New York v. U.S. Dep't of Commerce

JESSE M. FURMAN, United States District Judge:

*513 TABLE OF CONTENTS

BACKGROUND

A. History and Purposes of the Census...519

B. The Secretary's Authority Over the Census...521

C. The History of a Citizenship Question on the Census...524

D. Testing and Adding New Questions to the Census...526

E. Secretary Ross's Decision and This Litigation...527

SECRETARY ROSS'S DECISION...530

A. The Initial Administrative Record Submission...530

1. The December 12, 2017 DOJ Letter...530

2. The Census Bureau's Preliminary Analyses and Recommendations...532

3. Secretary Ross's February 12, 2018 Meeting with the Census Bureau...536

4. The Census Bureau's Analysis of Alternative D...536

5. Communications with Stakeholders...539

6. Secretary Ross's March 26, 2018 Memorandum...542

7. Secretary Ross's Testimony Before Congress...545

B. The Supplemental Administrative Record and the Trial Record...547

1. Secretary Ross's Early Interest in Adding the Citizenship Question...549

2. Comstock's Search for a Rationale and an Agency to Request the Question...550

3. Secretary Ross and His Aides Persist in Their Efforts...552

4. Secretary Ross's Intervention with the Attorney General...554

5. AAAG Gore Ghostwrites the DOJ Letter...555

6. The Attorney General Forbids DOJ to Meet with the Census Bureau...557

7. Efforts to Downplay Deviations from the Census Bureau's Standard Processes...558

a. Secretary Ross's Claim that the Question Was Well Tested...560

b. The Commerce Department Revises the Census Bureau's Description of the "Well-Established Process" for "Adding or Changing Content on the Census"...562

c. Secretary Ross's Description of His Dealings with Nielsen...563

d. Comstock's Testimony About the Census Bureau's Analyses...565

8. The Genesis of the DOJ Letter Was Kept from the Census Bureau...566

9. Findings Regarding the Timing of, and Reasons for, Secretary Ross's Decision...567

STANDING...572

A. General Legal Standards...573

B. Findings of Fact Related to Standing...576

1. Background...577

2. The Citizenship Question Will Cause a Differential Decline in Self-Response Rates...578

3. NRFU Operations Will Not Cure the Differential Drop in Self-Response Rates...583

4. Effects of the Citizenship Question on Apportionment Among and Within States...593 *514 5. Effects of the Citizenship Question on Funding to, and Within, States...596

6. Effects of the Citizenship Question on the Quality and Accuracy of Census Data...599

7. Secretary Ross's Decision Has Caused Plaintiffs to Divert Resources...600

C. Conclusions of Law Related to Standing...604

1. Associational Standing...604

2. Injury in Fact...606

a. Diminished Political Representation...607

b. Loss of Government Funds...608

c. Harm to the Quality and Accuracy of Data...610

d. Diversion of Resources...615

e. Loss of Privacy...618

3. Traceability and Redressability...619

RIPENESS...625

ADMINISTRATIVE PROCEDURE ACT CLAIMS...628

A. General Legal Standards...628

B. The Scope of Review...630

C. Discussion...635

1. Secretary Ross's Decision Was Not in Accordance with Law...636

a. The Section 6 Violation...636

b. The Section 141(f) Violation...641

2. Secretary Ross's Decision Was Arbitrary and Capricious...647

a. Secretary Ross's Explanations Ran Counter to the Evidence Before the Agency...647

b. Secretary Ross Failed to Consider Several Important Aspects of the Problem...651

c. Secretary Ross Failed to Justify Departures from the OMB Guidelines and the Census Bureau's Standards and Practices...654

3. Secretary Ross's Rationale Was Pretextual...660

THE DUE PROCESS CLAUSE CLAIM...664

A. Applicable Legal Principles...665

B. The Scope of Review...667

C. Discussion...669

REMEDIES...671

A. General Legal Principles...671

B. Discussion...673

1. Vacatur and Remand...673

2. Injunctive Relief...675

3. Declaratory Relief...678

CONCLUSION...679

The Constitution provides that "Representatives shall be apportioned among the several States according to their respective numbers, counting the whole number of persons in each State." U.S. Const. amend. XIV, § 2. To that end, it mandates that an "actual Enumeration" be conducted "every ... ten Years, in such Manner as [Congress] shall by Law direct," an effort commonly known as the census (or, more precisely, the decennial census). Id. art. I, § 2, cl. 3. By its terms, therefore, the Constitution mandates that every ten years the federal government endeavor to count every single person residing in the United States, whether citizen or noncitizen, whether living here with legal status or without. The population count derived from that effort is used not only to apportion Representatives among the states, but also to draw political districts and allocate power within them. And it is used to allocate hundreds of billions of dollars in federal, state, and local funds. Given the stakes, the interest in an accurate count is immense. Even small deviations from an *515 accurate count can have major implications for states, localities, and the people who live in them - indeed, for the country as a whole.

Since its inception in 1790, the decennial census also has been used for another purpose: to collect demographic data about the population of the United States, including information about respondents' race, sex, and age, and whether they own or rent their homes. Most relevant here, the government collected data about people's citizenship status from all households in the country in every census between 1820 and 1950 (with the exception of 1840). In 1960, however, the government stopped asking a citizenship question of every respondent, and for decades thereafter the official position of the Census Bureau was that reintroducing such a question was inadvisable because it would depress the count for already "hard-to-count" groups - particularly noncitizens and Hispanics - whose members would be less likely to participate in the census for fear that the data could be used against them or their loved ones. Every Secretary of Commerce (to whom Congress has long delegated significant authority over the census) adhered to that position - until early last year. On March 26, 2018, Secretary of Commerce Wilbur L. Ross, Jr. announced that he was reinstating the citizenship question on the 2020 census questionnaire, purportedly in response to a request from the Department of Justice ("DOJ") for better citizenship data to assist in its enforcement of Section 2 of the Voting Rights Act of 1965 ("VRA"). See 52 U.S.C. § 10301 . He did so over the strenuous objections of the Census Bureau itself, which warned that adding a citizenship question would harm the quality of census data and increase costs significantly and that it would do so for no good reason because there was an alternative way to satisfy DOJ's purported needs that would not cause those harms.

The question in these consolidated cases is whether Secretary Ross's decision to reinstate the citizenship question, and the process leading to that decision, violated provisions of statutory or constitutional law. Two sets of Plaintiffs - one, a coalition of eighteen states and the District of Columbia, fifteen cities and counties, and the United States Conference of Mayors (the "Governmental Plaintiffs"), and the other, a coalition of non-governmental organizations (the "NGO Plaintiffs") - challenge the decision on two principal grounds. First, they contend that the decision violated the Administrative Procedure Act ("APA"), which, among other things, prohibits federal agencies from acting in a manner that is arbitrary and capricious or not in accordance with law. Second, they allege that the decision violated the Due Process Clause of the Fifth Amendment because it was motivated in part by invidious discrimination against immigrant communities of color. Defendants - the United States Department of Commerce; Secretary Ross (the "Secretary"); the Bureau of the Census (the "Census Bureau"); and the Director of the Census, Dr. Steven Dillingham 1 - have tried mightily to avoid a ruling on the merits of these claims. They asserted a slew of unsuccessful jurisdictional arguments, raised multiple challenges to this Court's decisions authorizing discovery beyond the administrative record collected and filed in this litigation (one of which is still pending before the United States Supreme *516 Court), and tried no fewer than fourteen times to halt the proceedings altogether. Between November 5 and 27, 2018, however, this Court held - and completed - an eight-day bench trial to resolve Plaintiffs' claims, taking direct testimony by affidavit from many witnesses and orally from others.

This Opinion contains the Court's findings of fact and conclusions of law following that trial. Broadly speaking, the Court reaches three legal conclusions. First , the Court holds that most, if not all, of Plaintiffs have standing to bring their claims. Specifically, they have proved by a preponderance of the evidence that they will be harmed in various ways as a result of the addition of a citizenship question on the census and that a favorable ruling here will redress those harms. Defendants' own documents and expert witness confirm that adding a citizenship question to the census will result in a significant reduction in self-response rates among noncitizen and Hispanic households. And expert testimony, based in large part on the Census Bureau's own analyses of past censuses, indicates that the Census Bureau's "Non-Response Follow Up" procedures, extensive though they will be, are unlikely to remedy that reduction in self-response rates, which means that hundreds of thousands - if not millions - of people will go uncounted in the census if the citizenship question is included. The result will not only be a decrease in the quality of census data - something Defendants concede - but likely also a net differential undercount (that is, an undercount of certain sectors of the population, including people who live in households containing noncitizens and Hispanics, relative to others). That undercount, in turn, will translate into a loss of political power and funds, among other harms, for various Plaintiffs. In light of these and other factual findings, the Court holds that most, if not all, Plaintiffs have standing to bring their claims.

Second , the Court concludes on the merits that Secretary Ross violated the APA in multiple independent ways. Most blatantly, Secretary Ross ignored, and violated, a statute that requires him, in circumstances like those here, to collect data through the acquisition and use of "administrative records" instead of through "direct inquiries" on a survey such as the census. Additionally, Secretary Ross's decision to add a citizenship question was "arbitrary and capricious" on its own terms: He failed to consider several important aspects of the problem; alternately ignored, cherry-picked, or badly misconstrued the evidence in the record before him; acted irrationally both in light of that evidence and his own stated decisional criteria; and failed to justify significant departures from past policies and practices - a veritable smorgasbord of classic, clear-cut APA violations. On top of that, Secretary Ross acted without observing procedures required by law, including a statute requiring that he notify Congress of the subjects planned for any census at least three years in advance. And finally, the evidence establishes that Secretary Ross's stated rationale, to promote VRA enforcement, was pretextual - in other words, that he announced his decision in a manner that concealed its true basis rather than explaining it, as the APA required him to do. Notably, the Court reaches all of those conclusions based exclusively on the materials in the official "Administrative Record" - that is, the record of materials collected and submitted by Defendants that Secretary Ross allegedly considered, directly or indirectly, prior to making his decision. Looking beyond the Administrative Record merely confirms the Court's conclusions and illustrates how egregious the APA violations were.

*517 Third , on the merits of the constitutional claim, the Court concludes that Plaintiffs did not carry their burden of proving that Secretary Ross was motivated by invidious discrimination and thus that he violated the equal protection component of the Due Process Clause. In particular, although the Court finds that Secretary Ross's decision was pretextual, it is unable to find, on the record before it, that the decision was a pretext for impermissible discrimination. To be fair to Plaintiffs, it is impossible to know if they could have carried their burden to prove such discriminatory intent had they been allowed to depose Secretary Ross, as the Court had authorized last September. As defense counsel more or less conceded during closing arguments, a deposition of Secretary Ross would have been the best evidence of the question at the heart of the due process inquiry - namely, the true nature of Secretary Ross's intent in reinstating the citizenship question. But this Court's order authorizing such a deposition was stayed by the Supreme Court pending its further review, see In re Dep't of Commerce , --- U.S. ----, 139 S.Ct. 566 , 202 L.Ed.2d 400 , 2018 WL 5458822 (Nov. 16, 2018) ; In re Dep't of Commerce , --- U.S. ----, 139 S.Ct. 16 , 202 L.Ed.2d 306 (2018) (mem.), and Plaintiffs made the understandable decision to proceed with trial despite that stay (because, with the clock ticking on census preparations, waiting for a final ruling from the Supreme Court could have cost Plaintiffs a meaningful chance to obtain any relief). Be that as it may, it was - and remains - Plaintiffs' burden to prove discriminatory intent, and the evidence in the existing record does not support a conclusion that they carried that burden.

The Court's Opinion is, to put it mildly, long. But that is for good reasons. For one thing, the Court has taken care to thoroughly examine every issue because the integrity of the census is a matter of national importance. As noted, the population count has massive and lasting consequences. And it occurs only once a decade, with no possibility of a do-over if it turns out to be flawed. See Departments of Commerce, Justice, and State, The Judiciary, and Related Agencies Appropriations Act, 1998, § 209(a)(8), Pub. L. No. 105-119, 111 Stat. 2440 , 2480-81 (1997) ("1998 Appropriations Act") ("Congress finds that ... the decennial enumeration of the population is a complex and vast undertaking, and if such enumeration is conducted in a manner that does not comply with the requirements of the Constitution or laws of the United States, it would be impracticable for the States to obtain, and the courts of the United States to provide, meaningful relief after such enumeration has been conducted."). For another, time is of the essence because the Census Bureau needs to finalize the 2020 questionnaire by June of this year. See Docket No. 540, at 3; see also Brief for Petitioners at 45, Department of Commerce v. U.S. Dist. Ct. for S.D.N.Y. (Dec. 17, 2018) (No. 18-557), 2018 WL 6650094 , at *45 (noting "the need to finalize the census questionnaire by mid-2019"). 2 With time so short and the likelihood that one or both sides will seek appellate relief so high, it is critical to make a comprehensive record in order to facilitate higher court review and to minimize any potential need for a remand. That means reaching most, if not all, issues raised by the parties - even if, in other circumstances, it would be unnecessary or even inadvisable. See, e.g., Ashwander v. TVA , 297 U.S. 288 , 347, 56 S.Ct. 466 , 80 L.Ed. 688 (1936) (Brandeis, J., concurring) ("The Court will not pass upon a constitutional question although properly presented *518 by the record, if there is also present some other ground upon which the case may be disposed of."). And in light of Defendants' strenuous objections to consideration of any extra-record discovery - objections they are already pressing before the Supreme Court - it also means spelling out which facts are drawn exclusively from the Administrative Record rather than from other evidence, and which conclusions of law are based solely on the Administrative Record rather than on other evidence. 3

In short (or not, as the case may be), the Court concludes that Secretary Ross's decision to add the citizenship question to the 2020 census questionnaire, while not inconsistent with the Constitution, violated the APA in several respects. Those violations are no mere trifles. The fair and orderly administration of the census is one of the Secretary of Commerce's most important duties, as it is critical that the public have "confidence in the integrity of the process." Franklin v. Massachusetts , 505 U.S. 788 , 818, 112 S.Ct. 2767 , 120 L.Ed.2d 636 (1992) (Stevens, J., concurring in part and concurring in the judgment). And although some may deride its requirements as "red tape," the APA exists to protect core constitutional and democratic values: It ensures that agencies exercise only the authority that Congress has given them, that they exercise that authority reasonably, and that they follow applicable procedures - in short, it ensures that agencies remain accountable to the public they serve. That is not to say - and the APA does not say - that an agency cannot adopt new policies or otherwise change course. But the APA does require that before an agency does so, it must consider all important aspects of a problem; study the relevant evidence and arrive at a decision rationally supported by that evidence; comply with all applicable procedures and substantive laws; and articulate the facts and reasons - the real reasons - for that decision. The Administrative Record in these cases makes plain that Secretary Ross's decision fell short on all these fronts. In arriving at his decision as he did, Secretary Ross violated the law. And in doing so with respect to the census - "one of the most critical constitutional functions our Federal Government performs," 1998 Appropriations Act, § 209(a)(5), 111 Stat. at 2480-81, and a "mainstay of our democracy," Franklin , 505 U.S. at 818 , 112 S.Ct. 2767 (Stevens, J., concurring) - Secretary Ross violated the public trust.

BACKGROUND

The Court begins with relevant background concerning the history and purpose of the census, the Secretary's authority over the census, and the history of the citizenship question on the census. The relevant background is largely undisputed (for example, as reflected in stipulations of the parties) or drawn from materials of which the Court can take judicial notice. See, e.g. , Fed. R. Evid. 201(b)-(c) ; Effie Film, LLC v. Pomerance , 909 F.Supp.2d 273 , 298-303 (S.D.N.Y. 2012) (noting that a court may take notice of undisputed historical facts). To the extent that the Court cites trial testimony or exhibits in what follows, it is only by way of background and does not form a basis for any of the conclusions of law later in this Opinion. 4

*519 A. History and Purposes of the Census

Article I of the Constitution requires Congress to carry out an "actual Enumeration" every ten years, "in such Manner as [Congress] shall by Law direct." U.S. Const. art. I, § 2, cl. 3. The original and fundamental purpose of this "Enumeration," known as the decennial census, was to apportion congressional representatives (and, where necessary, direct taxes) among the states "according to their respective Numbers." Id. For the first eighty years of the country's history, the states' "respective Numbers" were calculated according to a formula mandated by the same constitutional provision's infamous Three-Fifths Clause, which reformulated the "actual Enumeration" established by the census by "adding to the whole Number of free Persons ..., and excluding Indians not taxed, three fifths of all other Persons" - "all other Persons" being the people then held as slaves. Id. (amended 1868). After the Civil War, that provision was superseded by the Fourteenth Amendment, which provides that "Representatives shall be apportioned among the several States according to their respective numbers, counting the whole number of persons in each State, excluding Indians not taxed." U.S. Const. amend. XIV, § 2 (emphasis added). The result is that the Constitution "requires the federal government to conduct a Decennial Census counting the total number of 'persons' - with no reference to citizenship status - residing in each state." Docket No. 480-1 ("Joint Stips."), ¶ 1.

