FOIA Reveals Actual Ballots for City Council, Confirming That a Recount Might Have Changed the Outcome

Above: Jessy Gregg, Mark Meadows, and Lisa Babcock on November 12, 2019, the day they were sworn in (photo by Raymond Holt)

Two days after the November 5 City Council election, we brought you analysis of voter choices based on the uncertified precinct results. Now an analysis of the actual ballots – made possible by the Freedom of Information Act – allows us to see how individual voters decided to cast up to three votes.

We already knew from the unofficial results that two of the four challengers, Jessy Gregg and Lisa Babcock, had each gotten support from more than 63% of the voters. By contrast, the two incumbents, Mark Meadows and Erik Altmann, were virtually tied for third place with support from only about 43% from voters. When the vote was certified, Meadows was awarded the third open seat after a machine count that showed him only two votes ahead of Altmann.

Here’s some of what we find now, with analysis of the actual, anonymous ballots:

A manual recount of the ballots might have given the third seat to Altmann, or it might have again found Meadows to be the third top vote-getter.

The individual ballots don’t show us how the machines read each one, but visual inspection and a comparison of the county-published results to our own human-reading results strongly suggests that a manual recount (by humans) might have led to slightly different final counts in many East Lansing precincts.

Based on what we found, those slight changes in precinct counts could have added up to Altmann (below) having a higher number than Meadows, or Meadows winning the third seat by a greater amount than just two votes.

(Photo by Raymond Holt)

In a way, we already knew this; the declared margin of just two votes separating Altmann and Meadows is so tiny that it is well within an expected margin of error.

That error rate is caused in part by voters not always marking their ballots the way the machines require – completely filling in the circle and not making any mark outside the circle. Here’s a sample of an actual ballot from this election where a machine’s read could well differ from a human’s read:

Of course, human ballot-reading also involves a margin of error, as humans must interpret unclear markings. Using the actual ballots from this election, we found that two samples of human-reads compared to the machine-read showed us a consistent “error” rate of about 1 in 100 ballots from one review to the next.

In other words, we disagreed with the machines about 1 in 100 ballots, and two sets of human reviewers disagreed with each other for about 1 in 100 ballots. That makes a two-vote difference between Meadows and Altmann statistically meaningless.

As it turned out, Altmann could have asked for a recount but did not for reasons he has not specified publicly. Since the time for a recount has now long passed, and the machine-count vote was certified, Meadows is, in fact, the third-seat winner of this election.

There was a pretty clear anti-incumbent/pro-challenger pattern. We already knew this, too, from the outcome, but a review of the actual ballots confirmed it.

Now we know, as we did not before, that the combination of three candidates that received the most votes was Babcock, Gregg, and John Revitte. All three of these candidates were not incumbents, and that combination was chosen by 580 voters, or 12.9% of the total.

More than 5% of the city’s voters voted for these four other combinations of three candidates, all of which included at least one incumbent:

• Altmann, Meadows, and Gregg (499 voters, 11.1%)
• Altmann, Meadows, and Babcock (311 voters, 5.9%)
• Altmann, Babcock, and Gregg (364 voters, 8.1%)
• Meadows, Babcock, and Gregg (351 voters, 7.8%)

Of those who used only two of their votes:

• 437 voters (9.7%) used their votes for the two highest vote-getters, namely challengers Gregg and Babcock,
• 131 voters (2.9%) voted for two of the four challengers in various other combinations,
• 130 voters (2.9%) voted for Altmann plus one of the challengers, and
• 116 voters (2.6%) voted for Meadows and one of the challengers.

Taken together, only 5.1% people voted only for one or both incumbents (Altmann and Meadows) and no challengers. Only 121 voters used only two votes to vote for both incumbents. Another 110 voters voted for either Meadows (79 people) or Altmann (31 people), using only one of their available votes.

All of this tells us that there was a strong wave towards voting for challengers as a class and against incumbents as a class.

Thirty percent of voters chose not to cast all three available votes, with a choice among six candidates. Whereas 3,164 voters (just over 70%) cast three votes, a total of 1,343 voters did not.

Among those who used fewer than three votes, 935 voters used only two of their votes, 378 voters used only one vote, and – surprisingly to us – 30 people did not vote for any Council candidate.

A note on sourcing: Local political pundit Mark Grebner filed a Freedom of Information Action (FOIA) request to inspect all the paper ballots at City Hall. He accepted an alternative offered by the City Clerk – photocopies of all the ballots – and was charged about $120 for paper copies of all of them. Grebner, in turn, shared the copied ballots with ELi at no charge after he was done entering all of the results into a spreadsheet.

We reviewed several hundred ballots and found Grebner’s tallies to be quite reliable. As noted above, there were a couple of small differences in our number in a small number of instances, presumably caused by different reads of the same ballots.

In the end, we can’t say why Grebner and we see slight differences in counts compared to the machine-read tallies, because the ballots don’t tell us how the machines read each ballot. The conclusions above are based on what we feel confident we can say taking into account what the certified results show and what tallies of the individual ballots show.

Read a follow-up to this story by clicking here.