HARTFORD – Post-election audits will analyze the accuracy of the new generation of electronic vote tabulators that were deployed statewide for the first time in this year’s local elections.
The routine drawing of 32 voting precincts Wednesday to have their machine totals audited took on some added significance following the replacement of the previous generation of optical scan ballot tabulators.
Post-election audits were mandated when the state switched from pull-lever voting machines to optical scanners in 2006. It was the first switchover to new voting technology since the 1930s.
At the time, there were concerns the optical scanners were susceptible to tampering. The auditing requirement was meant to protect the security of elections and reassure voters their votes will be recorded accurately.
“Audits are how we prove to voters that the system works,” said Anna Posniak, the assistant director of elections in the Office of the Secretary of the State. “They’re not about catching mistakes, they’re about showing that our machines are accurate and our processes are sound. That’s how you build trust.”
After every election, state law requires a manual recount of ballots from 5% of polling places to match against vote totals from voting counting machines, including absentee ballots. The ballots also will be tabulated again to ensure they were properly read. If a discrepancy is found, additional audits and possibly a recount may be required.
The audits are open to the public and are legally required to be bipartisan. Ballots from the selected precincts are manually reviewed and compared to the machine totals.
As an added precaution, the Center for Voting Technology Research at the University of Connecticut will analyze the results of the precinct audits and test memory cards that are used to count votes. This is another statutory requirement.
The UConn VoTeR Center submits a report of its findings to the secretary of the state’s office and the State Elections Enforcement Commission. Once submitted, the audit reports are public.
There were 785 voting precincts in the Nov. 4 elections in 168 of the state’s 169 cities and towns. The town of Union holds its local elections in May.
The random drawing this past week selected 32 voting precincts to be audited in a mix of 29 small towns, suburbs and cities. Alternative precincts in 15 other municipalities were also picked in case any of the selected voting districts cannot be audited.
Only 633 voting precincts were part of the drawing because 152 precincts with recounts were excluded from selection.
The state government spent $20 million on the new electronic ballot scanner and tabulator machines to replace the outmoded, worn down and discontinued optical scan machines that state and local election officials said had become unreliable and unserviceable.
The secretary of the state’s office purchased 2,699 purpose-built polling place scanner and tabulators from winning bidder Election Systems & Software, 38 high‐speed scanner and vote tabulators, and 13 a high-throughput central scanner and vote tabulators designed to process absentee, early vote and Election Day ballot scanning and sorting in less time.
About 2,500 of the new machines were used in the Nov. 4 elections, according to the secretary of the state’s office.
Secretary of the State Stephanie Thomas demonstrates one of the state’s 2,500 new voter tabulators before the Nov. 4 local elections. (Ken Dixon/Hearst Connecticut Media)
The optical scanner machines were first used statewide in a general election in 2008. The post-election audit initially identified some larger than expected discrepancies between the machine totals reported on Election Day and subsequent hand-count audits, but further investigation found many turned out to be the result of human error during the audit process.
The results of subsequent post-election audits reached the same conclusion.
The audits of the 2016 and 2020 presidential elections, for example, indicated small discrepancies consistent with anticipated levels of human error in counting or ballot handling, but revealed no conclusive signs the tabulators malfunctioned. The audit of the 2024 presidential election also revealed no such signs of tabulator malfunction.
Voter enrollment and turnout is highest in presidential elections every four years. Nearly 1.8 million of the more than 2.3 million registered voters cast ballots in the 2024 presidential election. That works out to a turnout rate of 76.2%.
In post-election testing, auditors classify ballots as undisputed or questionable based on their observations of how voters marked them.
An “undisputed mark” is a mark that covers the majority of the bubble and is dark enough that all auditors agree it should have been read as a mark by a working tabulator. A “questionable mark” is not large or dark enough to convince all of the auditors that a working tabulator would have recorded it as a mark.
More Information
Here are towns and voting place where audits are happening:
Scotland – Firehouse Community Center
South Windsor – Timothy Edwards School
New Haven – Atwater Senior Center
Orange – High Plains Community Center – Senior Cafeteria
Newtown – Middle School Gym
Torrington – City Hall
Torrington – Torrington School 2
Cornwall – Cornwall Town Hall
Greenwich – Bendheim Western Greenwich Civic Center
Stonington – Saint Michael Church
Bridgeport – Bridgeport Central High School
Brookfield – Candlewood Lake Elementary School
Stratford – Chapel School
New Britain – Chamberlain Elementary School
Glastonbury – Nayaug Elementary School
Waterbury – Waterbury City Hall (absentee ballots, early voting and same-day registration ballots)
Bristol – Bristol Eastern High School
Ridgefield – Yanity Gym
Vernon – Central Counting – Center 375 Gym
West Hartford – West Hartford Town Hall
West Hartford – Hall High School
East Hampton – East Hampton High School
New Haven – John S Martinez School
New Canaan – Central Counting – New Canaan Town Hall
Stonington – Stonington Fire House
Stratford – Central Counting – Town Hall, Town Council Chambers
Rocky Hill – Central Counting – Rocky Hill Town Hall, Council Chambers
New London – New London STEM High School
Bridgeport – Cesar Batalla School
Milford – Central Counting – Parson’s Government Center Gym
Woodstock – Woodstock Middle School
Danbury – War Memorial Gym
Source: Secretary of the state’s office
The auditors use the undisputed and questionable marked totals to determine an expected vote range equal to the undisputed mark count and the combined count.If a tabulator count falls within this range, the tabulated results and hand-counted results are considered consistent. If the tabulator total falls outside of this range, then it is considered an unexplained discrepancy.
Historically, the majority of observed individual discrepancies account for less than 1% of the total ballots cast.
Secretary of the State Stephanie Thomas plans to propose transitioning to a new method for checking reported election outcomes against voters’ ballots, known as “risk-limiting audits.”
“Risk-limiting audits use proven statistical methods to confirm results with greater efficiency and transparency,” she said. “They would move Connecticut to the forefront of election best practices.”
Unlike traditional audits that review a fixed number of ballots, risk-limiting audits scale their checks depending on the margin of victory. Essentially, the closer the race, the more ballots are manually reviewed. The initial sample is increased when discrepancies are found until either the level of confidence has been met or a full recount has been performed. This specified risk limit is intended to balance the need for accuracy with a full-hand recount’s time and financial cost.
Currently, the states of Rhode Island, Colorado, Pennsylvania, Georgia, Nevada and Virginia use risk-limiting audits, according to the National Conference of State Legislatures.
In 2021, the legislature established a working group to review risk-limiting audits and recommend how Connecticut could institute the practice. Implementing legislation has been introduced in every succeeding legislative session 2022, but none of the bill have advanced beyond the committee level. The 2025 version of the bill proposed a pilot program to test risk-limiting audits in the Nov. 4 municipal elections.
This article originally published at These 32 Connecticut voting precincts will be audited.
