What If a Citizenship Question Doesn't Meaningfully Reduce Census Response Rates After All?

Newly released data suggests Census analysts dramatically over-estimated the extent to which a citizenship question would discourage responses.

The Trump Administration's decision to include a citizenship question in the 2020 Census provoked substantial outrage. Many feared that inclusion of the question would depress response rates, leading to an inaccurate Census, and that it would produce a substantial undercount of people in immigrant communities, particularly in households containing noncitizens. Indeed, many suspected that was the whole point.

Fears of a significant undercount, and its consequent effects on political representation and the distribution of federal funds, prompted many states and interest groups to file suit against the Trump Administration. As the Supreme Court explained when it discussed the plaintiffs' claim of standing to bring the suit:

Respondents assert a number of injuries—diminishment of political representation, loss of federal funds, degradation of census data, and diversion of resources—all of which turn on their expectation that reinstating a citizenship question will depress the census response rate and lead to an inaccurate population count. Several States with a disproportionate share of noncitizens, for example, anticipate losing a seat in Congress or qualifying for less federal funding if their populations are undercounted. . . .

Several state respondents here have shown that if noncitizen households are undercounted by as little as 2%—lower than the District Court's 5.8% prediction—they will lose out on federal funds that are distributed on the basis of state population. . . .

Justice Breyer's separate opinion also stressed the arbitrariness of including a question that would produce a less accurate enumeration of people within the country.

How can an agency support the decision to add a question to the short form, thereby risking a significant undercount of the population, on the ground that it will improve the accuracy of citizenship data, when in fact the evidence indicates that adding the question will harm the accuracy of citizenship data?

As readers know, the Court decided, 5-4, that inclusion of the citizenship question was unlawful because the Administration's explanation for why the question was to be added seemed "contrived" and did not match the available evidence.

In 2019, the Census nonetheless conducted a significant test of the effects of including a citizenship question, and the results were not what anyone (including the Census's own experts) expected, as documented in a recently released report. As noted by Lyman Stone, the report tells a quite different story about the effects of a citizenship question than that  upon which the Census case was litigated.

Here is what the 2019 Census Test Report (from the Executive Summary):

In 2018, the U.S. Census Bureau decided to test the operational implications of a proposed question on citizenship status on the 2020 Census. In particular, experts and stakeholders raised concerns that such a question could depress self-response rates, increase cost, and reduce the quality of the 2020 population count. An indirect study by Census Bureau researchers predicted that "adding a citizenship question to the 2020 Census would lead to lower self-response rates in households potentially containing noncitizens…" compared to households with all citizens (Brown, Heggeness, Dorinski, Warren, & Yi, 2018). However, the authors recommended the ideal analysis would be to conduct a randomized controlled experiment to compare response rates on questionnaires with and without a citizenship question. . . .

The major finding of the 2019 Census Test was that there was no statistically significant difference in overall self-response rates between treatments. The test questionnaire with the citizenship question had a self-response rate of 51.5 percent; the test questionnaire without the citizenship question had a self-response rate of 52.0 percent. Although these results differ from the predicted rates in Brown's et al. study, the results of the two studies are not
comparable since this study benefits from the randomized controlled design, which isolates the treatment effect.

The Report also found statistically significant declines in response rates among some areas and among subgroups, including areas where more than are 4.9 percent noncitizens or more than 49.5 percent are Hispanic, as well as in the New York and Los Angeles metropolitan areas. This decline, however, was far lower than the Census and other experts had predicted. Whereas Census experts and the Administration's critics had expected a greater-than-five percentage point decline (or more) among particular subgroups, the actual drop as in the neighborhood of one percent. According to the report, these differences are small enough that they would not have required staffing changes for the Census' routine follow-up responses. It's also lower than the point at which plaintiff states claimed they would lose out on federal funding.

While these findings challenge some of the assumptions upon which the legal challenge to inclusion of the citizenship question was based, they don't undermine the Court's ultimate holding, which was based upon the arbitrariness and lack of candor of the Commerce Department (of which the Census is a part), not any particular assumptions about the effect of including the question. They do, however, challenge some of the assumptions that caused so many to file suit against the Census in the first place, and will make it easier for a future administration to include a citizenship question in the decennial Census, if it should so choose.