Trial Testimony Of Top Census Scientist Undercuts Ross Memo OKing Citizenship Question

NEW YORK – The testimony of the Census Bureau’s top scientist was used by those challenging the inclusion of a citizenship question on the 2020 survey to tear apart the official justifications offered by Commerce Secretary Wilbur Ross in adding question.

Census official John Abowd’s first day on the stand focused on internal analyses he and his bureau colleagues conducted in late 2017 and early 2018 about the effect of adding the question, which the Justice Department formally requested last December ostensibly for Voting Rights Act enforcement.

Abowd is serving as a witness for both the challengers and the administration, and is the only government official expected to take the stand at the ongoing trial in Manhattan.

The bureau’s internal analyses found that adding the question would be more costly and produce a less accurate count than other methods of providing the types data the Justice Department was allegedly seeking

ACLU attorney Dale Ho, an attorney for the challengers, asked Abowd if harming the quality of the data was a “bad thing.”

“It’s something we try to avoid,” Abowd said.

Abowd, who has served as an associate director for research and methodology and chief scientist since 2016 but has done work for the Census since the late ’90s, was mild-mannered but exacting, at times quibbling with the premises of Ho’s questions or how Ho employed methodological terms. Much of his testimony went into the excruciating detail of the Bureau’s approach to survey tools such as a “differential privacy” and “noise infusion.”

Yet, Abowd never strayed from the assessments made in those internal Bureau documents advising against the question.

You don’t think it’s a “good idea” to add the question, Ho asked Abowd.

“Correct,” Abowd answered.

Among the claims the challengers make in their lawsuit is that the Trump administration violated that the Administrative Procedure Act, which allows courts to strike down agency decisions if they are around to be “arbitrary” or “capricious.” The challengers sought to use the Abowd’s testimony to show that Ross ignored the advice of government experts in adding the question and misled the public in the reasons he offered to justify it. Ho’s pointed questions forced Abowd to rebut multiple claims that Ross made in his memo announcing the question and go into deep analysis into the flaws of the memo’s reasoning.

The Justice Department was barely able to begin its questioning of Abowd Tuesday, and his testimony will likely continue through most of Wednesday.

A Commerce Department spokesperson sent TPM a lengthy statement that said that, “While [Ross’] decision was ultimately different from Dr. Abowd’s recommendation, the Secretary reached his decision, in part, due to the Census Bureau’s assurances that any drop in self-response rates can and will be remediated by non-response follow operations.”

Ho sought to undercut claims made by Ross in a March 2018 memo announcing the addition of the question that the Census Bureau had not provided “definitive, empirical support” for the belief that adding the citizenship question would reduce response rates.

Abowd testified about a number of studies the bureau did on the response rates to the previous decennial census, which did not ask the question, and longer surveys sent to only a sample of the U.S. households, which did. One such comparison revealed that, by 5.8 percentage points, the nonresponse rates increased at a greater rate among households with at least one noncitizen for the longer survey, compared to the growth in nonresponse rates to the longer survey among all citizen households.

Abowd testified about focus groups conducted in 2018 on the environment around the census and their findings that certain groups are particularly sensitive to privacy issues around the census.

In some of those focus groups, the citizenship question was brought up before even the moderator asked about it, and the participants said that they knew people who would not fill out or return the surveys out of fear of deportation.

“It will be a very difficult group to count,” Abowd said of people who fear deportation.

Ho also asked Abowd about Ross’ suggestion that the Bureau had failed to provide evidence that the citizenship question was specifically causing the increase in nonreponse rates. Abowd testified the Bureau proposal to conduct controlled testing on the question had been rejected.

As part of their analyses, Census Bureau scientists recommended that Ross consider, in lieu of a citizenship question, offering to the Justice Department the administrative data on citizenship that the Bureau already collects. Ross, in his March 2018 memo, announced that he was combining that method with the move to add the question — a hybrid option known in the internal documents as “Alternative D.”

Ho asked Abowd whether “Alternative D” would be more accurate that just using than administrative data alone, as Ross suggested in a line in his memo.

Abowd testified that Ross never discussed the conclusion he had made in the memo with him, and that the Bureau had not analyzed that assessment.

Towards the end of the questioning, Ho returned to a line in Ross’ memo that the question was “necessary to provide complete and accurate data in response to the DOJ request.”

You don’t agree, Ho asked Abowd.

That is correct, he said.

Update: This story has been updated to include a comment from a Commerce Department spokesperson.