Rejecting improperly dated ballots disproportionately impacts communities of color in Pennsylvania, data shows

Entry type: Single project

Country/area: United States

Publishing organisation: Votebeat, Spotlight PA.

Votebeat and Spotlight PA have a collaborative reporting and content sharing relationship. Spotlight PA also operates similar to a wire service and so the story may have also been published in other outlets around the state.

Organisation size: Big

Publication date: 2022-11-28

Language: English

Authors: Carter Walker (Votebeat), primary reporter
Chad Lorenz (Votebeat), editor
Carrie Levine (Votebeat), editor
Jessica Huseman (Votebeat), editorial director
Sarah Hughes (Spotlight PA), editor
Aseem Shukla (Philadelphia Inquirer), contributing reporter
John Curiel (Ohio Northern University), assistant professor of political science who provided Bayesian Inference analysis
Thomas Wilburn (Chalkbeat), senior data editor


Carter Walker is an investigative reporter focused on voting and elections. Chad Lorenz is the editor-in-chief of Votebeat. Carrie Levine is Votebeat’s story editor. Jessica Huseman is the editorial director of Votebeat. Sarah Hughes is the state government editor for Spotlight PA. Aseem Shukla is a data reporter with The Philadelphia Inquirer. John Curiel assistant professor of political science at Ohio Northern University. Thomas Wilburn is the senior data editor at Chalkbeat.

Project description:

Whether a mail ballot needs a hand written date to be counted as valid has been a contentious issue in Pennsylvania. Shortly before the 2022 elections, Pennsylvania’s Supreme Court instructed counties not to count ballots without dates or with incorrect dates. This led to some counties publishing lists online of who had submitted improperly dated ballots. We wanted to know who this policy impacts, so we collected the data from various counties and analyzed the demographics of those voters using a variety of tools. What we discovered is that rejecting these ballots disproportionately impacts voters of color.

Impact reached:

This story was republished by news organizations around the state as well as by outlets focusing on multicultural issues. Pittsburgh’s NPR station also invited Carter on air to discuss the story. It was also featured as the front page story on Election Line, a news service focusing on election issue s.

Pennsylvania’s Department of State also said this data went a long way toward “confirming with empirical evidence what we understood to be the case anecdotally,” the agency wrote in a statement after viewing the findings. “This minor voter error appears to impact specific communities of voters more than others, including older voters, low-income voters and voters in communities of color.”

Techniques/technologies used:

The data for this story was released by county governments in an effort to have voters fix their errors and have their votes counted. Votebeat wanted to know who was impacted by the rejected improperly dated mail ballots. We collected the data from the county governments and standardized it so it could be viewed in one file. One county file had been taken down by the time we went to retrieve it and had to be sourced via the Wayback Machine. Voters were then grouped by zip code, and the dataset was combined with racial and poverty demographic data from the U.S. Census Bureau. This new combined dataset was used to determine if a voter submitting a flawed ballot was more likely to come from communities with higher than average non-white populations/poverty rates as compared to the voting population as a whole in the county. A more detailed explanation of the methodology used for this analysis can be found at the bottom of the story.

Pennsylvania’s voter roll does not contain racial information for voters, so knowing the exact number of minority voters impacted by this policy is not possible. This is why we used the Census data method. However, John Curiel of Ohio Northern University was able to use an R program to run a Bayesian Inference analysis on the voter’s names to estimate their races. This analysis provided further confirmation of our findings.

Votebeat/SpotlightPA maintains a collaborative relationship with the Philadelphia Inquirer. Reporter Aseem Shukla had run an age analysis on the data and kindly shared his results. This revealed an age disparity, as well.

Thomas Wilburn, of Votebeat’s sister publication Chalkbeat, turned our findings into user-friendly graphics with Datawrapper.

Context about the project:

Ever since Pennsylvania’s mail-in ballot law passed in 2019, it has been highly contentious. Several lawsuits have stemmed from how the statute was implemented, including over whether a date should be required for a ballot to be counted. This issue was ruled on by the state Supreme Court just days before the election, though they did not voice an opinion on whether the requirement violated the Civil Rights Act. A case raising that very question is currently pending in federal court, and we hope our story will help add quality information to that debate as it moves forward.

What can other journalists learn from this project?

I hope this story will show other journalists the value of cross-newsrooms collaborations and breaking down the barrier that competitiveness can put in front of the goal of informing the public. John’s contribution also shows the value of partnering with subject matter experts when your expertise is lacking.

Lastly, I feel strongly that all complex data stories should include a detailed explanation of the methodology, as is at the bottom of this story.

Project links: