Michael Pernick, a voting rights attorney with the NAACP Legal Defense Fund, relies on precinct-level election results to make many cases. He needs that detailed data to analyze whether local election rules have a racist impact. However, New York does not have a centralized database of results for the jurisdiction. When it comes to local elections, many jurisdictions maintain their own data and often don't report it to the state, or even the county.
Experts like Pernick have to travel from town to town, village to village, collecting and cleaning data themselves, putting in hundreds of hours of daily work just to obtain this basic information. I also often spend a lot of money. That is, if only local election offices were willing to provide it in the first place.
New York school districts are particularly vulnerable to discrimination because they often use at-large voting systems that can marginalize residents of color, Pernick explained. However, research is hampered by a lack of accessible data. “There are about 700 school districts in New York, and the only way to analyze them is to go through them one at a time and submit open records requests to each one,” Pernick said. Told. bolt. “Response to requests may be disrupted or delayed, and even the start of an investigation may be delayed by months.”
“Due to these challenges in identifying and collecting data, some jurisdictions teeth “Racism in voting is difficult or even impossible to prove,” he added.
Perry Grossman, director of voting rights at the New York Civil Liberties Union, also acknowledged that his team has spent countless hours collecting data and dealing with local offices that are often resistant or unresponsive. I share his frustration. “There shouldn't be a need to threaten to sue the county to get this data,” he says.
In January, the New York State Senate passed a bill requiring a central, publicly accessible state elections database. However, the bill has since stalled in Congress, leaving supporters worried that the issue will persist long into the next election this fall.
And the problem extends far beyond New York. Precinct-level election data powers the world of election analysis. This is the basis of Voting Rights Act lawsuits across the country.Impossible to prove how badly a map is gerrymandered We don't have this data because we need it to assess district partisanship. It is also used to create all kinds of maps, graphics, and tools that show partisan trends in neighborhoods. By combining district partisan data with demographic, geographic, and income data, we can also address a wide range of questions in political science, including showing that claims of voter fraud in the 2020 election were unfounded. can.
However, there is no organization in the United States that records election results or maintains boundary maps of the nation's 180,000 electoral districts. Many states do not even provide this data for the complete collection of areas within their state borders. Instead, universities, newsrooms, nonprofits, and volunteers collectively spend thousands of hours… every We collect the big elections ourselves.
This is a daunting task for organizations that often lack time and resources, and those who need precinct data are at the mercy of individual counties or local election offices, where data quality can vary widely. I am left alone. It also puts a strain on cash-strapped election officials who are asked for the same data over and over again.
In some states, this process is much easier than in others. For example, the Minnesota Secretary of State's Office compiles precinct-level election results across the state on its website. We also provide digital maps of school district boundaries, called shapefiles.
The changes between states are new york times After the 2020 election. A team of journalists, data scientists, and developers set out to create a map of U.S. electoral districts color-coded according to how they voted.
Four states on the map (Alabama, Alaska, Louisiana, and Virginia) are completely blank, as are large swathes of Idaho, Missouri, Kentucky, and Louisiana.
It took times Miles Watkins, who helped manage the project, said it took three months of full-time data and software development work to collect the data after the election, followed by several months of preparation leading up to election night. That's what it means. “We're pretty confident that when we released the national map, we used all the open data and FOIA possible information possible,” he said. bolt. But despite its efforts, one of the country's most staffed publishers was unable to obtain the data to complete the map.
In the end, more than 10 percent of all votes cast in the election did not appear on the map.
One problem that has proven to be intractable is that some counties are distributing mail-in, early, and provisional ballots rather than distributing ballots based on a voter's residence at the precinct level. The combination is reported as one total. This practice makes it impossible to analyze voting patterns at a more granular level, as was the case with the previous four completely blank states. of times'Map.
Zach Mahafuza, an analyst at the Southern Poverty Law Center who studies voting rights violations in the South, said this lack of precinct data often slows his efforts. In 2021, he was part of a team researching city council maps in Mobile, Alabama. The 2020 Census found that the majority of mobile users are Black. But on the city council's seven-district map, he had only three majority-black districts. Stand Up Mobile, a local voting rights group, has called on Mr. Mahafuza and the SPLC to work with at least four black-majority districts where black residents have a meaningful opportunity to elect the candidates of their choice. We considered whether a new parliamentary map could be drawn. This project required district-level election reports and district maps.
The city of Mobile provided the map in PDF format, forcing Mahafuza to jump through hoops to extract reliable data. “You're going to zoom in on this neighborhood in the north of the city and you're going to zoom in on another neighborhood in the south of the city, and it was very difficult to put all that back together,” he said. He estimated that it took him 80 to 85 hours to remap the city.
