Machines fail. We experience these malfunctions in our daily lives; washing machines spring leaks and automobiles break down on the side of the highway. When voting machines fail, the valid outcome of the election is at risk. Software malfunctions combined with human error can lead to the declaration of the losing candidate as the winner. That is why post-election audits are particularly important. A post-election audit is a check on the election process. Good post-election audits check machine results by inspecting some of the paper ballots manually. Well-designed and executed audits can catch and correct errors that make someone other than the true winner appear to win.
The Commission recommended that officials conduct post-election audits as a best practice of election administration. Post-election audits help to ensure that the outcome of the election is what voters intended -- and was not the result of machine malfunction or human error. The Commission endorsed two forms of audits: risk-limiting audits, which ensure the outcome of the election is correct, and machine performance audits, which spot-check machine function. Common Cause strongly endorses the risk-limiting audit as the most comprehensive audit system, because it provides strong statistical evidence that an outcome is correct. Moreover, the Commission strongly recommended that election jurisdictions publish data gathered on the audit outcome and on machine performance. With respect to machine performance, the Commissioners noted that "[i]t is very likely that a problem experienced by one jurisdiction is one soon to be experienced by another using the same or similar equipment."
Of the ten states, only Colorado is moving towards implementation of risk-limiting audits statewide, and has a law that requires their use in 2017. Florida and Colorado require post-election audits of voting equipment and mandate significant public disclosure of the audit results. Pennsylvania, Kentucky, Alaska, and North Carolina require post-election audits of voting equipment machine performance audits, but of these states, only Alaska requires that the results be publicized. Unfortunately, Arkansas, Georgia, Louisiana, and Michigan have no state requirements for post-election audits.
It is important to note, however, that none of the states surveyed with post-election audit requirements have elections which are fully auditable. Florida, Kentucky, and Pennsylvania still deploy voting systems that do not produce a voter-verifiable paper record and, therefore, a manual audit is not fully possible. In fact, one-third of counties in Kentucky use paperless voting machines that cannot be audited and checked against a paper record. North Carolina and Colorado allow military and overseas voters to cast ballots via the Internet, and Alaska allows any absentee voter to cast a ballot over the Internet. With internet voting, no voter-verified paper ballot is produced, meaning post-election audits in those three states are not fully possible.
The extent and nature of post-election audits varies greatly by state. In Colorado, the Secretary of State publicly and randomly selects at least five percent of the voting machines used in each county to be audited. Kentucky also has two distinct audit procedures. First, the law provides for a manual recount of randomly selected precincts representing between three and five percent of the total votes cast in each election. Second, the Kentucky Attorney General is required to make an independent inquiry for any irregularities that may have occurred in at least five percent of Kentucky's counties. In Florida, county canvassing boards can complete either a manual or automated audit with different requirements for each. North Carolina requires the number of ballots to be audited be chosen following consultation with a statistician.
|FINDINGS BY STATE|