Elaina King had mentally rehearsed her lines all night. Now, on what some called the most consequential Election Day, it was time to deliver.

“Hey, would you like to fill out our poll?” she asked a voter exiting the polling station. “It’s for a class project and can totally be anonymous.”

The 20-year-old handed the woman a five-page multiple-choice form.

King was supposed to be in a Spanish class. Connor Morris couldn’t stop thinking about an upcoming computer science exam. Allison Smock was itching to decompress from long weeks of midterms. But for the next few hours, these University of Maryland, Baltimore County students stood in front of a small fire station on Michigan Avenue that had temporarily become a polling place.

This stretch of Baltimore County, called the Baltimore Highlands, seemed to capture the many political heartbeats of Maryland and the United States. On one side of the street, a neighbor pitched a Trump-Vance sign in a small, trimmed front yard by a white Ford F-150. On the other side, a resident staked a Harris-Walz sign by a blue Tesla Model S.

Their job that evening was to take a final snapshot of these Baltimore County voices through an exit poll. Dozens of other UMBC students were doing the same at different stations, all randomly selected to curb selection bias. The poll results are scheduled to be released this week.

In the coming weeks, the results of exit polls will unspool some of the motivations behind local, state and national voting decisions. They will help political operatives mold future campaigns and improve messaging to voters. These results might lift shrouds of misinformation that politicos and pundits sometimes prematurely inject into the zeitgeist. They can also help communities understand what animated their neighbors’ choices at the polls.

This poll will examine the voting behaviors of Baltimore County residents. The region helped Democrat Martin J. O’Malley crush Republican former Gov. Bob Ehrlich in 2006, and cruise Republican former Gov. Larry Hogan to victory in 2014 and 2018. This year, the races here were the tightest among Maryland’s largest counties.

For the day, dozens of UMBC students were scattered across the county with one mission: to collect at least 1,000 responses by 8 p.m., when polls closed in Maryland. This sample size helps minimize the margin of error in polling results, said Ian G. Anson, a UMBC associate professor leading the exit poll project for the college’s Institute of Politics. A larger sample can tighten the margin but offers diminishing returns, he continued.

It was 4 p.m. and the students had gathered about 900 responses.

King repeated her spiel to another voter exiting the station. Anson told his students to emphasize the polls were partly for a class project. He found earlier that day from other students that voters were more likely to respond favorably then. “If you want,” King added, “you can scan our QR code and fill out the form when you get home.”

A woman wearing scrubs whipped out her phone and scanned the code. The link redirected her to a form with 30 questions.

Some of the questions probed voters’ identities. Over the past 12 months, what was your household’s total income? How often do you attend religious services? For about how long have you been a resident of Baltimore County?

Others asked voters about their perspectives on burning issues. In the past year, would you say the U.S. (and Maryland) economy has gotten better, worse or stayed about the same? On the issue of abortion, would you say you are more pro-choice or pro-life?

A few asked people who they voted for in national and local races.

These questions were a product of months of class sessions in which students mulled over every word in the packet, King said. “It was kind of a big brain dump. And then we revised them,” she said.

If these questions weren’t worded the right way, the class could end up with biased responses that would tarnish the outcome of the poll. They notably struggled with one question: As you may know, currently fewer than 1 in 3 members of the U.S. Congress are women. Do you feel satisfied or dissatisfied about this ratio?

“That was a huge emphasis in our class because especially with Kamala Harris and Angela Alsobrooks on the ballot,” Anson said. “We had a hard time trying to figure out how to ask a question that wasn’t ... just a straight up appraisal of Alsobrooks or Harris.”

The class had to cut many questions because they wanted people to complete the form in five minutes or less, Anson said.

“When you’re not paying somebody to complete a survey, if you go beyond something like four or five minutes, then the nonresponse becomes incredibly high.” Respondents, he added, might also rush through the tail end of a lengthy survey. “And so, the response quality starts to degrade really seriously, and you end up seeing results that are less meaningful.”

They closed the day with 1,231 responses.

The results of the UMBC exit poll will come out this week, trailing some national exit polls. This is intentional, Anson said. Early exit polls are notoriously unreliable. For instance, early exit polls showed a majority of white women voted for President Donald J. Trump in 2016. They also showed Hillary Clinton was ready to edge out Trump in the presidential race. Neither were true.

That’s partly because pollsters cannot adjust early data to the final outcome of a race, which is fully tabulated days or weeks after Election Day. So, Anson and his students waited for the final voting data before crunching their data.

While this poll skirts one of the largest limitations of an exit poll, the data isn’t perfect, Anson said. They could not account for mail-in voters who did not show up to the polling places on Election Day, an increasingly large voting bloc that leans Democratic. This year, Harris voters submitted more than three times as many mail-in ballots as Trump supporters, state election data shows, and mail-in ballots comprised about 27% of votes in the county this year.

“But we do weight for the mail-in population as well when we look at those, those precinct totals,” he continued. “So, we do have the ability to sort of massage the data back into shape.”

On a Thursday morning, just over a week after Election Day, Michelle Enomanna and a dozen other students stared at their laptops. Anson had just released a dashboard summarizing the exit poll results and gave them an assignment: “Play around with the dashboard, then we’ll share what patterns we see among voters.”

Enomanna and her partners wanted to see what various age groups thought about female representation in Congress. She fiddled with dashboard prompts for five minutes until she found her answer. “So, for people who are 18 to 24 who are unsatisfied or very unsatisfied with lack of representation of women in Congress, that is around 60%,” she said. Anson approached her and leaned in. “If we compare that to an older generation,” he added, “only about 40% were very unsatisfied or unsatisfied.”

“What are some patterns other people are finding?” Anson asked.

“Harris voters are more religious than Trump supporters,” one student chimed.

“Alsobrooks voters are more worried about guns. Hogan voters are more worried about crime,” another said.

By the end of the class, when Anson said they’d dig into explanations for these patterns, Enomanna was rethinking everything. She had come to UMBC years ago to set up a legal career. That meant investing years in undergrad studying for the law school admission test, then attending law school. Now, she was tempted to pursue a master’s in political science instead.

“Professor Anson, do you have time for a meeting next week?” she blurted at the end of class. “We need to talk.”