Secretary Wilbur Ross on Thursday pointed to the crappy performance of a citizenship question on a smaller scale survey to defend his move to add it to the 2020 census.
The smaller scale survey is the American Community Survey, a rolling survey the Census Bureau operates that goes out only to a small segment of the population. Ross and others have pointed to the question’s presence on the ACS to respond to concerns that the question was not tested for the decennial census before he added it.
The Trump administration has claimed that Ross added the question at the request of the Justice Department to improve its voting rights enforcement, which now uses the data from the ACS.
At the trial in New York in a case challenging Ross’ decision, a top Census Bureau scientist revealed that some 30 percent of noncitizen ACS-takers responded to the question inaccurately. The Census Bureau had discovered the question’s poor ACS performance while studying the Justice Department’s December 2017 request to add it.
The scientist, John Abowd, testified that the Bureau would be considering the evidence next time it does a review of the ACS, and that it would play a role in deciding whether they even keep it on the smaller scale survey.
Now in his opening remarks, Ross is trying to use that discovery to defend adding it to the Census. He suggested that the inaccuracy of the ACS citizenship showed why it would be helpful for the Justice Department on the census.
“Following receipt of DOJ’s letter, and during our review, Census Bureau officials recognized that current ACS data did not meet DOJ’s request for census block-level data,” he said, according to a prepared version of his remarks. “The Census Bureau analysis also showed that when non-citizens respond to the ACS question on citizenship, they respond incorrectly approximately 30% of the time.
The remarks did not address why he thought it would perform any better on the decennial census, where it is feared it will chill participation among immigrant communities. The Census Bureau during its review had advised him repeatedly that its inclusion would make the census more costly and its response’s less accurate.
At the trial, Abowd confirmed that the Bureau had no empirical evidence to lead it to believe that noncitizens are more likely to respond to it accurately on the census than they are on the ACS.
Abowd also testified that there were indications that responses to it on the census will be less accurate than the ACS responses.
Ross’ March 2018 announcement of the addition of citizenship question to the census said that the Census Bureau would use administrative records — i.e. data other agencies have on citizenship — to address the inaccuracies of the noncitizen response.
Abowd testified that that approach didn’t change the risks and costs of asking a citizenship question to the census.