Census 2020 Policy Update

September 30, 2020

As the potential end of census data collection draws closer, the state of the 2020 Census has claimed an increasingly prominent place in the public discourse. While many officials and stakeholders have ramped up their efforts to persuade people to participate in the count before it is too late, concerned municipalities, states, and organizations have also taken their case for an extended count and data processing period to court. In addition, lawmakers have introduced new legislation to address the urgent need for sufficient time for the Census Bureau to do its work as thoroughly as possible. Developments occurring during the last week of September will have a particularly pivotal effect on the accuracy and reliability of the census data products that will determine the division of public funding and political power, and the success of public and private investments in our economy, over the coming decade.

The Census in Court

As NALEO Educational Fund has reported in previous Census Policy Updates, there are several lawsuits now in progress that challenge policy decisions affecting the conduct of the 2020 Census, including the truncated timeline for data collection and processing; the Trump Administration’s memorandum setting forth the Administration’s policy of excluding undocumented residents from apportionment counts; and the July 2019 Executive Order directing the Bureau to produce data about residents’ citizenship. Fast-moving developments in these cases are likely, as litigants seek expedited consideration of claims that could result in immediate and significant changes to the Census Bureau’s current operational plans. 

Challenging the Truncated Timeline: A number of organizations, municipalities, and elected officials filed suit in California in mid-August, seeking an injunction prohibiting the Census Bureau from implementing the shortened timeline and accompanying operational changes it announced at the beginning of that month. In that matter, the judge issued a temporary restraining order to prohibit the Bureau from winding up its counting efforts at least until the court had enough time to view additional disclosures from the Commerce Department and Census Bureau, and then rule on plaintiffs’ request for a more permanent injunction against the early end of data collection and processing. After further review and argument, the judge issued a preliminary injunction against ending the census count and rushing the delivery of apportionment data, effective on September 24. Her decision requires the Bureau to continue counting and to refrain from delivering apportionment data at the end of December 2020 before it has been thoroughly processed and refined. As the government will appeal this decision and request a stay of the judge’s order, the judge’s ruling does not settle the matter; however, as of this writing, it ensures that field operations will continue until the Bureau has collected as much information as it can and that the Bureau has time to reinstate a full and credible data review process. 

Exclusion of Undocumented Immigrants from the Apportionment Count: On September 10, a three-judge panel hearing arguments in New York v. Trump, a challenge to the Administration’s intention to subtract undocumented individuals from apportionment counts, granted summary judgment and a permanent injunction to plaintiffs. The panel’s ruling prohibits the Census Bureau and Commerce Department from providing requested data about undocumented residents to the President and marks the memorandum setting forth the Administration’s policy of disregarding undocumented people for apportionment purposes as unlawful. An appeal from this decision is likely and would go directly before the Supreme Court. The Justices will have the discretion to determine whether to expedite their consideration of the case, so they could choose either to render a rapid decision or to add the matter to their regular calendar, in which case a decision might not be announced until 2021. While court proceedings continue, the Census Bureau is not prohibited from working on compiling data about undocumented people; however, for the present, it may not deliver any such information to the White House.

As timelines for remaining census cases are not certain at this point, additional developments that determine the deadlines for the contents and delivery of data products may occur at any time and will be discussed in future Census Policy Updates.

Census Legislation

During the week of September 14, Members of Congress introduced identical bipartisan bills in the U.S. House of Representatives and Senate that would extend deadlines for delivery of apportionment and redistricting data. The legislation would also prohibit the Census Bureau from releasing apportionment data any earlier than April 1, 2021, by which time it will have had at least five months following the last possible date for data collection to refine and process the information it has collected. The 2020 Census Deadline Extensions Act, H.R. 8250/S. 4571, is sponsored by Congressman Don Young (R-AK) and Congressman Ruben Gallego (D-AZ) in the House, and by Senators Brian Schatz (D-HI), Lisa Murkowski (R-AK), Dan Sullivan (R-AK), David Perdue (R-GA), Patrick Leahy (D-VT), and Gary Peters (D-MI) in the Senate.

These bills attracted new attention to the urgent need to ensure that the Census Bureau has the requisite time to collect as much response data as possible and carry out robust data evaluation and assessment to eliminate inaccuracies in resulting statistics. They would help the Bureau implement some components of the operational plan it initially developed in response to the COVID-19 pandemic. As discussed in detail in our September 2 Census Policy Update, top census officials had early on expressed concerns that the agency would not be able to provide quality Census 2020 data with the truncated deadline the Administration has implemented. In addition, several former Census Bureau Directors have been vocal about their concerns as well. The Administration initially supported extending the data delivery deadlines, but ultimately reversed its position.

In the days following their introduction, media reports noted that the provisions in the 2020 Census Deadline Extensions Act were being considered for inclusion in a Continuing Resolution to fund the federal government for the first several months of Fiscal Year 2021. Similar measures also were included in the House-passed HEROES-Act, H.R. 6800, and would be on the table during negotiations between Congressional chambers and the White House over additional COVID-19 relief legislation.

In addition, the 2020 Census Deadline Extensions Act is a response to growing evidence of the dangers of accelerating the census timeline, including the testimony elicited during the House Oversight and Reform Committee’s September 10 hearing on the state of the census. Witnesses included the Government Accountability Office’s Managing Director of Strategic Issues, J. Christopher Mihm, who shared auditors’ concerns that the intersection of pandemic-related challenges and truncated Nonresponse Followup (NRFU) and data processing could prevent the Bureau from producing data as accurate as the results of recent decennial censuses. Former Census Bureau Director John Thompson also reiterated his belief that the shortening of the final phases of the census is inconsistent with the goal of compiling the highest-quality information possible, stating, “the likelihood of a serious computer error that goes undetected is very high.” Stacey Carless, Executive Director of the NC Counts Coalition, and Stephen Roe Lewis, Governor of the Gila River Indian Community in Arizona, added warnings based on their direct observations of rushed operations in two states that have fallen behind national averages in terms of self-response and NRFU completion rates.

Senate Appropriations Committee Chairman Richard Shelby (R-AL) introduced legislation in July 2020 to provide an additional $448 million for the 2020 Census. The Senate, however, has not taken any additional action on this proposal and did not include any measures affecting the census in a package of COVID-19-related relief provisions that received a vote on September 10 but failed to pass out of the chamber.

Office of Inspector General Report on Risks Created by Accelerated Census Timeline

On September 18, the Office of Inspector General (OIG) of the Department of Commerce released a report detailing what the risks the accelerated timeline for Census 2020 operations implemented by the Bureau pose for an accurate and complete decennial count. The OIG is an independent and objective unit that provides oversight of the Department of Commerce’s programs and operations.

The report describes the lack of sufficient time for sound data collection; the threats posed by natural disasters such as hurricanes and wildfires, as well as the COVID-19 pandemic; and the lack of time for sound data processing and quality assurance after data collection. The report also notes that top Census Bureau officials, including the Director, did not know who ultimately made the August decision to accelerate the timeline, but confirmed that the decision was not made by the Bureau. Although the report does not identify the person or office responsible, its finding that the Bureau did not make the decision raises more troubling concerns about the Administration’s role in politicizing the 2020 Census.

A View of Census Operations from the Field

As NRFU operations have progressed and a sizeable temporary census workforce has been enlisted to enumerate the hardest-to-count households, an increasing number of census employees and observers have reported acute concerns about the consequences of shortening the time spent on remaining phases of the count. For example, in a September 16 opinion piece published in the Washington Post, Rosemary Armao, a University of Albany professor who also serves as a census enumerator, noted that “The administration says that stopping [NRFU] early is merely a matter of meeting deadlines, but to us [enumerators], it feels like sabotage.” 

 In confidential communications with census advocates, enumerators and field supervisors have disclosed that time and workload pressures have led to workers not completing full questionnaires, but instead, if they were able to conduct interviews, merely recording the number of household residents. In addition, employees have noted, and internal guidance from the Census Bureau has implicitly confirmed, that enumerators have not been making as many contact attempts with each household as plans originally required. For example, after the judge in the National Urban League case ordered the Bureau to revert temporarily to its April plan and timeline, the Bureau told its supervisors to resume assigning enumerators to do multiple visits, instead of just one, to housing units thought to be vacant.