Ultimately, Mahafuza's team showed that it is indeed possible to draw such maps. They warned local leaders that they could violate the VRA if they failed to ensure black voters were fairly represented when redistricting maps. The city ultimately adopted a map that included four districts where at least 53 percent of the voting age population was black.
To show that a map or election system violates the VRA's anti-discrimination laws, plaintiffs would need to prove several jurisdictional characteristics, such as racial polarization among the electorate. That is, whether different racial groups actually prefer different candidates. Experts perform these analyzes by looking at the demographic composition and election results of individual districts and inferring the voting trends of different populations. “Without precinct-level election results, there is no way to show how racially polarized the voting situation is,” said Ruth Greenwood, director of Harvard Law School's Election Law Clinic. bolt.
“We cannot enforce federal and state voting rights laws without precinct-level data,” Greenwood added.
In the absence of centralized information, many organizations have emerged to fill the gap. OpenElections, a volunteer group of journalists and software developers, the voting and election science teams at the University of Florida and Wichita State University, the Redistricting Data Hub, and the MIT Elections Lab all contribute resources to collect, standardize, and publish precinct-level elections. I'm pouring it into it.Compiling Results or Area Shapefile Maps.
After the 2020 presidential election, of times The MIT Elections Lab set out to map the nation's electoral districts and create its own database of electoral district results across the country. It took nearly two years for more than a dozen computer scientists and political scientists to complete the project. Just in time to restart after the 2022 midterm elections. He found that there were only about 10 states whose data was organized enough to be registered in the database without much effort. For the remaining states, we wrote code to clean up the data, call county offices to retrieve returns that are not posted online, and use optical character recognition software to retrieve PDF files and other images. We have designed a quality control process to read election returns from and check our work.
Samuel Baltz, a research scientist with the group, recalled that they even found precinct-level recount results in Bonner County, Idaho, from newspaper photos of numbers written on whiteboards.
Some researchers and lawyers, tired of hitting a wall every election cycle, are pushing for reforms that require states to organize complete precinct-level data and publish it in an accessible format. Some people do.
In early 2018, Kansas data scientist Peter Carman emailed state Rep. Boog Heiberger with just such a proposal. He spent months compiling the state's tract data by calling county offices, extracting data from PDF files, and tediously matching tract names between digital maps and spreadsheets. was. He thought the Kansas Secretary of State's office should be required to publish the data on its website.
Mr. Highberger introduced legislation to address these issues, and Mr. Cullman testified in both chambers. He also requested that the spreadsheet be published as a spreadsheet, not as an image of the spreadsheet, including a request regarding the format of the data. In less than four months, Gov. Jeff Collier signed a bill requiring the Secretary of State to post all federal, state, and Congressional district-level data on its website within 30 days of the last election campaign. Karman watched from the sidelines.
More recently, Connecticut mandated a statewide election database as part of its new voting rights law. The sweeping legislation is part of an effort by several states to recreate voting rights protections eroded by federal courts. Pernick, an attorney with the Legal Defense Fund, said the database portion of the law is modeled after a New York state legislative proposal.
The New York bill Pernick and Grossman helped author would establish a statewide database that would centralize and store election data and shapefiles down to the precinct level. The proposal, sponsored by Sen. Zelnor Miley, first passed the Senate in 2022, but did not receive a hearing in the state Legislature before the end of that session.
The latest version of the reform, Senate Bill S657A, is in danger of suffering the same fate, having passed the state Senate in January but now pending again in the House Election Law Committee. The committee will be chaired by Latrice Walker, who also sponsored the bill. Miley and Walker did not respond to requests for comment.
In the meantime, those who needed New York state data were left to their own devices to compile the data themselves. Ben Rosenblatt, an independent political consultant in New York who has worked on Democratic campaigns in the state, has decided to compile a shapefile of districts across the state in 2022. He posted his progress in the following article: twitter As he walked, he announced that the project was completed in December 2023 after a year of collecting and organizing data from each of New York's 62 counties. of times' The 2020 Mapping Project cites Rosenblatt's work as its work. only New York data source.
“We're trying to do it again this year for 2024,” Rosenblatt said. bolt.
Bills containing similar provisions have also been introduced in Michigan and New Jersey. Versions in Florida and Maryland did not advance by the end of this year's legislative session.
If this type of regulation existed in every state, and each state cleaned and posted its own school district data in an accessible format, the resources that outside groups would spend on this task would be greatly reduced, if not completely stopped. will change to
bolt Watkins asked how long it would take his team to complete the data. of times' Map the project with these conditions. It probably took “a few days,” he said. He paused to think, then added, “Maybe a week.”