Whether the result of machine malfunction, human error or even fraud, the unexplained voting discrepancies in Detroit last month were not sizable enough to affect the outcome in Michigan of the presidential election, according to a new Free Press analysis of voting precinct records.
In 248 precincts, there were a total of 782 more votes tabulated by voting machines than the number of voters listed as picking up ballots in the precincts’ poll books. That makes up just three-tenths of 1% of the total 248,211 votes that were logged in Detroit for the presidential election. That number was far too small to swing the statewide election results, even in this year’s especially tight race that saw a Republican win Michigan for the first time since George Bush in 1988.

Donald Trump carried Michigan by 10,704 votes, or 47.5% to 47.3%, according to the final results submitted to the Michigan Secretary of State. But in Detroit, Democrat Hillary Clinton trounced Trump, winning 95% of the vote to his 3%.
The Free Press analysis found there were 248 precincts in Detroit where voting machines tabulated more Election Day votes than people who were counted as checking in to vote. The affected precincts represent 37% of the city’s 662 precincts.
Most of those overages were by small amounts — on average about 3 votes — with the largest being 12 votes in a single precinct. Those small numbers, which add up to 782 total spread out across more than 200 precincts, tend to point to human or machine malfunction as the culprit, rather than widespread fraud.
In 158 precincts, the number of ballots tabulated by the optical-scanning voting machines was inexplicably less than the number of people who signed in to vote. At least 362 ballots were not counted in those precincts, even though the voters had been listed in poll books.
 
Altogether, the total of over-counted and under-counted ballots was about 1,144. As a result, nearly 60% of Detroit’s precincts weren’t eligible for recount because the number of ballots in the ballot box didn’t match the number of people listed as voting in the poll book.
The Free Press analysis came from handwritten tabulations logged by the Wayne County Board of Canvassers. The numbers are approximated because notes in eight precincts were illegible or unclear. This is the first time that actual figures for over-counted and under-counted votes have been reported
Detroit’s inability to reconcile its ballots with its voter lists was exposed in the recount requested by Green Party presidential candidate Jill Stein that was later ordered stopped by the Michigan Supreme Court. The discrepancy became national news, including headlines suggesting voter fraud.
Reasons for the under-counted and over-counted votes are unclear, although in some cases people may have signed in to vote, then left before casting their ballots because of long lines. Machine malfunctions also may have played a role; on Election Day, more than 80 optical vote scanners broke down in Detroit.
Detroit City Clerk Janice Winfrey and Elections Director Daniel Baxter could not be reached for comment Sunday regarding these latest findings.
Winfrey told the Free Press last week that the city’s decade-old voting machines broke down and caused problems throughout Election Day and that the city has struggled for years to recruit younger people to work the polls. Most Detroit poll workers are retirees with an average age of 68 and they typically work 15-18 hours on Election Day for a $150 paycheck.
Winfrey said Detroit will be getting new voting machines in time for the 2017 mayoral and City Council elections.
Under Michigan law, precincts cannot be recounted when the number of voters in the poll book doesn’t match the number of ballots in the ballot box. Almost 60% of Detroit’s precincts were mismatched — either having too many or not enough ballots to match poll books — and ineligible for recount, according to the Wayne County Clerk’s Office.

Michigan presidential election recount

 

Detroit wasn’t the only place in Michigan with recount problems. There was at least one ineligible precinct in each of the 22 counties where the recount had gotten under way before being halted by the court, according to Michigan Secretary of State records.
The state Bureau of Elections plans to conduct audits of about 20 Detroit precincts that couldn’t be recounted. Those ballots are to be brought to Lansing for an audit that should last for at least three weeks, said Chris Thomas, director of elections for the state.
“We don’t have any suspicion of fraud. We generally approach this as human error,” Thomas said last week. “We’re going to take a look at them to make sure there’s not a need for further explanations.”
Bill Ballenger, longtime Michigan political analyst and founder of the Ballenger Report, said Sunday that even though the number of questionable votes in Detroit was apparently too small to affect this election, the discrepancies are still disconcerting because the race was so close and they demonstrate the need for an audit.
“If there’s one thing good that came out of the recount petition by Jill Stein, it’s that it revealed there are some problems,” he said.
Ballenger noted how the outcome of the 2000 presidential election between George W. Bush and Al Gore hinged on just 537 votes in the state of Florida. That is fewer than the number of questionable Detroit votes.
“If this election had turned out to be as close as Florida in 2000, this would be a huge story right now,” he said.
Contact John Wisely: 313-222-6825 or jwisely@freepress.com. Follow him on Twitter @jwisely.

Election stats and Free Press findings: