The trial over the Trump administration's plan to ask about citizenship on the 2020 U.S. Census began on Monday with expert testimony explaining how the change may cause an undercount on the national survey among Hispanic households, as well as those of noncitizens.

Attorneys for the U.S. Department of Justice, meanwhile, began their effort to rebut the New York attorney general's challenge by trying to poke holes in the research and prior testimony about the effect of a question on immigration status.

The case is before U.S. District Judge Jesse Furman of the Southern District of New York.

Dr. Dione Sunshine Hillygus, a professor and researcher at Duke University, testified that research and past analyses suggest that Hispanic and noncitizen households will be less likely to be counted in the census if a citizenship question is added, despite follow-up efforts from the U.S. Census Bureau to include them.

Nearly six hours of testimony was led by John Freedman, a partner at Arnold & Porter, for the plaintiffs in the case. Freedman is part of the team representing the New York Immigration Coalition. The New York Civil Liberties Union and American Civil Liberties Union are also representing the group.

Their lawsuit was consolidated for trial with litigation over the same issue from New York Attorney General Barbara Underwood, who was present for about the first two hours of the trial on Monday. Underwood is leading a coalition of 18 states in the litigation.

Underwood said to reporters outside the courthouse during a break in the morning that the testimony from Hillygus was, so far, complementing the claims they've alleged in their lawsuit.

“I think so far the testimony is confirming what we have said in our complaint, and that's the purpose of the trial isn't it? To establish the facts you've alleged. So, we think that's happening here,” Underwood said.

Hillygus testified that there are several complications involved with adding a question about immigration status to the census.

She first claimed that asking about citizenship would decrease participation in the survey, which is a key part of the plaintiffs' case. They have argued that asking about citizenship will lead to an undercount in states with high immigrant populations, like New York. That could lead to fewer representatives in Congress and the Electoral College in those states. It could also reduce federal funding in areas like education and health care.

Hillygus said that both noncitizen and Hispanic households with legal status may be deterred from responding to the census if the citizenship question is added. The question would present a so-called “burden” to those residents, she said, who may view it as too sensitive to answer.

“We know there are a lot of different factors at play,” she said. “Broadly, the one that a lot of people talk about is the burden of the survey.”

Burden doesn't necessarily mean the number of questions a respondent has to answer, she said. It can also mean the stress caused to a respondent by answering a certain question. In the case of asking about citizenship, noncitizen or Hispanic households may interpret a certain level of stress from answering, Hillygus said.

“A citizenship question is going to make conducting the census a heck of a lot more difficult,” she testified. That's because, according to her, both independent research and analyses from the Census Bureau suggest that Hispanic and immigrant citizens—not just noncitizens—are wary about answering a question on citizenship because they fear how that information may be used. Even if the immigrant has legal status, Hillygus said, they may be reluctant to answer a question about citizenship.

A report from Dr. John Abowd, the chief scientist at the Census Bureau, proves as much, she testified. The report used data from the census bureau's American Community Survey—which already asks about citizenship—to estimate that such a question on the decennial census may depress the response rate of noncitizen households by 5.8 percent, which Hillygus said was a conservative estimate. The American Community Survey is conducted annually through a smaller sample of the population rather than every 10 years.

“If offers evidence from the American Community Survey that suggests that a citizenship question is sensitive for Hispanic and noncitizens and offers the likely impact on the response rate for Hispanic and noncitizen households,” Hillygus testified.

The Trump administration has rebutted that point in recent filings by saying even if Hispanic and noncitizen households do not initially respond to the census, the government has procedures in place to collect their data.

That begins with repeated attempts to have a household complete the census without additional measures. If the household chooses to forego the census, the government sends a person to their address to collect their responses. If that's unsuccessful, the government will then rely on someone else for the data, like a landlord or neighbor. The Census Bureau may also use other governmental administrative records to determine a household's characteristics.

Hillygus testified that if a household is unlikely to respond to the census because of a question about citizenship, they will also be unlikely to respond to the bureau's methods for following up, including an in-person visit. She also stressed that governmental administrative records may not exist for noncitizen households and that proxy respondents, like neighbors or landlords, may not report accurate data on a nonresponsive household.

“It's not just about getting inaccurate characteristics of that household, it's about underestimating its size,” Hillygus said.

She also criticized the Trump administration for not pre-testing the question among a sample before deciding to add it to the census, as is required whenever a question is added to the survey. The Commerce Department justified its decision not to pre-test the question by noting that it's already asked on the American Community Survey.

Tom Linson, an attorney with the DOJ, questioned Hillygus on her research during her cross-examination, repeatedly pointing out that she had not done any of her own, independent research on how the citizenship question would affect census participation rates.

He also noted during questioning that—aside from the Census Bureau's own research—the studies she used to compile her report for the plaintiffs did not directly review the impact of a question about citizenship. They, instead, looked generally at the confidentiality concerns of those asked to answer sensitive questions.

“You're not aware of any quantitative evidence about Hispanics and the citizenship question?” Linson asked Hillygus. She responded in the negative.

Kate Bailey, lead attorney for the Trump administration, also had her shot at cross-examining Jennifer Van Hook, another expert witness from the plaintiffs who testified by affidavit.

On the stand, Bailey went after Van Hook for a handful of errors that were included in her original report on the potential effect of a citizenship question, which was submitted in September. Van Hook used the Census Bureau's Current Population Survey to show nonresponse rates for different races. That survey also asks about citizenship at least once.

The problem, as Bailey pointed out, is that Van Hook initially mixed up some of the information about the Asian and Hispanic populations and their nonresponse rates. The errors were repeated in the body of her report and during her deposition. She has since submitted an amended version of her report and deposition, but Bailey didn't let her off the hook.

“Sorry, they look a lot alike,” Bailey said multiple times while Van Hook was trying to differentiate between a graph with errors and the corrected graph.

Freedman, arguing for the plaintiffs, noted during his questioning of Van Hook that the error was a small part of her report and did not affect her summary and conclusions.

Executive Deputy Attorney General Matthew Colangelo, who is leading the plaintiffs' case with senior trial counsel Elena Goldstein, said they expect to wrap their case by next Tuesday.

READ MORE: