People cast their ballots ahead of the Tuesday, Nov. 6, general election at Jim Miller Park, Saturday, Oct. 27, 2018, in Marietta, Ga.
Photo: Mike Stewart (AP Photo)

Voter suppression in Georgia was one of the most discussed issues during the 2018 midterm elections. Front and center in the debate stood Democratic gubernatorial candidate Stacey Abrams, a longtime voting-rights activist who had dedicated much of her career to fighting for every citizen’s right to vote. Abrams was also running to become the state’s first black governor and the first black female governor in the history of America.

As if it were a too on-the-nose movie plot, Abrams’ Republican opponent in the race for governor was Brian Kemp. As Georgia’s longtime secretary of state, Kemps presided over the state’s elections, voting machines and vote tallying. And as a vote suppressor, Kemp’s record of disenfranchising voters was unmatched.

Kemp wasn’t afraid to publicly warn his colleagues about the dangers of black voters. He disproportionately purged black voters from rolls. He threw out black absentee ballots. He used unauditable voting machines and left voter data unsecured. He essentially got to referee his own match.

Advertisement

And in the end, Brian Kemp won...

We think.

There were so many abnormalities during the Georgia election that many people questioned the validity of the results. There were numerous calls from black voters whose machines were misrecording their votes for some reason. Then, someone discovered a vulnerability in the secretary of state’s “My Voter” website that allowed anyone to have access to the entire voter registration database.

Advertisement

Since the election, Kemp has assumed his seat as governor. Abrams, a rising star in her party, gave the Democratic rebuttal to Donald Trump’s State of the Union address. Everyone has moved on.

But since the election, a watchdog group has been quietly gathering data about Georgia’s 2018 election. They have filed court cases, taken affidavits and analyzed the statistics in an attempt to verify the veracity of Georgia’s election results.

Advertisement

Then something weird happened.

The nonprofit group, Coalition for Good Governance, discovered that approximately 127,000 Georgia voters simply did not have a recorded vote for lieutenant governor. Officials claimed that most of these voters simply left that part of their ballot blank. And for some reason, the “drop-off” (the difference between people who voted and people who skipped one race) was disproportionately Democrat.

Advertisement

They had never seen it before. The drop-off rate in Georgia was higher than almost any statewide office drop-off rate they had ever seen. So they gathered the best minds in the election data field from across country to try to figure out the mystery of Georgia’s disappearing votes. It seemed so random. Researchers researched, voting machine experts looked at the machines’ known vulnerabilities and professors ran analytics and statistical regressions. But no one could figure out the pattern.

And then, just for kicks, they decided to run a statistical analysis by race. And viola, they discovered that an incredibly disproportionate number of Georgia voters in majority black precincts didn’t record a vote for the second-highest office in the state. They found the anomaly was incredibly high in precincts where there were high percentages of black registered voters.

Advertisement

And here is the troubling part: According to the report from Coalition for Good Governance (CGG) and the experts who spoke with The Root, the undervote wasn’t concentrated in Democratic areas. It seemed to specifically happen in black neighborhoods. Even stranger, the black voters’ absentee mail ballots didn’t reflect the drop-off, only the people who voted on election day and people who voted on machines in early voting.

The CGG’s report notes:

The extreme undervote issue occurred at statistically significant levels in 101 of Georgia’s 159 counties. However, the undervotes on voting machines are concentrated in precincts where African American voters make up the majority of the precincts’ registered voters. The rates of touchscreen machine–reported undervotes in such precincts in the Lt. Governor contest are far greater than the undervote rates in non–African American neighborhoods regardless of whether those neighborhoods lean Democratic or Republican. The undervote problem did not happen at the same exaggerated levels in many primarily White neighborhoods that overwhelmingly voted for Stacey Abrams and other Democrats, rebutting the argument that the difference can be explained by party-driven voter behavior.

Advertisement

The Root was given exclusive access to the analyzed data from the 2018 Georgia elections as well as the analyses conducted by some of the top experts in the area. We spoke to the researchers and none of them has a logical explanation for the statistical anomaly.

The statistics professor from Berkeley can’t explain it. The voting-machine expert at the University of Michigan doesn’t know how it happened. The data analyst from one of the leading analytics firm in the world has no answers. The political expert who specializes in black voters can’t explain it.

Advertisement

Either all the black people in Georgia collectively decided to skip a vote on their ballot or something happened with the voting machines. But only the machines where black people voted.

So what happened to all those black votes?


‘Drop-off Rate,’ Explained

In most elections, the voter selects a candidate for an office in a variety of different races. Sometimes inadvertently and sometimes intentionally, a voter won’t select any of the candidates in a race, leaving that race blank. The percentage of blank undervotes compared to the percentage of votes from the top-of-the-ballot race is the undervote or the drop-off rate.

Advertisement

In most places, the drop-off rate increases the further you go down the ballot. Essentially, the higher the position is, the more likely people are to vote in the race.

For instance in 2014, Georgia’s last election:

  • Almost everyone voted for governor.
  • 99.2% of voters selected a candidate for lieutenant governor (So the drop-off rate was .8%).
  • 99.1 percent voted for secretary of state (.9% drop-off).
  • 99 percent voted for attorney general (1% drop-off)
  • 98.4 percent voted for commissioner of agriculture (1.6% drop-off)

In the four previous elections (over 16 years), Georgia’s drop-off rate for lieutenant governor averaged about .8 percent. It had never been above 1.2 percent. In the 2018 election, the drop-off rate was 4 percent or about 159,000 votes. A conservative estimate suggests that that number should be about 32,000.

Advertisement

Graphic: Coalition for Good Government

But curiously, those people who skipped that lieutenant governors race, kept voting down ballot. They voted for the secretary of state. They voted for the attorney general. 63,718 more people voted in Georgia’s agriculture commissioner race than voted for the second most powerful political position in Georgia.

Advertisement

The 4 percent dip in votes for lieutenant governor is three times the normal drop-off rate. When the nonprofit group, Coalition for Good Governance, looked at majority African-American precincts, the drop-off rate was significantly higher, even reaching 13 percent in some areas. So why did so many people purportedly ignore this race?

And why was it mainly the people who live in black neighborhoods?

And almost exclusively the people who voted in person...

On unauditable machines.

It simply can’t be explained.

Why the Lieutenant Governor?

In most states, the lieutenant governorship resembles the vice presidency. It’s usually the second-highest office and the lieutenant governor is usually the governor’s running mate, with both candidates running on the same ticket. But in Georgia and 18 other states, the governor and lieutenant governor run on separate tickets, making it possible for the governor and lieutenant governor to be elected from different parties.

Advertisement

In the 2018 midterm elections, Georgia’s Democratic candidate for lieutenant governor was Sarah Riggs Amico. Amico’s Republican opponent was Geoff Duncan. Although the law requires candidates for governor to run a separate campaign from the lieutenant governor (even if they are in the same parties) as usual, Amico often campaigned with Abrams and Kemp made frequent appearances with Duncan, his GOP teammate.

The lieutenant governor in Georgia is largely ceremonial. He or she presides over the state Senate and casts tie-breaking votes if necessary. The only real power a lieutenant governor has in Georgia is that they get to appoint committee assignments in the Senate.

Advertisement

It’s the second-highest office in the state, but most people don’t think it’s very important. Others say Kemp wouldn’t want a Democrat as his legislative representative.

“There’s no reason anyone would rig an election for lieutenant governor” explained Jason Johnson, the politics editor at The Root. “But anyone who knows Georgia politics wouldn’t be surprised that there were questions about any election involving Brian Kemp.”

Advertisement

The Data

When Coalition for Good Governance first discovered the statistical anomaly, they had no idea it would be race related. They handed over the raw data, including the precinct-by-precinct voting results to Chris Brill and a team at TargetSmart Communications, an industry leader in analyzing data and statistics. Among other things, the company tracks voters, analyzes voter databases and integrates analytics for political campaigns. Brill said he had no idea about the racial component at first.

