People For the American Way has strongly condemned the Trump administration’s census citizenship question from the time it was first requested to its consideration by Congress and, through affiliate PFAW Foundation, the courts. Now that the proposal is open for public comment, PFAW is once again airing our deep concerns about its potential to jeopardize the accuracy of the census in all communities—the consequences of which the nation will have to live with for the next decade. Black and Latinx immigrants—indeed, all immigrants, both legal and undocumented—already feel demonized and their families are torn apart. What President Trump, Commerce Secretary Wilbur Ross, and the Department of Justice have proposed will make immigrants even more reluctant to participate. You can read PFAW’s public comment letter and the letter we joined with allied organizations in full below.
PFAW’s Public Comment Letter
Dear Ms. Jessup:
I am writing on behalf of People For the American Way (PFAW) to offer comments on the 2020 Census proposed information collection. Founded in 1981 by a group of civic, educational, and religious leaders, PFAW is a progressive advocacy organization grounded in the nation’s ideals of freedom, equity, opportunity, and justice. Over its history, PFAW has conducted extensive education, outreach, legislative and regulatory advocacy, and other activities to make these values a reality in the lives of all people.
PFAW and its hundreds of thousands of members nationwide strongly support the principle that a representative democracy requires meaningful public participation and a full, fair, and accurate census count. PFAW is concerned, however, about the addition of a census citizenship question and its potential to jeopardize the accuracy of the census in all communities—an outcome that the nation will have to live with for the next decade. We urge the Department of Commerce to remove this question from the 2020 form.
Even before Secretary Ross’s decision to include a census citizenship question, state and local officials and community leaders were deeply concerned about the difficulty of achieving robust participation from Black and Latinx households and other hard-to-count population groups. Black and Latinx people are no less deserving of freedom, equity, opportunity, and justice than any other segment of the American population. Yet the Black population was undercounted by more than 800,000 in 2010,1 including seven percent of young Black children.2 Latinx people have also been disadvantaged by decades of undercounts, including the 2010 Census,3 and Latinx youth are among the most undercounted of all.4
The likelihood of undercounts and the harms they will sow are already heightened in the current political environment where Black and Latinx immigrants—indeed, all immigrants, both legal and undocumented—feel demonized and their families are torn apart. The census citizenship question will make immigrants even more reluctant to participate; in addition, it will drive up costs as the Census Bureau struggles to develop new communications and outreach strategies with little time remaining, plan for an expanded field operation, and track down millions of nonresponsive households. In sum, asking about citizenship status in a climate of fear and mistrust can only heighten suspicions, depress response rates, cost additional taxpayer money, and thwart the 2020 enumeration.
Not surprisingly, the census citizenship question has drawn intense opposition from a broad array of business leaders, state and local officials, social scientists, and civil and human rights advocates who know how much is at stake. This groundswell of opposition includes Republican and Democratic mayors,5 six former directors of the Census Bureau6 and two former Commerce Secretaries7 from Republican and Democratic administrations, civil and human rights groups,8 faith leaders,9 members of Congress,10 and many others representing diverse communities and political ideologies. All are united in their deep-seated concern that an untested citizenship question will compromise the census and jeopardize the quality and accuracy of data for all communities.
The collection of useful, objective data about our nation’s people, housing, economy, and communities is vitally important. Not only is a nationwide census required by the Constitution, it is integral to our democracy, ensuring that district lines and political power are fairly drawn and allocated. The federal government uses census-derived data to direct federal assistance to states, localities, and families. Census data also guide important community decisions affecting schools, housing, health care services, business investment, and much more.
Simply put, a full, fair, and accurate census is essential for all basic functions of our society. The 2020 Census should not include a citizenship question when scientific evidence indicates that it will undermine a successful count.11 PFAW strongly urges the Department of Commerce to remove it from the data collection forms.
Senior Policy Analyst
PFAW and Allied Organizations
Dear Ms. Jessup:
On behalf of The Leadership Conference on Civil and Human Rights, a coalition charged by its diverse membership of more than 200 national organizations to promote and protect the civil and human rights of all persons in the United States, and the undersigned organizations, we appreciate this opportunity to provide comments in response to the Federal Register notice (the “Notice”).