Significantly, although the "initial constitutional purpose" of the census was to "provide a basis for apportioning representatives among the states in the Congress," it has long "fulfill[ed] many important and valuable functions for the benefit of the country." Baldrige v. Shapiro , 455 U.S. 345 , 353, 102 S.Ct. 1103 , 71 L.Ed.2d 199 (1982). In particular, it "now serves as a linchpin of the federal statistical system by collecting data on the characteristics of individuals, households, and housing units throughout the country." Department of Commerce v. United States House of Representatives , 525 U.S. 316 , 341, 119 S.Ct. 765 , 142 L.Ed.2d 797 (1999) (internal quotation marks omitted). See generally U.S. CENSUS BUREAU , MEASURING AMERICA: THE DECENNIAL CENSUSES FROM 1790 TO 2000 ("MEASURING AMERICA ") (2002), http://www2.census.gov/library/publications/2002/dec/pol_02-ma.pdf. "Today, policy makers at all levels of government, as well as private businesses, households, researchers, and nonprofit organizations, rely on an accurate census in myriad ways that range far beyond the single fact of how many people live in each state." COUNCIL OF ECONOMIC ADVISERS , THE USE OF CENSUS DATA: AN ANALYTICAL REVIEW (2000), https://clintonwhitehouse4.archives.gov/media/pdf/censusreview.pdf. Among other things, the data are used "for such varied purposes as computing federal grant-in-aid benefits, drafting of legislation, urban and regional *520 planning, business planning, and academic and social studies." Baldrige , 455 U.S. at 353 n.9, 102 S.Ct. 1103 .

Since 1790, the government has conducted the required "actual Enumeration" through questions - initially asked in person by U.S. Marshals and "specially appointed agents" and later by means of written questionnaire - about both the number and demographic backgrounds of those living in each American household. See MEASURING AMERICA 125-40. Congress provided for each of the first twelve censuses on an ad hoc basis; then, in 1902, Congress established the "Census Office" that it had organized for the twelfth census as a permanent office within the Department of the Interior, to be supervised by a "Director of the Census" appointed by the President and confirmed by the Senate. See Act of Mar. 6, 1902 §§ 1-3, Pub. L. No. 57-27, 32 Stat. 51 , 51. Shortly thereafter, Congress moved the Census Office into the newly created Department of Commerce and Labor. See Act of Feb. 14, 1903 § 4, Pub. L. No. 57-87, 32 Stat. 825 , 826-27. Ten years later, various parts of that combined department were transferred into the newly fashioned Department of Labor, and the Census Office was left behind at the slimmed-down (and renamed) Department of Commerce. See Act of Mar. 4, 1913 §§ 1, 3, Pub. L. No. 62-426, 37 Stat. 736 , 736-37. 5 (Today the "Census Office" is known as the "Bureau of the Census," or the "Census Bureau.") In 1954, Congress enacted the various census-related statutes codified in Title 13 into positive law, see Act of Aug. 31, 1954, Pub. L. No. 83-740, 68 Stat. 1012 , and has enacted all subsequent revisions to the census statutes as amendments to Title 13, which, as a result, is known colloquially as the "Census Act."

The modern decennial census is administered exclusively through a "short-form" questionnaire - a short questionnaire containing only a handful of questions. This is a relatively recent phenomenon. Beginning in 1960, and until 2000, each census also included a "long-form" questionnaire, which contained many additional questions but was sent to only a sample fraction of the population. See MEASURING AMERICA 72. In 1960, twenty-five percent of households received the "long-form" questionnaire, while the remainder received the "short-form." Id. In the 1970 and subsequent censuses, approximately one-sixth of all households received the "long-form" questionnaire. See Joint Stips. ¶ 31. During that time, none of the short-form questionnaires included a question about citizenship or birthplace, although the long-form questionnaires each did. See Joint Stips. ¶¶ 30, 35. After the 2000 census, the Census Bureau introduced a new survey instrument, the American Community Survey ("ACS"). Joint Stips. ¶ 37. Unlike the decennial census questionnaires, the ACS is conducted annually and not used to enumerate the population for apportionment purposes. It is distributed to about 3.5 million households (approximately two percent of households in the country) each year, for the sole purpose of collecting demographic data about the population. Since its inception, the ACS has included a question on citizenship. Joint Stips. ¶¶ 38-41. With the advent of the ACS, the Census Bureau phased out the "long-form" decennial census questionnaire. Joint Stips. ¶ 37; see JENNIFER D. WILLIAMS , THE 2010 DECENNIAL CENSUS: BACKGROUND AND ISSUES 3 (2011), https://www.census.gov/history/pdf/2010-background-crs.pdf. Thus, *521 the 2010 decennial census included only a single "short-form" questionnaire, distributed primarily by mail. Id. So too, the 2020 census will be conducted with a single short-form questionnaire, which, for the first time, many respondents will complete online. See Tr. 1091.

The modern decennial census begins with a Master Address File ("MAF"), a database containing every known housing unit in the country. See 83 Fed. Reg. 26643 , 26644 (June 8, 2018). Every household is then given the opportunity to self-respond to the census questionnaire. Joint Stips. ¶ 7. A majority of households self-respond to that questionnaire. To attempt to count the households that do not self-respond, the Census Bureau uses a set of procedures known as "Non-Response Follow-Up" or "NRFU." Joint Stips. ¶ 8. The first step in NRFU is an in-person visit from a census enumerator. Assuming that a household is listed in the MAF - a precondition for any NRFU efforts - a NRFU enumerator will visit any nonresponding household in person and, if possible, conduct the census survey face-to-face. In the 2020 census, if the NRFU enumerator's first visit is unsuccessful, but the Census Bureau believes the housing unit to be occupied, the Census Bureau will then refer to "administrative records" - data collected from other federal or state entities - to enumerate that household, assuming administrative records of sufficient quality exist. Joint Stips. ¶¶ 9-10; 83 Fed. Reg. at 26649 . If the household cannot be enumerated with high-quality administrative records, a NRFU enumerator will return to the household for at least two more in-person attempts. Joint Stips. ¶¶ 10-11. After three unsuccessful attempts to contact a member of the household in person, the NRFU enumerator will return and attempt to gather information from a "proxy," such as a neighbor or landlord, who can report what he or she knows about the household and its members. Joint Stips. ¶¶ 11-13; 83 Fed. Reg. at 26649 . Finally, if enumeration-by-proxy fails, the Census Bureau will then "impute" either the number of household members or their characteristics (or both) based on already-existing data from the area. See AR 1281-82, 1304. In other words, the Census Bureau will use a formula to extrapolate what it does not know about the population from what it already knows. See Tr. 1351. Because NRFU data is less accurate than self-response data, see AR 1281, the Census Bureau places a high priority on obtaining self-responses from as many households as possible, see AR 163-65. To that end, the Census Bureau partners with local organizations (which it refers to as "Trusted Partners") to encourage local households to self-respond. Joint Stips. ¶¶ 26, 28.

B. The Secretary's Authority Over the Census

Since Congress first delegated its census-related authority to an Executive Branch official, it has retained some control over the design and administration of the census. The first permanent delegation, for example, provided that the decennial census "shall be restricted to inquiries relating to the population, to mortality, to the products of agriculture and of manufacturing and mechanical establishments" and that the tabulations of population "shall comprehend for each inhabitant the name, age, color, sex, conjugal condition, place of birth, and place of birth of parents, whether alien or naturalized, number of years in the United States, occupation, months unemployed, literacy, school attendance, and ownership of farms and homes." Act of Mar. 3, 1899 § 7, 30 Stat. 1014 , 1015. Within those broad confines, however, Congress provided that "the Director of the Census may use his discretion *522 as to the construction and form and number of inquiries necessary to secure information." Id.

In 1976, Congress amended Title 13 substantially. See Act. of Oct. 17, 1976 ("1976 Census Act"), Pub. L. No. 94-521, 90 Stat. 2459 (codified in scattered sections of 13 U.S.C.). Among other things, the 1976 Census Act amended Section 141(a) of Title 13 to update and consolidate its delegation of authority over the census to the Secretary of Commerce. See id. § 7(a), 90 Stat. at 2461 (codified at 13 U.S.C. § 141 (a) ). Section 141(a) now provides:

The Secretary shall, in the year 1980 and every 10 years thereafter, take a decennial census of population as of the first day of April of such year, which date shall be known as the "decennial census date," in such form and content as he may determine, including the use of sampling procedures and special surveys. In connection with any such census, the Secretary is authorized to obtain such other census information as necessary.

13 U.S.C. § 141 (a). That delegation now sits alongside Section 5 of Title 13, which authorizes the Secretary to "prepare questionnaires," including but not only for the decennial census, and to "determine the inquiries, and the number, form, and subdivisions thereof." 13 U.S.C. § 5 .

Along with - and within - that broad delegation, however, the 1976 Census Act also constrained the Secretary's delegated authority over the decennial census and its questionnaires in several significant ways. First, by its terms, Section 141(a) itself authorized the Secretary to collect information "other" than population information only "as necessary." Id. § 141(a). Second, and significantly, Congress added a new subsection to Title 13's Section 6, which had previously merely authorized the Secretary to acquire and use "pertinent" information from other federal, state, and local authorities for the purpose of gathering census-related data. See 13 U.S.C. § 6 (1970). The new subsection - Section 6(c) - added that, "[t]o the maximum extent possible and consistent with the kind, timeliness, quality and scope of the statistics required, the Secretary shall acquire and use information available from any source referred to in subsection (a) or (b) of this section instead of conducting direct inquiries." 1976 Census Act § 5(a), 90 Stat. at 2460 (codified at 13 U.S.C. § 6 (c) ) (emphases added).

At the same time, Congress also cabined the Secretary's authority to collect data - other than for the straightforward purpose of counting whole persons for apportionment purposes - through nationwide inquiries of the whole population. Whereas Section 195 of Title 13 had previously merely authorized data collection through statistical sampling, the 1976 Census Act amended that provision to state that, "[e]xcept for the determination of population for purposes of apportionment of Representatives in Congress among the several States, the Secretary shall , if he considers it feasible, authorize the use of the statistical method known as 'sampling' in carrying out the provisions of this title." 1976 Census Act § 10, 90 Stat. at 2464 (codified at 13 U.S.C. § 195 ) (emphasis added); see Department of Commerce , 525 U.S. at 341 , 119 S.Ct. 765 (noting that the new Section 195"changed a provision that permitted the use of sampling for purposes other than apportionment into one that required that sampling be used for such purposes if 'feasible' "). Thus, the "broad grant of authority given in § 141(a) is informed ... by the narrower and more specific § 195," Department of Commerce , 525 U.S. at 338 , 119 S.Ct. 765 , and is similarly limited by the narrower and more specific Section 6(c). Thus, together, *523 Sections 6(c) and Section 195 effectively established a new default rule for the collection of census data other than for apportionment purposes: first, the Secretary was to "acquire and use" administrative record data instead of conducting direct surveys "to the maximum extent possible," if consistent with the type of data required, 13 U.S.C. § 6 (c) ; and, second, when conducting surveys, he was required to use statistical sampling "if ... feasible," instead of asking a question of everyone, id. § 195.

That was not all. The 1976 Congress also enacted a new reporting requirement, mandating that the Secretary report to the relevant congressional committees, at least three years before the "census date" for a given census, all "subjects proposed to be included, and the types of information to be compiled." 13 U.S.C. § 141 (f)(1). Further, no later than two years before the given census date, the Secretary must report to the same congressional committees all " questions proposed to be included in such census." Id. § 141(f)(2) (emphasis added). Congress authorized the Secretary to diverge from the proposals set forth in those reports, but only if he "finds new circumstances exist which necessitate that the subjects, types of information, or questions contained in reports so submitted be modified," and he submits another report "containing the Secretary's determination of the subjects, types of information, or questions as proposed to be modified." Id. § 141(f)(3).

Finally, to the extent relevant here, the 1976 Census Act's new constraints on the Secretary's authority built on another important, longstanding constraint: a sharp restriction on the authority to share any information gathered in any given data collection effort. With certain limited exceptions, Section 9 of Title 13 provides:

Neither the Secretary, nor any other officer or employee of the Department of Commerce or bureau or agency thereof, ... may ...

(1) use the information furnished under the provisions of [Title 13] for any purpose other than the statistical purposes for which it is supplied; or

(2) make any publication whereby the data furnished by any particular establishment or individual under [Title 13] can be identified; or

(3) permit anyone other than the sworn officers and employees of the Department or bureau or agency thereof to examine the individual reports.

13 U.S.C. § 9 (a). Moreover, Section 9 categorically forbids anyone , "except the Secretary in carrying out the purposes of [Title 13]," from ordering the production of census reports that have been retained by the people who submitted them. Id. And it provides that copies of such census materials are both "immune from legal process" and unusable "for any purpose in any action, suit, or other judicial or administrative proceeding," without the person's consent. Id. Notably, the Secretary's authority to share data with other federal agencies, including DOJ, is "[s]ubject to the limitations contained in" Section 9 ( and Section 6(c) ). Id. § 8(b). More specifically, although the Secretary is authorized to "furnish copies of tabulations and other statistical materials" to other federal agencies, those materials may "not disclose the information reported by, or on behalf of, any particular respondent." Id.

In sum, as befits a subject over which the Constitution assigns Congress "virtually unlimited discretion," Wisconsin v. City of New York , 517 U.S. 1 , 19, 116 S.Ct. 1091 , 134 L.Ed.2d 167 (1996), Congress has been judicious in its delegation of that authority to the Executive Branch, and it *524 has retained an active role in policing the form, content, and methodology of the census through these and other provisions. That is, even as it has delegated broad authority over the census to the Secretary, Congress has taken care to limit that authority and, with respect to a few topics that it has deemed especially worthy of restraint - such as the use of survey questions instead of administrative records or the practice of asking survey questions of all respondents as opposed to sampling - Congress has enacted clear instructions for the Secretary to follow in carrying out his statutory duties. Cf. Department of Commerce , 525 U.S. at 337 -39 , 119 S.Ct. 765 .

C. The History of a Citizenship Question on the Census

As the Court described at length in an earlier Opinion and Order, see New York v. U.S. Dep't of Commerce , 315 F.Supp.3d 766 , 776-79 (S.D.N.Y. 2018), the questions posed on the census have ebbed and flowed since the first census in 1790 asked each household about "the sexes and colours of free persons," as well the age of each resident, see Act of March 1, 1790 § 1, 1 Stat. 101 , 101-02 (1790). Most relevant for present purposes, a question regarding citizenship appeared for the first time on the fourth census in 1820, when Congress directed enumerators to tally the number of "Foreigners not naturalized." Act of March 14, 1820 § 1, 3 Stat. 548 , 550 (1820). With one unexplained exception (the 1840 census), a question about citizenship status or birthplace appeared on every census thereafter through 1950. See New York , 315 F.Supp.3d at 776 -79 . That changed in 1960 - the first census after Congress authorized the use of sampling. Id. at 778 . That year, only five questions were posed to all respondents, concerning the respondent's relationship to the head of household, sex, color or race, marital status, and month and year of birth. 6 In a review of the census several years later, the Census Bureau explained the decision not to ask all respondents about citizenship as follows: "It was felt that general census information on citizenship had become of less importance compared with other possible questions to be included in the census, particularly in view of the recent statutory requirement for annual alien registration which could provide the Immigration and Naturalization Service, the principal user of such data, with the information it needed." U.S. BUREAU OF THE CENSUS , 1960 CENSUSES OF POPULATION AND HOUSING: PROCEDURAL HISTORY ("1960 CENSUSES OF POPULATION AND HOUSING") 194 (1966), http://www2.census.gov/prod2/decennial/documents/1960/proceduralHistory/1960proceduralhistory.zip.

Between 1970 and 2000, the Census Bureau used both a short-form questionnaire (containing only a handful of questions), which was distributed to the vast majority of the population, and a long-form questionnaire (containing both the inquiries on the short-form questionnaire as well as additional questions), which was distributed to only a sample of the population. During that time, the long-form questionnaires contained a citizenship question, but the short-form questionnaires did not. See MEASURING AMERICA 77-78, 84-85, 91-92, *525 100-101. In 2010, after the advent of the ACS, the Census Bureau dropped the long-form questionnaire entirely. The 2010 census asked about such matters as "the age, sex, race, and ethnicity (Hispanic or non-Hispanic) of each person in a household," but did not ask about citizenship. WILLIAMS , THE 2010 DECENNIAL CENSUS 3. Thus, the last time that the census asked every respondent about citizenship was sixty-nine years ago, in 1950. Notably, that is before the VRA was enacted in 1965. In other words, for all fifty-four years that the VRA has existed, the federal government has never had a "hard-count" tally of the number of citizens in the country. Instead, consistent with the requirement to use statistical sampling techniques "if ... feasible" for everything other than the constitutionally mandated "actual Enumeration," see 13 U.S.C. § 195 , the federal government has extrapolated from citizenship data collected from a subset of the population to model data for the population as a whole, see WILLIAMS , THE 2010 DECENNIAL CENSUS 3.

