In recent months, the Trump administration's efforts to add a citizenship question to the 2020 decennial United States census have provoked fierce criticism and several ongoing lawsuits. The administration argues that the question will help the government comply with the Voting Rights Act; critics worry it will depress response rates among immigrants, thus resulting in an undercounting of such residents. Census experts have warned that the addition of a citizenship question could lead to "a bad count."
In a new working paper, five Census Bureau economists provide more evidence against the citizenship question. The paper compares household response rates for the 2010 census and the 2010 American Community Survey. (The ACS is a more extensive survey that's sent to a small subset of American households and does include a citizenship question.) The researchers find lower response rates for the ACS, an effect that is larger for households with a non-citizen member. Similarly, the researchers examine response rates to several other surveys that contain citizenship questions and find evidence "consistent with the possibility that households with noncitizens are more sensitive to the inclusion of citizenship questions."
Thanks to non-citizen households' reluctance to respond to the census, the researchers estimate that the addition of the question could result in hundreds of thousands (or, under one set of assumptions, 1.75 million) fewer correct enumerations. They also estimate that the addition of the question would increase the costs of the 2020 census by at least $91.2 million, and further stress that "this cost estimate is a lower bound." In other words, adding a citizenship question to the 2020 census will increase costs, reduce accuracy, and likely will not produce a more accurate count of the citizen voting-age population.
In some respects, the paper's conclusions are not surprising. In fact, they're in line with both independent expert's predictions and an internal memo sent by John Abowd, the Census Bureau's chief scientist, to Ross back in January. In it, Abowd warns that adding a citizenship question would be "very costly, harms the quality of the census count, and would use substantially less accurate citizenship status data than are available from administrative sources."