The Trump administration decision to ask U.S. residents whether they are citizens as part of the 2020 census will result in the collection of another level of data about American demographics, but it may also threaten the overall quality of data used by federal, state and private institutions for decades.

“Experts at the Census Bureau warned, and I quote, ‘that it harms the quality of the census count,’” said Rep. Elijah Cummings, D-Md., at a House Oversight and Reform Committee hearing March 14.

The choice to ask about citizenship in the 2020 census has proven controversial, as some see it as an attempt to depress response rates by U.S. immigrant populations and impact congressional reapportionment based on the survey’s responses.

Commerce Secretary Wilbur Ross previously testified that the question was included due to a Department of Justice memo stating that citizenship data would better help them enforce the Voting Rights Act, but communication documents between Ross and his staff revealed that the commerce secretary was interested in getting a citizenship question included before the DOJ request was made.

“I testified truthfully to the best of my ability on my understanding of what the questions were,” Ross said at the March 14 hearing.

Thus far, two federal judges have declared the inclusion of such a question illegal, and the Supreme Court is set to hear and rule on the issue in the coming months.

But according to experts, regardless of the rationale for why the question will be asked, its inclusion is likely to increase costs for the Census Bureau and decrease data accuracy, because those afraid of disclosing their citizenship status will be for less likely to respond.

“Most participants across all audiences said they believe the purpose of the citizenship question is to deport undocumented people, and, for that reason, they think most undocumented immigrants will not participate in the census,” the 2020 Census Barriers, Attitudes and Motivators Study, released in January 2019, said.

“Citizens reported what they believed would be the reaction of undocumented immigrants — namely, that they would either skip the question or ignore the form entirely. Many stated that they would not complete the census, despite being citizens themselves, if people in their household did not have U.S. citizenship.”

The Census Bureau is not allowed to disclose the personally identifiable information it collects to other agencies, such as for law enforcement purposes, but experts say that the perception of the Trump administration as anti-immigrant will lead people to not respond anyway.

“The addition of the citizenship question — and this is the Census Bureau’s data — is going to result in a lower self-response rate for all the households with one or more noncitizens. A lower response rate by five or six percentage points, and that’s a conservative estimate,” Bill O’Hare, president of O’Hare Data and Demographic Services with over 40 years of experience working with the Census Bureau, told Federal Times.

“If you have a lower response rate from people, self-response, that means you have to send out enumerators, which is the most costly part of the census. So adding the citizenship question means that the bureau is going to have to spend more money going after households that don’t respond.”

Despite being the most expensive option, enumerator-based responses are also historically less accurate than self-responses, and enumerators are still not able to reach all of the households that do not respond on their own.

On top of that, citizen children with non-citizen family members in their households are less likely to be counted with the inclusion of a citizenship question.

“If you look at the demography of who lives in a household with one or more noncitizens … the biggest group percentage-wise are young children, zero to four,” said O’Hare.

“The reality is almost all of those are citizens. It’s going to have a big impact on children who are 99 percent [of the time] citizens.”

If the 2020 census data is inaccurate enough, it can have an impact on resource allocation, political representation and other demographic surveys conducted by both the government and private sector.

“It might not be immediately apparent to a lot of people, but almost every survey … uses the decennial census data to benchmark. So when that benchmark is off because of undercounting and because of the citizenship question, it’s going to affect the quality of all these other surveys,” said O’Hare.

He estimated that the inclusion of the question could impact approximately 10,000 political districts, and significant percentages of federal grant dollars.

“Between 2021 and 2031, when the 2020 census data will be used, the federal government will give out somewhere in the neighborhood of $26 trillion to states and localities,” said O’Hare. “So small errors when you multiply by $25 trillion [sic] become big numbers.”