Since 1950, the Census Bureau and former Census Bureau officials have consistently opposed periodic proposals to resume asking a citizenship question of every census respondent. In 1980, for example, several plaintiffs (including the Federation for American Immigration Reform, which appears here as amicus curiae in support of Defendants, see Docket Nos. 75, 179) sued the Census Bureau, contending that the census was constitutionally required to count only citizens. See Fed'n for Am. Immigration Reform v. Klutznick , 486 F.Supp. 564 (D.D.C. 1980) (" FAIR ") (three-judge court). In that litigation, the Census Bureau argued that reinstating a citizenship question for all respondents would "inevitably jeopardize the overall accuracy of the population count" because noncitizens would be reluctant to participate, for fear "of the information being used against them." Id . at 568. Likewise, in Congressional testimony prior to the 1990 census, Census Bureau officials opposed reinstating a citizenship question for all respondents, opining that it could cause legal residents to "misunderstand or mistrust the census and fail or refuse to respond." Exclude Undocumented Residents from Census Counts Used for Apportionment: Hearing on H.R. 3639, H.R. 3814, and H.R. 4234 Before the Subcomm. on Census & Population of the H. Comm. on Post Office & Civil Serv. , 100th Cong. 47-51 (1988) (statement of John G. Keane, Director, Bureau of the Census); see also Census Equity Act: Hearings on H.R. 2661 Before the Subcomm. on Census & Population of the H. Comm. on Post Office & Civ. Serv. , 101st Cong. 42-44 (1989) (statement of C. Louis Kincannon, Deputy Director, Bureau of the Census). Before the 2010 census, former Bureau Director Kenneth Prewitt testified before Congress to the same effect. See Counting the Vote: Should Only U.S. Citizens Be Included in Apportioning Our Elected Representatives?: Hearing Before the Subcomm. on Federalism & the Census of the H. Comm. on Gov't Reform , 109th Cong. 72 (2005) (statement of Kenneth Prewitt). Just two years ago, four former Census Bureau Directors wrote in an amicus curiae brief to the Supreme Court (in a case about the use of total population in intrastate redistricting) that a "citizenship inquiry would invariably lead to a lower response rate to the Census." Brief of Former Directors of the U.S. Census Bureau as Amici Curiae in Support of Appellees at 25, Evenwel v. Abbott , 136 S.Ct. 1120 (2016) (No. 14-940), 2015 WL 5675832 , at *25. 7

*526 D. Testing and Adding New Questions to the Census

The Census Bureau's longstanding opposition to a citizenship question on the census is consistent with a more general reluctance to tinker with the questionnaire unnecessarily. That is, although early census questionnaires changed regularly, a more sophisticated understanding of statistics and survey design in the modern era has caused the Census Bureau to approach any changes to the questionnaire with great care. For instance, after the 1990 census, the Census Bureau considered adding a question regarding respondents' Social Security Numbers ("SSNs") to the "short-form" questionnaire. See Tr. 998-99. Before deciding to add such a question, however, the Census Bureau conducted a randomized controlled trial comparing a version of the questionnaire that asked about SSNs to one that did not in order to assess the question's impact on self-response rates. See id. at 999 . Overall, the Census Bureau observed a 3.4% decline in self-response rates attributable to the question, a decline that was not evenly distributed among subpopulations. See id. at 999-1000 . In part due to these results, the Census Bureau did not - and does not to this day - ask a question about SSNs on the decennial census. See id. at 999 .

As the SSN example reflects, in recent decades, the Census Bureau has followed a fairly robust process in evaluating whether to add a new question to a survey such as the census. AR 9865, 9867; AR 3560; AR 3890-91; Docket No. 516-1 ("Thompson Decl."), ¶¶ 45, 47-49. The process usually begins with a request from Congress or an Executive Branch agency to add a question. AR 3890; see AR 2304. After receiving such a request, the Census Bureau works with the Office of Management and Budget ("OMB") to ensure that the proposed data collection would comply with applicable legal and regulatory requirements. AR 3890. If the Census Bureau determines that adding the new question is "warranted," the Secretary of Commerce notifies Congress of his intent to add the question - first by including the subject of the question in the Section 141(f)(1) report to Congress, at least three years before the census date and, later, by reporting the question itself in the Section 141(f)(2) report, at least two years before the census date. Id. The Census Bureau must then test the wording of the new question. AR 3891. Pre-testing requires approval from OMB and a process that includes notifying the public and inviting comment through a notice in the Federal Register. Id. After the Census Bureau has responded to comments, OMB can approve the test. Id. Once the question has been tested, the Census Bureau must redesign the questionnaires (or internet collection systems), including translation into non-English languages, and redevelop training procedures for enumerators. Id. Finally, the Census Bureau must submit the final questionnaire to OMB for approval. Id.

This process is subject to several sets of guidelines and standards governing collection of statistical data. First, since 2006, the design and administration of governmental surveys - including the census - have been subject to OMB's Standards and Guidelines for Statistical Surveys . PX-260, at ii; see 79 Fed. Reg. 71610 (Dec. 2, 2014) ; PX-359; see also Docket No. 498-11 ("Habermann Aff."), ¶ 20. Several provisions of the OMB Standards and Guide lines *527 are relevant here. First, Statistical Directive Number 1 requires that "a Federal statistical agency must be independent from political and other undue external influence in developing, producing, and disseminating statistics." 79 Fed. Reg. at 71612 . Second, Standard 2.3 states that "[a]gencies must design and administer their data collection instruments and methods in a manner that achieves the best balance between maximizing data quality and controlling measurement error while minimizing respondent burden and cost." PX-359, at ii. Finally, Guideline 2.3.1 states: "Design the data collection instrument in a manner that minimizes respondent burden, while maximizing data quality." Id. at 11.

Additionally, in 2010, the Census Bureau adopted a comprehensive set of "Statistical Quality Standards." PX-260, at vii. The Standards require pre-testing of any questions to be added to data-collection products such as the census questionnaire. See id. at 8. Sub-Requirement A2-3.3 of the Standards requires that "[d]ata collection instruments and supporting materials must be pretested with respondents to identify problems (e.g., problems related to content, order/context effects, skip instructions, formatting, navigation, and edits) and then refined, prior to implementation, based on the pretesting results." Id. at 8 (emphasis added). Sub-Requirements A2-3.3-1c and A2-3.3-1d further provide that pretesting must be performed when "[r]eview by cognitive experts reveals that adding pretested questions to an existing instrument may cause potential context effects" and when "[a]n existing data collection instrument has substantive modifications (e.g., existing questions are revised or new questions added)." Id. The Standards note that, "[o]n rare occasions, cost or schedule constraints may make it infeasible to perform complete pretesting. In such cases, subject matter and cognitive experts must discuss the need for and feasibility of pretesting. The program manager must document any decisions regarding such pretesting, including the reasons for the decision. If no acceptable options for pretesting can be identified, the program manager must apply for a waiver." Id. The Standards provide for another exception to the pretesting requirement: "Pretesting is not required for questions that performed adequately in another survey." Id.

E. Secretary Ross's Decision and This Litigation

As noted above, the Census Act requires the Secretary of Commerce to submit a report to the relevant congressional committees at least three years before any given census listing the "subjects proposed to be included, and the types of information to be compiled" on the census. 13 U.S.C. § 141 (f)(1). Consistent with that requirement, in March 2017 - approximately one month after his confirmation by the Senate - Secretary Ross submitted a report to Congress titled "Subjects Planned for the 2020 Census and American Community Survey." See AR 194-270. The report listed as the planned subjects for the 2020 census questionnaire the very same subjects that had appeared on the 2010 census questionnaire: age, gender, race/ethnicity, relationship, and tenure (that is, whether the respondent's home in question is owned or rented). See AR 204-13. The list of subjects did not include citizenship status.

On March 26, 2018, however, Secretary Ross issued a memorandum directing the Census Bureau to reinstate a question about citizenship status on the 2020 census questionnaire. See AR 1313-20 ("Ross Memo"). In his memorandum, Secretary Ross asserted that his decision was *528 prompted by a letter from DOJ, dated December 12, 2017, which requested reinstatement of the citizenship question to facilitate enforcement of Section 2 of the VRA. See Ross Memo 1, at AR 1313. A few days later, Secretary Ross submitted another report to Congress (of which the Court can and does take judicial notice) titled "Questions Planned for the 2020 Census and American Community Survey." See PX-489. The report included the following planned question about citizenship:

Id. at 7 . It stated that a question about citizenship had been "asked since 1820." Id. ; see id. n.1 ("Citizenship asked 1820, 1830, 1870, and 1890 to present."). And it asserted that the question is "USED TO CREATE STATISTICS ABOUT CITIZEN AND NONCITIZEN POPULATIONS," which "are essential for enforcing the Voting Rights Act" and "is of interest to researchers, advocacy groups, and policymakers." Id. (capitalization in original).

Eight days after Secretary Ross's March 26, 2018 memorandum announcing his decision, the first of these cases - brought by a coalition of states and local governmental entities (the "Governmental Plaintiffs") - was filed. See Docket No. 1. The Governmental Plaintiffs are comprised of eighteen states (New York, Colorado, Connecticut, Delaware, Illinois, Iowa, Maryland, Massachusetts, Minnesota, New Jersey, New Mexico, North Carolina, Oregon, Pennsylvania, Rhode Island, Vermont, Virginia, and Washington) and the District of Columbia, fifteen cities and counties (the cities of Central Falls, Chicago, Columbus, New York, Philadelphia, Phoenix, Pittsburgh, Providence, and Seattle; the city and county of San Francisco; and the counties of Cameron, El Paso, Hidalgo, and Monterey), and the United States Conference of Mayors. See Governmental Plaintiffs' Second Amended Complaint ("SAC"), Docket No. 214. On June 6, 2018, the second case - brought by a coalition of nongovernmental organizations (the "NGO Plaintiffs") - was filed. See NGO Plaintiffs' Complaint ("NGO Compl."), 18-CV-5025, Docket No. 1. The NGO Plaintiffs are comprised of New York Immigration Coalition ("NYIC"), CASA de Maryland ("CASA"), American-Arab Anti-Discrimination Committee ("ADC"), ADC Research Institute ("ADCRI"), and Make the Road New York ("MRNY"). See id. In each case, Plaintiffs alleged that Secretary Ross's decision to reinstate a citizenship question violated both the APA and the Enumeration Clause of the Constitution. The NGO Plaintiffs argued as well that the decision violated the equal protection component of the Fifth Amendment's Due Process Clause.

*529 The procedural history of these cases has been somewhat unusual because, among other things, Defendants filed multiple petitions for mandamus challenging discovery orders issued by the Court and corresponding applications - to this Court, the Second Circuit, and the Supreme Court - to stay proceedings pending adjudication of those petitions. See New York v. U.S. Dep't of Commerce , No. 18-CV-2921 (JMF), 2018 WL 6060304 , at *1 n.1 (S.D.N.Y. Nov. 20, 2018) (detailing this history). For present purposes, two pieces of that procedural history warrant mention. First, in May and June 2018, Defendants moved to dismiss all of Plaintiffs' claims. See Docket No. 154; 18-CV-5025, Docket No. 38. On July 26, 2018, the Court granted those motions in part and denied them in part. See New York , 315 F.Supp.3d at 811 -12 . The Court denied the motions with respect to Plaintiffs' claims under the APA and the Due Process Clause, finding, among other things, that the NGO Plaintiffs had alleged a plausible claim of invidious discrimination in violation of the Fifth Amendment's Due Process Clause. See id. at 806-08 . By contrast, the Court held that Plaintiffs failed to state a claim under the Enumeration Clause. "That conclusion," the Court reasoned, was "compelled not only by the text of the Clause, which vests Congress with virtually unlimited discretion in conducting the census, but also by historical practice," namely "that, since the very first census in 1790, the federal government has consistently used the decennial exercise not only to obtain a strict headcount ..., but also to gather demographic data about the population on matters such as race, sex, occupation, and, even citizenship." Id. at 774 .

Second, around the same time, Plaintiffs moved for relief related to the "Administrative Record" - the record, compiled and submitted by Defendants, of materials "upon which the Secretary of Commerce based his decision." Docket Nos. 173, 173-1; see Docket No. 193. To the extent relevant here, Plaintiffs moved for two forms of relief: first, an order compelling Defendants to "complete" the Administrative Record; and second, an order authorizing discovery "outside" the Administrative Record. See Docket No. 193, at 1. In an oral ruling on July 3, 2018, the Court granted both requests. With respect to the former, the Court found that the Administrative Record did not constitute the "whole record" - namely, the "full scope of" materials that Secretary Ross had considered, whether directly, "indirectly," or "constructively." Docket No. 208 ("July 3rd Tr."), at 79-82. And with respect to the latter, the Court found that Plaintiffs had "made a strong preliminary showing or prima facie showing that they will find material beyond the Administrative Record indicative of bad faith" or pretext. Id. at 85. Notably, Defendants did not immediately challenge the Court's ruling authorizing discovery beyond the Administrative Record (and have never challenged its ruling with respect to completing the Administrative Record). Several months later, however, after the Court authorized depositions of Secretary Ross and a DOJ official, Defendants challenged those rulings by way of petitions for mandamus, tacking on a challenge to the Court's initial discovery Order. Eventually, the Supreme Court agreed to hear one such challenge (treating Defendants' mandamus petition as a petition for certiorari), In re Dep't of Commerce , --- U.S. ----, 139 S.Ct. 566 , 202 L.Ed.2d 400 , 2018 WL 5458822 (Nov. 16, 2018) ; and stayed the deposition of Secretary Ross pending its decision, but otherwise allowed these proceedings to continue, see In re Dep't of Commerce , --- U.S. ----, 139 S.Ct. 16 , 202 L.Ed.2d 306 (2018) (mem.). 8 The case proceeded to trial, without *530 a jury, on November 5, 2018. After extensive post-trial briefing (Plaintiffs' briefs, alone, totaled 502 pages), the Court held closing arguments on November 27, 2018. 9

SECRETARY ROSS'S DECISION

With that as background, the Court turns to the process and basis for Secretary Ross's March 26, 2018 decision. In light of the pending Supreme Court challenge to the Court's decision authorizing extra-record discovery and the limited time to resolve Plaintiffs' claims before the 2020 census questionnaires need to be printed, the Court begins with an account that is based exclusively on the initially filed Administrative Record and then turns to what the evidence beyond that portion of the Administrative Record reveals. 10

A. The Initial Administrative Record Submission

1. Defendants first filed what they characterized as the Administrative Record on June 8, 2018. See Docket No. 173. That submission was 1,320 pages and included the December 12, 2017 DOJ letter requesting addition of a citizenship question to the census questionnaire; various analyses of that request by the Census Bureau; and Secretary Ross's March 26, 2018 memorandum. These pages reveal that the Census Bureau repeatedly and consistently recommended against addition of a citizenship question to the census questionnaire based on its assessment that adding the question would reduce self-response rates, thereby increasing costs and harming the overall data and integrity of the census, and that DOJ's stated interest in having more granular citizenship data could be satisfied in a less costly, more effective, and less harmful manner. More importantly, for the purposes of this Court's review, the initial Administrative Record submission alone contains overwhelming evidence to that effect, and none that contradicts it.

1. The December 12, 2017 DOJ Letter

2. Secretary Ross asserted in his March 26, 2018 memorandum that his decision to add the citizenship question to the census questionnaire was prompted by a December 12, 2017 letter from DOJ. The letter came from Arthur E. Gary of the Justice Management Division; it was addressed to then-Acting Director Jarmin. See AR 663-65 (the "Gary Letter"); AR 1525-27 (same). In the letter, DOJ "formally *531 request[ed] that the Census Bureau reinstate on the 2020 Census questionnaire a question regarding citizenship, formerly included on the so-called 'long form' census." Gary Letter 1, at AR 663. "This data," the Letter stated, "is critical to the Department's enforcement of Section 2 of the Voting Rights Act and its important protections against racial discrimination in voting. To fully enforce those requirements, the Department needs a reliable calculation of the citizen voting-age population in localities where voting rights violations are alleged or suspected." Id.

3. More specifically, the Gary Letter noted that the Supreme Court had "held that Section 2 of the Voting Rights Act prohibits 'vote dilution' by state and local jurisdictions engaged in redistricting, which can occur when a racial group is improperly deprived of a single-member district in which it could form a majority." Gary Letter 1, at AR 663 (citing Thornburg v. Gingles , 478 U.S. 30 , 50, 106 S.Ct. 2752 , 92 L.Ed.2d 25 (1986) ). Further, "[m]ultiple federal courts of appeals have held that, where citizenship rates are at issue in a vote-dilution case, citizen voting-age population" - often referred to as "CVAP" - "is the proper metric for determining whether a racial group could constitute a majority in a single-member district." Id. (citing cases). "These cases," the Gary Letter reasoned, "make clear that, in order to assess and enforce compliance with Section 2's protection against discrimination in voting, the Department needs to be able to obtain citizen voting-age population data for census blocks, block groups, counties, towns, and other locations where potential Section 2 violations are alleged or suspected." Gary Letter 2, at AR 664.

