Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 1 of 126 1 2 3 4 5 6 7 UNITED STATES DISTRICT COURT 8 NORTHERN DISTRICT OF CALIFORNIA 9 10 STATE OF CALIFORNIA, et al., Plaintiffs, 11 United States District Court 12 FINDINGS OF FACT AND WILBUR ROSS, et al., 15 Defendants. 16 --------------------------------------------------- 17 CITY OF SAN JOSE, et al., Plaintiffs, 18 19 v. 20 21 WILBUR L. ROSS, et al., 22 Defendants. 23 24 25 26 27 28 18-cv-02279-RS v. 13 14 Case No. 18-cv-01865-RS CONCLUSIONS OF LAW Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 2 of 126 TABLE OF CONTENTS 1 2 I.  SUMMARY OF DECISION ........................................................................................... 3  3 II.  BACKGROUND .............................................................................................................. 6  4 III.  5 6 STANDING .................................................................................................................. 8  A.  Legal Standard ........................................................................................................... 8  B.  Findings of Fact Related to Standing ........................................................................ 9  C.  Conclusions of Law Related to Standing ................................................................ 49  7 8 9 1.  Injury-in-Fact ....................................................................................................... 50  10 2.  Traceability .......................................................................................................... 55  11 3.  Redressability ....................................................................................................... 57  United States District Court 12 IV.  APA CLAIM ............................................................................................................... 58  13 A.  Legal Standard ......................................................................................................... 58  15 B.  Scope of Review ...................................................................................................... 58  16 C.  Findings of Fact Based Exclusively on the Administrative Record ........................ 60  D.  Findings of Fact Based on Extra-Record Evidence ................................................. 87  E.  Conclusions of Law ............................................................................................... 105  14 17 18 19 20 V.  CONSTITUTIONAL CLAIMS ................................................................................... 118  21 A.  Legal Standard ....................................................................................................... 119  22 B.  Scope of Review .................................................................................................... 120  C.  Conclusions of Law ............................................................................................... 121  23 24 VI.  REMEDIES ............................................................................................................... 124  25 26 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 2 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 3 of 126 I. SUMMARY OF DECISION 1 The formal decision by Secretary of Commerce Wilbur L. Ross, Jr. on March 26, 2 3 4 5 6 7 8 9 2018 to add a citizenship question to the 2020 Decennial Census violated the Administrative Procedure Act (“APA”) and the Enumeration Clause of the United States Constitution. Nearly a year before issuing that decision, on May 2, 2017, Secretary Ross sent an email to Deputy Chief of Staff Earl Comstock stating in part “I am mystified why nothing [has] been done in response to my months old request that we include the citizenship question. Why not?” What ensued was a cynical search to find some reason, any reason, or an agency request to justify that preordained result. As to the APA, one need look no further than the Administrative Record1 to conclude 10 11 United States District Court 12 13 14 15 16 17 18 19 20 that the decision to include the citizenship question was arbitrary and capricious, represented an abuse of discretion, and was otherwise not in accordance with law. In response to Secretary Ross’s demand, Comstock began to search for an agency that would be willing to request the inclusion of the citizenship question in the 2020 Census. When initially approached by Comstock about the citizenship question, the Department of Justice (“DOJ”) opted not to request its inclusion in the census. Comstock then reached out to the Department of Homeland Security, which similarly declined to request the addition of the question. Only after Secretary Ross personally interceded with then Attorney General Jeff Sessions did the DOJ switch its position and request the inclusion of a citizenship question, ostensibly to assist in the enforcement of Section 2 of the Voting Rights Act (“VRA”). Despite unrefuted evidence produced by the professional staff of the Census Bureau 21 22 23 24 25 26 27 that inclusion of a citizenship question would likely result in a significant differential decline in self-response rates within noncitizen and Latino communities and that the requested data 1 The Administrative Record comprises all documents identified in the parties’ stipulation, specifically: AR 1 through AR 13024 (PTX-001 through PTX-014); PTX-016 through PTX152; PTX-154; PTX-156; PTX-157; PTX-164 through PTX-170; PTX-172 through PTX182; PTX-184; PTX-185; PTX-227; PTX-228; PTX-244; PTX-264; PTX-274; PTX-362; PTX-363; PTX-370; PTX-374 through PTX-390; PTX-397; PTX-399 through PTX-433; PTX-435 through PTX-452. Joint Pretrial Statement 11-12. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 3 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 4 of 126 1 could be obtained by other means, Secretary Ross insisted upon adding the citizenship 2 question to the census. When Census Bureau staff offered to meet with DOJ staff to ascertain 3 if other available data could be used to meet their VRA enforcement needs, DOJ took the 4 unprecedented step of refusing to allow even such an inter-agency meeting to take place. United States District Court 5 These facts and other evidence contained in the Administrative Record, along with all 6 reasonable inferences to be drawn therefrom, demonstrate that Secretary Ross’s reliance on 7 VRA enforcement to justify inclusion of the citizenship question was mere pretext and the 8 definition of an arbitrary and capricious governmental act. Moreover, Secretary Ross’s 9 conclusion that adding the citizenship question would enable the Census Burau to obtain 10 more “complete and accurate data” in response to the DOJ’s request is not only unsupported, 11 it is directly contradicted by the scientific analysis contained in the Administrative Record. 