Additionally, the Census Bureau’s April plan calls for employees to attempt to contact a household several times before asking a third party, such as a neighbor, for information about the household. However, stakeholders have heard from people who were asked for information about another household very early on in the door-knocking operation and before enumerators are likely to have had enough time to complete multiple visits to the household in question. Enumerators have also reported that they were prevented from handing off case assignments to linguistically-skilled enumerators when they encountered households with which they believed they did not speak a language in common.

In several parts of the country, stakeholders have learned that enumerators visited households that already submitted responses multiple times under the mistaken impression that those households remained uncounted. The geographic diversity of such reports indicates that the Bureau’s data processing mechanisms may be experiencing systemic failures in correctly matching responses received within their source address file. Enumerators have also been reassigned between Area Census Offices, resulting in some being asked to work in communities with which they are not familiar, and in which they may lack the necessary cultural fluency to communicate successfully with non-responding households. With some enumerators having expressed concerns about being given an inadequate supply of personal protective equipment, it is unsurprising that stakeholders have cataloged numerous reports of census employees approaching individuals they hope to interview while not wearing masks. In sum, these issues and other breaches of protocol have alarmed many of the residents upon whom our nation counts for census participation, likely weakening the integrity and accuracy of 2020 Census data products if not corrected. NALEO Educational Fund is working with partners to bring these issues to the attention of Members of Congress and Census Bureau officials.

Response Rate Update

NALEO Educational Fund continues to track census self-response rates both for the overall population and areas with large concentrations of Latinos. In addition, the Bureau is publishing information about the total share of housing units enumerated, which is the total share of housing units that have self-responded and the total share of housing units “enumerated” during NRFU. The Bureau’s reported self-response rates, NRFU enumeration rates, and total housing unit enumeration rates, which are provided for the nation and each state daily, are found here. The following sections also describe some limitations of the enumeration rates reported by the Bureau. As of September 27, the total share of housing units enumerated was 97.9 percent, with 66.4 percent having self-responded and 31.5 percent enumerated through NRFU.

Self-response:  We continue to see several indicators that Latino self-response rates are lower than the national rate. For example, a larger share of households in tracts that received English-language census mailings have responded than the share of households in tracts that received bilingual English and Spanish mailings. Additionally, in census tracts in which Latinos are the most numerous population group, the average response rate is 56.2 percent, more than 10.2 percentage points behind the national average.

Our internal analysis also reveals that as of September 27, 2020, on average, the higher the Latino share of a county’s population is, the lower its self-response rate. Thus, counties whose populations are less than 20 percent Latino tend to have notably higher census response rates than counties whose populations are majority-Latino, and counties whose populations are 75 percent or more Latino tend to be ones with the lowest self-response rates. Self-response rates indicate only the rate of census participation of known households (addresses that are on the Census Bureau’s Master Address File); they do not affirm the comprehensiveness or accuracy of information collected from participating households.

NRFU and Total Housing Units Enumerated: While housing units enumerated in NRFU have contributed to the increase of the total enumerated housing unit share, New Mexico and Arizona, both with large Latino populations, continue to have lower housing unit enumeration rates compared to other states. The number of housing units enumerated during NRFU includes all units where the Bureau has completed its casework, which can be accomplished in various ways. For example, the Bureau considers a unit completed if it has verified that the unit is vacant or nonexistent. If the Bureau obtains information about a household by conducting an interview with a neighbor or landlord, it also considers the housing unit completed and enumerated. In addition, the Bureau anticipates that it will enumerate at least 6.2 million households using federal administrative records, which often do not contain accurate or complete information about Latinos and their household members. 

Participation in the census by a household through either self-response or a NRFU interview does not ensure that everyone residing in the household was included in the census questionnaire or the household member’s NRFU response. For example, analysis of the 2010 Census revealed that the majority of very young children not counted in that census lived in households that were, in fact, “enumerated.” All of the limitations discussed in “Self-response” and “NRFU” above should be taken into account when assessing the extent to which the Bureau’s enumeration data reflect the progress made in enumerating the nation’s population, particularly in areas with hard-to-count communities.