4. The Gary Letter noted that DOJ had previously relied on citizenship data acquired through either the "long form" census questionnaire or the ACS. See id. The Gary Letter asserted, however, that citizenship data acquired through the ACS is not "ideal ... for such purposes" for four reasons. Gary Letter at 2-3, at AR 664-65. First, "[j]urisdictions conducting redistricting, and the Department in enforcing Section 2, already use the total population data from the census to determine compliance with the Constitution's one-person, one-vote requirement," such that "using the ACS citizenship estimates means relying on two different data sets, the scope and level of detail of which vary quite significantly." Id. at 2, at AR 664. Second, ACS surveys are conducted annually and produce data that are "aggregated into one-year, three-year, and five-year estimates," and thus "do not align in time with the decennial census data," whereas "[c]itizenship data from the decennial census ... would align in time with the ... data from the census that jurisdictions already use in redistricting." Id. at 3, at AR 665. Third, "ACS estimates are reported at a ninety percent confidence level" (that is, as a statistical sample with a margin of error that "increases as the sample size - and thus, the geographic area - decreases"), whereas "decennial census data is a full count of the population." Id. And finally, decennial "Census data is reported to the census block level, while the smallest unit reported in the ACS estimates is the census block group," requiring DOJ in some instances to "perform further estimates" of CVAP at the census block level. Id.

5. "For all of these reasons," the Gary Letter concluded, DOJ "believes that decennial census questionnaire data regarding citizenship, if available, would be more appropriate for use in redistricting and in Section 2 litigation than the ACS citizenship estimates." Id. On that basis, the Gary Letter "formally request[ed]" that the Census Bureau "reinstate" a citizenship question on the 2020 census. Id. "At *532 the same time," however, the Gary Letter "request[ed] that the [Census] Bureau also maintain the citizenship question on the ACS, since such question is necessary, inter alia , to yield information for the periodic determinations made by the Bureau under Section 203 of the Voting Rights Act." Id. 11

2. The Census Bureau's Preliminary Analyses and Recommendations

6. Following receipt of the Gary Letter, experts at the Census Bureau - acting under the supervision of Dr. John M. Abowd, the Bureau's Chief Scientist and Associate Director for Research and Methodology - sought to determine the effects of adding a citizenship question to the census questionnaire and whether there were alternative means to satisfy DOJ's interest in more granular citizenship data. The Census Bureau's initial analyses and recommendations were memorialized in memoranda dated December 22, 2017, see AR 11634-45 (the "December 22 Memo"); see also AR 5500-11 (copy of the same); January 3, 2018, see AR 5473-75 (the "January 3 Memo"); 12 and January 19, 2018, see AR 1277-85 (the "January 19 Memo"). 13

7. The December 22 Memo included two significant conclusions. First, the Memo opined that the Census Bureau could "potentially" obtain "a more accurate measure of citizenship" in a more "cost efficient" manner by using means other than a question on the census. December 22 Memo, at AR 11644. The Memo identified four existing surveys, including the ACS, through which the Census Bureau obtained "directly reported" citizenship data, as well as a slew of "administrative record[s]" (that is, records from other federal agencies or states) with similar information that the Census Bureau had or could acquire. Id. at AR 11634-35. The administrative records included Numident, a Social Security Administration ("SSA") dataset that represents "the most complete and reliable administrative record source of citizenship data currently available," and citizenship data maintained by the U.S. Customs and Immigration Service ("USCIS"). Id. at AR 11636, 11643-44.

8. Second, the December 22 Memo concluded that including a citizenship question in the 2020 census questionnaire was likely to depress self-response rates, particularly among noncitizen households; increase costs; and produce lower quality citizenship data. Id. at AR 11639-45. The Memo explained that including a citizenship question could lower the rate of voluntary compliance, which would require expanded field operations that "require additional unnecessary costs and burden to the Bureau." Id. at AR 11644-45.

9. The December 22 Memo recommended that the best way to "support[ ]

*533 redistricting in the manner requested by the Department of Justice is" not to add a citizenship question to the census, but "to make a citizenship variable available on the 2020 Census Edited File (CEF)." Id. at AR 11644. The CEF is "the internal, confidential data file" that underlies the census data set used by DOJ in redistricting, known as the "PL94 tabulations." Id . If citizenship were available on the CEF, the Memo explained, the PL94 tabulations could be restructured to include direct estimates of CVAP by race and ethnicity at the census block level. See id . "These tabulations would have essentially the same accuracy as current PL94 and Summary File 1 (SF1) data." Id .

10. On the afternoon of December 22, 2017, Dr. Jarmin emailed Arthur Gary at DOJ, stating that "the best way to provide PL94 block-level data with citizen voting population by race and ethnicity would be through utilizing a linked file of administrative and survey data the Census Bureau already possesses. This would result in higher quality data produced at lower cost." AR 3289.

11. The January 3 Memo from Dr. Abowd to Dr. Jarmin expanded on the December 22 Memo's analyses and recommendations. See January 3 Memo, at AR 5473-75. The January 3 Memo described "three alternatives for meeting the DoJ request." Id. at AR 5473.

12. Under Alternative A, the Census Bureau would "[m]aintain the status quo for data collection, preparation, and publication," but would prepare a "special product" for DOJ containing the Census Bureau's "best estimate of block-level citizen voting age population by race and ethnicity," similar to what the Census Bureau prepares for use in connection with Section 203 of the VRA. Id. Alternative A would cost an extra $200,000 and would produce similar-quality data to that produced for Section 203 purposes. Id. at AR 5474.

13. The January 3 Memo concluded that Alternative B - adding a citizenship question to the 2020 census questionnaire - would "most likely deliver higher quality block-level citizen voting age population by race and ethnicity data than Alternative A." Id. But it would cost an estimated additional $27.5 million and, based on an estimated minimum 5.1% decline in self-response among noncitizen households, would lead to an estimated minimum "154,000 fewer correct enumerations." Id. 14

14. Finally, the January 3 Memo addressed Alternative C, the option the December 22 Memo had recommended (although not by that name). See December 22 Memo, at AR 11644. Under Alternative C, the Census Bureau would not add a citizenship question to the questionnaire, but would "[a]dd the capability to link an accurate, edited citizenship variable from administrative records to the final 2020 Census microdata files." January 3 Memo, at AR 5473. In that way, Alternative C would provide block-level tabulations of CVAP by race and ethnicity. Id. It would deliver "higher quality data than Alternative B for DoJ's stated uses" because the "primary data sources for the administrative record citizen variable require proof of *534 citizenship" and, thus, are "very accurate." Id. at AR 5475.

15. Based on this analysis, the January 3 Memo opined that "Alternative A is not very costly and does not harm the quality of the census count"; that "Alternative B better addresses DoJ's stated uses," but "is very costly and does harm the quality of the census count"; and that "Alternative C even better meets DoJ's stated uses, is comparatively far less costly than Alternative B, and does not harm the quality of the census count." Id. "For these reasons," the Memo concluded, "we recommend Alternative C for meeting the Department of Justice data request." Id.

16. The January 19 Memo contained a "Technical Review" of the facts and the Census Bureau's analysis of Alternatives A, B, and C. January 19 Memo, at AR 1277-85.

17. The January 19 Memo described Alternative A in more detail, explaining that existing "redistricting and CVAP data are used by the Department of Justice to enforce the Voting Rights Act" and "by state redistricting offices to draw congressional and legislative districts that conform to constitutional equal-population and Voting Rights Act nondiscrimination requirements. Because the block-group-level CVAP tables have associated margins of error, their use in combination with the much more precise block-level census counts in the redistricting data requires sophisticated modeling. For these purposes, most analysts and the DoJ use statistical modeling methods to produce the block-level eligible voter data that become one of the inputs to their processes." Id. at AR 1279. Without modifying any of its other procedures, the January 19 Memo explained, the Census Bureau could deploy "a small team of Census Bureau experts similar in size and capabilities to the teams used to provide the Voting Rights Act Section 203 language determinations" to assist DOJ in making those calculations. Id. The January 19 Memo updated the cost of Alternative A to $350,000, and estimated that it would "have no impact on the quality of the 2020 Census." Id.

18. The January 19 Memo also presented an updated analysis of Alternative B. Id. at AR 1279-82. It explained bluntly that adding a citizenship question to the census "is very costly, harms the quality of the census count, and would use substantially less accurate citizenship status data than are available from administrative sources." Id. at AR 1277. The Memo summarized "[t]hree distinct analyses" showing that the addition of a citizenship question would lead to "an adverse impact on self-response" to the census questionnaire and, consequently, "on the accuracy and quality of the 2020 Census." Id. at AR 1280. First, the Memo described evidence that a citizenship question would increase "item nonresponse rates" - that is, omitted answers to the question - and would do so disproportionately among Hispanics. Id. Second, the Memo described evidence that a citizenship question would decrease "self-response rates" - that is, the rate at which households voluntarily respond to the questionnaire - and would do so disproportionately among noncitizen households. Id. at AR 1280-81. Third, the Memo described evidence that a citizenship question would increase "breakoff rates" - that is, the rate at which households stop answering the questionnaire when they come to a particular question - and would do so disproportionately among Hispanics. Id. at AR 1281.

19. The updated analysis of Alternative B included more detail on its high cost. Id. at AR 1281-82. The reduction in self-response rates caused by the citizenship question, the January 19 Memo explained, would lead to more people being counted *535 through the Census Bureau's NRFU (that is, Non-Response Follow-Up) procedures. Id. at AR 1280. The greater use of NRFU procedures produces, in turn, lower-quality data because the NRFU procedures yield more "erroneous enumerations" and "whole-person imputations" than self-responses do. Id. at AR 1281. 15 Using the "conservative" estimate that addition of a citizenship question would cause a 5.1% decline in self-response among noncitizen households, the January 19 Memo explained that adding a citizenship question would produce 432,000 fewer correct enumerations and would cost at least $27.5 million more. Id. at AR 1282. The Memo stressed that its estimates of the decline in data quality and increase in cost associated with Alternative C were "conservative" and that the true numbers could be "much greater." Id.

20. The January 19 Memo also discussed Alternative C in more depth. It summarized the Census Bureau's evidence that Alternative C - linking a citizenship variable from administrative records to the census microdata files - "is comparatively far less costly than Alternative B, does not increase response burden, and does not harm the quality of the census count." AR 1277. The Memo noted that the Census Bureau had been testing its ability to link administrative data to census data in that manner since the 1990 census. Id. For the 2020 census, the Census Bureau had already begun "regularly ingesting and loading administrative data from the Social Security Administration, Internal Revenue Service and other federal and state sources into the 2020 Census data systems." Id.

21. Significantly, in its analysis of Alternative C, the January 19 Memo again noted that Alternative C was likely to yield more accurate citizenship data than Alternative B would yield. See id. at AR 1277. First, the Memo explained that comparisons of ACS data and federal administrative records revealed that somewhere between 23.8% and 30% of people whom administrative records indicate are noncitizens report that they are citizens on the ACS. See id. at AR 1283. This suggests that self-reported citizenship status is of dubious accuracy. Second, the sampling techniques used for ACS data may not accurately predict the citizenship status of those who do not respond. The sampling techniques assume that those who do not respond to any given question are statistically similar to those who do - an assumption contradicted by the data. See id. at AR 1283-84. As a result, the ACS imputes citizenship responses based on data that is not predictive of the missing set. See id .

22. The January 19 Memo acknowledged that the Numident database might be missing citizenship data for older citizens who obtained SSNs before the SSA required proof of citizenship, for naturalized citizens who have not communicated the fact of their naturalization to the SSA, and for noncitizens who do not have a SSN or other taxpayer identification number. Id. at AR 1285. "All three of these shortcomings," however, "are addressed by adding data from [USCIS]." Id. Accordingly, the Memo concluded, "the administrative records citizenship data would most likely have both more accurate citizen status and *536 fewer missing individuals than would be the case for any survey-based collection method." Id. (emphasis added). The January 19 Memo estimated that the cost of using administrative records to obtain block-level CVAP data would be between $500,000 and $2 million. Id.

3. Secretary Ross's February 12, 2018 Meeting with the Census Bureau

23. On February 12, 2018, Secretary Ross met with Dr. Abowd and others from the Census Bureau to discuss the various alternatives. See AR 9450. The Administrative Record contains only one document concerning what was discussed at the meeting: an email dated February 13, 2018, from an attendee at the meeting, "identif[ying]" action items and items for future consideration that were discussed at the meeting. AR 9450. The Administrative Record otherwise contains no record - contemporaneous or otherwise - of what was discussed during the meeting. Nor is there is any indication in the Administrative Record of any other meeting between Secretary Ross and Census Bureau officials.

24. The Administrative Record does reveal that sometime before March 1, 2018, Secretary Ross asked the Census Bureau to analyze a fourth option - "Alternative D" - that would combine Alternative B (adding a citizenship question to the 2020 census) and Alternative C (relying on administrative records to generate block-level CVAP data). See AR 1308-12 ("March 1 Memo"), at AR 1309.

4. The Census Bureau's Analysis of Alternative D

25. Dr. Abowd transmitted a detailed analysis of Alternative D to Secretary Ross in a memorandum dated March 1, 2018. The March 1 Memo explained at length that Alternative D (combining the addition of a citizenship question to the census with the use of administrative records) would entail "all the negative cost and quality implications of Alternative B" (that is, simply adding the citizenship question) and would still produce "poorer quality citizenship data than Alternative C" (that is, linking data from administrative records to data from the census). See March 1 Memo at AR 1312.

26. The March 1 Memo described, first, how the various sources of administrative records interact to supplement and correct for one another's "gaps." See id. at AR 1309-10. It then identified seven "[r]emaining citizenship data gaps" in the administrative records: (1) "U.S. citizens from birth with no SSN or U.S. passport"; (2) U.S. citizens from birth born outside the United States without a passport who applied for an SSN before they were required to provide proof of their citizenship to do so; (3) U.S. citizens naturalized before 2001 who did not inform the SSA of their naturalization; (4) U.S. citizens who were automatically naturalized if they were under eighteen when their parents were naturalized after 1999, but did not inform USCIS or receive a U.S. passport; (5) lawful permanent residents ("LPRs") who obtained that status before 2001, but lack an SSN or received an SSN before the SSA asked about citizenship; (6) noncitizen, non-LPR residents who do not have an SSN (or other taxpayer identification number) and who have not applied for a visa extension; and (7) persons for whom citizenship does appear in administrative records, but for whom it is not possible to link those records with decennial census data. Id. at AR 1310-11.

27. The March 1 Memo explained that, perhaps counterintuitively, "survey data" (that is, data from a citizenship question) would not "help fill the ... gaps" in the administrative records. Id. at AR 1311.

*537 This is because a "significant, but unknown, fraction" of people whose citizenship status is unknown fall into "Category Six": noncitizen, non-LPR residents without SSNs or taxpayer IDs. These people have a "strong incentive to provide an incorrect answer" or no answer to a citizenship question. Id . And, because "there is no feasible method of independently verifying their non-citizen status," identifying false reports of citizenship on the census would be "an inexact science." Id . Instead of doing that inexact science, the "survey response[s] of 'citizen' would be accepted as valid" for large numbers of people who are not, in fact, citizens. See id. Making matters worse, if a household does not self-respond in order to protect one household member in Category Six, this problem would apply to the entire household. See id. at AR 1311-12.

28. Second, the March 1 Memo explained that, like Alternative B, Alternative D would lower self-response rates and "push[ ] more households into ... NRFU." Id. at AR 1311. That would lower the quality of the data in two ways. First, as explained above, NRFU procedures yield more "erroneous enumerations" and "whole-person imputations" than self-responses do. January 19 Memo, at AR 1281. Second, NRFU procedures produce more responses that cannot be linked to administrative records (due to lower-quality personal identifying information) and linking to administrative records helps improve the accuracy of the survey data. See March 1 Memo, at AR 1311. For responses gathered through NRFU procedures, only 81.6% can be linked to people recorded in administrative records, compared to 96.7% for self-responses in the 2010 census. Id .

29. Additionally, the March 1 Memo concluded that Alternative D would likely increase the number of proxy responses required, another less accurate type of enumeration. See id. This is because those who do not self-respond because of the presence of the citizenship question are especially likely not to respond in NRFU as well. See id. Proxy responses also lead to even lower rates of linkage to administrative data on citizenship: only 33.8%. See id.

