Trust in election results is vital for a healthy democracy. Without it, Portland State University researcher Stephanie Singer said power struggles could lead to rioting and violence.
“If people don’t accept the outcome of an election, that’s a very bad situation,” said Singer, a newly hired assistant professor with PSU’s Hatfield School of Government Center for Public Service. “People can get hurt or even die.”
Singer received a $300,000 grant from the National Science Foundation to study the role of data science in election verification and build tools for election officials to use. She said while it’s not the first time NSF has funded research tied to elections, funding this kind of election verification research marks a new direction for the federal government agency.
Ultimately, Singer plans to build a sustainable ecosystem of research groups, software tools and data banks that allows candidates and the electorate to look at election results and reaffirm trust in the election process. That ecosystem includes a common language used by academics, nonprofits and government organizations, and it provides a framework to investigate results that look irregular.
“If we communicate with each other, and use good data science, this can become just a regular part of American political life and the democratic process,” Singer said.
But Singer acknowledges there are limited resources for election officials to investigate anomalies.
“So how do you figure out what to investigate?” she said.
The project will use data analysis from a hotly contested and controversial North Carolina congressional race as the launch point to develop an automatic analysis. That analysis can indicate if investigations into fraud or ballot discrepancies may be worth considering.
During that 2018 race, officials in North Carolina noticed an anomaly: Republican absentee ballots in one congressional district were unusually high. Election officials found proof of fraud, which led to a new election in that district.
Singer said investigations such as the North Carolina race are part of the necessary process to protect elections.
Over the next year, Singer will take the analysis conducted by North Carolina elections officials and build a tool that can automatically duplicate the investigation on a routine basis.
Further, she plans to merge data published by state election boards with census data and information about election technology — meaning various voting machines — and develop some predictive modeling.
“We can apply predictive analytics to precinct-level data — historical election data, census data and data about voting technology, for starters — in order to identify expected patterns within the current election results,” she said. “Just as an accountant knows to ask questions when actual expenditures don’t match budgeted expenditures, election officials and candidates will know to ask questions when the actual voting breaks an expected pattern.”
Birol Yesilada, director of PSU Mark O. Hatfield School of Government at the College of Urban and Public Affairs, said the election verification project is also important for PSU.
“It marks an important step in the Hatfield School’s research interest in cybersecurity and will enhance efforts for future plans for collaboration with other colleges and departments across PSU,” he said, adding Provost Susan Jeffords asked him to organize and lead campus discussion on cybersecurity and related projects.
The Hatfield School and the Fariborz Maseeh College of Engineering and Computer Science have been pursuing such collaboration and are interested in creating a research center focused on cybersecurity, disaster recovery and collaborative governance as well as undergraduate and graduate degree programs, Yesilada said. This joint-effort will work especially well for this research project.
“There are a lot of universities jumping into this area, but what’s missing generally is the link between hardware, software and public policy,” he said. “In this field, there’s a need for a bridge.”
PSU will also partner with the nonprofit Verified Voting for the project.
“If you want to know what election technology is going to be used in Georgia, for example, there’s only one place that collects that information and that’s Verified Voting,” Singer said.
Marian Schneider, president of Verified Voting, said their technology is a one-of-a-kind resource.
“We’re excited to be leveraging both the additional election technology data we’ve been monitoring and the information out there to shed greater light on how data science can protect our elections,” Schneider said.