and stopping, the obstruction using ballot boxes, and the refusal to allow monitors to see completed tally sheets.

Rebecca Kerlin, also of IBIP, had worked in software quality assurance since 1999. Her testimony was the most technically detailed. Her affidavit covered six paper ballot items, fourteen electronic paper tape items, and one work stoppage item. She described systemic procedural failures: paper ballots not visually audited, no machine-to-machine comparisons performed, double-marked ballots fed into machines without separation, inconsistent methods for reading votes aloud (switching between first names, last names, and nicknames), no consistent method for marking place when taking breaks, and teams sometimes re-reading votes from the beginning after breaks. Kerlin’s testimony framed the problems not as isolated incidents but as systemic quality assurance failures.

William Shipley, an attorney licensed in New York and a Chicago resident, observed on March 28. His testimony was the most legally precise. He organized his observations around three specific tables, each exhibiting a different method of forcing the hand count to match the official result (erasure, Post-it notes, forced recounts). He offered photographic evidence. He also identified a structural weakness in the audit sample itself: some of the machines selected for audit were standby or reserve machines that had recorded zero votes — meaning they diluted the audit without providing any verification of machines that had actually been used.

These were not casual observers filing complaints. They were credentialed monitors with relevant professional backgrounds — a physician, a political scientist, a software quality assurance professional, and an attorney — who spent multiple days observing a specific government process and documented their observations in affidavits submitted to the body responsible for that process.

10.6 — The Board’s Defense

The Board’s response came through two channels: General Counsel James Scanlon’s statements at the April 5 meeting and Communications Director Jim Allen’s post-meeting interviews with CounterPunch.

At the meeting, Scanlon denied that monitors had been obstructed, stating that Chamberlain “had poll watchers at every single station.” He characterized the 5% audit as an equipment test rather than a recount, and emphasized that the audit results could not change the election outcome. He did not specifically address the witnesses’ descriptions of erasures, Post-it notes, or forced recounts. His rebuttal was general — “there was no intention to manipulate results” — rather than responsive to the specific irregularities described.

After the meeting, Allen spoke to CounterPunch. He admitted that “the numbers didn’t match” initially but described this as a “perception issue.” His explanation: the tally sheets had not been resized to accommodate the higher volume of votes on early voting machines, so counters ran out of space. He offered a checkbook analogy: “If you are going through your checkbook and the first time is mismatched, you don’t immediately yell ‘the bank is ripping me off.’”

What the Board never explained:

Why the official machine-generated vote counts were printed on the tally sheets before the hand count began. If the purpose of the audit is to independently verify the machine count, providing the “correct answer” to the people conducting the verification defeats the purpose. The Board offered no justification for this procedure.

Why the hand tallies were conducted in pencil. An independent verification process should produce a permanent record. Pencil tallies can be erased — and, according to multiple witnesses, were erased.

Why monitors were obstructed from observing the final tallies. If the audit was conducted properly, there would be no reason to prevent credentialed observers from seeing the results. The Board denied the obstruction occurred, but the 7th Circuit opinion, accepting the plaintiffs’ allegations as true, stated that Board employees “actively attempted to obscure Plaintiffs’ view of the tally sheets.”

What procedure existed for handling discrepancies between the hand count and the official result. Allen acknowledged the numbers didn’t match. The monitors described erasures, Post-it notes, and forced recounts as the methods used to resolve discrepancies. The Board never described an alternative procedure — what should have happened when the hand count didn’t match.

Commissioner Cressy, a professor of auditing, acknowledged the concerns and promised to “spend more time during the 5% audit out at the warehouse” after the November election. This was the most substantive response from any Board member — and it implicitly acknowledged that the audit process warranted closer oversight.

10.7 — The Lawsuit

Six plaintiffs — William Shipley, Nina Marie, Katherine Wuthrich, Rebecca Kerlin, Michelle Gale, and Claire Tobin — filed suit in federal court in July 2016, naming the Chicago Board of Election Commissioners and General Counsel Scanlon as defendants. The case was filed as Kerlin v. Chicago Board of Election Commissioners in the Northern District of Illinois, assigned to Judge John Robert Blakey. The complaint asserted three counts under 42 U.S.C. § 1983: violation of the right to vote, freedom of association, and right to petition.

The district court dismissed the case. Its reasoning was direct: “No matter how flagrantly [the Board] might have doctored the numbers during its 5% test, that test’s results could not possibly have affected the election results under Illinois law.”

Three plaintiffs — Shipley, Marie, and Wuthrich — appealed. (Kerlin, Gale, and Tobin withdrew on appeal.) The case became Shipley v. Chicago Board of Election Commissioners, No. 17-3511, 947 F.3d 1056, before the Seventh Circuit Court of Appeals. The panel — Judges Brennan, Scudder, and St. Eve — heard argument on January 7, 2020, and issued its opinion on January 27, 2020.

The Seventh Circuit affirmed the dismissal. The court’s reasoning turned on the statutory framework. Illinois Election Code § 24C-15 provides that the 5% audit results “shall be treated in the same manner and have the same effect as the results of the discovery procedures set forth in Section 22-9.1.” And Section 22-9.1 provides that those results “shall not be certified, used to amend or change the abstracts of the votes previously completed, used to deny the successful candidate… his certificate of nomination or election, nor used to change the previously declared result.”

The court stated explicitly: “No matter how factually detailed, egregious, or willful the misconduct alleged, the five percent audit cannot affect the election results under Illinois law.”

The court acknowledged that “this does not mean that all allegations of misconduct during a five percent audit are without recourse entirely” — suggesting that state law claims might exist. But in federal court, on a constitutional theory, the plaintiffs had no viable claim. The case was dismissed with prejudice.

10.8 — The Legal Paradox

The Seventh Circuit’s ruling exposes a structural paradox in the Illinois election code.

Illinois requires a 5% post-election audit. The stated purpose is to verify the accuracy of the voting equipment. But the statute simultaneously provides that the results of this audit cannot change the election outcome. The audit exists as a check on the system, but the check has no enforcement mechanism.

This creates a perverse incentive structure. If the audit results cannot change anything, there is no institutional consequence for conducting a bad audit. The Board can run the audit with official results printed on the tally sheets, with pencil tallies, with obstructed monitors, with forced recounts — and even if every one of these things is documented, photographed, and testified to under oath, the legal framework provides no remedy for the election at issue.

Dan Flatley, in his notes on this case, drew an analogy to financial auditing: “Imagine, if you will, the SEC watching the CEO and CFO and Chair of the Board of Directors of a public corporation reacting this way to an audit of a PCAOB-certified auditor. The SEC would be all over them.” The comparison is instructive. In financial regulation, the audit has teeth — a failed audit triggers consequences. In the Illinois election code, the audit is mandatory but toothless.

The paradox is not unique to Illinois. Many states have post-election audit requirements that are structured as equipment tests rather than result-verification mechanisms. The Chicago 2016 case is distinctive because the paradox was litigated to a clear appellate ruling — the Seventh Circuit stated plainly that no amount of misconduct during the audit can affect the result. But the structural problem exists wherever an audit is required but its findings are legally inconsequential.

10.9 — What We Know and What We Don’t

What is established:

Multiple credentialed monitors, working independently across four days, described the same pattern of irregularities at multiple tables: official results preprinted on tally sheets, pencil tallies, erasures to match official counts, excess votes recorded on Post-it notes, forced recounts when hand tallies didn’t match, replacement of counters who insisted their count was correct, and active obstruction of monitors. Their descriptions are specific, consistent with each other, and documented in affidavits submitted to the Board.

The Board’s own Communications Director admitted that “the numbers didn’t match” during the audit. He attributed this to a tally-sheet sizing issue rather than fraud, but he did not deny the discrepancies.

The Board never specifically rebutted the monitors’ descriptions of what happened at individual tables. General Counsel Scanlon’s defense was general (“there was no intention to manipulate results”) rather than responsive to specific observations. The Board never described a procedure for resolving discrepancies between the hand count and the official result — which is what the monitors were describing.

One monitor — Shipley — has photographic evidence of erasures on tally sheets.

What we do not know:

The full extent of the discrepancies. The monitors did not have every table covered every day. As Chamberlain acknowledged, “we didn’t have every table covered every day” and “there was a lot of blocking behavior.” The irregularities documented are a sample of a sample — observations from some tables during some days of a 5% audit of all machines.

