Shorter Lines, Stronger Democracy: UML Researchers Take on Voting Inequities

In the 2020 U.S. presidential election, 22% of Black voters waited in line for more than 30 minutes to cast their ballots in person compared with 17% of white voters, according to a Pew Research Center report. 

A multidisciplinary team of UMass Lowell faculty researchers is working on a way to address this disparity while reducing wait times for everyone.

Belleh Fontem, an assistant professor of operations and information systems (OIS) in the Manning School of Business, is the principal investigator on a project that is developing a mathematical model to improve poll worker assignment and voting machine allocation across electoral precincts. 

The two-year project, which includes faculty from the Political Science and Mathematics & Statistics departments, as well as an undergraduate student, recently received an Internal Seed Grant of nearly $11,000 from UML’s Office of Research Development.

“There is documented evidence to show that waiting times are quite high in traditionally minority-dominated areas,” Fontem says. “We are looking for ways to have a more equitable allocation of voting resources.”

The project’s findings could help public officials run elections that are more fair and efficient. That’s important, Fontem says, because voter disenfranchisement can erode faith in democracy.

“If people are waiting too long to vote, it's going to leave a bitter taste in their mouths, and they might be less likely to vote the next time. Or they might think that the system is rigged,” he says. “For social cohesion, it’s best that everyone participates in the electoral process and feels like they had a say in the outcome.”

Leading the project along with Fontem is OIS Asst. Prof. Nichalin Summerfield. Starting this fall, she will collaborate with Political Science Prof. Joshua Dyck to develop a queueing simulation model that shows the relationship between voting resource allocation and average wait times.

“We have to make sure the model is an accurate representation of the real world,” Summerfield says. “How long does it take for a person to check in? How long do they interact with the voting machine?”

Jacob Milligan, a rising junior mathematics major, will then implement the simulation model in MATLAB, a programming platform.

Milligan learned about the project after a chance encounter with Fontem at the North Campus Starbucks, where he was working. Fontem was reading a mathematics book that interested Milligan, and they struck up a conversation. Fontem wound up inviting Milligan to join the research project, a paid opportunity that the Bellingham, Massachusetts, native “couldn’t refuse.”

“I love the aim of the project. Helping make elections run more smoothly is something that I think anyone would want to do,” Milligan says. “And this complements my studies because it teaches me more mathematics. You can never know too much mathematics.” 

Next, Fontem and Milligan, along with Mathematics & Statistics Assoc. Prof. Hung Phan, will develop an optimization algorithm in MATLAB that will iteratively improve the allocation strategy.

“With a simulation optimization project like this, you model something and then pass that information to another model, which tries to make it better,” Fontem explains. “Then it proposes a new configuration, which must be simulated. You pass the information back and forth between the models, and with each iteration, you hopefully improve until you converge on a stable configuration.”

While the algorithm will be customized for voting resources, Fontem says it could be used to optimize other processes such as inventory management or electronic circuit design.

The third step of the project involves sensitivity analysis, which tests the model against potential variables in the voting process.

“Let’s say a district is redrawn and the number of people that go to a precinct increases tremendously. What would happen?” Summerfield says. “Or what if there is a change to the voting machine interface and it takes people less time to vote? We can play with an increase or decrease in those parameters.”

The final phase will be a return-on-investment analysis, which will investigate how the varying budget levels of voting precincts influence the efficiency and fairness of the model.

Although the mathematical model won’t be finished in time to help shorten wait times for this November’s election, Fontem hopes that when the research is eventually published, it will make an impact. 

Fontem, who co-teaches a course with Summerfield called Prescriptive Analytics and Simulation, says their papers “tend to be quite mathematical, with a very narrow audience.” So they plan to work with Dyck on an executive summary geared toward policymakers.

“Hopefully it’s a conversation starter that gets them thinking about these issues,” says Fontem, who hopes to apply for additional funding from the Public Interest Fellowship’s National Security Grant Program.

Fontem notes that the number of democracies around the world has been on the decline over the past decade.

“One of the ways that you can weaken your democracy is to decrease the citizens’ confidence in the process,” he says. “Even if you live in a place with enough access to voting resources and you think it’s not your problem, you’ve got to think again, because if there are people in your state who can’t trust the electoral process, that’s going to weaken social cohesion and trust in public institutions. And that doesn’t bode well for anyone.”