Advertisement

“I started at the county level,” Chris Brill, TargetSmart’s senior data analyst, told the Root. “I noticed that the undervote was happening all over. There was no particular geographic level that this was happening.”

“It wasn’t until I drilled down to the precinct level and started looking at the undervote in every single precinct, that’s when I started to see a strong correlation between the share of the African-American voters in a precinct and the undervote,” Brill explained. “Basically, as the percentage of the African-American vote would get higher, the undervote would also get higher.”

Advertisement

Illustration: The Root/GMG

In an affidavit, Brill called the Georgia results “extremely suspect and irregular and cast a serious doubt over the accuracy of the final vote and the certified outcome of the lieutenant governor’s contest,” adding that he “finds no reasonable, plausible explanation other than machine malfunction.”

Advertisement

One of the most noted expert in the world on these kinds of statistical anomalies is Phillip Stark, professor of statistics and associate dean of mathematical and physical sciences at the University of California, Berkeley. According to court documents obtained by The Root, Stark believes the results of Georgia’s 2018 election data is highly suspect. The noted expert in post-election election auditing and voting statistics expert often consults in state and federal courts, testified before committees of the U.S. House of Representatives and Senate, and published more than 190 articles and books.

“The disparity in undervote rates by voting technology strongly suggests that malfunction, misconfiguration, bugs, hacking, or other error or malfeasance caused some DREs [direct recording electronic voting machines] not to record votes in the lieutenant governor’s contest,” Stark explained, according to court documents.

Advertisement

But neither Brill nor Stark could definitively say what caused the undervote. They have heard various theories, none of which makes any sense.

One theory is that a racial discrimination lawsuit against someone at the trucking company run by Democratic lieutenant governor candidate Amico caused black voters to ignore her. But Brill doesn’t buy that theory because the stark drop-off rate mostly shows up in precincts where there are large numbers of black registered voters who voted in-person, on DREs.

Advertisement

“The big red flag in the data is the difference in the vote method,” Brill told The Root. “When you look at the absentee ballots, even the African-American precincts, you don’t see this drop-off.”

Other naysayers tried to say that many of the people who voted for Stacey Abrams didn’t care about the other candidates. However, the data show that they voted for lesser candidates down ballot.

Advertisement

“The state commissioner of insurance got more votes than the lieutenant governor,” Marilyn Marks said. “That just doesn’t happen anywhere.”

“I have not seen an undervote pattern that looks like this,” added Brill, who pointed out that the state of Georgia seems to oppose any investigation into the matter. “They just don’t seem to want to investigate it. They don’t even seem curious about why this happens.”

Advertisement

Stupid Voting Machines?

Days before the midterm elections in Georgia, The Root spoke with Khyla D. Craine, the NAACP’s assistant general counsel, who said that black voters who used voting machines were experiencing weird malfunctions. At the time, we wrote:

In some cases, eyewitnesses report that when they attempted to vote for gubernatorial candidate Stacey Abrams, the machines switched their vote to the Republican nominee for governor, Brian Kemp, who is also Georgia’s chief election officer in his current capacity as Secretary of State. Other voters have reported incidences of machines spitting their ballot card out when they inserted it into the electronic voting machines.

“It’s not something that’s new, unfortunately,” explained Craine. “These machines are old but it’s incumbent upon the people running the election to ensure that machines are fully functional.”

Advertisement

The Root also reviewed a pile of affidavits filed by Georgia voters who swore that something funny happened to their voting machines, including some who say the machines switched votes.

Advertisement

There is only one person who spoke to The Root who has seen something like Georgia’s mysterious undervote.

Well...kinda.

“We have seen this disparity in the number of undervotes before,” explains voting-systems expert Matt Bernhard, pointing to a recent Florida case of undervoting in the same midterm election. “It’s not that unusual.”

Advertisement

“But in other cases, like what happened in Florida, there’s normally a readily available, easily explainable explanation for why that undervote occurs. In Georgia that’s not the case,” Bernhard told The Root.

