The evidence shows that the routine use of statistically driven risk-limiting audits would be the most effective way to ensure the accuracy and security of elections. With the increase in the use of technology to record and tally votes, auditing election results before they are certified can provide additional assurance of the integrity of the results.
While election officials may conduct a number of procedural audits, absent any concerns or with a sufficient margin of victory, election results are often certified without any audit of votes.
Risk-limiting audits are designed to provide statistical evidence of whether the outcome of the election is accurate with a high level of confidence. An initial random sample of ballots is examined based on the margin of victory in an election, the total number of cast ballots, and a predetermined level of certainty that the audit will detect and correct an incorrect outcome. If necessary, additional ballots would be examined until there is statistical support for the accuracy of the election outcome. Risk-limiting audits are meant to be a more efficient and statistically sound process than traditional post-election audits that tally a fixed percentage of ballots. Risk-limiting audits require an auditable paper trail, highlighting one of the major weaknesses of paperless direct recording electronic systems.
The National Academies of Sciences, Engineering, and Medicine 2018 report, Securing the Vote: Protecting American Democracy, and general scientific consensus support audits as an essential component of ensuring accurate, secure elections. Risk-limiting audits are endorsed by the American Statistical Association.
Currently, the only state that conducts risk-limiting audits for statewide elections is Colorado, where voters primarily vote by mail. In 2017, Rhode Island enacted legislation mandating risk-limiting audits by 2020. Washington and Ohio’s audit procedures now allow for the option of risk-limiting audits while California, Indiana, and Georgia have passed legislation implementing risk-limiting audit pilots. Michigan and New Jersey also began administering risk-limiting audit pilots. These pilot programs are an effective way to introduce risk-limiting audits and begin training officials without requiring substantial investments.
The specific characteristics of a risk-limiting audit will depend on the voting systems in use but considerations include:
- Risk-limiting audits require a paper trail and are not possible for paperless voting systems
- Risk-limiting audits require a high degree of preparation and coordination among election officials including the use of a ballot manifest, a record of where ballots are physically stored
- Poor ballot design can increase voter mistakes that increase the workload of a risk-limiting audit
- Risk-limiting audits require software to conduct the audit and technical support to ensure the procedures are done properly
- The workforce for the election must be familiar with the election procedures and requirements of the audit to ensure the election is conducted in a manner that fits the planned risk-limiting audit method
Election administrators must also implement rigorous chain of custody and ballot reconciliation practices for protecting the paper audit trail.
To learn more about risk-limiting audits, see the two-part guide, Knowing It's Right, by Jennifer Morrell, a consultant leading the Election Validation Project at Democracy Fund.