The Leadership Conference provides a powerful unified voice for the many constituencies of the coalition: persons of color, women, children, individuals with disabilities, gays and lesbians, older Americans, labor unions, major religious groups, civil libertarians, and human rights organizations. Our coalition views an accurate and fair census, and the collection of useful, objective data about our nation’s people, housing, economy, and communities generally, to be among the most important civil rights issues of our day. We have a long record of first-hand experience working in support of previous censuses. For the 2010 Census, we undertook the most comprehensive and extensive effort by a stakeholder organization to promote participation in historically hard-to-count communities and to mobilize local advocates in support of the census by highlighting the civil rights and social justice implications and community benefits of an accurate count.
As a coalition representing a broad, diverse set of communities, The Leadership Conference and its members are best positioned to offer a civil rights perspective on issues presented for comment in the Federal Register notice. Given the unique circumstances surrounding the final determination of proposed questions for the 2020 Census questionnaire, we have decided to offer comments in this letter exclusively on the proposed citizenship status question. The Leadership Conference will submit additional comments addressing a broader set of design and operational issues set forth in the Notice.
Overview and summary
There is one constitutional purpose for the decennial census: to apportion seats in the U.S. House of Representatives among the 50 states, based on an enumeration of the “whole number of persons in each state.”1 The Supreme Court recently confirmed in a unanimous opinion in Evenwel v. Abbott, that “representatives serve all residents, not just those eligible or registered to vote.”2 To realize the Constitution’s “principle of representational equality,”3 the overarching goal of the decennial census must be an accurate count of all persons residing in the country. Any element of the census design and plan that might undermine or detract from the Census Bureau’s ability to achieve that goal simply cannot stand.
With this fundamental, constitutional purpose in mind, we urge the Census Bureau, in the strongest possible terms, to remove the proposed citizenship question from the 2020 Census form. This question will unnecessarily and, perhaps significantly, increase the cost of the 2020 Census, while collecting data at the census block level for which there is no credible or widely-accepted evidence of utility. More specifically, and in response to the formal Request for Comments on four questions (section IV of the Notice), we offer the following conclusions and recommendations:
- The proposed citizenship question does not have practical utility and should be removed from the 2020 Census questionnaire.
- Adding a citizenship question will increase the cost of the 2020 Census significantly but unnecessarily—costs that will be borne by American taxpayers.
- Assuming for argument’s sake only that the Department of Justice requires block level citizen voting age population data to implement federal statutes, the Census Bureau can improve the quality of the requested data, protect the quality and accuracy of all other data collected in the 2020 Census, and minimize the burden of information collection on the public, by using administrative records data.
We explain our conclusions and recommendations in more detail below.
Questions for discussion
Is the information collected through the citizenship question necessary to facilitate the functions of the agency, and does it have “practical utility?”
The decision to add a citizenship question to the 2020 Census allegedly was based on a December 12, 2017 request4 from the Department of Justice (DOJ) to the Census Bureau to “reinstate a citizenship question on the decennial census to provide census block level citizenship voting age population (“CVAP”) data that are not currently available from government survey data (“DOJ request”),” according to the March 26, 2018 memorandum from Secretary of Commerce Wilbur Ross to Under Secretary for Economic Affairs Karen Dunn Kelley.
The bureau’s March 2017 submission to Congress of the subjects planned for the 2020 Census and related American Community Survey (ACS), which itself includes a question on citizenship, describes a multi-year process for evaluating proposed content for these two surveys.5 The bureau launched a comprehensive review of federal data needs in partnership with the U.S. Office of Management and Budget (OMB), beginning with creation of the Interagency Council on Statistical Policy Subcommittee on the ACS and the 2014 ACS Content Review. The effort, in part, was designed “to confirm and update the statutory and regulatory authority for the questions with federal agencies.”
The Census Bureau has set forth guidelines to determine whether it should collect data in the decennial census and ACS (previously the census “long form”). Those protocols are meant to ensure that the census only collects data “required by federal programs.” More specifically, the census and ACS will include topics because “federal law or regulation specifically mandates their inclusion in the census, the legal system requires that the data be collected, or federal law requires it for program implementation and the census is the only source for the information.”6
However, we now know from documents that have come to light through litigation challenging the decision7 that DOJ officials were not the genesis of the late request for block-level citizenship data in order to facilitate enforcement of section 2 of the Voting Rights Act (VRA). Rather, Secretary Ross acknowledged that, after administration officials—including the president’s chief strategist Steve Bannon—raised the idea of collecting citizenship data in the 2020 Census, he and his senior political staff asked the DOJ if it would request such a question.8 This revised explanation not only demonstrates that the decision to add the citizenship question circumvented a longstanding, comprehensive process for determining the content of the decennial census, it calls into question the truthfulness of the explanation in DOJ’s December 2017 letter of request. In fact, documents released as part of the litigation reveal that Secretary Ross and his staff suggested to DOJ that it should assert the usefulness of data from a census citizenship question for enforcement of the VRA, and then worked with DOJ officials to facilitate the formal request to the Census Bureau.
The practical utility of block-level citizenship status data for enforcement of the VRA is further called into question by historical precedent with respect to enforcing this seminal statute and the experience of experts who have worked to ensure compliance with the Act since its enactment in 1965. In a “friend of the court” (amicus) brief in State of New York, et al., v. U.S. Department of Commerce, et al., more than 150 organizations noted that neither the DOJ nor private litigators seeking to uphold minority voting rights have ever sought block-level citizen voting age population data from the decennial census to enforce the VRA. In relevant part, the amici wrote:
While CVAP data may be useful in vote dilution cases, in the 53 years that the Department of Justice (DOJ) and private plaintiffs have enforced § 2 of the VRA, they have never tried to obtain CVAP data from the decennial census. That is because such data can be reliably obtained from other sources—without the negative effects of including a citizenship question, described above. In particular, from 1970 to 2005, litigants bringing § 2 claims could obtain CVAP data from the “long form” census, and from 2005 to the present, CVAP data has been obtainable from the ACS. See Levitt Testimony 16. That data has amply sufficed to facilitate VRA enforcement without running the risk of suppressing census response rates from under-represented communities.9
The Census Bureau has a statutory obligation under the Paperwork Reduction Act to minimize the burden of information collection on the public. The agency itself assures the public that it will “determine whether the information [it collects] is truly necessary.”10 The clear standards for including topic and questions in the census, described in numerous bureau documents, have not been met, and there is evidence to suggest intentional circumvention of those standards. Therefore, the proposed citizenship status question does not have practical utility and should be removed from the 2020 Census questionnaire.
Is the agency’s estimate of the response hours and cost of including a citizenship status question accurate?
The Census Bureau attaches no cost to the proposed collection of data on citizenship status in the 2020 Census. While we understand that there is no cost to individual respondents to participate in the census, we believe this assertion is inaccurate and misleading, especially with respect to adding an untested, controversial question that will increase the cost of the 2020 Census significantly but unnecessarily—costs that will be borne by American taxpayers.
Associate Director for Research and Methodology John Abowd concluded in his January 19, 2018 memorandum to Secretary Ross11 that adding a citizenship question to the 2020 Census would have an “adverse impact on self-response and, as a result, on the accuracy and quality of the 2020 Census.” He then determined that the new question would increase the cost of the 2020 Census “by at least $27.5 million, a “lower bound” and “conservative” estimate based on the likelihood that nonresponse among noncitizen households would exceed their projections. With a minimum $55 million cost attached to each one percent of households that does not self-respond, the additional cost to the taxpayer could skyrocket if the citizenship question drives millions of people further into the shadows and away from the census, as we believe it will.
It also is worth noting that while Dr. Abowd said the bureau “assume[s] that citizens would be unaffected by the change,” we believe that assumption fails to account for the millions of citizens, especially children, living in mixed status households, who easily could decline to respond or be left out of the census if others in the household (e.g. Temporary Protected Status holders; undocumented residents; “Dreamers”) fear the consequences of filling out the form. Therefore, self-response among households comprised of both citizen and non-citizen members (headed either by non-citizens or citizens) could decline, as well, further putting data quality and census accuracy at risk.
Dr. Abowd and his staff concluded that the estimated cost of producing citizenship data exclusively from administrative records is between $500,000 and $2 million, a fraction of the likely cost of proceeding with an untested new question on the 2020 Census form. In a subsequent analysis prepared for the Secretary of Commerce on March 1, 2018, Dr. Abowd evaluated the possibility of combining administrative records citizenship data with survey-collected data.12 He concluded that this additional option for meeting DOJ’s request still “would result in poorer quality citizenship data” than using administrative records alone to produce the desired dataset.
