Wisconsin’s first statewide presidential recount found no major problems with the state’s voting system, but it did reveal several errors affecting thousands of ballots that could spur local clerks to tighten procedures, according to a Wisconsin State Journal review of the results.
The recount revealed a yawn-inducing shift in the presidential election results — President-elect Donald Trump extended his lead over Democrat Hillary Clinton by 131 votes and total votes increased by about 400 out of nearly 3 million cast. Recount proponents had raised concerns about Russia possibly tampering with election results, but the recount found no evidence to support such claims.
However, the small net change in votes obscured that there were thousands of both positive and negative swings in the final totals.
At least 9,039 presidential votes weren’t counted correctly on Election Night, and only were added to the official results because of the recount, the State Journal review found. Another 2,161 votes were originally counted but later tossed out for reasons including to square vote totals with the number of voters who signed the poll book.
The more than 11,000 changes to the original vote total represent a minor 0.38 percent error rate out of the nearly 3 million votes counted.
Lessons in errors
State election officials, who cast the 12-day recount as a complete audit of the state election system at no cost to the public, say the changes are mostly the result of human error, not a problem with the voting equipment.
The analysis found one type of voting machine — the Optech Eagle, which processed about 10.6 percent of the ballots in the state — produced an error rate of 0.8 percent, likely because some voters didn’t comply with instructions to use a certain kind of ink or pencil.
Eau Claire County, one of the few still using the machines, is planning to replace them.
“Is an election ever perfect? Are there voters that mark their ballots incorrectly?” Eau Claire County Clerk Janet Loomis said. “But is anything ever intentional by either side, I’d say no.”
Andrea Kaminski, executive director of the League of Women Voters of Wisconsin, which monitors state elections, said the recount showed most discrepancies resulted from errors committed by voters or election officials.
“The problems that took place are things that can be dealt with through better training of local election officials and better design of ballots and better voter education,” Kaminski said.
UW-Madison political science professor Barry Burden said the statewide error rate might represent a small change in the eyes of the public or candidates, but for an election official it reveals areas where there’s room for improvement.
“The lesson is more important for the officials to improve their practices behind the scenes than for elected officials or voters to be alarmed,” Burden said.
Burden added that the error rate was slightly higher than the 0.18 percent rate identified in the 2011 Supreme Court recount. However, that could be the result of more candidates, more write-ins and more voters who don’t vote in every election.
This year’s recount was requested by Green Party presidential candidate Jill Stein, who paid $3.5 million to cover the estimated cost.
As of Friday, with 69 of 72 counties reporting, actual costs totaled $1.8 million, or about 53 percent of what those counties originally estimated, according to the Wisconsin Elections Commission.
Stein unsuccessfully asked a court to mandate the recount be conducted entirely by hand, rather than letting local boards of canvassers decide the method.
The State Journal review found more than 55 percent of the ballots were recounted by hand. The error rate was 0.44 percent in wards that conducted the recount by hand and 0.30 percent in wards that recounted by machine.
The difference in error rate between wards that Trump and Clinton won was negligible — 0.38 percent versus 0.37 percent.
The method for determining error rate is imprecise. The exact total variation can’t be gleaned from the ward-level results reported by the Elections Commission because each ward reports the net number of votes per candidate. For example, if Donald Trump gained three votes and lost two in a ward, the commission reported the result as one net vote for Trump, even though it should count as five separate errors.
Of the more than 11,000 errors identified in the available data, the most common error perhaps can be explained by the 6,481 votes that were deducted from the total number of write-ins.
Under a new state law, election officials don’t count votes for non-registered write-in candidates, but the recount identified several cases in which a write-in vote was cast for a registered write-in candidate or for an official candidate.
Officials counting those ballots on Election Night likely didn’t count them correctly, but recount officials corrected the mistake.
Ink and missing signatures
Other reasons votes were added or subtracted include math errors by poll workers conducting hand counts on Election Night, absentee ballots inadvertently left unopened, disagreements about voter intent and, in the case of the Optech Eagle, voters using the wrong type of ink.
Elections Commission spokesman Reid Magney explained that the Optech Eagle requires ballots be filled in using a pencil or special marker with ink containing carbon.
There are some cases in which voters will either mail in an absentee ballot using the wrong type of ink, or at the voting booth use their own pen that doesn’t contain the correct ink instead of a provided pencil because they are worried someone will erase their mark and change their vote, Magney said.
Election officials are supposed to remake an absentee ballot if it uses the wrong ink, but that didn’t always happen, Magney said.
Other, newer machines don’t have a problem reading the wrong type of ink. Ballots with the wrong ink would likely be counted by a hand recount.
In some cases, such as when the number of ballots exceeded the number of voters who signed a poll book or when an absentee ballot envelope didn’t have complete information such as a witness address or signature, recount officials had to randomly discard a ballot. That’s because ballots can’t be traced to a specific voter.
Dane County Clerk Scott McDonell said one of the common errors he saw was voters filling in an oval for a candidate, and then writing in that candidate’s name in the write-in line, which a machine would have rejected as an invalid double-vote. Recount officials were able to determine a voter’s intent and count that ballot.
McDonell said the key lesson learned from the recount was how accurate the state’s election system is.
“It really aggravates me when people suggest there was illegal voting or the Russians hacked,” McDonell said. “I know that’s not true because of the systems we have in place.”