Bernhard is a Ph.D. candidate at the University of Michigan’s Center for Computer Security and Society and serves as the technical expert in another voter suppression lawsuit against Brian Kemp and the state of Georgia, Curling v. Kemp. He explained that the drop-off rate in Florida was due to the layout of the paper ballots. In that case, when elections officials went back and reviewed the data, they saw that the design of the ballot may have confused voters.

Advertisement

In most states, whenever there is a question about the results of elections, the problem can easily be solved by viewing the paper trail. Even electronic voting machines used across the country produce some form of a paper ballot.

Not in Georgia.

Georgia is one of five states that exclusively uses unauditable machines with no paper trail. The majority of Georgia’s machines are touch-screen voting systems. Not only are they old and out of date, they are cumbersome and easily hackable. For each election, Georgia’s voting machines are loaded with software that is controlled by the secretary of state’s office. At the end of the election, the voters have no way of verifying the results.

Advertisement

In 2017, after a security flaw was found in Georgia’s election servers, Coalition for Good Governance and other parties filed a lawsuit against Kemp and other Georgia entities on behalf of five Georgia voters. One of the most important goals of the lawsuit was to examine the data on the servers to see if they had been tampered with.

On July 7, 2017, according to court documents in the case, Curling v. Kemp (pdf), someone wiped the state’s election server clean.

Advertisement

Then they wiped the backup server.

Like Marks, Bernhard believes the only way to find out what happened in the election is to complete a forensic analysis of the voting machines. He has repeatedly called for Kemp, Brad Raffensperger—the current secretary of state—and others to allow professionals to examine the servers, databases, code and voting machines. Referring to one of his favorite quotations, “Never attribute to malice what can easily be explained by stupidity,” Bernhard cautions that the occurrence might not have anything to do with a rigged election.

Advertisement

“I simply haven’t seen the level of sophistication from anyone in Georgia’s government to pull that off,” Bernhard explained. “I don’t have enough information to determine whether this was a bug or an intentional act. But something happened. What I will say is that the nature of Georgia’s system easily allows coverups.”

But no one, not Marks, the fighter; Not Brill, the analyst; not Johnson, the political pundit; and certainly not Stark, the professor and mathematician, believes the Georgia election results are trustworthy.

Advertisement

“Based on my analysis, described above, and my knowledge of Georgia’s DRE voting system used in the November 6, 2018 election,” Stark said in an affidavit, “it is my opinion that the certified results of the lieutenant governor’s race are in substantial doubt.”

And asked to put a number on the likelihood that the state of Georgia will conduct a thorough investigation, Bernhard, the scientist who always speaks in fact and not speculation, quickly offered a mathematical guess:

“Zero percent.”


Here’s what we don’t know:

  • Were other races and ethnic groups also subjected to disparate treatment?
  • Was it limited only to the lieutenant governors race?
  • Was this a malfunction or was it intentional?
  • How can we be sure any other elections weren’t affected?

“Time and time again, Georgia’s state elections officials have proven to be untrustworthy,” Marks told The Root. “It is impossible to say whether the discrepancies are intentional of the result of ineptitude. We know the system has flaws. We know how to fix them. But the state of Georgia refuses to listen.”

Advertisement

Brill, having worked with Georgia’s elections systems for years, says that Georgia officials “just don’t seem to want to investigate.” (When contacted by The Root, the Secretary of State would only say that the “office of the Secretary of State remains committed to fair and secure elections.”)

If CGG’s effort is successful, the group may force the state to allow experts to determine what happened in the election. They are also fighting to force Georgia into buying a voting system utilizing hand-marked paper ballots counted with optical scanners. The secretary of states office has a record of destroying records and obstructing investigation with legal challenges.

Advertisement

If CGG fails, 4 million Georgia voters may never know if the 2018 election was legitimate. Thousands of black Georgians might not have had their votes counted. Oh, there’s one other thing you should know about the lieutenant governor’s race where roughly 127,000 disproportionately African-American votes seem to have vanished into thin air:

Sarah Riggs Amico lost by 123,172 votes.