12 PTX-26 at 1, 7. While it is of course appropriate for an incoming cabinet member to 13 advocate for different policy directions, to solicit support for such views from other agencies, 14 and to disagree with his or her professional staff, this record reflects a profoundly different 15 scenario: an effort to concoct a rationale bearing no plausible relation to the real reason, 16 whatever that may be, underlying the decision. 17 Again confining review solely to the Administrative Record, it is evident that the 18 inclusion of the citizenship question on the 2020 Census violated both Sections 6(c) and 19 141(f)(3) of the 1976 Census Act. Section 6(c) mandates that, to the maximum extent 20 possible, the Secretary use administrative records as opposed to additional census questions 21 to obtain secondary data, such as demographic information. Section 141(f) mandates certain 22 timely reports to Congress regarding the subject and questions to be included on the census 23 and limits the Secretary’s ability subsequently to modify the contents of the census absent 24 new circumstances that necessitate a change. Quite simply, Secretary Ross ignored these 25 statutory requirements in issuing his March 26, 2018 decision. 26 27 While finding a violation of the APA logically flows from the Administrative Record in this action alone, the facts here satisfy the requisite standard warranting consideration of 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 4 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 5 of 126 1 extra-record evidence. Such evidence includes the absence of any effort to test the impact of 2 the addition of the citizenship question to the census, the deviation from the Census Bureau’s 3 usual process for adding new questions to the census, the troubling circumstances under 4 which the DOJ’s request letter was drafted and procured, and Sessions’ order prohibiting 5 DOJ staff from meeting with Census Bureau officials to discuss alternative sources of data 6 that could meet DOJ’s VRA enforcement needs. Going beyond the Administrative Record, 7 in short, confirms that the decision to include a citizenship question runs afoul of the APA. 8 United States District Court 9 The analysis of the Enumeration Clause claim similarly involves evidence beyond the four corners of the Administrative Record. As a general proposition, the decision to 10 include a specific question on the census is committed to the discretion of the Commerce 11 Secretary and does not implicate the constitutional command that all persons in each state be 12 counted every ten years. However, if the Secretary’s decision to include a question 13 affirmatively interferes with the actual enumeration and fulfills no reasonable governmental 14 purpose, it may form the basis for a cognizable Enumeration Clause challenge. 15 The evidence admitted in the trial of these actions demonstrates that a significant 16 differential undercount, particularly impacting noncitizen and Latino communities, will 17 result from the inclusion of a citizenship question on the 2020 Census, compounded by 18 macro-environmental factors arising out of the national immigration debate. Efforts to 19 ameliorate these effects through Non-Response Follow-Up (“NRFU”), the evidence showed, 20 would not remediate and could in fact exacerbate the differential undercount of noncitizens 21 and Latino persons. While a citizenship question had been included in the decennial census 22 in 1950 and before, the analysis now must turn on the impact of that question on the prospect 23 of achieving the central constitutional purpose of an actual enumeration in 2020. Viewed 24 through that lens, the inclusion of the question is contrary to the Constitution. 25 Plaintiffs in each of these actions satisfied their burden of demonstrating standing 26 under Article III of the Constitution. The State of California demonstrated that it will suffer a 27 loss of federal funding and face a substantial risk of losing political representation directly 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 5 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 6 of 126 1 traceable to the inclusion of the citizenship question on the census. California established 2 that the inclusion of this question will also require the expenditure of additional funds to 3 attempt to mitigate the effects of the question and minimize the resulting undercount of 4 California vis-à-vis other states. Similarly, the City of San Jose and the Black Alliance for 5 Just Immigration (“BAJI”) each established injury directly flowing from the addition of the 6 citizenship question. In the case of San Jose, it showed the negative impact on federal 7 funding it would receive for various programs dependent on census data and the additional 8 resources that would be required to attempt to mitigate those effects. As to BAJI, the 9 organization demonstrated that it will be obliged to commit additional time and resources to 10 address the specific effects of the citizenship question on its constituents and to encourage 11 them to participate despite the perceived risks. 12 In light of the statutory and constitutional violations outlined above, the issue 13 becomes the appropriate remedy. With respect to the APA claim, consistent with and for the 14 reasons stated in New York v. United States Dep’t of Commerce, No. 18-cv-2921 (S.D.N.Y. 15 Jan. 15, 2019) (the “New York matter”), vacatur of the Secretary’s decision, remand to the 16 agency, and an injunction against inclusion of the citizenship question on the 2020 census is 17 warranted and will be ordered. As to the Enumeration Clause violation, an injunction is the 18 proper relief. The Department of Commerce urges that any relief should be limited to the 19 particular plaintiffs before the Court. While mindful of the concerns regarding individual 20 district courts issuing orders of national scope, the limitation advanced by Defendants here is 21 simply impractical in light of the nationwide nature of the questionnaire at issue. 22 Accordingly, no such limitation will be included in the injunctive relief ordered by this 23 Court. 