An LPM News investigation published last week found that 1,800 households may have been given the wrong ballot during the May primary election, in some cases wrongly preventing people from voting in state and local races.
The findings were a result of a combination of data analysis and interviews with voters and local officials. Among other things, the story highlighted multiple examples where voters were given ballots that did not include races they should have been allowed to vote in. There was one instance — a state House primary in Shively — where the five registered Democrats prevented from voting could have impacted the outcome.
After the investigation was published, Jefferson County Clerk David Yates held a press conference where he announced his office would be conducting a house-by-house audit ahead of the general election in November.
Data reporter Justin Hicks said the problems his analysis uncovered are not an example of voter fraud: No one intentionally cast an incorrect ballot and there's nothing to suggest malicious intent in the clerk's office's errors. He also pointed out that precinct assignment mistakes like those in Louisville have been seen in Virginia and Georgia, as well.
"If you're concerned about election fraud, I would say, also be concerned about data accuracy, and be concerned about all the things that lead up to you casting a vote and making sure all those processes are right," Hicks said.
LPM News Managing Editor Amina Elahi spoke with Hicks and city government reporter Roberto Roldan about how one persistent voter who spoke out about her experience at the ballot box, sparked a journalistic investigation and an official audit of voter precinct assignments.
A transcript of that interview, edited for length and clarity, is below:
AE: Let's set the table with what we found. Justin, your and Roberto's reporting showed that some 1,800 Louisville households may have been assigned to the wrong voting precincts. What does this mean?
JH: So, precincts are the building blocks of elections. You can kind of think of them as the legos that snap together to form political districts, and that's Metro Council districts, state House districts, any of your political districts, they're made up of precincts. Precincts tell you where to vote. They tell you what kind of ballot you can get. And so, what we found was that some people were getting ballots for the wrong precinct. Even though they lived in one precinct, they would be getting a ballot for another, so that's what we would consider to be an incorrect ballot.
AE: Roberto, how did you first learn this was a problem?
RR: Yeah, it all sort of came to us from a resident in the Wilder Park neighborhood, Ann Gilly. She came to us around midday on primary day, and we went back and forth with the clerk's office. She said she was given a ballot that didn't include her Metro Council race, she was told by multiple people that she might have just been mistaken, that she was in the wrong race, but she felt really strongly that she was correct and that she was being wrongly prevented from voting for her Metro Council member in District 21. Eventually, after going back and forth, the clerk's office did acknowledge that there were issues with her address, that it was assigned to the wrong precinct and that she was prevented from voting.
We had another resident from Wilder Park reach out via Facebook the day after the primary. We did some additional digging and realized it was an issue with 44 addresses in that Wilder Park neighborhood, in that area. But it didn't make sense that this issue would just be limited to Wilder Park and nowhere else. So, that's when we decided to do even more digging and see if this was a county-wide issue.
AE: Yeah, I remember speaking with both of you after election night, and we all had this sort of lingering question: What if more Louisville voters were being sent to the wrong precincts? Then our next question was: How can we figure this out?
So, Justin, as our data reporter, a lot of that fell to you. Can you walk us through how you untangled this mystery?
JH: So, we used a tool from the clerk's office called "Where Do I Vote?" It's the tool that the clerk's office has used for a while. You enter your address, and then it tells you where to go vote, and a couple other things. We realized that you could see your precinct based upon that "Where Do I Vote?" tool and that would be the precinct that matched up with the clerk's data, which was errant data. So we knew that we had this tool.
The next thing we needed was a whole bunch of addresses to enter into this tool, so it turns out our emergency management agency and LOJIC, our mapping agency here in Louisville, actually have a database already that was public of about 450,000 addresses in Louisville. So, we took those addresses and wrote a program that one by one entered each address into that "Where Do I Vote?" tool, and then it logged what precinct the clerk's office said it was supposed to be in, and then we compared those to other data that was a little bit more accurate about the location of those houses relative to precincts to see if there was any discrepancies. That's how we got to the 1,800 number, because there were 1,800 households where the clerk's data just didn't match the address data.
AE: Roberto, how did the Jefferson County Clerk's office respond to your findings, first when you brought them in to talk to them about it, and later when the story actually went public?
RR: Yeah, we did an interview with Clerk [David] Yates, and, even before that, we brought in some folks from the Clerk's Office's IT and media relations staff. They came into the newsroom, and we presented them with our findings and explained how we did our analysis. They did seem genuinely surprised by what we had found.
We provided them with a few examples to look into, including the big one, which was voters in a tight House district race in Shively who were prevented from voting in that election. We met with Clerk Yates the day that we published, and he confirmed some of the examples, including that state House race in Shively, and a handful of voters in Algonquin who were wrongly allowed to vote in a Metro Council race for a district that they weren't actually in.
Clerk Yates went on to hold an emergency press conference the next day to alert not just us, but all the news stations that they were going to do this audit and he thanked Justin and I, and LPM, for bringing this issue to him.
AE: So, it was both that people were prevented from voting in certain elections and also voting in elections that they shouldn't have been?
RR: Correct.
JH: Yeah.
AE: What did you learn from the County Clerk's office about how long this might have been a problem?
RR: According to Clerk Yates, this issue goes back to roughly like 2022. The city, every 10 years based on the U.S. Census, redraws political boundaries and precinct boundaries in Louisville to make them all roughly equal in population. What it sounds like happened is that someone was manually inputting these changes to addresses and the precincts that they were assigned to, a lot of these errors were just sort of both human error and maybe an error in some of the mapping that they have.
AE: There are a few months to go until the general election. So, what's next here?
RR: The clerk's office is promising a thorough audit. They say they're going to go house by house. Right now, they say they have eight people working on that, but as we get closer to November, those people have to do their actual jobs. So, Yates says he plans to issue an RFP, a request for proposals, to try to get a third-party company to come in and help them sort this out.
The election is in November, but actually they really have to sort this issue out by October, because that's when the clerk's office generally sends out those postcards that tell you where to vote. That's based on your precinct. Your precinct decides where you vote, so they obviously don't want to send out a postcard that tells you to vote in a location that's not the location you should be actually voting in.
AE: Roberto, this issue of voters being sent to potentially the wrong precincts, why is it important?
RR: I think there's a couple reasons. The first being that we see in the state House race that was decided by just five votes, these relatively small discrepancies can have an actual impact on potentially the outcome of close races.
But even when it doesn't affect the outcome, residents being allowed to vote for people that represent them is the most fundamental right, and it's what makes a representative democracy work. Whether it's a local Metro Council race, an election for state representative or state senator, residents have the right to make themselves heard at the ballot box, and I think just saying, 'Well, it's a dozen people here,' or 'It's residents of one apartment complex there,' does a real disservice to democracy. I mean, if we say, as a country, that every vote matters, then disenfranchising one person through these data errors has to be fixed.
AE: What's next for you guys in reporting this storyline?
RR: I think at this point it's mostly just looking at what is the reaction from folks at the state level and what appetite there might be to try to do some things that could prevent this from happening in the future. One of the things that we found in our reporting is there isn't a lot of guidance, there isn't a lot of administrative regulations, and there isn't really any state law that says clerks need to be checking the accuracy of their voter precinct assignments after you have these big changes that happen during the redistricting. So, we're sort of taking a look and seeing what might be helpful to prevent something like this from happening again.
JH: And specifically in the case of House District 44, you know, waiting to see if there's anything that could be done legally to somehow verify the results of that election. There was a five vote difference between the winner and loser, and there was a five-voter gap in who was allowed to vote, so it'd be interesting to see if there is anything that can be done to solve that past election, as well.
Copyright 2026 LPM News