SAN FRANCISCO (CN) - The U.S. Census Bureau's top scientist on Friday disputed claims that U.S. Secretary of Commerce Wilbur Ross decided to add a citizenship question to the 2020 Census before weighing all the evidence.
"He did not appear to have made up his mind," U.S. Census Bureau chief scientist John Abowd said in court Friday, referring to a January 2018 meeting with Ross.
Called to testify as the first witness for the defense, Abowd explained how Ross asked "detailed" and "sophisticated" questions about finding the best source of citizenship data for voting rights enforcement at an over-hour-long meeting last year.
Emails and records that came to light through litigation show Ross was pushing for a citizenship question to be added since he joined the Trump administration, and he asked the Justice Department to write a December 2017 letter justifying the need for citizenship data.
U.S. District Judge Richard Seeborg, who is overseeing the bench trial in San Francisco, noted in a ruling last month that the documents leave "little doubt" that Ross "entered office with an interest in reinstating the citizenship question."
On Friday, Seeborg asked Abowd how he came to the conclusion that Ross was keeping an open mind.
"We spent a lot of time on the costs and consequences of asking the question," Abowd said. "He didn't appear to me to have made up his mind."
Speculating about another person's state of mind is not usually permitted in witness testimony, but lawyers for the state of California, San Jose and other cities challenging the census policy raised no objections.
California claims the citizenship question will discourage census participation in a state where Latinos make up nearly 40 percent of the population. An undercount could lead to a reduction in federal funds and congressional representation for the state and its cities.
Abowd, who also testified at another census trial in New York last November, acknowledged that adding a citizenship question will lower self-response rates and increase costs for census agents that have to visit non-responding households.
However, the chief scientist insisted that follow-up efforts could mostly mitigate any potential harm to the accuracy of the count.
Abowd testified that in a worst-case scenario, based on the bureau's conservative estimate for a drop in self-responses, it would cost the government an extra $137.5 million, well below its nearly $2 billion in contingency funds.
The scientist also argued that the Census Bureau would only have to use the last-resort, imperfect method of "imputation" for up to 0.4 percent of U.S. households. Imputation is a process that substitutes the size of a non-responding household, if no records can verify who lives there, with the number of people residing in a similar and nearby home.
Earlier this week, Matthew Barreto, a survey expert for the plaintiffs, testified that using imputation for homes that don't respond for fear of deportation leads to an undercount of Latinos and non-citizens. That's because the non-responding households tend to be larger than responding ones, according to Barreto's research.
Barreto also warned about "geographic clustering," a phenomenon that occurs when several homes in the same area don't respond. That makes fewer similarly situated homes available for imputation, resulting in biased outcomes, Barreto explained.
On cross examination, Abowd acknowledged that imputation can produce biased results in those scenarios, but he insisted that was merely a "theoretical consideration" not based on evidence.
Representing San Jose, attorney John Libby, of the Los Angeles-based firm Manatt Phelps & Philips, pressed Abowd about a research study that found 53 percent of respondents thought census data was used to target undocumented immigrants, even though census answers are supposed to be confidential and not shared with other government agencies.
"Isn't that significant," Libby asked.
"It's a very important issue," Abowd conceded. "It shows we have more work to do."
When Libby asked if that startling statistic might serve as a "potential barrier" to the success of the 2020 Census, the witness and his inquisitor began quibbling over semantics.
"I would agree that it is a major challenge," Abowd said.
Abowd is scheduled to return to the witness stand for further cross examination on Monday. The trial is expected to wrap up on Tuesday.
Read the Top 8
Sign up for the Top 8, a roundup of the day's top stories delivered directly to your inbox Monday through Friday.