30. The March 1 Memo concluded that Alternative D was plainly inferior to Alternative C: It "would result in poorer quality citizenship data than Alternative C," and it "would still have all the negative cost and quality implications of Alternative B" summarized in the January 19 Memo. Id. at AR 1312. The March 1 Memo remarked that "[u]sing the 2020 Census data only to fill in gaps for persons without administrative data on citizenship would raise questions about why 100 percent of respondents are being burdened by a citizenship question to obtain information for the two percent of respondents where it is missing." Id .

31. In another memorandum, undated but obviously written in this same March 2018 timeframe, Dr. Abowd presented Secretary Ross with numerical estimates illustrating the differences between Alternative C and Alternative D. See AR 1304-06 ("Key Differences Memo"). The Key Differences Memo illustrated that, under Alternative C, the Census Bureau expected to link 295 million people - 89.4% of the population - to high-quality citizenship data. Id. at AR 1306. It would be unable to identify the citizenship status of 35 million people through linking, leaving those to be modeled. Id. The Key Differences Memo illustrated these numbers for Alternative C in the following flowchart:

*538 AR 1306.

32. The Key Difference Memo explained that, under Alternative D, the Census Bureau expected to receive responses to the citizenship question from 294.6 million people. See id. at AR 1307. Of those responses, 263 million could be confirmed through linkages to administrative records; 9.5 million responses could be linked to, but would be inconsistent with , citizenship data in administrative records; and 22.2 million responses would not be linked to administrative records at all. See id. On top of that, the Census Bureau expected that approximately 35.4 million people's citizenship would not be measured by the 2020 census, either due to non-response or NRFU failure. See id. Approximately 21.5 million non-respondents could be linked to administrative records containing citizenship data, while approximately 13.8 million would have to have their citizenship status modeled. See id. Once again, the Key Differences Memo illustrated the numbers for Alternative D with a flowchart:

AR 1307.

33. These numbers reveal that Alternative D would produce more people who *539 could not be linked to administrative records: 36 million people in Alternative D as compared to 35 million in Alternative C. See id. at AR 1306-07. Of the unlinked people in Alternative D, 13.8 million would have their citizenship modeled (or "imputed"), see id. at AR 1307, while 22.2 million would have their citizenship status evaluated through their census responses, despite the known likelihood that a high rate of noncitizens (just under 500,000, according to the Census Bureau's estimate) would be incorrectly enumerated as citizens through the survey process. See id. at AR 1305-07; March 1 Memo, at AR 1311. That would leave 9.5 million people whose census responses would conflict with administrative records. See Key Differences Memo, at AR 1307. For these people, the "[h]istoric Census Bureau practice is to use self-reported data in these situations," even though "the Census Bureau now knows from linking ACS responses on citizenship to administrative data that nearly one third of noncitizens in the administrative data respond to the questionnaire indicating they are citizens, indicating that this practice should be revisited in the case of measuring citizenship." Id. at AR 1305.

34. Put differently, Alternative D would provide no improvement to the citizenship data available under Alternative C for 90.4% of the population. For 6.7% of the population, Alternative D would produce lower quality data than Alternative C because the Census Bureau would have to use survey responses that are, generally speaking, less accurate than the imputation methods the Census Bureau would deploy under Alternative C. See id. And finally, for 2.9% of the population, Alternative D would create a problem that would not exist under Alternative C: conflicts between the survey data and the administrative data, with no reliable method for discerning accurate data amidst the conflict. See id.

35. The Key Differences Memo summarized the implications of these comparative predictions. He acknowledged that "all possible measurement methods will have errors" and that the Census Bureau "cannot quantify the relative magnitude of the errors across the alternatives at this time." Id. at AR 1305. Nevertheless, Alternative C would involve some risk of error in the administrative record data, but that would be "relatively limited" thanks to the procedures used by the SSA, USCIS, and the State Department. Id. Alternative C would also be subject to some prediction error in the 35 million cases that would have to be modeled, which would be a similar, but lesser issue in Alternative D. Id. By contrast, Alternative D would be subject to an error that is "only an issue in Alternative D": the "response error." Id. Dr. Abowd noted that while "[s]tatisticians often hope" that such response errors "are random and cancel out," the Census Bureau "know[s] from prior research" that they are "systematically biased for a subset of noncitizens." Id.

5. Communications with Stakeholders

36. While the Census Bureau was analyzing the citizenship question, the Commerce Department was communicating with stakeholders about the question. At the February 12, 2018 meeting with Dr. Abowd and other representatives of the Census Bureau, Secretary Ross requested a list of stakeholders with whom to discuss the addition of the citizenship question to the census. See AR 9450. During the outreach to stakeholders after the meeting, however, Commerce Department officials struggled to find anyone willing to express support for adding the question. See AR 3274-75 ("Email re AEI"); AR 4853-55 (same).

*540 37. For example, on February 13, 2018 - only one day after the meeting with Secretary Ross - Dr. Jarmin emailed someone at the American Enterprise Institute ("AEI"): "We are trying to set up some meetings for Secretary Ross to discuss the proposed citizenship question on the 2020 Census with interested stakeholders. Most stakeholders will speak against the proposal. We're looking to find someone thoughtful who can speak to the pros of adding such a question ." Email re AEI, at AR 3275 (emphasis added). Later the same day, Michael Strain of the AEI responded: "None of my colleagues at AEI would speak favorably about the proposal. Is it important that the person actually be in favor of the proposal?" Id .

38. Dr. Jarmin responded the same day, noting: "We are trying to find someone who can give a professional expression of support for the proposal in contrast to the many folks we can find to give professional statements against the proposal. Interesting, but perhaps not so surprising, that no one at AEI is willing to do that." AR 8325. Less than two minutes later, Dr. Jarmin forwarded his email exchange with the AEI to then-Under Secretary for Economic Affairs, Karen Dunn Kelley, noting: "Appears no one at AEI willing to speak in favor of putting question on the 2020." Id. at AR 3274. 16 The Administrative Record reflects that only two organizations supported the citizenship question: the Center for Immigration Studies and the Heritage Foundation. See AR 1206; AR 8325.

39. By contrast, between December 12, 2017 (the date of the Gary Letter) and March 26, 2018 (the date of Secretary Ross's decision), the Commerce Department received many communications opposing addition of a citizenship question to the census, from former directors of the Census Bureau, business groups, civil rights groups, social science groups, members of Congress, and state and local officials. See generally Plaintiffs' Joint Proposed Post-Trial Findings of Fact, Docket No. 545 ("Pls.' Proposed Findings"), ¶¶ 674-724.

40. For instance, on January 5, 2018, the American Sociological Association ("ASA") wrote to Secretary Ross to urge him to reject DOJ's proposal to add a citizenship question. See AR 787. The ASA letter predicted that if a citizenship question were added, "the integrity of the 2020 Census data will be fundamentally compromised." Id. "Including a citizenship question," the ASA continued, "is likely to keep some people from responding to the questionnaire and others from responding truthfully, thereby undermining the accuracy of the data. In addition ..., [w]ith little time left before the 2020 launch, a new question could not be subject to standard rigorous testing, which would further undermine the quality of the data." Id.

41. Similarly, on January 9, 2018, members of the Census Scientific Advisory Committee ("CSAC") wrote to then-Attorney General Jeff Sessions, Secretary Ross, and others, calling inclusion of a citizenship question "a serious mistake which would result in a substantial lowering of the response rate." AR 794. The CSAC members noted that "[a]dding a citizenship question to the main Census questionnaire is almost certain to jeopardize the cooperation of at least some community partners and lead to a lower response rate, hurting the reputation of the Census Bureau." Id. at AR 794-95.

*541 42. On January 10, 2018, the Leadership Conference on Civil and Human Rights (the "Leadership Conference") wrote to "urge" Secretary Ross "to reject the Department of Justice's untimely and unnecessary request for a new citizenship question on the 2020 Census, which would threaten a fair and accurate decennial census." AR 798. Adding a citizenship question "would destroy any chance for an accurate count, discard years of careful research, and increase costs significantly." Id. Moreover, the Leadership Conference continued, adding a citizenship question was "unnecessary." Id. at AR 799. "The Justice Department has never needed to add this new question to the decennial census to enforce the Voting Rights Act before, so there is no reason it would need to do so now." Id .

43. On January 26, 2018, Secretary Ross received a letter from six former Directors of the Census Bureau - John Thompson, Robert Groves, Steven Murdock, Kenneth Prewitt, Martha Farnsworth Riche, and Vincent Barabba - expressing concern about the possible addition of a citizenship question. See AR 8555-56. The former Directors said they were "troubled to learn that the Department of Justice has recently asked the Bureau to add a new question on citizenship to the 2020 census." Id. at AR 8555. They were "deeply concerned about the consequence of this possible action and hope[d] that [their] objective observations provide a useful perspective before a final decision is made on this issue." Id.

44. In particular, the former Directors expressed concern that the question had not been appropriately tested. They noted that "[t]here 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." Id. at AR 8556. That is why "[t]here is a well-proven multi-year process to suggest and test new questions." Id. at AR 8555. They wrote that they "strongly believe[d] that adding an untested question on citizenship status at this late point in the decennial planning process would put the accuracy of the enumeration and success of the census in all communities at grave risk." Id.

45. On March 19, 2018, Secretary Ross received a letter from the Latino Community Foundation, which "strongly urge[d Secretary Ross] to reject the inclusion of a citizenship question to the 2020 Census." AR 1222. The Foundation explained: "Including a citizenship question to the census will add to an extensive list of concerns that can and will suppress Latino participation. Increased immigration enforcement, anti-immigrant rhetoric in our political discourse, and privacy concerns have already meshed together to create a climate of fear and aversion of the federal government." Id.

46. On March 22, 2018, Secretary Ross received a letter from Ready Nation, a council of "American business leaders," who wrote "to express [its] deep concern about the Department of Justice's request that the Census Bureau include an untested question about citizenship in the 2020 Census questionnaire." AR 3608. The business leaders were concerned because "[t]he decennial Census provides critical data that informs decisionmaking in both the private and public sectors"; they expressed the worry that "[a]dding a new question this late in the decennial Census process could reduce the accuracy" of the census. Id.

47. On March 23, 2018, Secretary Ross received a letter from members of the scientific community, including the Acting President of the Federation of American *542 Scientists, the Director of 2020 Census Counts, and the President of the American Political Science Association. See AR 1269-71. The letter stated that the DOJ request "is ill-conceived for a number of reasons. We have more accurate methods for measuring and studying non-citizenship, for example through anonymous surveys. Imposing a citizenship question would lead to a lower participation rate and substantial undercount of certain geographic regions and demographic populations, undermining the scientific integrity of the entire project." Id. at AR 1269.

48. That same day, Secretary Ross spoke with Christine Pierce, Senior Vice President of Data Science for Nielsen, a private survey company. See AR 1276. According to the Commerce Department notes of the conversation, Pierce expressed concern that the addition of the citizenship question "could lead to a lower response rate," which "is very important." Id. Pierce said that "including a question on citizenship could make people less likely to respond, but that there is no data to predict how much lower the response rate might be." Id. Pierce explained that at Nielsen, when the company was considering asking a sensitive question that was expected to depress response rates, the company "conducts a cost-benefit analysis to determine whether it is worth asking the question." Id. She "noted the importance of testing questions." Id. With regard to the federal government's questions, Pierce described that Nielsen once added the ACS questions about birthplace and date of arrival in the United States to one of its surveys. Id. She explained that although the company was "concerned about response rates declining, ... Nielsen did not observe lower response rates to the survey." Id. But, Pierce noted, Nielsen would have offered a cash reward to incentivize participation in that survey. See id.

6. Secretary Ross's March 26, 2018 Memorandum

49. As noted, on March 26, 2018, Secretary Ross issued his memorandum announcing the decision to adopt Alternative D (which he referred as "Option D"), which included adding the citizenship question to the 2020 census. See Ross Memo 1, 4, at AR 1313, 1316. Secretary Ross wrote that "[f]ollowing receipt" of the Gary Letter on December 12, 2017, he had "set out to take a hard look at the request and ensure that [he] considered all facts and data relevant to the question so that [he] could make an informed decision on how to respond." Id. at AR 1313. "To that end," he wrote, the Department of Commerce "immediately initiated a comprehensive review process led by the Census Bureau." Id.

50. Secretary Ross said that his decision "prioritized the goal of obtaining complete and accurate data ." Id. (emphasis in original). He wrote that "Congress has delegated to me the authority to determine which questions should be asked on the decennial census, and I may exercise my discretion to reinstate the citizenship question on the 2020 decennial census, especially based on DOJ's request for improved CVAP data to enforce the VRA." Id. at AR 1314. Secretary Ross did not cite any authority for the proposition that the DOJ's request expanded his discretion to determine the content of the decennial census questionnaire. See id . Nor did Secretary Ross address whether he was required to comply with DOJ's request in the first place. See id.

51. Secretary Ross noted that "collection of citizenship data by the Census [Bureau] has been a long-standing practice." Id . He gave as examples of this "long-standing practice" the fact that a citizenship question had appeared on decennial *543 censuses up until 1950; that the 2000 "long form" survey "distributed to one in six people in the U.S.[ ] included a question on citizenship"; and that the ACS (which replaced the "long form sample") had "included a citizenship question since 2005." Id. (internal quotation marks omitted). From this, he concluded that "the citizenship question has been well tested." Id. Secretary Ross did not, however, detail any such testing or address how the testing would apply to a question in 2020 on a census questionnaire distributed to the entire population.

52. Secretary Ross explained that DOJ had stated "that the current data collected under the ACS are insufficient in scope, detail, and certainty to meet its purpose under the VRA." Id. For that reason, DOJ sought "CVAP data for census blocks, block groups, counties, towns, and other locations where potential Section 2 violations are alleged or suspected." Id. Secretary Ross explained that the "Census Bureau ha[d] advised [him] that the census-block-level citizenship data requested by DOJ are not available using the annual ACS, which ... does ask a citizenship question and is the present method used to provide DOJ and the courts with data used to enforce Section 2 of the VRA." Id.

53. Secretary Ross then discussed the four alternatives he had considered in reaching his decision: Alternatives A, B, C, and D (albeit referring to them as "Options" A, B, C, and D). Id. at AR 1314-17.

54. The Ross Memo described Alternative A as rejecting the DOJ request entirely and relying entirely on existing citizenship data. See id. at AR 1314; January 19 Memo, at AR 1279. Secretary Ross rejected this option because, "[u]nder Option A, the 2020 decennial census would not include the question on citizenship that DOJ requested and therefore would not provide DOJ with improved CVAP data." Ross Memo 2, at AR 1314. The existing block-level CVAP data obtained through the ACS, he noted, "has associated margins of error because the ACS is extrapolated based on sample surveys." Id. Thus, Secretary Ross reasoned, "[p]roviding more precise block-level data would require sophisticated statistical modeling," requiring "the Census Bureau ... to deploy a team of experts to develop model-based methods that attempt to better facilitate DOJ's request for more specific data." Id. Secretary Ross's problem with this option was that "the Census Bureau did not assert and could not confirm that such data modeling is possible for census-block-level data with a sufficient degree of accuracy," and, "[r]egardless, DOJ's request is based at least in part on the fact that existing ACS citizenship data-sets lack specificity and completeness," such that "[a]ny future modeling from these incomplete data would only compound that problem." Id. at AR 1314-15. In summary, Secretary Ross wrote that he rejected Alternative A because it "would provide no improved citizenship count, as the existing ACS sampling would still fail to obtain actual , complete number counts, especially for certain lower population areas or voting districts, and there is no guarantee that data could be improved using small-area modeling methods." Id. at AR 1315.

55. Secretary Ross then turned to Alternative B, "which would add a citizenship question to the decennial census." Id. Secretary Ross acknowledged that "[t]he Census Bureau and many stakeholders expressed concern that Option B ... would negatively impact the response rate for non-citizens," which "could reduce the accuracy of the decennial census and increase costs for non-response followup ('NRFU') operations." Id. (emphasis omitted). "However," he continued, "neither the Census Bureau nor the concerned *544 stakeholders could document that the response rate would in fact decline materially." Id. Secretary Ross cited three sources for this conclusion. First, he stated that he had "discuss[ed] the question with the national survey agency Nielsen" and that Nielsen had "stated that it had added questions from the ACS on sensitive topics such as place of birth and immigration status to certain short survey forms without any appreciable decrease in response rates." Id. Second, "the former director of the Census Bureau during the last decennial census" said "that, while he wished there were data to answer the [citizenship] question, none existed to his knowledge." Id. Third, "Nielsen's Senior Vice President for Data Science and the former Deputy Director and Chief Operating Officer of the Census Bureau under President George W. Bush also confirmed that, to the best of their knowledge, no empirical data existed on the impact of a citizenship question on responses." Id.

