Citizenship and the census, in context

Editor's Note: The Boston Indicators research center at the Boston Foundation is currently working on a report exploring the history, uses and impacts of the decennial census; from which this piece is excerpted. 

The recent news that Secretary of Commerce Wilbur Ross had approved the inclusion of a question on citizenship status for the 2020 Census has brought the Census effort to the forefront this week. The process of proposing and designing the question, however, has been months in the making. 

In late 2017, the Department of Justice (DOJ) under Attorney General Jeff Sessions requested that the Census Bureau, which is overseen by the Commerce Department, incorporate into the 2020 census a question on citizenship status. The DOJ argued that the more granular data allowed by the census would be useful in enforcement of Section 2 of the Voting Rights Act, which prohibits discrimination against any citizen’s voting rights on account of race, color, or membership in a language minority group..

The hypothetical situation posited by the DOJ argues that minority communities, drawn as majority-minority districts using census population counts, may not be able to elect their preferred representatives if the district is made up of non-citizens and thus, ineligible voters. And on the surface, it may not seem like a big change – citizenship status was asked from the 1860s through 1950, when it was dropped on the standard census form. Citizenship questions are asked in other demographic surveys, as well.

Others see a different motive – to either surface or intimidate the counting of noncitizens. A number of previous census directors have written to Ross opposing the addition of a citizenship question, while 14 states led by California’s Attorney General Xavier Becerra are suing the Trump administration over the inclusion of the question. The Attorneys General charge that introducing a question of citizenship goes against the constitutional requirement to “count each person in our country – whether citizen or noncitizen – ‘once, only once, and in the right place.’”

The Census, of course, doesn’t just help with the drawing of Congressional and legislative districts. It’s used in myriad budgetary and public policy decisions, the vast majority of which affect citizens and non-citizens, and the communities in which they live, without regard to citizenship or voting status. Critics say it’s a virtual lock that in the current political climate that non-citizens would think twice about taking part in the Census if they had to openly declare their citizenship status. That would exacerbate already significant issues with undercounting minorities that has an impact on funding and representation in past Census efforts.

The Commerce Department has one estimate for that. The Secretary of Commerce, Wilbur Ross, notes that the Bureau estimated such a question could require census canvassers to revisit an additional 630,000 households to ensure a complete census. The Bureau freely admits this is a rough estimate, but believes that follow-up operations would be well positioned to respond to this increase.

The Census Bureau won’t get a chance to test that estimate, though. While the question is a familiar one (pulled from the American Community Survey), it was released too late to be tested as part of the decennial census in next month’s Providence, Rhode Island end-to-end test.

That’s a real concern for data scientists.

Introducing a new question to the census is generally not an easy process. For all its perceived simplicity, the census is an incredibly complex social science instrument. The process to add a question generally takes years – sometimes decades, and often includes multiple rounds of testing and revisions. This is all to ensure that wording choice or question order doesn’t impact the respondent’s interest in completing the census and returning it to the Bureau.

In a report presented at the Bureau’s National Advisory Committee on Racial, Ethnic and Other Populations fall meeting, researchers reported an increase in confidentiality concerns raised by respondents during tests of other census processes. Respondents feared having their information passed on to the Department of Homeland Security (DHS) or Immigration and Customs Enforcement (ICE), even though by statute census data is confidential, even to law enforcement. Some provided incomplete or incorrect information to interviewers and tried to break off or walk out of interviews. In one of the more extreme cases, an interviewer returned to a Hispanic community after dropping off interview literature, and found one of the potential respondents moving due to a fear of being deported.

Despite the Census Bureau’s relative optimism that it can adjust to deal with an increase in non-response, the full impact of a citizenship question is unknown. Even on the tests noted by the researchers above, respondents were paid for their time, and were walked through the process by interviewers who explained the ways individual responses are legally protected. A true census, where the bureau must rely on the self-response, will likely have even greater difficulty getting the response rate it’s looking for from some groups.

With a court challenge and negative reactions to navigate, the citizenship question is not a sure bet for the Census 2020 questionnaire. If it is included, however, the uncertain impact the question on the data could have much more real (intended and unintended) consequences – on funding, representation, and the rights of the same groups the question itself is purported to protect.

To receive the Boston Indicators Census 2020 report when it is released, as well as the Boston Indicators monthly summary of interesting data and research, please sign up for our mailing list.