24 25 II. BACKGROUND This action, comprising two related cases, arises from the U.S. Census Bureau’s 26 decision to include a question regarding citizenship status on the 2020 Census questionnaire. 27 Plaintiffs in Case No. 18-cv-1865 are the State of California, the County of Los Angeles, the 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 6 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 7 of 126 1 City of Los Angeles, the City of Fremont, the City of Long Beach, the City of Oakland, the 2 City of Stockton, and the Los Angeles Unified School District (collectively, “California 3 Plaintiffs”). Plaintiffs in Case No. 18-cv-2279 are the City of San Jose and the Black 4 Alliance for Just Immigration (“BAJI”) (collectively, “San Jose Plaintiffs”). Defendants in 5 both matters are Wilbur L. Ross, Jr., in his official capacity as Secretary of the U.S. 6 Department of Commerce; the U.S. Department of Commerce; Stephen Dillingham, in his 7 official capacity as Director of the United States Census Bureau; and the U.S. Census 8 Bureau. During the relevant period Dr. Ron Jarmin served as Acting Director of the United 9 States Census Bureau. On March 26, 2018, Secretary Ross issued a memorandum (the “Decision Memo”) United States District Court 10 11 directing the Census Bureau to add a question on citizenship status to the 2020 Census. PTX- 12 1 at 1313-20. Plaintiffs contend the decision to include this question violated the 13 Constitution and the APA. They specifically argue Secretary Ross’s decision violated the 14 Enumeration Clause, U.S. Const. art. I, § 2, cl. 3, and was “arbitrary, capricious, [and] 15 otherwise not in accordance with law” under the APA, 5 U.S.C. § 706(2)(A).2 The San Jose 16 Plaintiffs also allege a violation of the Apportionment Clause of the Constitution. U.S. 17 Const. amend. XIV, § 2. The process by which the decennial census is taken has changed significantly over 18 19 the years, as have the questions asked in the census instrument. From 1790 to 1960, the 20 Bureau collected data directly from households through in-person interviews. Undisputed 21 Fact (“UF”) 75.3 Moreover, from 1820 to 1950, with the exception of 1840, respondents 22 were asked a question concerning citizenship or birthplace. UF 67. The Census Bureau 23 24 25 26 27 2 There is significant overlap in the arguments and evidence presented by the San Jose Plaintiffs and the California Plaintiffs. Accordingly, certain portions of this order will treat Plaintiffs arguments jointly. 3 The parties have stipulated to over one hundred undisputed facts, which can be found under Exhibit A to the Joint Pretrial Statement (ECF 144 in Case No. 18-cv-01865). 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 7 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 8 of 126 1 subsequently transitioned to a mailed questionnaire, which involved sending a “short form” 2 questionnaire to most residences, and a “long form” questionnaire with significantly more 3 questions to the remaining households. UF 77-78. The long form questionnaires used in 1970, 1980, 1990, and 2000 included a 4 5 question about citizenship status, whereas the short form questionnaires did not. UF 80. 6 After the 2000 Decennial Census, the functions performed by the long form questionnaire 7 were replaced by the American Community Survey (“ACS”). UF 83. The ACS is a yearly 8 survey of approximately 2% of households—about 3.5 million—across the United States. 9 UF 85. A question concerning citizenship status currently appears among more than 50 10 questions on the ACS questionnaire. UF 86. In keeping with recent practice, the 2020 Census will be “short form only.” UF 102. United States District Court 11 12 The ACS will continue to be distributed as usual and will continue to include a citizenship 13 question. UF 103. Per Secretary Ross’s Decision Memo, the 2020 Census will also include a 14 citizenship question. The text of this newly added question will read, “Is this person a citizen 15 of the United States?,” with the answer options “Yes, born in the United States”; “Yes, born 16 in Puerto Rico, Guam, the U.S. Virgin Islands, or Northern Marianas”; “Yes, born abroad of 17 U.S. citizen parent or parents”; “Yes, U.S. citizen by naturalization – Print year of 18 naturalization”; and “No, not a U.S. citizen.” UF 104. As in past years, the 2020 Census 19 questionnaire will also pose questions regarding sex, Hispanic origin, race, and relationship 20 status. UF 106.4 21 III. STANDING 22 A. Legal Standard 23 In order to establish standing under Article III of the Constitution, a “plaintiff must 24 have (1) suffered an injury in fact, (2) that is fairly traceable to the challenged conduct of the 25 26 27 4 Several entities and individuals have filed motions for leave to file amicus briefs in this matter. Although these amicus briefs do not form the basis of this opinion, each of these motions is hereby granted. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 8 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 9 of 126 1 defendant, and (3) that is likely to be redressed by a favorable judicial decision.” Spokeo, 2 Inc. v. Robins, 136 S. Ct. 1540, 1547 (2016) (citing Lujan v. Defs. of Wildlife, 504 U.S. 555, 3 560-61 (1992)). As the party invoking federal jurisdiction, the plaintiff bears the burden of 4 establishing all three requirements by a preponderance of the evidence. Lujan, 504 U.S. at 5 561. 6 B. Findings of Fact Related to Standing5 7 1. Inclusion of the Citizenship Question on the 2020 Census Will Cause a Differential Decline in Self-Response Rates 8 9 10 1. Undisputed evidence in this case shows that adding a citizenship question to the 2020 Census will cause a differential decline in self-response rates for noncitizen and Hispanic households. 11 United States District Court 12 13 14 15 16 17 18 19 2. Defense expert Dr. John Abowd, Chief Scientist and Associate Director for Research and Methodology at the Census Bureau, testified credibly that the Census Bureau has produced quantitative evidence that adding a citizenship question to the 2020 Census will lower self-response rates. Tr. 797:20-25 (Abowd). He specifically endorsed the Census Bureau’s finding that the citizenship question will lead to a lower self-response rate in both noncitizen and Hispanic households, New York Tr. 881:19-882:1 (Abowd); id. at 918:3919:1 (Abowd), and that this lower response rates will harm the quality of census data, id. at 882:2-5 (Abowd).6 The Plaintiff experts similarly endorsed these conclusions. See Part III.B.1.e, infra. 20 a. December 22 Census Bureau Memo 21 3. The analysis that underpins Dr. Abowd’s testimony is set forth in three 22 23 24 25 26 27 5 Any conclusion of law that is mistakenly characterized as a finding of fact in this order may be recharacterized as a conclusion of law, and vice versa. 6 The parties have stipulated to the admission of Dr. Abowd’s trial testimony in the New York matter, subject to Defendants’ standing objection to the consideration of evidence outside the Administrative Record in adjudicating the merits of this action. Stip. & Order re Abowd Trial Transcript (ECF 172 in Case No. 18-cv-01865). 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 9 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 10 of 126 1 memoranda issued by the Census Bureau: the December 22 Memo, PTX-148; the January 19 2 Memo, PTX-22; and the Brown, et al. Memo, PTX-160. New York Tr. 896:7-15 (Abowd); 3 Census Bureau 30(b)(6) Dep. Vol. II 353:2-6, 353:19-21. 4 5 Census Bureau (nicknamed the “SWAT Team”). They found that, based on a comparison of 6 self-response rates to the 2010 Census and the 2010 ACS (which included a citizenship 7 question), noncitizen households were 5.1 percent less likely than all-citizen households to 8 respond to a survey with a citizenship question. PTX-103 at 6-7; PTX-148 at 6-7. This 9 finding is “consistent with citizenship questions being more sensitive for households with 10 11 12 United States District Court 4. The December 22, 2017 Memo was authored by senior professional staff at the 13 14 noncitizens.” PTX-103 at 7; PTX-148 at 7. b. Dr. Abowd’s January 19 Memo 5. Dr. Abowd’s January 19, 2018 Memo conveyed the 5.1 percent differential selfresponse estimate to Secretary Ross. PTX-22 at 4. 6. This finding was the result of just one of the “[t]hree distinct analyses” in the 15 January 19 Memo that “support the conclusion of an adverse impact on self-response and, as 16 a result, on the accuracy and quality of the 2020 Census.” Id. The other two analyses focused 17 on indicators that suggest that Hispanic households are disproportionately less likely to 18 respond to a survey with a citizenship question. 19 7. The first of these anlyses focused on the item nonresponse rates—the rate at 20 which respondents do not answer a particular survey question. New York Tr. 905:10-24 21 (Abowd). The Census Bureau found that item nonresponse rates for the citizenship question 22 on the ACS were more than twice as high for Hispanics as for non-Hispanic whites from 23 2013 through 2016, and that the nonresponse rate for Hispanics increased by 2.5 percent 24 relative to non-Hispanic whites over that span. PTX-22 at 4; see also New York Tr. 906:12- 25 908:6 (Abowd); Tr. 156:4-157:19 (O’Muircheartaigh). 26 27 8. The second analysis considered breakoff rates—the rate at which respondents stop completing a survey when presented with a particular question. New York Tr. 913:13-24 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 10 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 11 of 126 1 (Abowd). It found that the breakoff rate for the citizenship question on the 2016 ACS was 2 more than eight times higher for Hispanics than for non-Hispanic whites. PTX-22 at 5; see 3 also New York Tr. 914:5-8 (Abowd); Tr. 158:4-21 (O’Muircheartaigh). Similarly, the 4 breakoff rate for three related questions on immigration status (citizenship, place of birth, 5 and year of entry) on the 2016 ACS was more than three times higher for Hispanics than for 6 non-Hispanic whites. PTX-22 at 5; see also New York Tr. 915:9-13 (Abowd). 7 9. Based on the Census Bureau’s analysis of item nonresponse rates and breakoff 8 rates, Dr. Abowd testified credibly that a citizenship question would be sensitive for 9 Hispanics, and that the sensitivity of the question is increasing for Hispanics (but not for 10 non-Hispanic whites). New York Tr. 917:4-918:2 (Abowd). c. Brown, et al. Memo 11 United States District Court 12 10. The Brown, et al. Memo builds upon and updates the analysis in Dr. Abowd’s 13 January 19 Memo. New York Tr. 896:7-12 (Abowd). This memo represents the Census 14 Bureau’s best analysis of the consequences of adding a citizenship question to the 2020 15 Census. Census Bureau 30(b)(6) Dep. Vol. II 355:15-356:15; New York Tr. 897:4-15 16 (Abowd). 17 11. The Brown, et al. Memo summarized its findings as follows: 18 This paper’s examination of several Census Bureau surveys with and without citizenship questions suggests that households that may contain noncitizens are more sensitive to the inclusion of citizenship in the questionnaire than allcitizen households. The implication is that adding a citizenship question to the 2020 Census would lead to lower self-response rates in households potentially containing noncitizens, resulting in more nonresponse follow-up (NRFU) fieldwork, more proxy responses, and a lower-quality population count. 19 20 21 22 23 24 PTX-160 at 54. 12. The memo presented data showing that citizenship-related questions are more 25 sensitive for Hispanics and that, because Hispanics have higher rates of nonresponse for 26 citizenship than for sex or age, they could be disproportionately impacted by adding a 27 citizenship question to the 2020 Census questionnaire. PTX-160 at 7-10. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 11 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 12 of 126 1 2 the self-response rate of noncitizen households to 5.8 percent. PTX-160 at 39; Census 3 Bureau 30(b)(6) Dep. Vol. II 372:2-12; New York Tr. 897:16-20 (Abowd); Tr. 161:13-21 4 (O’Muircheartaigh). This revised estimate is the result of a natural experiment that compared 5 response rates on the 2016 ACS, which included a citizenship question, to response rates on 6 the 2010 Census, which did not incorporate a citizenship question, and then compared the 7 change in response rates between all-citizen households and all other households (i.e., 8 households that contain or may contain one or more noncitizens). PTX-160 at 33-34; Census 9 Bureau 30(b)(6) Dep. Vol. II 373:9-15, 374:10-16; New York Tr. 898:2-899:6 (Abowd); Tr. 10 11 United States District Court 13. The Census Bureau also updated the estimated 5.1 percent differential decline in 161:22-164:17 (O’Muircheartaigh). 14. The 5.8 percent figure represents the Census Bureau’s best conservative estimate 12 of the differential effect of the citizenship question on noncitizen household self-response. 13 New York Tr. 894:17-895:2, 897:9-12 (Abowd). 14 15. The Brown, et al. Memo emphasized that the 5.8 percent estimate is 15 “conservative.” PTX-160 at 39; New York Tr. 900:21-25 (Abowd); Tr. 164:21-24 16 (O’Muircheartaigh). The Bureau acknowledged that this figure may underestimate the 17 impact of the citizenship question on census self-response rates for two reasons: (1) the 18 question will be more prominent on the 2020 Census questionnaire, which has just ten other 19 questions, than it was on the ACS questionnaire, which has 75 questions, PTX-160 at 39; 20 New York Tr. 901:22-902:10 (Abowd); Tr. 164:25-165:14 (O’Muircheartaigh), and (2) 21 given “the level of concern about using citizenship data for enforcement purposes,” the 22 macro-environment at the time of the 2020 Census may be worse than it was when the ACS 23 data were collected, PTX-160 at 39; see also New York Tr. 902:11-24 (Abowd); Tr. 165:15- 24 21 (O’Muircheartaigh). 25 16. The 5.8 percent estimate is also conservative because of limitations in the design 26 of the natural experiment. For example, the natural experiment assumed that individuals 27 whose citizenship information was missing from administrative records were citizens, which 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 12 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 13 of 126 1 had the effect of reducing the estimated difference between the response rates of all-citizen 2 households and noncitizen households. Tr. 165:25-166:15 (O’Muircheartaigh). 3 17. The Brown, et al. Memo also confirmed the findings in Dr. Abowd’s January 19 4 Memo showing that (1) Hispanics were more than twice as likely as non-Hispanic whites to 5 skip the citizenship question on the ACS and that the differential in such item nonresponse 6 rates increased between 2013 and 2016, and (2) the breakoff rate for the citizenship question 7 on the 2016 ACS was more than eight times higher for Hispanics than for non-Hispanic 8 whites. PTX-160 at 8-11. Based on this data, the Census Bureau concluded that Hispanics 9 are more sensitive to survey questions about citizenship than they were a few years ago but 10 that non-Hispanic whites are not. Census Bureau 30(b)(6) Dep. Vol. II 369:1-19. This 11 suggests that nonresponse rates to the citizenship question on the 2020 Census will be higher 12 for Hispanics than for non-Hispanic whites. New York Tr. 910:7-13, 914:9-12 (Abowd). 13 18. Recent Census Bureau data show that the differential breakoff is escalating. 14 After the January 19 Memo and the Brown, et al. Memo were issued, the Census Bureau 15 made the 2017 ACS breakoff data publicly available. New York Tr. 915:19-916:3 (Abowd). 16 That data, which was reviewed by the SWAT team, showed that the breakoff rate for the 17 citizenship question on the 2017 ACS is now twelve times higher for Hispanics than for non- 18 Hispanic whites. New York Tr. 916:4-917:3 (Abowd). 19 20 d. CSM Memo and CBAMS Results 19. Recent Census Bureau qualitative research suggests that the citizenship question 21 will cause an even greater differential decline in self-response rates than estimated by 22 Brown, et al. The macro-environment, particularly the political environment around 23 immigration, has the potential to amplify the negative effect of the citizenship question on 24 self-response rates. New York Tr. 926:21-927:10 (Abowd). 25 20. This research includes the Census Bureau’s Center for Survey Measurement 26 (“CSM”) focus group testing in 2017, which revealed increased concern among immigrants 27 about the confidentiality of their survey responses, PTX-157 at 1, and the Census Barriers, 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 13 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 14 of 126 1 Attitudes, and Motivators Study (CBAMS) conducted in 2018, which revealed concerns 2 among Spanish language respondents about the citizenship question, PTX-153 at 21-22. i. CSM Findings 3 4 5 observed in a September 20, 2017 memo for the Associate Directorate for Research and 6 Methodology at the Census Bureau, PTX-157, and in presentations of their findings to the 7 American Association of Public Opinion Research (AAPOR), PTX-158, and to the National 8 Advisory Committee on Racial, Ethnic, and Other Populations, PTX-326. 9 United States District Court 21. CSM researchers summarized the respondent confidentiality concerns they 22. During the pretesting studies conducted in 2017, CSM researchers “noticed a 10 recent increase in respondents spontaneously expressing concerns about confidentiality” and 11 “reported that respondents’ fears, particularly among immigrant respondents, have increased 12 markedly this year.” PTX-157 at 1. 13 23. For example, CSM researchers observed Spanish-speaking respondents who 14 were “uncomfortable ‘registering’ other household members,” who “left three or four 15 roomers off the roster” and “mentioned being worried because of their ‘[immigration] 16 status,’” and who stated that “the Latino community will not sign up because they will think 17 that Census will pass their information on and people can come looking for them.” Id. at 2. 18 24. CSM researchers observed that “this level of deliberate falsification of the 19 household roster, and spontaneous mention of concerns regarding negative attitudes toward 20 immigrants, is largely unprecedented in the usability interviews that CSM has been 21 conducting since 2014 in preparation for the 2020 Census.” Id. at 3. CSM researchers 22 worried that the concerns expressed by immigrant respondents might be “even more 23 pronounced” during the 2020 Census, because respondents are generally more willing to 24 participate in pretesting surveys “given that they are being paid a cash incentive for their 25 participation and [are] being interviewed by a researcher with whom they have established 26 rapport.” Id. 27 25. During focus group testing, respondents similarly expressed “fear of 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 14 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 15 of 126 1 deportation[] [and] concern about how the data are used[] and which agencies can see it,” 2 specifically asking whether the Department of Homeland Security (“DHS”) or Immigration 3 and Customs Enforcement (“ICE”) would have access to their data. PTX-326 at 9. ii. CBAMS Findings 4 5 26. The CBAMS is a survey of 50,000 households in a series of 42 focus groups 6 designed to inform the integrated partnership and communications program for the 2020 7 Census about the macro-environment. Census Bureau 30(b)(6) Dep. Vol. II 437:17-438:6; 8 New York Tr. 927:22-928:6 (Abowd). The Census Bureau finds CBAMS research 9 sufficiently reliable to provide actionable information for the integrated partnership and 10 United States District Court 11 communications program. Census Bureau 30(b)(6) Dep. Vol. II 438:7-11. 27. After Secretary Ross announced that the 2020 Census would include a 12 citizenship question, Census Bureau researchers began asking for feedback about the 13 question from 30 of the 42 focus groups, including Spanish-language groups. PTX-161 at 6; 14 see also New York Tr. 930:16-19 (Abowd). 15 28. The CBAMS found that in the Spanish-language (U.S. Mainland) focus groups, 16 the citizenship question was a “determining factor for participation.” PTX-153 at 22. 17 Although most participants said that they were not afraid to answer the citizenship question 18 because they are citizens or legal residents, they knew many others who would not 19 participate in the 2020 Census “out of fear.” Id. While all participants wanted to participate 20 in the 2020 Census, “fear of deportation outweighs any benefit.” Id. 21 29. The Census Bureau views the results of the Spanish-language focus groups with 22 respect to the citizenship question as “extremely problematic.” Census Bureau 30(b)(6) Dep. 23 Vol. II 450:16-451:1; New York Tr. 934:8-12 (Abowd). Other immigrant and non-white 24 groups raised similar concerns. New York Tr. 930:9-24, 938:22-939:17, 940:4-941:14 25 (Abowd). 26 30. Census Bureau researchers ultimately concluded that “[t]he citizenship question 27 may be a major barrier” to participation in the 2020 Census because respondents, including 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 15 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 16 of 126 1 citizens and legal residents, believed that the census’s purpose “is to find undocumented 2 immigrants” and because “[t]he political discourse is targeting their ethnic group.” PTX-465 3 at 43. 4 31. The CBAMS results suggest that the citizenship question is sensitive in the 5 current macro-environment and is a “major concern” for the Census Bureau’s efforts to 6 encourage participation in the 2020 Census within Hispanic communities. New York Tr. 7 944:7-24 (Abowd). Moreover, the increased sensitivity to the citizenship question that was 8 observed in the 2018 CBAMS results was likely not captured in Brown, et al.’s 5.8 percent 9 estimate, which was based on 2016 data. Id. at 944:25-945:4 (Abowd). 10 United States District Court 11 e. The Plaintiff Experts’ Testimony 32. The Plaintiff experts’ testimony further supports the conclusion that the 12 citizenship question will cause a greater differential decline in self-response rates than 13 estimated by Brown, et al. 14 33. Dr. Colm O’Muircheartaigh, professor in the Harris School of Public Policy and 15 senior fellow at the National Opinion Research Center (NORC) at the University of Chicago, 16 testified that he agrees with the Census Bureau research discussed above. Tr. 33:4-17; 17 18 19 20 21 22 23 24 25 26 27 145:15-166:15 (O’Muircheartaigh). Dr. O’Muircheartaigh also cited additional factors that will exacerbate the effects of the differential decline in self-response rates caused by the citizenship question on the ultimate enumeration. Id. at 166:16-174:20 (O’Muircheartaigh). 34. First, missing units in the Census Bureau’s Master Address File (MAF) contain a disproportionate number of immigrant and noncitizen households. Tr. 166:21-25 (O’Muircheartaigh). The MAF is the “first building block” of census data collection. Id. at 122:4-6. The MAF is constantly updated throughout the census-taking process. Tr. 803:23805.7 (Abowd). In general, the census is unlikely to count persons whose households do not appear on the MAF. Id. at 46:1-6 (O’Muircheartaigh). Dr. O’Muircheartaigh testified that social science research, including recent research on Mexican immigrants, has observed that the Census Bureau has particular difficulty identifying household addresses for immigrants 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 16 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 17 of 126 1 and noncitizens. Id. at 122:7-123:13, 124:7-17 (O’Muircheartaigh). To the extent that 2 immigrant and noncitizen households are not identified by the Census Bureau and included 3 in the MAF, and the residents of these households choose not to come forward to be counted 4 because of the citizenship question, such households and their residents will not be included 5 in the 2020 Census despite the Census Bureau’s NRFU efforts. Id. at 166:21-167:14 6 (O’Muircheartaigh). United States District Court 7 35. Second, respondents, especially those that live in households containing 8 noncitizens, may omit certain household members on the census questionnaire because of 9 fears generated by the citizenship question. Id. at 167:15-20 (O’Muircheartaigh). In 10 particular, the 2017 CSM research observed that Spanish-speaking respondents were 11 reluctant to provide a complete roster of household members. Id. at 147:18-148:16 12 (O’Muircheartaigh) (citing PTX-157). Dr. O’Muircheartaigh testified that such rostering 13 omissions are a particularly problematic form of nonresponse because “[t]he quality of the 14 census is fundamentally dependent on complete rostering of individuals within households,” 15 and “the census protocol has no mechanism for remediating such a response.” Id. at 147:10- 16 16, 148:8-149:9 (O’Muircheartaigh); Census Bureau 30(b)(6) Dep. Vol. II 397:19-399:2, 17 459:21-460:7. 18 36. Dr. O’Muircheartaigh credibly testified to the following conclusions relating to 19 the impact of the citizenship question on self-response: (1) current survey methodology 20 research, primarily by the Census Bureau, has observed that Latinos and immigrants hold 21 considerable fears about participating in the 2020 Census, (2) the citizenship question will 22 increase the Census Bureau’s misidentification of households as unoccupied, particularly 23 among Latinos and households with noncitizens, (3) the citizenship question will depress 24 self-response rates, particularly for Latinos and households with noncitizens, and the Census 25 Bureau’s conservative estimate is that the self-response rate for households containing a 26 noncitizen will be 5.8 percent lower than for all-citizen households, and (4) factors such as 27 rostering errors will exacerbate the difference in the effective self-response rates of 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 17 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 18 of 126 1 2 37. Dr. Matthew Barreto, a professor of political science and Chicano studies at the 3 University of California, Los Angeles, Tr. 366:13-17 (Barreto), similarly testified that 4 adding a citizenship question to the 2020 Census will reduce self-response rates, particularly 5 among immigrants and Latinos, Id. at 374:7-15 (Barreto). Dr. Barreto’s findings were based 6 on a comprehensive literature review of research publications and reports, including those 7 produced by the Census Bureau, related to response rates (as well as NRFU and imputation); 8 an original survey he fielded in which he asked people about whether they intend to 9 participate in the 2020 Census; and his expertise and years of experience implementing 10 surveys in Latino and immigrant communities. Id. at 375:18-376:4, 379:19-380:7 (Barreto) 11 (citing PTX-499). 12 United States District Court noncitizens versus citizens. Tr. 175:1-19 (O’Muircheartaigh). 38. Dr. Barreto identified three interrelated factors that affect survey participation: 13 (1) trust, (2) sensitive questions, and (3) the macro-environment in which the survey is 14 administered. Tr. 380:19-381:7, 383:13-16 (Barreto). Applying the literature on these factors 15 to the citizenship question, Dr. Barreto concluded that the citizenship question will cause a 16 significant decline in self-response rates on the 2020 Census because it is a sensitive 17 question that will exacerbate trust issues in the current macro-environment, particularly for 18 immigrants and immigrant-adjacent communities. Id. at 386:21-25, 411:5-14 (Barreto). Dr. 19 Barreto defined “immigrant-adjacent communities” as communities with mixed-status 20 households, where one family member is a U.S. citizen and another family member is not, 21 and communities in which residents would interact with immigrants daily at work, school, or 22 in other similar environments. Id. at 387:1-14 (Barreto). 23 39. A consistent finding in the social science research is that “if a potential 24 respondent does not trust the survey taker to keep their information confidential and not put 25 them at risk, then the survey respondent won’t participate in the survey at all.” Tr. 381:17-23 26 (Barreto). With regard to census participation specifically, Dr. Barreto observed that the 27 Census Bureau, particularly in Manuel de la Puente’s ethnographic studies of the 1990 and 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 18 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 19 of 126 1 2000 Censuses, found that “immigrant and undocumented populations in particular [] don’t 2 trust the federal government to fully protect or keep in confidence their information.” Id. at 3 385:3-19, 390:12-395:7 (Barreto) (citing PTX-308 and PTX-309), see also id. at 388:11- 4 389:8 (Barreto) (citing PTX-339). To break down the barriers he observed in his studies, Dr. 5 de la Puente recommended that the Census Bureau work with community groups to assure 6 them that the Census Bureau isn’t seeking information about respondents’ citizenship status. 7 Id. at 393:25-394:15 (Barreto). 8 United States District Court 9 40. Moving on to the second factor, Dr. Barreto testified that “a sensitive question is one that asks a respondent for some very personal information that they may be 10 uncomfortable revealing.” Tr. 383:2-6 (Barreto). Social science research suggests that survey 11 takers should “reduce unnecessary sensitive questions because they do create considerable 12 trust issues with respondents.” Id. at 383:10-12 (Barreto). Whether a question is sensitive 13 varies in different environments and contexts and across subpopulations. Id. at 383:13-20, 14 384:19-385:2 (Barreto). Dr. Barreto observed that the citizenship question is likely to be 15 most sensitive to “those who are closer to the immigrant experience or closer to [] immigrant 16 communities,” particularly “in the Latino community where there have been concerns over 17 immigration-related issues over the past few years.” Id. at 387:15-23 (Barreto). 18 41. The third factor, macro-environment, is “the context in which any survey is 19 being implemented,” including “the social and political environment, the atmosphere that is 20 present when the survey is being administered.” Id. at 395:11-19 (Barreto). A respondent 21 “may be more willing to participate if the context or the environment seems very agreeable 22 and welcoming, and they may be far less likely to participate if the environment seems 23 threatening or concerning.” Id. at 395:20-25 (Barreto). Dr. Barreto observed that social 24 science research has found that “[i]mmigrants and mixed-status households are likely to 25 avoid government contact when they suspect it is unsafe to participate.” Id. at 397:19-398:2 26 (Barreto). This observation holds true for a census with a citizenship question, because the 27 question will be asked in a macro-environment that is perceived by many immigrants to be 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 19 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 20 of 126 1 “threatening or negative.” Id. at 396:3-13 (Barreto). 42. To evaluate participation in the 2020 Census, Dr. Barreto conducted a large United States District Court 2 3 national survey that inquired about people’s attitudes and behaviors. Id. at 411:15-23 4 (Barreto). Within the scientific community, survey research is considered reliable and has 5 predictive value. Id. at 414:2-7 (Barreto). 6 43. Dr. Barreto conducted his survey on a sample of 6,309 respondents from across 7 the United States, including oversamples of Latinos nationwide and residents of the State of 8 California, the City of San Jose, and two border counties in Texas. Id. at 424:8-19 (Barreto). 9 Respondents were randomly chosen, and weighting was applied to balance out the 10 demographic characteristics of the sample. Id. at 415:19-418:12, 434:2-435:19 (Barreto). In 11 addition, the survey response rate—28.1 percent—was within the American Association of 12 Public Opinion Research (AAPOR) response rate guidelines (at least 20 to 30 percent) for 13 telephone surveys. Id. at 425:2-23 (Barreto). 