56. Secretary Ross noted that "the Census Bureau attempted to assess the impact that reinstatement of a citizenship question on the decennial census would have on response rates by drawing comparisons to ACS responses," as the Census Bureau had done in the January 19 Memo. Id. ; see January 19 Memo, at AR 1279-80. Secretary Ross wrote that "such comparative analysis was challenging, as response rates generally vary between decennial censuses and other census sample surveys." Ross Memo 3, at AR 1315. For example, the Census Bureau attributed higher response rates on the decennial census to "greater outreach and follow-up associated with the ... decennial census" and to the comparatively short length of the decennial census questionnaire compared to the ACS. Id.

57. Secretary Ross also assessed the empirical data showing that Hispanics were disproportionately likely not to self-respond to citizenship questions. See id. at AR 1315-16. With regard to the ACS, he pointed out that many questions other than the citizenship question also have significant nonresponse rates. Id. at AR 1315. But in citing those other questions, he did not acknowledge any comparative response rates by race or ethnicity, citing only average nonresponse rates. See id. With regard to the decennial census, Secretary Ross acknowledged the Census Bureau's data showing that noncitizen households (compared to citizen households) were disproportionately less likely to self-respond to the 2000 long-form census questionnaire, which included a citizenship question, than to the short-form questionnaire, which did not. See id. at AR 1316. But, Secretary Ross wrote, the Census Bureau "was not able to isolate what percentage of decline was caused by the inclusion of a citizenship question rather than some other aspect of the long form survey." Id. He concluded that "while there is widespread belief among many parties that adding a citizenship question could reduce response rates, the Census Bureau's analysis did not provide definitive, empirical support for that belief." Id.

58. Secretary Ross next evaluated Alternative C, which would use administrative records instead of adding a citizenship question to the decennial census. Id. He acknowledged that the Census Bureau had used administrative record data "since the early 20th century" and that administrative records often yield "more accurate" citizenship data than self-responses do. Id. "However," he wrote, "the Census Bureau is still evolving its use of administrative records, and the Bureau does not yet have a complete administrative records data set for the entire population." Id. He also acknowledged that the Census Bureau was able to match 88.6% of the population "with what the Bureau considers credible *545 administrative record data" in the 2010 census. Id. But, he wrote, "[w]hile impressive, this means that more than 10 percent of the American population - some 25 million voting age people - would need to have their citizenship imputed by the Census Bureau. Given the scale of this number, it was imperative that another option be developed to provide a greater level of accuracy than either self-response alone or use of administrative records would presently provide." Id.

59. Secretary Ross explained that he had "asked the Census Bureau to develop" an Alternative D, which "would combine" Alternatives B and C. See id. His "judgment," the Memo announced, was that Alternative D "will provide DOJ with the most complete and accurate CVAP data in response to its request." Id. at AR 1317. Secretary Ross explained:

Asking the citizenship question of 100 percent of the population gives each respondent the opportunity to provide an answer. This may eliminate the need for the Census Bureau to have to impute an answer for millions of people. For the approximately 90 percent of the population who are citizens, this question is no additional imposition. And for the approximately 70 percent of non-citizens who already answer this question accurately on the ACS, the question is no additional imposition since census responses by law may only be used anonymously and for statistical purposes. Finally, placing the question on the decennial census and directing the Census Bureau to determine the best means to compare the decennial census responses with administrative records will permit the Census Bureau to determine the inaccurate response rate for citizens and non-citizens alike using the entire population. This will enable the Census Bureau to establish, to the best of its ability, the accurate ratio of citizen to non-citizen responses to impute for that small percentage of cases where it is necessary to do so.

Id.

60. Secretary Ross addressed the possible effect on self-response rates of the citizenship question as set forth in Alternative D. See id. at AR 1319. He opined that "even if" addition of a citizenship question had "some impact on responses, the value of more complete and accurate data derived from surveying the entire population outweighs such concerns. Completing and returning decennial census questionnaires is required by Federal law, those responses are protected by law, and inclusion of a citizenship question on the 2020 decennial census will provide more complete information for those who respond." Id. Secretary Ross then stated, without reference to any evidence or to the Census Bureau's conclusion that adding a citizen question would produce "substantially less accurate citizenship status data than are available from administrative sources," January 19 Memo, at AR 1277, that "[t]he citizenship data provided to DOJ will be more accurate with the question than without it," Ross Memo 7, at AR 1319. He wrote that this "accura[cy]" was "of greater importance than any adverse effect that may result from people violating their legal duty to respond." Id.

61. Secretary Ross concluded: "I have determined that reinstatement of a citizenship question on the 2020 decennial census is necessary to provide complete and accurate data in response to the DOJ request." Id. at AR 1320. He also announced his decision to place the question "last on the decennial census form" to "minimize any impact on decennial census response rates." Id.

7. Secretary Ross's Testimony Before Congress

62. As the foregoing makes clear, the chronology of events in the initial Administrative *546 Record produced on June 8, 2018, largely conformed to the chronology in the Ross Memo. According to the Ross Memo, the decision to add a citizenship question to the 2020 census began with the December 12, 2017 request from Arthur Gary, a career official at DOJ. See Ross Memo 1, at AR 1313. "Following receipt" of the Gary Letter, Secretary Ross claimed, he had "set out to take a hard look at the request." Id . The Memo did not mention any earlier discussions and efforts to propose addition of a citizenship question or any role in those discussions by White House personnel, political appointees at DOJ, or political appointees at the Commerce Department. See id. at AR 1313-14.

63. In sworn testimony before Congress around the same time, Secretary Ross repeated that the decision to add a citizenship question began with the Gary Letter and denied White House involvement in the decision and discussions leading to the decision.

64. Before the House Appropriations Committee on March 20, 2018, for example, Representative José Serrano asked Secretary Ross whether "the President or anyone else in the White House [had] directed [him] to add this or a similar question to the 2020 census." Hearings Before Subcomm. on Commerce, Justice, Science, and Related Agencies of the H. Comm. on Appropriations , 115th Cong. 15 (2018) (admitted as an audio file at PX-491). Secretary Ross responded that the Department of Commerce was "responding solely to Department of Justice's request." Id. (emphasis added). Later in the same hearing, Representative Grace Meng asked Secretary Ross whether "the President or anyone in the White House discussed with you or anyone on your team about adding this citizenship question." Id. at 38 (admitted as an audio file at PX-493). Secretary Ross answered: "I am not aware of any such." Id .

65. Secretary Ross testified similarly before the House Ways and Means Committee on March 22, 2018. See Hearing with Commerce Secretary Ross: Hearing Before the H. Comm. on Ways and Means , 115th Cong. 1 (2018) (admitted as PX-480). Most relevant here, Representative Judy Chu asked if Secretary Ross could tell her "whether the Department of Commerce plans to include the citizenship question in the 2020 Census." Id. at 51. Secretary Ross responded that the "Department of Justice, as you know, initiated the request for inclusion of the citizenship question" to the decennial census. Id. (emphasis added).

66. Finally, before the Senate Subcommittee on Commerce, Justice, Science, and Related Agencies on May 10, 2018, Secretary Ross testified again that DOJ had initiated the request. See Hearing on the F.Y. 2019 Funding Request for the Commerce Dep't Before the S. Appropriations Subcomm. on Commerce, Justice, and Science and Related Agencies , 2018 WL 2179074 (May 10, 2018). Senator Patrick Leahy questioned whether a citizenship question was actually "necessary to enforce the Voting Rights Act." Id. He asked Secretary Ross: "[W]hy this sudden interest in that when the department that's supposed to enforce violations doesn't see any problems?" Id . Secretary Ross responded: "Well, the Justice Department is the one who made the request of us." Id .

67. In a brief filed with the Supreme Court in December 2018, Defendants argue that "[v]iewed in context," these statements should not be understood to imply that the Department of Commerce had not previously considered the issue or spoken with others in the Administration about it.

*547 Brief for the Petitioners at 27, Dep't of Commerce v. U.S. Dist. Ct. for S.D.N.Y. (Dec. 17, 2018) (No. 18-557), 2018 WL 6650094 , at *27. Instead, they argue that Secretary Ross's response to Representative Serrano should be understood to mean that "no outside political parties or campaigns had made a request to which the Department of Commerce was responding"; that his statement to Representative Meng was in reference to an email purportedly sent by the Trump Reelection Campaign and thus only about "whether any political actors in the White House had made a formal request" about the citizenship question; that his response to Representative Chu pertained only to the "formal process" that began with the Gary Letter; and that his assertion to Senator Leahy was meant only to "rebut[ ]" Senator Leahy's suggestion that DOJ did not see a problem with current VRA enforcement. Id. at 27-31 (emphasis omitted).

68. Defendants' post hoc interpretations of Secretary Ross's sworn statements to Congress are unconvincing. Viewed individually, even in context, none of the statements are limited in the ways Defendants suggest. And viewed together, the statements afford only one conclusion: Secretary Ross intended to convey the impression that the Gary Letter - and the Gary Letter alone - prompted consideration of whether to add a citizenship question to the census; that neither he nor anyone else at the Commerce Department prompted DOJ's request; and that he had not discussed the matter with White House officials before 2018. Moreover, nothing in Secretary Ross's March 26, 2018 Memorandum, in the rest of the Administrative Record, or in the relevant statutes and regulations supports the purported distinction between an "informal" and "formal" process. 17

B. The Supplemental Administrative Record and the Trial Record

69. Secretary Ross's first version of events, set forth in the initial Administrative Record, the Ross Memo, and his congressional testimony, was materially inaccurate.

70. The first concrete sign of the inaccuracy came on June 21, 2018, when - "to provide further background and context regarding" Secretary Ross's decision - Defendants added a one-page "Supplemental Memorandum" signed by Secretary Ross himself, to the Administrative Record. See Docket No. 189. 18 In the Supplemental *548 Memorandum, Secretary Ross admitted that "[s]oon after [his] appointment as Secretary of Commerce," he had begun considering "whether to reinstate a citizenship question." AR 1321 ("June 21 Supplemental Memo"). He also acknowledged that "other senior Administration officials had previously raised" the issue, but he did not identify the officials by name. Id . He continued: "My staff and I thought reinstating a citizenship question could be warranted, and we had various discussions with other governmental officials about reinstating a citizenship question to the Census." Id. He then disclosed that, "[a]s part of that deliberative process," he and his staff had "inquired whether the Department of Justice ... would support, and if so would request , inclusion of a citizenship question as consistent with and useful for enforcement of" the VRA. Id. (emphasis added).

71. After receiving Secretary Ross's Supplemental Memorandum, the Court ordered Defendants to complete the Administrative Record and authorized discovery beyond the Administrative Record. See Docket No. 199; see also July 3rd Tr. 76-89. 19

72. The evidence disclosed as a result reveals a very different set of events from the one described in the initial Administrative Record, the Ross Memo, and Secretary Ross's congressional testimony. In particular, as outlined in depth below, the evidence shows that shortly after his confirmation as Secretary of Commerce, Secretary Ross discussed the addition of the citizenship question with then-White House advisor Steve Bannon, among others; that Secretary Ross wanted to add the question to the 2020 census prior to, and independent of, DOJ's December 12, 2017 request; that the Secretary and his aides pursued that goal vigorously for almost a year, with no apparent interest in promoting more robust enforcement of the VRA; that, believing they needed another agency to request and justify a need for the question, Secretary Ross and his aides worked hard to generate such a request for the citizenship question from DOJ; that these efforts included a direct intervention by Secretary Ross with Attorney General Sessions; and that these efforts ultimately *549 succeeded, resulting in DOJ's request for a citizenship question. The evidence also reveals that DOJ deliberately (and unusually) did not explore whether there was a way to obtain the data it purportedly needed that would not involve a citizenship question on the census. The Court now turns to this more complete story of Secretary Ross's decision.

1. Secretary Ross's Early Interest in Adding the Citizenship Question

73. The Senate confirmed Secretary Ross as Secretary of Commerce on February 27, 2017. See 163 CONG. REC . S1421, S1455 (2017). Two days later, on March 1, 2017, the Census Bureau briefed Secretary Ross on the census and the upcoming deadline to notify Congress about the proposed subjects for the census questionnaire. See AR 1410, 3685-86; PX-193 ("Ross Calendar"), at 1; Docket Nos. 510-2, 510-3 (together, "Langdon Dep.") at 81, 93-98; see also 13 U.S.C. § 141 (f)(3).

74. At some point between February 27 and March 10, 2017, Secretary Ross asked his Deputy Chief of Staff and Director of Policy, Earl Comstock, why there was no citizenship question on the census. See Docket No. 490-2 ("Comstock Dep.") at 55. Comstock responded that he did not know, but would "check." Id. On March 10, 2017, Comstock sent Secretary Ross an email with the subject line: "Your Question on the Census." AR 2521. The email reported that the Census Bureau's webpage on apportionment was "explicit" that "all people (citizens and noncitizens) with a usual residence in the 50 states are to be included in the census and thus in the apportionment counts." Id . The email also included the text of a Wall Street Journal article titled "The Pitfalls of Counting Illegal Immigrants," which, Comstock noted, "confirms that neither the 2000 nor the 2010 Census asked about citizenship." Id.

75. Around this same time, Secretary Ross's interest in adding a citizenship question to the census first surfaced. Comstock, for example, first heard "about the notion of adding a question about citizenship to the decennial census" from the Secretary himself, "shortly after the confirmation." Id. at 54 . In fact, some evidence suggests that Secretary Ross went even further in these early weeks and had actually decided to add the citizenship question already. According to Comstock, for example, Secretary Ross actually made a "request" to add the citizenship question "sometime in the spring" - "[p]otentially" as early as March 10, 2017 (when Comstock had emailed him the Wall Street Journal article). Id. at 146 .

76. Although Secretary Ross acknowledges speaking with various "senior Administration officials" and "other governmental officials" about adding a citizenship question to the census around this time, AR 1321, the record is largely (and somewhat surprisingly) void of details regarding when these conversations occurred and with whom. But it does reflect that Secretary Ross discussed the topic with at least three outside officials. First, Secretary Ross discussed the topic with Attorney General Sessions "in the Spring of 2017 and at subsequent times." See Docket No. 379-1 ("Defs.' Second Supp. Interrog."), at 2-3 (admitted as PX-302). Second, on or about April 5, 2017, Secretary Ross spoke with White House advisor Steve Bannon, who asked "if he would be willing to speak to Kansas Secretary of State Kris Kobach about Secretary Kobach's ideas about a possible citizenship question on the decennial census." Id. at 3 ; AR 763-64, 2561. Third, complying with Bannon's request, Secretary Ross spoke with Kobach (who also served as Vice Chair of the since aborted Presidential Commission on Election Integrity). See AR 763-64, 2561; Defs.'

*550 Second Supp. Interrog. at 2-3; Docket Nos. 509-2, 509-3 (together, "Teramoto Dep.") at 38-47. During that discussion, Secretary Ross and Kobach discussed the potential effect on "congressional apportionment" of adding "one simple question" to the census. AR 763-64.

77. After these conversations, Secretary Ross and his staff began to take action on the citizenship question, in part by contacting Mark Neuman, who was not then a government official but had served as the point person on census-related issues for the Trump Administration's transition team. Teramoto Dep. 126-27. On April 11, 2017, Neuman emailed Comstock a link to the Supreme Court's decision in League of United Latin Am. Citizens v. Perry , 548 U.S. 399 , 126 S.Ct. 2594 , 165 L.Ed.2d 609 (2006) (" LULAC "), which considered CVAP data in assessing claims under Section 2 of the VRA. See PX-188 (AR); see also Comstock Dep. 155-56. On April 13, 2017, Comstock asked Neuman by email when the Census Bureau would need to "notify Congress" regarding census questions. AR 3709. Neuman responded the next day, informing Comstock that the notification deadline for census topics had already passed, but there would "be another opportunity next year" when the report of specific questions was due to Congress. AR 3709.

78. Secretary Ross soon began to express frustration with his staff's lack of progress on the citizenship question issue. On April 20, 2017, for example, he emailed Comstock, copying Wendy Teramoto, who was then his Senior Advisor and Chief of Staff. See AR 3694; Teramoto Dep. 78. 20 In the email, he noted that then-Director of the Census Bureau John Thompson was scheduled to meet with the Census National Advisory Committee on Racial, Ethnic and Other Populations on April 29, 2017, and he noted: "We must get our issue resolved before this!" AR 3694 (emphasis in original); see Comstock Dep. 137-41.

79. On May 2, 2017, Secretary Ross expressed even greater frustration. He emailed Comstock that he was "mystified why nothing ha[s] been done in response to [his] months old request that we include the citizenship question." AR 3710 (emphasis added).

80. A few hours later, Comstock responded. "On the citizenship question," he declared, "we will get that in place." Id. ; PX-298(R), at RFA 63; 21 see Comstock Dep. 151-52. He explained that the specific questions on the census questionnaire were not due to Congress until March 2018. See AR 3710. In the meantime, Comstock wrote, "[w]e need to work with Justice to get them to request that citizenship be added back as a census question, and we have the court cases to illustrate that DoJ has a legitimate need for the question to be included." Id . Comstock promised to "arrange a meeting with DoJ staff this week to discuss." Id.

