During a Thursday meeting between top Census Bureau officials and advisers, one expert began her presentation on the Trump administration’s last-minute decision to add a citizenship question to the 2020 survey by asking: “What the hell?”
D. Sunshine Hillygus, a member of the the Census Scientific Advisory Committee and a political science professor at Duke University, offered a scalding critique of the decision, blasting the administration for what she says puts census data at risk.
“I want to say in no uncertain terms that I think this is an absolutely awful decision,” said Hillygus, who is on the panel of about 15 outside experts that advise the Census Bureau on data collection, methodology and analysis. “I am dumbfounded that this decision is coming in at such a late date. My view is that this is going to have severe negative implications for data quality and costs.”
Civil rights groups have loudly opposed the Commerce Department’s decision, which was announced Monday evening, saying it was not needed and would discourage minorities from responding to the survey. Undercounting the population could have severe consequences because census data is used to determine how electoral districts are drawn and how billions of dollars in federal funding is allocated.
Hillygus’ comments on Thursday highlight a different type of concern, a worry that the reliability and accuracy of census data, relied on by policymakers, academics and businesses to inform their work, could be compromised.
She directly challenged the Trump administration’s justification for adding the citizenship question, addressing Commerce Secretary Wilbur Ross′ memo on the question. The memo highlighted that the bureau had asked the citizenship question until 1950 and that it was already asked on the American Community Survey (ACS), a more extensive survey that goes out to a smaller portion of the population.
Hillygus responded that the presence of a question on the 1950 census said nothing about how people would respond to it in 2018. As an example, she noted that another survey she’s done work with, the American National Election Study, had an 85 percent response rate in 1956, but researchers were thrilled when they got a 50 percent response rate in 2016.
“The notion that we can look at history and say that there was once a citizenship question on the census form and so it will be an easy thing to slip in, just doesn’t hold water,” she said.
She added that the bureau spent years testing a question about race, so it was remarkable that the bureau would add a question about citizenship without the same level of thorough vetting. She went on to say that the decision had tarnished decades of work to show that the Census Bureau was a nonpartisan entity. Seeing the latest decision destroy some of that reputation, she said, was “depressing.”
Ron Jarmin, the acting director of the Census Bureau, defended adding the question to the 2020 census.
“The citizenship question is in the field constantly in the ACS. The response pattern on that question and other questions looks pretty good. We don’t have a lot of prima facie evidence to say that people don’t respond to a citizenship question,” he said. He added that the bureau would try and incorporate some testing ahead of the 2020 decennial census to try and get a better sense of how people would respond. But he also said there’s no way to really test what the citizenship question would look like in the decennial census because it takes place in an environment unlike any other survey.
Barbara Anderson, a sociology professor at the University of Michigan and chair of CSAC, questioned how one could say that asking about citizenship on the ACS was an adequate substitute for testing it on the decennial census.
“I’m not actually certain at all how relevant that is because the ACS is a sample and the census is everybody. I think people are acutely aware of this difference and the greater potential of what you might call misuse or something else of data from the census than from the ACS,” she said.
Hillygus also asked how the question addition could have impacted morale among career employees in the Census Bureau.
Al Fontenot, the associate director of decennial census programs, said he convened an all staff meeting the day after the Commerce Department announced its decision. He said most of the questions he got from employees were technical and he reminded them of their duty to carry out a good census.
“In a very transparent way, [I] talked to them about our role as federal employees and servants of the people of the United States to ensure that we continue to do a good census regardless of the conditions that we do it under out in the community,” he said.