March 1, 2024
Connecticut Government Administrations and Elections Committee
Legislative Office Building, Room 2200
Hartford, CT 06106
Via online portal
Dear Committee Members:
On behalf of Verified Voting, I write in support of S.B. No. 254 regarding risk-limiting audits. Verified Voting is a nonpartisan nonprofit organization whose mission is to strengthen democracy for all voters by promoting the responsible use of technology in elections. Since its founding by computer scientists in 2004, Verified Voting has promoted voter-verified paper ballots and routine, rigorous post-election audits—especially risk-limiting audits—to check the accuracy of computerized voting systems.
As you know, the United States confronts unprecedented security threats to election systems—and to public confidence in election outcomes. To counter these threats, tabulation audits, such as Connecticut’s existing manual audit, are needed to check the results against the actual ballots cast. Risk-limiting audits (RLAs) are robust tabulation audits designed to efficiently confirm that election outcomes match what a full hand count of those ballots would reveal by manually examining a representative random sample of the ballots. RLAs have been recommended by the American Statistical Association, U.S. Department of Homeland Security, the U.S. Senate Select Intelligence Committee, and many other experts as one element of a strong and resilient election infrastructure.
Election officials across the country have taken the lead in piloting and implementing RLAs. Colorado, Georgia, Pennsylvania, and Rhode Island regularly conduct RLAs prior to the finalization of election results, and their experience offers valuable lessons learned for Connecticut.
SB 254 would allow for the successful implementation of RLAs in Connecticut, providing a strong basis for public confidence in election results. The bill gives the Secretary of State the time and authority to develop procedures and regulations related to RLAs, without overspecifying the details in legislation. The adaptive comparison audits developed by the University of Connecticut also bring the goals of this legislation within reach.
We also offer a few suggestions on how this legislation might be improved. We would recommend that the bill require a pilot program in 2025 as preparation for RLA implementation following the November 2026 election. Pilots would provide registrars additional experience with the RLA process and allow them to test their new voting equipment—expected in 2025—as part of an RLA.
One of the strengths of the bill is that it requires a range of contests to be audited, but the state may benefit from phasing in the set of contests for audit. The bill could instead only require state offices to be audited to a risk limit in 2026. Other contests in 2026—one randomly selected office for U.S. representative in Congress and five percent of the offices of state senator and state representative—could also be audited but not necessarily to a risk limit. For instance, these contests could be audited as they appear on the ballots selected as part of the statewide RLA sample. In 2028, RLAs could then be required of all the contests currently listed in the bill. We would also suggest exempting from the RLA contests that are “uncontested,” where a candidate runs unopposed for an office.
We also suggest postponing until 2028 the requirement to audit U.S. House and General Assembly contests already subject to a recanvass, because of a close vote or a tie. If two candidates are separated by only a handful of votes, an RLA would likely result in a full hand count, which could prove challenging for registrars to complete prior to certification.
Thank you for considering our comments and suggestions. RLAs are an important tool for verifying election outcomes, and we applaud your efforts to implement them in Connecticut. We would be happy to discuss further any provisions of the bill or any of the suggestions we have outlined.
Respectfully submitted,
Chrissa LaPorte
Senior Policy & Technical Associate