Whether the irregularities were directional. Chamberlain stated that “the inaccuracies were mostly in favor of Hillary” but immediately caveated the claim. The specific example she provided (erasing 21 Sanders votes, adding 49 Clinton votes) is directional. But a comprehensive assessment of directionality across all audited machines is not available from the evidence.

Whether the audit irregularities reflect counting-layer problems in the original vote. The audit was supposed to check the machines. If the audit was compromised, we don’t know whether the machines were accurate — which was the question the audit was designed to answer. The irregularities may reflect employees masking real discrepancies, or they may reflect employees taking shortcuts on an audit they considered a formality. The evidence is consistent with either interpretation.

Whether any of this would have changed the election outcome. Clinton’s margin in Chicago was approximately 59,000 votes. Even if every irregularity documented by the monitors reflected a real vote-count error, a 5% sample cannot establish the magnitude of discrepancy across the full jurisdiction. The case study does not and cannot support a claim that the primary result would have been different.

10.10 — What Actual Vote Would Have Contributed

The Chicago 2016 case is different from prior case studies in this manual — Butler County, Shelby County, Gaston County — where the discrepancy existed between precinct-level counts and official results. In those cases, AV’s poll-tape comparison would detect the reporting-layer failure directly. In Chicago, the problem was not (or not only) a reporting-layer discrepancy. It was the failure of the audit itself — the process designed to check the results was compromised from within.

AV operates differently from a post-election audit in ways that are directly relevant to the Chicago failure.

AV generates evidence before the audit, not during it. Poll tape photographs are taken at poll closing, at the precinct, by volunteers who are present for the closing of the polls. This evidence is generated before vote data reaches any central system, before any audit is planned, and before anyone has an opportunity to structure the verification process around a predetermined result. In Chicago, the audit began eight days after the election. AV’s evidence would have been generated on election night.

AV evidence is public and distributed. Each poll tape photograph is held by the individual volunteer who took it and is published through AV’s public platform. The Board cannot control this evidence, obstruct its collection, place ballot boxes between volunteers and their photographs, or turn tally sheets away from observers. In Chicago, the monitors had to fight for the ability to see the tally sheets. AV’s evidence is in the hands of the volunteers the moment it is created.

AV does not rely on Board employees to do the counting. The comparison is between a photograph of the poll tape (taken by a volunteer) and an official result (published by the Board). No Board employee is in a position to erase tallies, record excess votes on Post-it notes, or force recounts until the numbers match. The comparison is between two independent records, not between a hand count and the “answer” sitting on the same table.

AV’s evidence has no statutory limitation on its impact. The 5% audit legally cannot change election results in Illinois. AV’s published discrepancies are a different kind of evidence — they enter the public record, the press, and potentially the courts through different channels than a statutory audit. They are not constrained by § 22-9.1.

What AV would catch: If the electronic voting machines produced official results that diverged from the poll tape totals — whether through a tabulation error, a reporting-layer manipulation, or any other cause — AV’s comparison would detect the discrepancy.

What AV would not catch: If the DREs recorded votes incorrectly at the counting layer — before the paper record was printed — AV’s poll tape would reflect the same error as the machine totals, and no discrepancy would be detected. AV audits the reporting layer. It does not audit the counting layer.

The broader point: AV is not a substitute for a properly conducted post-election audit. But AV generates independent evidence that cannot be manipulated by the entity being audited. The Chicago 2016 case demonstrates, in granular detail, what happens when the entity responsible for verifying election results is also the entity being verified. Independence is not an abstract principle — it is the specific, structural feature that distinguishes an audit from a confirmation exercise. AV is designed around that principle.

The defensive frame: the broader Sanders-camp narrative

The Cook County 2016 audit fraud documented in this case is the offensive AV illustration: the audit failed in real time, in front of credentialed observers, in ways the Seventh Circuit ultimately ruled could not affect the certified result. The case’s defensive frame sits beside that documentation. The Cook County episode was central to a broader Sanders-aligned narrative that emerged across the 2016 Democratic primary cycle: that Hillary Clinton had stolen the nomination through a combination of state-party rule-bending, superdelegate maneuvering, and — most directly relevant to AV — vote-counting irregularities at the precinct level in states ranging from Illinois to Nevada to New York to Arizona to California. The narrative was intra-Democratic but consequential. It contributed to the legitimacy gap that affected Democratic mobilization in the 2016 general election, and it shaped the post-2016 left’s understanding of party-internal verification of electoral results in ways that persist.

