A few months ago, you’d have been forgiven for assuming that Wednesday, May 25, would be a day for picking up the pieces of a brutal Republican primary in Georgia. Gov. Brian Kemp was seeking reelection but had clashed with then-President Donald Trump over the latter’s failed efforts to subvert the 2020 presidential election results in the state. Secretary of State Brad Raffensperger, the guy responsible for confirming Trump’s loss in Georgia, was on the ballot on Tuesday, too. And Trump had endorsed former professional football player Herschel Walker for the U.S. Senate, a decision that many Republicans questioned.
So the primaries on Tuesday, May 24, seemed likely to be the culmination of several brawls and fights that would be resolved only after equally rough runoff contests. But the opposite happened: Kemp and Raffensperger locked up majorities right out of the gates, and Walker never faced any significant competition.
It was not the result Trump hoped to see. Yes, he got Walker, but his endorsed candidate against Kemp, former senator David Perdue, was obliterated. Perhaps more frustrating for Trump, Georgia primary voters gave Raffensperger their votes, a lower-ticket race that might have seemed riper for Trump to steer. And while Raffensperger’s margin was narrower than Kemp’s — in the sense that a roadside ditch is narrower than the Grand Canyon — he still pulled out a win.
Here’s an interesting question that has emerged: Does Raffensperger have Democrats to thank?
Consider how the results of these three races overlapped. The Washington Post’s newsroom engineering team compiled voter registration data (including modeled party registration from the data firm L2) and compared it with actual results by precinct in the state. If we compare Walker’s results to Kemp’s, you can see a pattern: Walker did better in more rural, more Trump-friendly parts of the state, while Kemp did better in more urban, less-Trumpy areas.
This overlaps with other factors, like the quite-important metric of education, but let’s just focus on population density for now.
Let’s look next at how Raffensperger’s results compare with Kemp’s. Kemp did better than Raffensperger in basically every precinct, but notice the difference from Walker: Raffensperger’s results generally aligned with Kemp’s. Better in urban places, less well in rural ones.
If we compare Walker with Raffensperger, the contrast is stark. Raffensperger outperformed Walker in most urban precincts. Walker outperformed him in more than 9 in 10 rural ones.
This has a lot to do with Trump. More-rural places have consistently been more Trump-friendly (again, in part because it correlates with education and other factors). But more-urban places are also more-Democratic places, and there are some indications that this was important to Raffensperger’s win.
Analysis from The Post’s Lenny Bronner suggests that perhaps of this year’s 77,000 Republican primary voters had voted in the Democratic primary in 2020. These aren’t all Democrats, certainly; voters could request whichever party’s primary ballot they wanted. There was no real presidential primary for Republicans to vote in that year, so they may have voted in the Democratic primary (as is allowed in the state) simply for lack of another option.
This matters not because 77,000 votes gave Raffensperger a win; he’s up by nearly 220,000 as of this writing. Instead, it matters because Raffensperger is about 54,000 votes over the 50 percent margin he needed to avoid a runoff contest. In other words, take away 54,000 votes and suddenly there’s a faceoff between Raffensperger and Rep. Jody Hice (R-Ga.), with Hice getting Trump’s endorsement and the two having to battle with the spotlight entirely on their race.
Again, we don’t know that Raffensperger was the beneficiary of 77,000 votes from Democratic primary voters. We don’t know that there were precisely 77,000 people who flipped from voting in the Democratic contest in 2020 to the Republican one this week. We don’t know that all of those people would have voted for Raffensperger.
We do, however, see some signs that places with more people who voted previously in the Democratic primary were more favorable to Raffensperger, particularly compared with Kemp and Walker.
Below, we broke out all three candidates’ results in two ways. On the left is the percentage of support in a precinct (vertical axis) relative to the modeled density of Democrats in the precinct (horizontal axis). On the right, we put each precinct into one of five buckets — quintiles — based on the density of Democratic and Republican primary voters.
There’s a lot here. Take a look.
Notice that right-hand chart for Raffensperger. The precincts in the lowest quintile of Republican voters gave the incumbent secretary of state 58 percent of their votes. In the highest quintile of Democratic voters, he pulled 54 percent of the vote. (Two-party density doesn’t add to 100 percent in every precinct, so the quintile groupings are not perfect mirror images.) In the fifth of precincts with the lowest Democratic density and the fifth with the highest Republican density, Raffensperger got less than half of the vote.
Georgia doesn’t require registration by party, making this a difficult question to pin down. But there are important clues: better performance in urban (that is, more-Democratic) precincts, better performance in precincts with more Democratic primary voters and tens of thousands of past Democratic primary voters voting in the Republican primary this year. There was a contested Democratic primary for secretary of state this year, but primary voting is party-specific. That the Democrats had no competitive primaries for governor or Senate may also have spurred more participation in the Republican primaries.
Of course, it’s also true that a margin of 54,000 votes out of 1.1 million cast can have a lot of causes. Even in rural precincts, Raffensperger got a majority of the vote. No doubt some consolation for a candidate who will not be eager to be seen as having earned his party’s nomination thanks to assistance from the other party.
Lenny Bronner contributed to this report.