2. Comstock's Search for a Rationale and an Agency to Request the Question

81. By May 2, 2017, Comstock had come to believe that the Commerce Department *551 would need another agency to request addition of the citizenship question on the census because OMB and the Paperwork Reduction Act, 44 U.S.C. §§ 3501 et seq. , required the Commerce Department to "justify" why a citizenship question was "need[ed]." Comstock Dep. 153-54. Comstock acknowledged that simply "say[ing] the Secretary wanted" the question added would not "clear [the] legal thresholds." Id. at 154 .

82. With this understanding, Comstock set out to find a "legal rationale" to support the Secretary's request to add a citizenship question, id. at 266 , and to "find an agency that would have a reason" to do so, id. at 181 . Comstock testified that he viewed it as his job to "help [the Secretary] find the best rationale" for adding the question, because "[t]hat's what a policy person does." Id. at 267 . In his view, he did not "need to know what" the Secretary's actual "rationale might be, because it may or may not be one that is ... legally-valid." Id.

83. On May 4, 2017, two days after promising Secretary Ross that he would "get [the citizenship question] in place," Comstock emailed Eric Branstad, the Senior White House Advisor at the Department of Commerce, asking him to identify the "best counterpart ... at DOJ" to speak with "[r]egarding [the] Census." AR 3701; see also AR 12755; AR 2458 (same as AR 12755 but with more redactions); AR 12756 (similar to AR 12755). Branstad advised Comstock to contact Mary Blanche Hankey, who had previously served as legislative counsel to then-Senator Jeff Sessions and, at the time, was the White House liaison at DOJ. See AR 3701, 12755.

84. Later that day, Comstock emailed Hankey and asked to set up a meeting. AR 2462; see also AR 12755; PX-298(R), at RFA 68. At some point that month, Comstock and Hankey spoke in person; "[a] few days later," Hankey "directed [Comstock] to James McHenry in the Department of Justice." AR 12755. McHenry was the Director of DOJ's Executive Office for Immigration Review; he had no responsibility for enforcement of the VRA. See Attorney General Sessions Announces Appointment of James McHenry As Director of the Executive Office for Immigration Review, Department of Justice (Jan. 10, 2018), https://www.justice.gov/opa/pr/attorney-general-sessions-announces-appointment-james-mchenry-director-executive-office; PX-298(R), at RFA 71; Docket No. 491-2 ("Gore Dep."), at 65.

85. Comstock and McHenry "spoke several times ... by phone." AR 12755. After these calls and "considering the matter further," McHenry told Comstock that "Justice staff did not want to raise the [citizenship] question given the difficulties Justice was encountering in the press at the time (the whole Comey matter)." Id . 22 McHenry referred Comstock instead to Gene Hamilton at the Department of Homeland Security ("DHS"). Id. Comstock and Hamilton held "several phone calls to discuss the matter," but Hamilton "relayed that after discussion DHS really felt that it was best handled by Department of Justice." Id. Faced with these rejections by DOJ and DHS, Comstock turned to James Uthmeier, a lawyer at the Commerce Department. Id. He asked Uthmeier "to look into the legal issues and how Commerce *552 could add the question to the Census itself." Id.

86. On May 24, 2017, Secretary Ross met "all afternoon" with various aides, including David Langdon, a Policy Advisor who reported to Comstock. AR 12542; see also AR 3702-04 (same as AR 12541-43 but with more redactions). During the meeting, Secretary Ross "seemed ... puzzled why citizenship is not included in [the] 2020" census. AR 12541. At about 11 p.m. that night, Langdon emailed Lisa Blumerman, Acting Associate Director of the 2020 Decennial Census, asking her to respond immediately - "[i]deally this evening" - to his inquiry about a citizenship question. Id. ; see also Langdon Dep. 172-74. Langdon reported to Comstock that he made that request. PX-543 (AR).

3. Secretary Ross and His Aides Persist in Their Efforts

87. Throughout July and August 2017, Secretary Ross and his staff continued to work internally, and with Kobach, to arrange for the addition of the citizenship question.

88. On July 14, 2017, Kobach emailed Secretary Ross to follow up on their April telephone conversation. See AR 763-64. Kobach wrote that the lack of a citizenship question on the census "impairs the federal government's ability to do a number of things accurately," and "also leads to the problem that aliens who do not actually 'reside' in the United States are still counted for congressional apportionment purposes." Id. at AR 764; PX-298(R), at RFA 81-82. Kobach stated that it was "essential that one simple question be added to the upcoming 2020 census," urging, in particular, a "slight variant" of the citizenship question that appears on the ACS. AR 763-64, at AR 764. Kobach did not mention the VRA or any rationale tied to VRA enforcement. Id .

89. On July 21, 2017, Kobach called Teramoto. See id. at AR 763. He also forwarded her his July 14, 2017 email to Secretary Ross. See id. In a note above the forwarded email, Kobach wrote that he had "spoken" to Secretary Ross about the addition of citizenship question to the census "at the direction of Steve Bannon." Id. ; see also PX-298(R), at RFA 83. He asked "to schedule a short call." AR 763. In response to this email, Teramoto arranged a call between Kobach and Secretary Ross. See id . Teramoto would later testify that she had "no recollection of ever speaking" with Kobach. Teramoto Dep. 40-45. She also testified that she had "no idea" who Kobach was, id. , even though the email he forwarded to her began with the line: "Kansas Secretary of State Kris Kobach here." AR 764. 23 On or about July 25, 2017, Secretary Ross spoke with Kobach about the addition of a citizenship question to the decennial census. See id. ; Ross Calendar. at 8, 40; PX-298(R), at RFA 84.

90. On August 8, 2017, Representative Mark Meadows and Secretary Ross spoke by telephone. Ross Calendar at 9. Later that day, Secretary Ross emailed Comstock to ask "where" DOJ was "in their analysis" of whether to request the addition of a citizenship question. See AR 4004; PX-298(R), at RFA 91; see also Comstock Dep. 213. Secretary Ross advised: "If [DOJ] still have not come to a conclusion please let me know your contact person and I will call the AG." AR 4004; PX-298(R), *553 at RFA 92; Comstock Dep. 214. Comstock immediately responded to Secretary Ross, writing that he would "be back shortly with an update." AR 4004; PX-298(R), at RFA 93.

91. The next day, August 9, 2017, Comstock responded again by email to Secretary Ross about their internal analysis of the citizenship question. See AR 12476; see also AR 3984 (same as AR 12476 but with more redactions). He wrote: "[W]e are preparing a memo and full briefing for you on the citizenship question. The memo will be ready by Friday." AR 12476. Critically, Comstock cautioned: "Since this issue will go to the Supreme Court we need to be diligent in preparing the administrative record." Id. Secretary Ross responded that he "would like to be briefed on Friday by phone" and added that "we should be very careful, about everything, whether or not it is likely to end up in the SC." Id.

92. On August 11, 2017, Comstock and Uthmeier exchanged edits on briefing materials regarding the citizenship question for Secretary Ross. In one email, Uthmeier shared "recommendations on execution." PX-607 (AR), at 2; AR 11343-45 (redacted version). Uthmeier stated that "our hook" was "[u]ltimately, we do not make decisions on how the [citizenship] data will be used for apportionment, that is for Congress (or possibly the President) to decide." PX-607 (AR), at 2. Later the same day, Comstock emailed Secretary Ross and Teramoto a memorandum prepared by Uthmeier regarding the addition of a citizenship question to the census. See AR 2461; see also AR 11362. (The Memorandum itself was withheld by Defendants - with the Court's blessing, see Docket No. 361 - on the basis of attorney-client privilege and, thus, is not part of the record.)

93. On August 21, 2017, senior Commerce Department personnel met regarding the citizenship question. See AR 2461. Attendees included Teramoto; Comstock; Peter Davidson, the newly appointed Commerce Department General Counsel, see 163 CONG. REC . S4781, S4897 (2017); and Under Secretary Kelley, who oversaw the operations of the Census Bureau, see id . See AR 2461; Docket No. 493-2 ("Kelley Dep."), at 25. During their respective depositions, however, Comstock, Teramoto, and Under Secretary Kelley denied having any recollection of the August 21, 2017 meeting. See Comstock Dep. 221; Teramoto Dep. 54-58; Kelley Dep. 91-92.

94. On September 1, 2017, Secretary Ross complained to Comstock and Teramoto that he had "received no update" on "the issue of the census question." AR 2424; see also AR 4002-03, at AR 4002. Comstock responded: "Understood. Wendy and I are working on it." AR 4002-03, at AR 4002.

95. On September 6, 2017, Secretary Ross met with various aides, including Under Secretary Kelley, Teramoto, Comstock, Davidson, and Uthmeier, to discuss addition of the citizenship question. See AR 1411-12; AR 1996-97; AR 2426-28; Ross Calendar 11 ("Staff Briefing: Census Legal Question"); see also AR 1998-99 (same email chain as AR 1996-97); AR 2429-30 (same email chain as AR 2426-28). Uthmeier prepared a briefing book for the meeting, which he gave to at least Under Secretary Kelley. See AR 1996-97, at AR 1996; Docket No. 253 ("Uthmeier Decl."), ¶ 3. The briefing book is not part of the record, however, apparently because none of the participants in the meeting retained a copy. See Pls.' Proposed Findings ¶ 383. At their depositions, Under Secretary Kelley, Comstock, and Teramoto denied having any recollection of the meeting. See Comstock Dep. 221; Teramoto Dep. 58-61; Kelley Dep. 105-07.

96. The day after the meeting with Secretary Ross, Comstock emailed Uthmeier *554 and Davidson that Secretary Ross "would like an update on progress since the discussion yesterday regarding the citizenship question." AR 2034; see also AR 2395-96; AR 2459-60. Later that day, Davidson wrote to Comstock, Uthmeier, and Teramoto expressing "concern[ ] about" directly contacting Kobach - whom Secretary Ross had mentioned in the September 6, 2017 meeting. PX-614 (AR). Instead, Davidson recommended contacting a "trusted" advisor, such as Neuman, "before we do anything externally." Id.

97. Consistent with that recommendation, Uthmeier sent Neuman an email on September 8, 2017, with the subject line "Questions re Census," and asking if Neuman had a "few minutes" that morning "to discuss." AR 2051. That same day, Comstock sent Secretary Ross a memorandum reporting on his - as yet unsuccessful - efforts to find an agency that would request addition of a citizenship question to the 2020 census. See AR 12755; PX-298(R) at RFA 96. The memorandum summarized Comstock's prior discussions with Hankey and McHenry of DOJ, and with Hamilton of DHS, and noted that none had expressed interest in requesting the addition of the citizenship question on the census. See AR 12755.

4. Secretary Ross's Intervention with the Attorney General

98. By sometime in late summer 2017, Secretary Ross was plainly out of patience with Comstock's failed efforts to get DOJ to request the citizenship question and decided to take matters into his own hands by contacting Attorney General Sessions directly. The precise date of his first communication on the census topic with the Attorney General during that period is unclear. See AR 1321; Gore Dep. 83-84; see also Defs.' Second Supp. Interrog. 2-3. But around Labor Day, the Attorney General spoke then-Acting Assistant Attorney General for Civil Rights John Gore ("AAAG Gore") about requesting the addition of a citizenship question. Gore Dep. 83. AAAG Gore learned that Secretary Ross had "initiated" an earlier discussion with Attorney General Sessions about a citizenship question. Id. at 83-84 .

99. Until September 13, 2017, the conversations between the Commerce Department and DOJ about the citizenship question had been "initiated by the Department of Commerce." Gore Dep. 67-68, 91-98. But on that date, AAAG Gore initiated contact with Teramoto. He sent Teramoto an email introducing himself and asking "to talk about a DOJ-DOC" issue, referring to the addition of a citizenship question to the 2020 census. AR 2628-29, at AR 2629; AR 2634-35 (same email chain); Gore Dep. 95-97. And for the next couple months, AAAG Gore served as the primary contact at DOJ about the citizenship question issue. Gore Dep. 91-92, 94-95.

100. A few days later, on or about September 16, 2017 (a Saturday), AAAG Gore and Teramoto spoke by telephone. See AR 2628-29, at AR 2628; AR 2639. Later that day, AAAG Gore put Teramoto in touch with Danielle Cutrona, an aide to Attorney General Sessions. See AR 2639. In his email introducing them, AAAG Gore explained that "Danielle is the person to connect with about the issue we discussed earlier this afternoon," namely the citizenship question. Id .; AR 2651-52; AR 2653-54; Gore Dep. 102-03.

101. Teramoto asked Cutrona to "let [her] know when the AG is available to speak to Secretary Ross." AR 2639. In response, Cutrona sent Teramoto Attorney General Sessions' cellphone number and stated: "From what John [Gore] told me, it sounds like we can do whatever you all need us to do and the delay was due to a *555 miscommunication. The AG is eager to assist." AR 2651; Gore Dep. 105-06.

102. On September 18, 2017, Secretary Ross spoke again to Attorney General Sessions about adding a citizenship question to the census. See AR 2528; AR 2636; Gore Dep. 111-12. The next day, on September 19, 2017, Secretary Ross sent an email with the subject "Census" to Davidson. AR 2528. The email said: "Wendy [Teramoto] and I spoke with the AG yesterday. Please follow up so we can resolve this issue today." Id .

5. AAAG Gore Ghostwrites the DOJ Letter

103. Over the next three months, relying heavily on the VRA rationale first proposed by the Department of Commerce, political appointees in DOJ - led by AAAG Gore - drafted the letter requesting addition of the citizenship question on the decennial census.

104. Notably, in drafting the letter, AAAG Gore relied not only on the Commerce Department's proposed rationale, but also on its work product and its advisors. For example, he spoke with Uthmeier, the Commerce Department lawyer who had drafted the August 11, 2017 legal memorandum on the issue for Secretary Ross. See Gore Dep. 117-19. After they spoke, AAAG Gore received a copy of the memorandum and a handwritten note from Uthmeier. See id. ; see also AR 2461 (email referencing August 11 memo). At his deposition, AAAG Gore stated that "[t]he note contained information regarding [the citizenship question] issue that was considered by the Department of Justice in drafting its request," that is, the December 12, 2017 Gary Letter. Id. 123-24 . AAAG Gore acknowledged, however, that Uthmeier had no experience with, or responsibility for, enforcement of the VRA. See id. at 117-18 .

105. AAAG Gore also communicated directly with Neuman, who was advising Secretary Ross and his aides on the issue and who appears to have been the first to float the VRA rationale for adding the question. See Gore Dep. 437-38; see also PX-188 (AR); Comstock Dep. 155-56. (Neuman, in turn, kept the Commerce Department abreast of his dealings with DOJ. For example, on Sunday, October 8, 2017, Secretary Ross emailed Davidson with the subject line "Letter from DoJ," asking "what is its status?" AR 2482. Davidson responded: "I'm on the phone with Mark Neuman right now ... he is giving me a readout of his meeting last week." Id .) AAAG Gore knew that Neuman "was advising the Department of Commerce and the Census Bureau with respect to this issue" as well. Gore Dep. 437-38.

106. By November 1, 2017, AAAG Gore had completed a draft of what would become the Gary Letter. See Gore Dep. 126-27. He solicited feedback on the draft from only a few people in DOJ. AAAG Gore emailed the draft letter to Chris Herren, Chief of DOJ's Voting Section, asking for Herren's input. Gore Dep. 126-27. AAAG Gore knew that Herren was a career DOJ official. See Gore Dep. 151-53. AAAG Gore also copied Ben Aguiñaga on his email, and authorized Aguiñaga to forward it to Bethany Pickett. See Gore Dep. 133. AAAG Gore knew that Aguiñaga and Pickett were both political appointees. See id. AAAG Gore also knew that both Aguiñaga and Pickett had recently graduated from law school and that neither had any experience as counsel in VRA cases or in assessing the reliability of CVAP data used in VRA litigation. See id. at 134-35 .

107. Herren, Aguiñaga, and Pickett provided substantive feedback on the initial draft of the letter. See id. at 136-37 . AAAG Gore also received edits on a "near-final" version of the draft letter from Rachael *556 Tucker, then counsel in the front office in the Office of the Attorney General, and Robert Troester, then Associate Deputy Attorney General. See id. at 139-42 . Like Aguiñaga and Pickett, neither Trucker nor Troester had any experience as counsel in VRA cases or in assessing the reliability of CVAP data used in VRA litigation. See id. at 140 .

108. Tucker's and Troester's edits were the last substantive edits made to the letter before it was sent. See id. at 142, 146 . AAAG Gore did not recall receiving any input or edits from career DOJ Civil Rights Division staff other than the first round of edits from Herren. See Gore Dep. 152-53.