Most of the broader claims were not validated by subsequent investigation. Some had real referent (the Cook County audit fraud documented here is one) but did not extend to the underlying primary results. Some rested on contested inferences from anomalies that other analysts characterized as routine. None of them — including the Cook County case — definitively established that the certified primary results in any state were wrong. But the narrative did not require validation to operate. It operated because the underlying evidentiary record, in 2016 as in 2004 and 2000 and 2002, was insufficient to test specific claims against arithmetic. The claims about reporting-layer manipulation in specific Cook County precincts could have been confirmed or refuted from independent precinct-level evidence. That evidence did not exist outside the Board’s custody. The Board’s custody was the very thing being challenged.

What AV evidence would have contributed in 2016 is what it would have contributed in 2004 and 2000: an independent precinct-level record against which the broader Sanders-camp narrative could have been tested, claim by claim, against the actual tape totals. The Cook County audit fraud — the offensive case — would still have been visible (and AV would have helped surface it earlier and more authoritatively). The broader claim that Cook County’s underlying primary results themselves were fraudulent could have been tested, precinct by precinct, against citizen-held photographs. Some of the broader claim might have been validated. Most likely would not have been. Either way, the dispute would have been resolvable on its merits rather than inheritable as settled belief into the 2016 general election and beyond.

The collection’s companion analyses of Florida 2000, Baldwin County 2002, and Warren County 2004 describe the same pattern operating in general elections rather than primaries. The structural problem — bad-faith and good-faith accusations alike growing in evidentiary vacuums, becoming permanent because the evidence to settle them never existed — is the same. The structural defensive solution AV provides is the same.

10.11 — Why It Matters

The Chicago 2016 case matters for Actual Vote not because it proves an election was stolen — it doesn’t — but because it demonstrates three things about election verification that are directly relevant to AV’s mission.

First, the existence of an audit requirement is not sufficient. Illinois has a 5% audit statute. The statute was complied with. The Board conducted the audit. And by multiple credentialed accounts, the audit was structured to confirm a predetermined result rather than independently verify it. The statutory requirement, by itself, did not produce a meaningful check.

Second, the “answers on the test” problem is structural, not incidental. When the people conducting a verification process can see what the result is supposed to be, the process is no longer independent. This is a well-understood principle in every field that relies on auditing — financial auditing, scientific research, medical testing. Blinding is standard practice precisely because humans, even well-intentioned ones, adjust their work to match expectations. AV’s design avoids this problem: poll tape evidence is generated at poll closing, before official results are known, by volunteers who have no stake in the outcome.

Third, the legal framework can render an audit toothless even when the audit reveals problems. The Seventh Circuit confirmed that no amount of misconduct during the 5% audit can affect the election result. This is not a critique of the court — the judges applied the statute as written. It is a critique of a system in which the verification mechanism has no enforcement authority. AV’s evidence is not constrained by this statutory framework. It enters the public record as independent evidence, available to voters, journalists, candidates, and courts through whatever channels they choose to pursue.

The monitors who observed the Chicago 2016 audit — Dr. Chamberlain, Michelle Gale, Rebecca Kerlin, William Shipley, and their colleagues — did exactly what concerned citizens are supposed to do. They showed up. They watched. They documented. They testified under oath. They filed suit. And the system’s response, from the Board meeting through the Seventh Circuit, was that none of it could change anything.

AV does not claim to solve this problem. But AV generates a kind of evidence that the Chicago case did not have: independent, publicly held, precinct-level records that exist outside the control of the election authority being audited. That evidence does not depend on being allowed to observe the audit. It does not depend on the Board’s cooperation. And it does not depend on a legal framework that permits audit findings to matter.

10.12 — Further Reading

Primary documentation

Litigation

Statutory framework

  • Illinois Election Code § 24C-15 (5% audit requirement); § 22-9.1 (audit results cannot change outcomes).

Cross-references in this collection