ATLANTA — Georgia election officials released the results Friday of the statewide risk-limiting audit of the Nov. 8 election. The Secretary of State's Office said the audit confirmed those results.
The audit is required under state law in every even election year. It is performed on one race - in this instance, Sec. Brad Raffensperger chose his own election contest against Democrat Bee Nguyen.
The audits function by having counties pull random batches of ballots and hand-counting them. The overall sample is compared against official machine-counted totals to see if it remains below a risk limit, which in this case was set at 5% - essentially making sure there's no extreme discrepancy revealed by the hand count.
According to results from the Secretary of State's Office, there was an overall discrepancy of just three votes (231,072 counted by hand, 231,069 in the machine total). The hand count produced 21 more votes for Raffensperger than the official count, 18 fewer votes for Nguyen and an exact match in the total for Libertarian Ted Metz.
State officials stress that machine counts are more reliable than hand counts, with most discrepancies a product of human error in the hand count process.
A more detailed summary produced in an Excel sheet by the Secretary of State's Office showed few counties found any notable discrepancy between the hand and machine counts. The Secretary of State's Office said in a release that roughly 85% of batches produced no discrepancy at all.
Of the 49 that did, the office said all but one of the discrepancies were "within an expected margin of error."
In that one instance reported by the Secretary of State's Office, the hand count saw an 11 vote increase for both Raffensperger and Nguyen and a four-vote increase for Metz.
The Excel sheet showed that occurred in a Clay County batch of votes. The Secretary of State's release said that was a likely product of "test ballots getting mixed in with voted ballots from the election."
"Properly-marked test ballots are required to have an identifier, such as a watermark, that allow them to be distinguished from official voted ballots; however, these did not," the release said.
The Excel sheet also showed a 20-vote higher total for Raffensperger in a batch hand-counted in Hall County, as well as one fewer vote for Nguyen and six fewer votes for Metz. The Secretary of State's release did not mention Hall County.
Smaller discrepancies were also reported in a Calhoun County batch (nine fewer votes for Raffensperger in the hand count) and a Fayette County batch (11 fewer votes for Nguyen in the hand count).
The audit process also resulted in Cobb County officials finding a memory card with roughly 800 votes from Kennesaw that had not been uploaded into the county's total, resulting in a change of winner in a close City Council race.