In light of Dr. Abowd’s conclusions that producing the data DOJ requested using administrative records “would most likely have both more accurate citizen status and fewer missing individuals than would be the case for any survey-based collection method” and was the “best option for block-level citizenship data, quality much improved [compared to the options of not producing the data or of collecting the data through a new question on the 2020 Census form],” there is no acceptable rationale for potentially increasing the cost of the 2020 Census by tens of millions of dollars by including the proposed citizenship question on the 2020 Census form.
How can the Census Bureau enhance the quality, utility, and clarity of information on citizenship, and minimize the burden of data collection on respondents?
The Notice poses two questions that, we believe, can be answered together based on research and analysis the Census Bureau’s own staff has conducted.
First, as discussed previously, and accepting only for argument’s sake the Justice Department’s assertion that it requires block level data on citizenship data to enforce section 2 of the VRA, Dr. Abowd’s memoranda clearly suggest that the Census Bureau can minimize the burden on respondents of collecting the desired information by using administrative records and statistical modeling techniques to produce estimates at the block level. These methods, Dr. Abowd concluded, would produce more accurate data than collecting citizenship data through a question on the 2020 Census form without increasing costs significantly.
Second, the Secretary’s decision memo directing the addition of a citizenship question does not acknowledge urgent concerns from the Census Bureau’s Center for Survey Measurement, describing an “unprecedented” level of “deliberate falsification of the household roster and spontaneous mention of concerns regarding negative attitudes toward immigrants” in 2020 Census pretesting.13 Because respondents in the pretesting surveys had participated in other bureau surveys and were being paid to participate in the pretests, research staff warned that people asked to participate in the actual census (the “production survey”) might have an even greater level of fear and reluctance to respond. The staff recommended scientifically robust research on the significant fears about confidentiality of census responses—driven by respondents’ perception of anti-immigrant policies—that field employees documented in pretesting.
Census Bureau research staff conducted subsequent qualitative evaluations of the 2017 multilingual pretesting studies and additional studies done in 2018 (including in-language focus groups).14 Notably, the phenomenon of fear census employees encountered occurred before public discussion of adding a citizenship question to the census form. Preliminary message testing found that many Spanish speakers appeared to be reassured that the 2020 Census would not ask questions related to immigration status, a finding that now must be considered unreliable. Furthermore, other Spanish speaking respondents were not swayed by statements that, by law, the Census Bureau could not share their answers with immigration enforcement agencies. And researchers noted that confidentiality concerns “may have a disproportionate impact on an already ‘hard to count’ population: immigrants.”
While the research cited above was limited, it highlighted the likelihood that fears about how census participation could be used to harm immigrants and their families could adversely affect response rates and data quality—concerns that service providers and advocates who work closely with immigrant populations have raised since the possibility of adding a citizenship question to the census surfaced in December 2017. Incorrect or incomplete household rosters, in particular, are likely to reduce data quality, at best, and increase omissions and, therefore, the undercount, at worst. Research staff noted these possible implications and recommended systematic, quantitative, and more current analysis of the reactions they had already documented.
Yet, the March 26, 2018 decision to add the question failed even to acknowledge the alarm sounded by Census Bureau research staff. Instead, the Secretary’s memorandum stated that “neither the Census Bureau nor concerned stakeholders could document that the response rate would in fact decline materially,” and went on to note that a former Census Director and Deputy Director were not aware of empirical evidence of such an effect on response rates. Shifting the burden of proof for adverse consequences on census accuracy to stakeholders, and suggesting that the absence of traditional in-depth Census Bureau research to assess the possibility of these consequences is evidence of no consequences, flies in the face of OMB’s own standards and guidelines for statistical surveys.15 In relevant part, those guidelines provide:
Survey Response Rates
Standard 1.3: Agencies must design the survey to achieve the highest practical rates of response, commensurate with the importance of survey uses, respondent burden, and data collection costs, to ensure that survey results are representative of the target population so that they can be used with confidence to inform decisions. Nonresponse bias analyses must be conducted when unit or item response rates or other factors suggest the potential for bias to occur. (emphasis added)
Pretesting Survey Systems
Standard 1.4: Agencies must ensure that all components of a survey function as intended when implemented in the full-scale survey and that measurement error is controlled by conducting a pretest of the survey components or by having successfully fielded the survey components on a previous occasion.
