“Is this person a citizen of the United States?”

In December 2017, the Department of Justice formally proposed adding a citizenship question to the 2020 census. This question would ostensibly help to enforce the Voting Rights Act of 1965.

However, census experts, over 100 national scientific and civil rights organizations, the Congressional Hispanic Caucus, The Leadership Conference on Civil and Human Rights and Democratic senators and House members protested vehemently.

I am a social scientist who studies immigration. I have used census data on immigration and citizenship in my research for over two decades, and I have urged government statistical agencies before to collect more data about immigrants. But I don’t think it’s wise to collect citizenship status in the 2020 census. Doing so would not only raise the risk of collecting inaccurate data, but also reduce public confidence in the census itself.

Tracking citizenship

On the one hand, data on citizenship is valuable. In any modern democracy, statistical data is essential for informing policy debates and guiding the implementation of governmental programs. Without it, decisions would almost certainly be too easily shaped by anecdotal evidence and personal biases.

Citizenship data has been used to track political participation and inclusion of immigrant groups. Citizenship is strongly associated with access to public assistance, health care and jobs. Social scientists and policy analysts rely heavily on survey items on citizenship to understand immigrants’ well-being and their impact on host societies.

What’s more, the U.S. Census Bureau has successfully collected confidential information on citizenship status in the past. The citizenship question was first introduced in the 1870 census and was part of all censuses from 1890 through 1950. It was included in the “long” form of the census – administered to 1 in 6 households – as late as 2000. It’s also asked in the American Community Survey, a survey that Census Bureau conducts every year.

Immigrants tend to be willing survey respondents. In a 2010 study, Hispanic immigrants were more likely than U.S.-born Hispanics to agree that the census is good for the Hispanic community. They were also more likely to correctly understand that the census cannot be used to determine whether a person is in the country legally, and that the bureau must keep their responses confidential.

In another study I published in 2014 with two colleagues, James Bachmeier and Frank Bean, we found that nearly all immigrants answered questions about their immigration and documentation status. These response rates are on par with or better than typical survey questions on health or income. Moreover, immigrants’ responses to these questions appeared to be fairly accurate.

Harming the data

However, the political climate surrounding immigration has changed in the last year.

Not all immigrants have been cooperative respondents in the past. Those who are more likely to be undocumented have been undercounted in past censuses and were more likely to incorrectly report themselves as U.S. citizens.

The Trump administration’s anti-immigrant rhetoric and policy may have increased mistrust among all immigrants, not just those who are undocumented. During focus group interviews conducted by the Census Bureau roughly six months into Trump’s presidency, immigrants appeared anxious and reluctant to cooperate with Census Bureau interviewers. They mentioned fears of deportation, the elimination of DACA, a “Muslim ban” and ICE raids. One respondent walked out when the questionnaire turned to the topic of citizenship, leaving the interviewer alone in his apartment. Respondents even omitted or gave false names on household rosters to avoid “registering” with the Census Bureau. Interviewers remarked that it was much easier to collect data on immigration and citizenship just a few years ago than it is now.

It’s not yet clear whether the fears seen in the focus group interviews are widespread or how such fears would affect response rates if the citizenship question were added to the 2020 census. Additionally, researchers haven’t yet worked out a way to ask the citizenship question so it’s not perceived as threatening.

Unfortunately, there’s not enough time to find out. A finalized questionnaire must be submitted to Congress by the end of March.

What to do in 2020

I served on the Census Advisory Board from 2008 to 2011 and have personally witnessed the time and effort it takes for the Census Bureau to develop questions for the census. Officials must pay meticulous attention to the exact question wording, response categories, ordering and questionnaire layout.

I believe adding a citizenship question without adequate testing could severely reduce participation in the 2020 census among the country’s 44 million immigrants and the additional 32 million U.S.-born people who live with them.

The social and economic consequences of a low response rate for the 2020 census would be severe. Even small errors in coverage could shift the distribution of political power and federal funds, as well as reduce the effectiveness of public health systems and other government functions.

Perhaps even worse, high coverage error in the 2020 census could undermine the public’s trust in the census as the nation’s source of information on the size, growth and geographic distribution of the U.S. population.

This occurred a century ago, as historian Margo Anderson described in her book, “The American Census.” The 1920 census revealed dramatic shifts in population from rural to urban areas, as large waves of Eastern and Southern European immigrants settled predominantly in American cities. Congress, fearing the political ramifications of these changes, rejected the results of the 1920 census and voted not to redistribute the seats of the House according to the most recent census data. A similar rejection of the results of the 2020 census would likely result in a constitutional crisis today.

Citizenship data would be valuable. But the risks of poor data quality – or the erosion of public trust in the census and other governmental institutions – far outweigh the potential benefits. Given that there are other current data available on citizenship, why take unnecessary risks when the stakes are so high?

Jennifer Van Hook receives funding from the National Institutes of Health. She is Roy C. Buck Professor of Sociology and Demography at the Pennsylvania State University, and is the nonresident fellow of the Migration Policy Institute.