Many democracies around the world require every eligible voter to fulfill their civic duty and cast their votes during national elections. If you were an Australian, Belgian, Singaporean, or Uruguayan citizen eligible to vote in your country’s elections, you could even expect to incur a fine from the government for failing to do so. Some countries set their election dates on weekends or declare national holidays so that more voters can participate.
The United States doesn’t mandate voting or fine voters for not participating, and our election day is on a business-as-usual Tuesday. Even so, in 2020, we saw the highest rate of voter turnout—66%—in a national election for more than a century.
It’s great news for a democracy when more eligible voters vote! But why do some people sit out elections—and why aren’t all ballots from eligible voters counted? With the 2024 presidential election just weeks away, UCS Research Associate Liza Gordon-Rogers and her team with the Center for Science and Democracy got their hands on some rare precinct-level data, which they’re using to make the case for a more inclusive democracy, where voters show up and their votes are counted.
AAS: You and your team analyzed data from the 2020 presidential election from eleven counties in seven battleground states that are likely to be significant in the upcoming election. Since your goal has been to inform science-based best practices for future elections, why did you start with this data from 2020?
Liza Gordon-Rogers: Election data transparency is among the core issues we’re working on at UCS, in service of equitable and fair elections. The reason it’s so important is that we can’t know what’s working, or not working, in our elections if we don’t have the data to figure that out. Without data, we don’t know whose ballots aren’t being counted and why not. So, before we can remediate the problem or suggest solutions, we need to know what the problem is. One of our goals for this project has been to use this data to try to put the upcoming 2024 election into context, regardless of what plays out.
AAS: And what did you learn from the data you analyzed? What stood out?
Liza Gordon-Rogers: Well, unfortunately, our results confirmed what we had anticipated. For the states and counties we looked at, we found that precincts that were majority Black, majority Hispanic had lower rates of turnout than majority-White precincts. And this is significant even when we control for county and state-level fixed effects—which means it’s not just characteristics that are unique to the state or county. It’s not just election law. It’s not just income.
Here’s one example: there’s a precinct in Wisconsin, in Milwaukee County, that was 91% White in 2020. That precinct had 84% of its registered voters turn out that year, the highest in the county. Four miles east, in a precinct with far more Black and Hispanic voters, only 12% of registered voters cast ballots in 2020. We also found that voters living in those low-turnout precincts—again, those that are disproportionately majority-Black and majority-Hispanic—were more likely to have their provisional or absentee ballots rejected. And majority-Black and majority-Hispanic precincts were twice as likely to have a higher incidence of ballot rejections compared to majority-White precincts.
AAS: Do you have thoughts on what can depress voter turnout? And what’s going on with ballots being rejected?
Liza Gordon-Rogers: I have so many thoughts. A lot of people think voting is easy. But you have to be registered to vote. You have to make sure your name is still on the rolls, during a time when it’s easier than ever to kick people off. You have to find your polling place. You have to go to that polling place during the hours they’re open. You may have to take off work, and if you’re paid by the hour, that can be really difficult. If you have kids, you have to find childcare. If you don’t drive, you have to find transportation some other way.
Maybe a state has a voter ID law in place, and a registered voter who’s just gone through all that trouble to get to their polling place forgets their ID and has to fill out a provisional ballot. And maybe since it was already such a hassle to get there, the voter couldn’t come back with their ID, and then that provisional ballot doesn’t get counted down the line.
Or maybe they don’t even have the right kind of ID. People assume that everyone has a government-issued ID, and that’s just not the case. If you live in a city, you might not need to drive, so you never get a driver’s license. Maybe you don’t have the resources to travel, so you don’t have a passport. So you want to vote anyway, and now you have to apply for an ID. And maybe the office isn’t open every day, or maybe they’re only open when you work. You have to have all your paperwork in order, your birth certificate, or whatever required paperwork there is, and go through that whole process. It takes cumulative, additive resources to show up and vote. And if people don’t have those resources, it’s difficult.
On the other side of the equation, the people who run elections locally are also dealing with a lack of resources for maintaining efficient, well-staffed operations on Election Day. Those resources come from states to counties to the precinct level, and federal funding can be unreliable.
When it comes to ballot rejections, some of what’s happening is because of the lack of centralized laws around provisional ballots and ballot curing [fixing errors such as missing signatures, or incorrect addresses]. These are precinct- and county-level policies to decide whose votes get counted when there are errors or challenges; there’s no statewide policy on that. And because of that, for counties that allow ballot curing, every election year, it’s just chaos: court cases trying to decide whether election officials should allow people to fix errors, and if they do, what errors count as being allowed to be fixed and what aren’t. It’s a mess.
AAS: I’m glad you and your team are doing this work to find an evidence-based path out of the mess. Speaking of, apparently it was a feat to get this precinct-level data that you’ve laid out in the map accompanying your research. You’ve explained why the data is important. Can you explain why it was important to view this information—voter turnout, and percent of ballots rejected—at this granular level? And why is it so hard to access this data?
Liza Gordon-Rogers: What we decided to do was focus on the most pivotal states in the last few election cycles and zoom in on large populous counties. So, for example, in my home state of Pennsylvania, we looked at Allegheny County and Philadelphia County. The county-level data give us more information than just cities; for example, focusing on Allegheny County allowed us to view suburban and rural data as well. It’s a better sample of people and communities. We need to get to that granular level to understand people’s real experiences when they go to vote.
There are national studies that find that people of color are disenfranchised in our election system, there are state- and county-level studies that find the same results. This may be the first precinct-level study that confirms we see the same outcomes at that level.
And yes, it was a struggle to get the data we did. It really shows the need for better election data transparency. We examined 11 counties in seven states for turnout, but we could only look at data on rejected ballots for eight counties across five states, because we just couldn’t get that data. Even working with election officials, we couldn’t get it. That’s not to say it doesn’t exist. But there are so many hurdles for sharing and generating data when it comes to elections.
Again, decentralization plays a role, as well as a lack of training and resources for local administrators to store and generate the data. Sometimes even if administrators do collect data on the variables we’re interested in, they are only able to share the data as PDF images which, as many scientists know, are difficult to work with.
In some cases, even when election officials have this data, they’re not sure how to present it or use it in helpful ways. For example, data can be compiled in spreadsheets that are totally unintelligible or unusable to the public. This is part of what we’re working on with the Election Science Task Force—to be able to collect this data and then represent it in a way that people can understand.
AAS: What would you say to any dispirited poll workers, election administrators, and/or volunteers who’ll be staffing the polls on Election Day and may be feeling some kind of way about your findings?
Liza Gordon-Rogers: Well, first, I don’t in any way want anyone to think that election administrators, workers, and volunteers are intentionally gumming up the system and not counting people’s votes! That is not happening. It’s a function of how complex and difficult the work is, how decentralized and dispersed the tasks are.
I also wouldn’t want to discourage anyone from signing up to work polls on Election Day—which is an interesting way to get hands-on experience in how our democracy functions. The US Election Assistance Commission provides information on how to sign up and the rules in various jurisdictions for doing so.
Most election administrators are just trying to do a good job. And unfortunately, the barriers they face to doing a good job have increased as legislators in some states pass increasingly restrictive voting laws. UCS is using science to fight unfair restrictions on voting, and this effort is something our supporters can work with us on now and in advance of future elections.
For more of Gordon-Rogers and her team’s findings, and to explore voter data using the interactive map they created, check out Race and Representation in Battleground Counties.