109. On or about November 26, 2017, Secretary Ross spoke with President Trump. PX-298(R), at RFA 103. The next evening, Secretary Ross sent Davidson an email with the subject "Census Questions." AR 11193. The email said that the Census Bureau "is about to begin translating the questions into multiple languages and has let [sic] the printing contract. We are out of time. Please set up a call for me tomorrow with whoever is the responsible person at Justice. We must have this resolved." Id. ; PX-298(R), at RFA 105. AAAG Gore called Davidson that afternoon, but the record does not reflect whether they spoke. See AR 2496.

110. Ultimately, Attorney General Sessions made the decision that DOJ would send the letter to request that the Census Bureau add a citizenship question to the census. See Gore Dep. 442. The final authorization to AAAG Gore to send the letter came from Tucker or Troester on behalf of the Attorney General. See id. at 158-60 .

111. Although AAAG Gore - a political appointee - was the principal drafter of the letter, he did not sign or send it; instead, those tasks fell to Arthur Gary, a career DOJ official. On December 8, 2017, AAAG Gore emailed Gary a copy of the letter "with leadership's final changes." See id. at 145 . He wrote that "[w]ith these changes, we are authorized to send. Sending on Monday is fine." Id. at 146-47 . On Tuesday, December 12, 2017, Gary sent the final version of the letter to Dr. Jarmin. AR 663-65; see also AR 5489-91; Gore Dep. 155.

112. The Gary Letter was the first time that DOJ communicated to the Census Bureau that its existing CVAP data was not ideal for VRA enforcement purposes. See Tr. 996. Dr. Abowd testified at trial that "prior to December 2017 ..., the Census Bureau had never heard from the Department of Justice that existing CVAP data ... was not ideal for purpose of DOJ's VRA enforcement work." Id. AAAG Gore understood that before the Gary Letter was sent, DOJ had not reached out to the Department of Commerce to initiate a conversation "for the purposes of obtaining better data to enforce the Voting Rights Act." Gore Dep. 67-68. He was also aware that DOJ "staff did not want to raise the [citizenship] question" in the fall of 2017. Id. at 68-69 .

113. AAAG Gore, the primary drafter of the DOJ letter requesting improved data, admitted that he believes "that CVAP data collected through the census questionnaire is not necessary for DOJ's VRA enforcement efforts." Id. at 300 . AAAG Gore also testified that he had "no understanding of what the Census Bureau is going to do or what data it's going to provide [DOJ] in the future related to this request," id. at 215-16 ; that he did not "understand ... any margin of error" associated with the data the Census Bureau will obtain through the citizenship question on the census, id. at 224-25 ; and that he did not "know whether or not CVAP data produced from responses to the citizenship *557 question ... will, in fact, be more precise than the CVAP data on which the DOJ is currently relying for purposes of VRA enforcement," id. at 232-33 . AAAG Gore testified that he is not aware of any communications between DOJ and the Census Bureau on these subjects. Id. at 228, 233-34 .

114. In short, although the letter AAAG Gore drafted states that the "decennial census questionnaire is the most appropriate vehicle" for collecting "reliable" CVAP data for purposes of enforcing the VRA, Gary Letter at AR 663, AAAG Gore admitted that he did not know whether citizenship data obtained through the census would in fact be "more precise than the CVAP data on which DOJ is currently relying for purposes of VRA enforcement," Gore Dep. 233.

6. The Attorney General Forbids DOJ to Meet with the Census Bureau

115. Throughout December 2017 and January 2018, Census Bureau staff sought to meet with DOJ officials to better understand their request and to discuss other ways to satisfy DOJ's interest in more granular CVAP data. See AR 3289; AR 8651; Docket No. 511-2 ("Jarmin Dep.") at 64. The DOJ rebuffed these attempts.

116. On December 22, 2017, for example, Dr. Jarmin emailed Gary that "the best way to provide PL94 block-level data with citizen voting population by race and ethnicity would be through utilizing a linked file of administrative and survey data the Census Bureau already possesses. This would result in higher quality data produced at a lower cost." AR 3289; see also PX-297, at RFA 184; Tr. 961-62. Dr. Jarmin suggested a "meeting of Census and DOJ technical experts to discuss the details of this proposal." AR 3289.

117. Dr. Jarmin repeatedly followed up with Gary. On January 2, 2018, Dr. Jarmin sent Gary a follow-up email requesting a meeting the following week. See AR 5490. Gary, in turn, informed AAAG Gore that Census Bureau officials were attempting to arrange a meeting between the Census Bureau and DOJ. See Gore Dep. 262-63. AAAG Gore also learned from Gary that the Census Bureau had an alternative means for providing the DOJ with block-level CVAP data. See id. at 265-66 . AAAG Gore told Gary that he "would think about the issue and discuss it further with others." Id. at 264 . AAAG Gore did not ask Gary to get more information about the specifics of the Census Bureau's alternative proposal. See id. at 268 .

118. AAAG Gore then discussed the matter with several DOJ officials, including Attorney General Sessions at an in-person meeting. See id. at 265, 268-69 . Attorney General Sessions decided "not to pursue the Census Bureau's alternative proposal." Id. at 271-72 . To AAAG Gore's knowledge, the reasons for the Attorney General's decision not to pursue the alternatives were not memorialized anywhere. Id. at 272 . To this date, Defendants have not identified any reason for DOJ's decision not to meet with the Census Bureau.

119. DOJ eventually "communicated that it did not want to meet with the Census Bureau to discuss alternative sources of block-level CVAP data other than a citizenship question on the decennial census questionnaire." PX-297, at RFA 200. On February 6, 2018, almost two months after first suggesting a technical meeting to DOJ, Dr. Jarmin reported to the Census Bureau and Commerce staff that DOJ did not want to meet. See AR 3460; AR 9074. He wrote that he had "spoken with DOJ leadership. They believe the letter requesting citizenship be added to the 2020 Census fully describes their *558 request. They do not want to meet." AR 3460; AR 9074; Docket Nos. 502-2, 502-4 (together, "Census Bureau 30(b)(6) Dep."), at 98. 24

120. Thus, no meeting between the Census Bureau's and DOJ's technical experts took place before Secretary Ross issued his March 26, 2018 Memorandum announcing his decision to add the citizenship question. See Census Bureau 30(b)(6) Dep. 96; Gore Dep. 259; PX-297, at RFA 170, 195, 196.

121. DOJ officials' refusal to meet with the Census Bureau to discuss their request for data was highly "unusual." Census Bureau 30(b)(6) Dep. 98-99. It is standard operating procedure for the Census Bureau to hold a technical meeting with the agency requesting additional data to discuss the best way to deliver usable data for a particular use. See Tr. 1248; see also Thompson Decl. ¶ 60. As Dr. Abowd testified, when "an agency has requested a statistical product that cannot be produced with current public estimates, we would normally expect to meet with that agency to determine ... the use case, the application, what they wanted to do, and why they felt that our current products did not serve that need. That would be normal." Tr. 1248. And as former Census Bureau Director Thompson explained, input from subject matter experts - usually consisting of "staff from both the Census Bureau and the requesting agency" - is "essential to the development of a new question" for a survey. Thompson Decl. ¶ 60. That is because "[t]hese experts help ensure that the Census Bureau has a clear understanding of the desired uses of the new data so that the new question can be worded to achieve the desired outcome." Id. ; see also Habermann Aff. ¶ 29 ("[A meeting between the Census Bureau and the requesting agency] allows the technical experts to better understand how the Census Bureau can meet the needs of the proposers. It also allows for a discussion of alternative ways of meeting a request.").

122. Neither the Census Bureau nor AAAG Gore is aware of any instance, other than the request to add a citizenship question to the decennial census questionnaire, in which DOJ has requested data from the Census Bureau and then declined to meet to discuss that request. PX-297, at RFA 201; Gore Dep. 282.

123. The fact that the Attorney General himself made the decision not to allow DOJ officials to meet with the Census Bureau is also highly unusual. Dr. Abowd, the Census Bureau's Chief Scientist and Defendants' own expert, testified that he was unaware of any other circumstance in which a Cabinet Secretary personally directed agency staff not to meet with the Census Bureau and that DOJ's refusal in this case was thus "unusual." Tr. 962-65. Dr. Abowd opined that Attorney General Sessions's decision constituted improper "political influence" on the decision-making process. Tr. 1267-68.

7. Efforts to Downplay Deviations from the Census Bureau's Standard Processes

124. As noted above (by way of background only), in recent decades, the Census Bureau has adhered to a fairly robust process in evaluating whether to add new questions to any data-collection instrument, including the census. AR 9865, 9867; AR 3560; AR 3890-91; Thompson Decl. ¶¶ 45, 47-49. That process includes rigorous *559 pre-testing. In fact, the Census Bureau's "Statistical Quality Standards," issued in 2010, explicitly require pre-testing of any question to be added to a survey such as the census, unless the Census Bureau obtains a waiver or uses a question that has "performed adequately in another survey." PX-260, at vii, 8; see also PX-364 ("If there is insufficient evidence about how well a question performs, the question must be subjected to some form of questionnaire pretest.").

125. Initial plans for the 2020 census adhered to the Census Bureau's historical practices and Statistical Quality Standards. All questions on the 2010 census had been the subject of extensive cognitive testing and field testing. See Tr. 997. And the Census Bureau began testing potential improvements for the questions on race and ethnicity for the 2020 census more than ten years before the census, in 2008. See Thompson Decl. ¶ 49. The Census Bureau sought community feedback on the proposed changes in 2014 and 2015, and then conducted additional testing in 2015. See id. ¶¶ 50-53 . Despite that extensive testing, the Census Bureau opted not to make the proposed changes to the questions on race and ethnicity because a final decision had not been made as of December 31, 2017, leaving inadequate time to deliver the final question wording to Congress two years prior to the census, as required by Section 141(f)(2). See id. ¶ 54 .

126. Additionally, on April 29, 2016, Lisa Blumerman, Associate Director of Decennial Census Programs, issued a memorandum "officially document[ing] the U.S. Census Bureau's plan to develop and transmit to Congress" the questions planned for the 2020 decennial census. PX-271, at 1; see also Tr. 995. To the extent relevant here, the Memorandum invited "[f]ederal agencies with known uses of the 2020 Census or ACS content" to submit any requests for data collection by July 1, 2016. PX-271, at 3. Blumerman explained that "[f]inal proposed questions are based on the results of extensive cognitive testing, field testing, other ongoing research, and input from advisory committees." PX-271, at 4; Tr. 995-96.

127. Despite these plans and the Census Bureau's Statistical Quality Standards, neither the Census Bureau nor the Commerce Department conducted any pretesting of the citizenship question before Secretary Ross made the decision to add it to the 2020 census questionnaire. There was no cognitive testing, field testing, or randomized control testing of the question, nor was there any testing of the question within the context of the entire questionnaire or consultation with the Census Bureau's advisory committees or outside researchers with relevant expertise. See Tr. 156-57, 736-37, 925-26, 997-98, 1279-80; Thompson Decl. ¶¶ 60, 71, 77; Census Bureau 30(b)(6) Dep. 142-43, 426-27; Habermann Aff. ¶¶ 56-58. The one "end-to-end" test - in essence, a form of dress rehearsal - conducted for the 2020 census did not include the citizenship question. Tr. 92, 155-56, 998, 1096; Census Bureau 30(b)(6) Dep. 225.

128. Nor did the Census Bureau apply for, let alone receive, a waiver from the pre-testing requirement for the citizenship question. Tr. 1279. And while Secretary Ross opted to use a question that had previously appeared on the ACS, the record establishes that it has not "performed adequately" within the meaning of the Census Bureau's Statistical Quality Standards. For example, 32.7% of all people identified as noncitizens by administrative records reported themselves as citizens on the 2010 ACS. See January 19 Memo, at AR 1280. For the 2016 ACS, that figure was 34.7%. See id. Defendants' own expert witness, Dr. Abowd, acknowledged that *560 the Census Bureau views this "disagreement" between the ACS survey responses and administrative records - which are generally viewed as more accurate - as a "problem with the ACS citizenship question." Tr. 1282. In fact, he opined that, in light of the disagreement rate, the citizenship question on the ACS has not "performed adequately." Id. at 1286-88 ; see also Defendants' Post-Trial Proposed Findings of Fact and Conclusions of Law Regarding Plaintiffs' Claims, Docket No. 546 ("Defs.' Post-Trial Br."), at 62, ¶ 420 (conceding that "the citizenship question does not appear to be performing adequately on the ACS").

129. Thus, the failure to conduct any pretesting of the proposed citizenship question on the decennial census questionnaire was a "significant deviation" from the Census Bureau's historical practices, its own mandatory Statistical Quality Standards, and its previously announced plans for the 2020 census. See Thompson Decl. ¶ 96; AR 3890-91; AR 2304; AR 9865, 9867; Tr. 1264. Yet, Secretary Ross and the Commerce Department tried to downplay, if not conceal, the degree of that deviation. The Court will address four examples of these efforts.

a. Secretary Ross's Claim that the Question Was Well Tested

130. First, Secretary Ross described the citizenship question as "well tested" in his March 26, 2018 Memorandum. Ross Memo 2, at AR 1314. He did not, however, describe that testing or even define what he meant by "well tested." See id. at AR 1313-20. Nor did Secretary Ross make any effort to reconcile his characterization of the question as "well tested" with the poor performance of the question on the ACS - that is, the high disagreement rate between survey responses and administrative records. See id. at AR 1316. Ironically, Secretary Ross himself noted this high disagreement rate when explaining his reasons for rejecting Alternative C. See id.

131. There is no basis in the record to dispute Dr. Abowd's assessment - which he conveyed to Secretary Ross in advance of the March 26, 2018 Memorandum - that the citizenship question was "well tested" for purposes of the ACS. See Tr. 1254-55, 1288; see also Tr. 568 (testimony of Thompson, former Census Bureau Director, that testing of the citizenship question for use on the ACS was "complete and thorough"). But the fact that the question was "well tested" for purposes of the ACS does not mean that the question is well tested for purposes of the decennial census. Nor does it mean that the testing was consistent with the Census Bureau's own standards and historical practices.

132. To the contrary, the record supports the conclusion of experts in the field that the question was not well - or even adequately - tested for purposes of the decennial census questionnaire. That view is shared by: (1) six former Census Bureau Directors, in both Republican and Democratic Administrations, including Dr. Thompson (whom Defendants' own expert described as "the expert in this trial who's most deserving of weight on his testimony), AR 8555-56; Tr. 1192; Thompson Decl. ¶¶ 93-96; (2) Plaintiffs' experts, Dr. Hillygus and Dr. Barreto, Tr. 157, 737-38; (3) the National Academies of Sciences, Engineering, and Medicine's Committee on National Statistics' Task Force on the 2020 Census, PX-539; and (4) the American Statistical Association, the American Sociological Association, and the Population Association of America, "three leading national associations of professional and academic statisticians, sociologists and demographers," Brief of the American Statistical Association et al. as Amici Curiae Supporting Plaintiffs, Docket No. 420-1, at 1, 4-9.

*561 133. Defendants did not present any evidence to dispute the conclusion that the citizenship question has not been "well tested" for purposes of the 2020 census questionnaire. Instead, their own expert, Dr. Abowd, agreed that "[i]t would not be appropriate to describe it as well-tested in the context of the 2020 questionnaire. That is simply not true.... It hasn't ever been tested in that context." Tr. 1330.

134. There are several reasons that the testing conducted for the citizenship question on the ACS is not adequate for purposes of the decennial census questionnaire. First, although superficially similar, there are salient differences between the ACS and the decennial census. See Docket No. 456 ("Defs.' Pretrial Reply Br."), at 5 (conceding that the ACS is "a different instrument with different considerations and goals"). The primary purpose of the census is to obtain an "actual Enumeration" of the entire population of the country, see U.S. Const. art. I, § 2, cl. 3, while the ACS is "intended to provide information on the characteristics of the population, and the social and economic needs of communities," Thompson Decl. ¶ 32. And while the census strives for an "actual Enumeration" through an actual count of every household in the country, ACS estimates are statistical estimates based on a sample of United States households. Thompson Decl. ¶¶ 32-33; Tr. 805, 1027-28. Additionally, the ACS is a much longer questionnaire than the decennial census questionnaire. It includes dozens of questions - depending on household composition and housing status, a household might have to answer more than seventy questions - while the 2020 census questionnaire will contain either ten or eleven questions (depending on whether the citizenship question appears on it). See Tr. 87, 889, 1146; Thompson Decl. ¶ 63. A question concerning citizenship may take on added significance in the context of the much shorter decennial census questionnaire. See Thompson Decl. ¶ 63; Tr. 87-88. Thus, a test of the question in the context of the ACS questionnaire has only limited relevance to how the question will perform in the context of the census questionnaire.

135. Second, and related, the question order and context for the question in the ACS (as well as on the 1950 census questionnaire and the long-form questionnaire, both of which were also referenced by Secretary Ross in his March 26, 2018 Memorandum) are different than they would be on the census questionnaire. On the online ACS (and the prior census questionnaires), the question was preceded

This text is long and has been trimmed here. Open the source document for the complete record.