The process of verifying the results of Colorado’s 2022 primary election began this week with the rolling of a 10-sided die.
On Monday, officials in the secretary of state’s office used 20 separate dice to generate a random sequence of numbers, or “seeds,” to determine which ballots from the June 28 Primary must be double and in some cases triple checked by counties.
These designations are the first step in what is called a risk mitigation audit – a process that takes place at every election before the results are certified. The procedure uses software to compare the votes of a selected group of paper ballots with the votes recorded by each county’s electronic tabulation machines.
Counties must report initial results to the state by the end of the day Wednesday. If a sufficient number of discrepancies are found, a second random audit is triggered.
“As long as the ballots and votes match, we know that overall winners are winners and losers are losers,” said Peg Perl, deputy chief electoral officer for Arapahoe County, who concluded the first round of its audit on Tuesday.
Colorado was the first state to require this type of audit after lawmakers passed an election security law in 2009. The first risk mitigation audit took place after the November 2017 general election due to delays in the development of technology.
Now more than one dozen states require audits or test them in pilot programs to help ensure the accuracy of elections.
Many counties record and even livestream the process online to help reassure the public that their community’s election results are accurate.
“It’s a really important part of our process that shows that nothing disappears into a black hole during the election,” Perl said. “We keep all the paper ballots and we can check them one by one.”
During each county’s audit, bipartisan teams of judges work together to draw the random sample of paper ballots. Then the teams enter the votes into a portal on the Secretary of State’s website, who cross-checks them against the voting marks he has on file.
The software then spits out a report of any anomalies.
If no anomalies are found, it is a sign that there is a high level of statistical confidence in the accuracy of the initial ballot count. If a sufficient number of errors are detected, verifiers should draw more ballots to ensure that the “risk limit” of the election being miscounted is small enough.
In the case of Colorado, the risk limit is about 3%. If additional auditing detects further errors, the process triggers a full recount.
That never happened, however, according to the secretary of state’s office.
“The department has never found a single instance where the tabulation equipment did not work properly while performing the RLA,” a spokeswoman wrote in an email.
However, isolated errors sometimes appear. In the case of the Arapahoe County audit this week, the software found a discrepancy on 145 ballots.
The ballot in question was one where a voter had filled in the bubble with an uncontested race, but then crossed out both the candidate’s name and the filled-in bubble with two large ‘x’s.
Under the crosses they drew an arrow to another bubble for a different race.
“You can guess pretty well that the voter said, ‘I don’t want that choice,'” Perl said, reviewing the ballot with a bipartisan team of judges. “I guess the (audit) team probably marked it as a choice.”
The discrepancy was likely due to human error and not a problem with the county’s tabulation machines, Perl said.
“In the actual election, that ballot would have gone to arbitration and a bipartisan team would have made a decision based on the state’s voter intent guideline not to count that,” Perl explained. “Then today, when a different team looked at the same ballot using voter intent guidance, they would also make a decision. So those two decisions just didn’t match. And that’s the gap we have here.
In the end, confusion alone was not enough to trigger another audit of Arapahoe’s election results.
After reviewing the discrepancy, Perl and representatives from local Republican and Democratic web councils endorsed the audit’s findings.
Audit results for all counties will begin to be posted on the Secretary of State’s website. website this week. State officials will also review any discrepancies found during the audit.
The deadline for counties to submit their final results to the state is July 19.