The Washington PostDemocracy Dies in Darkness

Census Bureau says new test results show citizenship question would not have affected response rate

The 2020 U.S. Census will not contain a citizenship question. (John Amis/AP)

Preliminary results of a test conducted over the summer found that including a citizenship question on the 2020 census would not have changed the self-response rate, according to a blog post the Census Bureau posted Thursday.

However, the bureau said that for Hispanics, there was a statistically significant difference of 0.3 percent between households that received the form with the question and those that received the form without it.

The test, which was sent to 480,000 housing unit addresses across the country, was conducted as the administration was embroiled in litigation challenging the addition of the citizenship question, which Commerce Secretary Wilbur Ross announced last year would be added to the test.

One criticism from census experts and civil rights groups was that the question had not gone through the rigorous field testing that census questions typically undergo before being added to the survey. By the time Ross said the question would be added, field testing of the census was already underway.

The testing of forms with the citizenship question began in June, even though the results would not be available before the July deadline to begin printing the actual forms. At the time, it was not clear whether the question would appear on the survey. But in late June the Supreme Court effectively ruled against the question, and in July the administration said it would drop efforts to include it on the 2020 census and would seek citizenship data through existing administrative records.

The government had said it needed citizenship information to better enforce the Voting Rights Act, but census experts, civil rights organizations and the bureau’s chief statistician said inclusion of the data would likely suppress the response rate among immigrants and minorities. Census data is used to determine congressional apportionment, redistricting and hundreds of billions of dollars in federal funding each year.

The bureau did not release the full report of the test results, saying it will do so “when [it is] available.”

When testifying before Congress, Ross had struggled to defend his decision to add the question. He said in a statement Thursday that the test results were “gratifying news to those who supported its inclusion.”

But Terri Ann Lowenthal, a former staff director of the House census oversight subcommittee, said Ross’s comment ignored the bigger picture.

“Self-response rates are one narrow slice of what it takes to produce an accurate census, and the secretary’s suggestion that the test proves that a citizenship question would not have affected accuracy is misleading,” she said.

Noting that the bureau would need to follow up with roughly half of all households going door to door, she said, “I still think it’s very likely that those households would be harder to count accurately and that the citizenship question would have led to a significant disproportionate undercount of immigrants, self-response rates aside.”

Dale Ho, director of the ACLU’s Voting Rights Project and an attorney for plaintiffs who challenged the question, said there was not enough information in the bureau’s announcement to vindicate Ross’s decision, noting that internal bureau analyses had found the response rate among Hispanics and noncitizens would suffer if the question were included.

“The Census Bureau indicated that its ‘preliminary analysis suggests that in some areas and for some subgroups, there were lower self-response rates for the test form with the citizenship question than the test form without the citizenship question,’” Ho said in a statement. “There’s not enough information in its blog post to understand what precisely that means. But all other research to date by the Census Bureau has indicated that adding a citizenship question to the census would depress responses among noncitizens and Hispanics.”

The Leadership Conference on Civil and Human Rights, which opposed the question, said the environment in which the test was conducted may have affected results.

“The test took place in a confusing environment, given the timing of the Supreme Court ruling on the legal challenges to the citizenship question, and the president’s reluctance to accept the Court’s ruling,” the organization said in a statement. “We don’t know how that aftermath of the Court’s ruling affected willingness to respond or not — and how it will play out in 2020.”