The inclusion of a citizenship question on the American Community Survey cannot substitute for the pre-production testing for the decennial census these principles envision, which would require testing of the question in the context of a completely redesigned 2020 Census and the contemporary environment of fear in many communities.
The Census Bureau’s 2020 Census Operational Plan (v. 3) envisioned a multi-year, iterative process for testing questionnaire design and content.16Among the lessons learned from the 2010 Census, according to the plan: “Conduct comprehensive testing of optimized content in the usability lab and in a field test to prevent unanticipated negative impacts on data quality.” That means questions must be tested in the context of a well-designed and well-tested census form that considers “design elements (size, color, spacing implications, etc.), mode, and language when finalizing questionnaire content and design. Also test for successful data capture before implementation.”17 The process started with cognitive testing of content and question wording in 2015, followed by the 2015 National Content Test (1.2 million homes), two large field tests in 2016 (Los Angeles County, CA, and Harris County, TX), and a national sample (80,000 addresses) in 2017. Unfortunately, the final opportunity for robust testing has come and gone, as the 2018 End-to-End Census Test (the ‘dress rehearsal’) started in March with a questionnaire and associated materials (e.g. letters, postcards, inserts, language assistance guides, enumerator scripts, etc.) that did not include or contemplate a citizenship question.
As six former census directors wrote to Secretary Ross in January 2018, every census is different, and the environment in which a census occurs is a significant factor.18 “There are sound reasons that the Census Act requires the Bureau to submit to Congress the topics and actual questions it will include, three and two years, respectively, before Census Day. It is highly risky to ask untested questions in the context of the complete 2020 Census design. There is a great deal of evidence that even small changes in survey question order, wording, and instructions can have significant, and often unexpected, consequences for the rate, quality, and truthfulness of response,” the directors said.
Further amplifying the federal government’s principles for collection of statistical data, the Census Bureau’s Census Scientific Advisory Committee also opposed “last-minute inclusion” of a citizenship question in the 2020 Census in its recommendations to Acting Census Director Ron Jarmin following the committee’s 2018 spring meeting (March 29-30). Among the committee’s concerns were: (1) “the lack of adequate testing, about the implications for nonresponse (unit and item), implications for the cost, and implications for attitudes about the Census Bureau and concerns about confidentiality.” (2) “ … just because there is not clear evidence that adding the question would harm the census accuracy, this is not evidence that it will not. [T]he empirical evidence that was discussed by Sec. Ross came from data collected in a different data collection context, in a different political climate, before anti-immigrant attitudes were as salient and consequential.” (3) “The Census tradition has always been to collect evidence about the impact of a question before the question is added to the Census. … [B]orrowing the question from the ACS—which has question wording that reflected a different set of motivations and uses—results in a question wording that is puzzling in its specificity distinguishing U.S. territories.”
Therefore, assuming for the sake of argument only that the Department of Justice requires block level citizen voting age population data to implement federal statutes, the Census Bureau can improve the quality of the requested data, protect the quality and accuracy of all other data collected in the 2020 Census, and minimize the burden of information collection on the public, by using administrative records data on citizenship. The Census Bureau estimates the workload, success, and cost for each census operation based on years of research and rigorous, iterative testing that builds on lessons learned and refines methods and operations based on weaknesses identified in each environment.
Adding a new, controversial citizenship question likely will affect in known and unknown ways assumptions and outcomes from response rates, to load capacity for Census Questionnaire Assistance (telephone assistance), to length and cost of paper forms, to recruitment and training of census takers, to effectiveness of messages and segmentation of target audiences (which have been researched and tested without a citizenship question in the mix)—in other words, the cost, quality, and accuracy of the entire decennial census. The Census Bureau should remove the citizenship question from the 2020 Census form and consult further with the Secretary of Commerce and with career professionals in the Justice Department about ways to produce data that clearly and demonstrably meet the bureau’s own standards, as well as broader federal standards and guidelines, for collecting data from the American public in federal surveys and censuses.
Thank you for the opportunity to comment on the questions to be asked in the 2020 Census and to ensure that the voices of the civil and human rights community continue to be heard in this important ongoing national conversation. If you have any questions about these comments, please contact Corrine Yu, Managing Policy Director, at 202-466-5670 or [email protected].