44. Dr. Barreto set forth the results of his survey in a number of tables. PTX-499A, 14 15 PTX-863 through PTX-890. He estimates that, because of the citizenship question, census 16 response rates are likely to decline between 6.3 and 8.0 percent nationally and between 10.5 17 and 14.1 percent in the State of California. PTX-870; PTX-871; see also Tr. 457:17-458:3 18 (Barreto) (explaining PTX 870); id. at 461:9-20 (Barreto) (explaining PTX-871). The 19 nonresponse rate attributed to the citizenship question in California is statistically higher 20 than the nationwide average. Tr. 463:8-464:15 (Barreto) (explaining PTX-873).7 45. Based on the Census Bureau’s most current data, the average Latino household is 21 22 larger than the average non-Latino household. Id. at 1036:12-1037:6 (Abowd). By factoring 23 in the difference in average household size between Latino households and other households, 24 25 26 27 7 Removing California from the national self-response estimates reveals just how severe the differential decline in self-response rates are likely to be in California relative the rest of the country. Dr. Barreto estimates that the decline in self-response rates in the rest of the country, excluding California, is likely to be 6.5 percent, as compared to an estimated decline of 12.3 percent in California. PTX-871. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 20 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 21 of 126 1 Dr. Barreto estimated that Latinos would constitute approximately 35 percent (over 10 2 million) of the total number of persons (approximately 28 million) that would not self- 3 respond to the 2020 Census because of the citizenship question, far surpassing the rate of 4 Latinos in the national population (18 percent). Id. at 478:1-481:6 (Barreto) (explaining 5 PTX-880 and PTX-881). This evidence further supports the conclusion that Latinos will be 6 disproportionately affected by the citizenship question. Id. at 480:9-14 (Barreto). 46. Although Dr. Barreto’s study provides credible evidence that the inclusion of the United States District Court 7 8 citizenship question on the 2020 Census is likely to cause a decline in self-response rates 9 among certain demographic groups relative to the rest of the population, some aspects of the 10 survey design and methodology limit the weight the Court affords to this evidence. In 11 particular, Dr. Barreto asks in Question 2 whether respondents would participate in the 2020 12 census if the federal government were to include a citizenship question on the questionnaire. 13 Tr. 576:4-13.8 By contrast, Question 1 specifically referred to the Census Bureau as the 14 agency responsible for the census and asked respondents, without mentioning the citizenship 15 question, whether they would participate in the 2020 census. Id. at 440:13-441:2.9 It is 16 plausible that respondents who are more distrustful of the federal government writ large than 17 they are of the Census Bureau in particular may have responded negatively to Question 2 in 18 part because of this difference in terminology. While this hardly represents a glaring flaw, it 19 does diminish somewhat the weight to be afforded to the drop in willingness to respond 20 21 22 23 24 25 26 27 8 Question 2 of Dr. Barreto’s survey reads: “In 2020, the federal government is adding a new question to require you to list whether you, and every person in your household is a U.S. citizen, or not a citizen. With the addition of a citizenship question, will you participate and submit your household information, or not?” Tr. 441:4-11 (Barreto). 9 Question 1 read: “The Census is an official population count that is conducted every 10 years by the federal government. It requires all households to list the name, age, and race or ethnicity of every person living in the home and provide that information to the Census Bureau either online, by mail, or in-person with a census taker. The Census is required to keep this information confidential, and every single household in the country is required to participate. In March 2020 you will receive an invitation from the U.S. Census to fill out the census form. Do you plan to participate and submit your household information?” Tr. 440:13-441:2. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 21 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 22 of 126 1 United States District Court 2 between Question 1 and Question 2. 47. Defendants’ remaining arguments that the decline in self-response rates between 3 Question 1 and Question 2 should not be credited are unpersuasive. In particular, the fact that 4 a randomized controlled trial (“RCT”) may produce more accurate results than a survey does 5 not automatically render Dr. Barreto’s survey unreliable. See Tr. 874:10-19. 6 48. Ultimately, respondents’ increased reluctance to participate in the census 7 between Question 1 and Question 2 of Dr. Barreto’s survey provides credible evidence that 8 the addition of the citizenship question is likely to result in a significant decline in self- 9 response rates in California and within the Latino population relative to the public at large. 10 2. NRFU Will Not Remediate the Differential Decline in Self-Response Rates 11 49. In keeping with recent practice, the Census Bureau will implement a series of 12 NRFU operations to attempt to count the significant number of persons who do not self- 13 respond to the 2020 Census, UF 39-47, including the millions who will not self-respond 14 because of the citizenship question, PTX-22 at 6; PTX-160 at 42; New York Tr. 894:1-16 15 (Abowd). All available evidence indicates that at every NRFU stage, including the 16 imputation phase, the Census Bureau will be differentially less effective at counting 17 noncitizens and Latinos—the very subpopulations most likely not to respond to the 2020 18 Census because of the citizenship question. Tr. 175:20-218:6 (O’Muircheartaigh). 19 20 a. Hard-to-Count Populations 50. The Census Bureau has always struggled to count hard-to-count subpopulations, 21 including noncitizens and Latinos, even when the census count for the national population 22 has been fairly accurate. Tr. 57:17-60:8 (O’Muircheartaigh). For example, as measured in 23 the Census Bureau’s post-enumeration surveys, Hispanics have been differentially 24 undercounted compared to non-Hispanic whites in each of the last three censuses. Id. at 25 55:2-15, 56:11-57:5 (O’Muircheartaigh); UF 61-62. In the 2010 Census, Hispanics were 26 undercounted by 1.54 percent and non-Hispanic whites were overcounted by .84 percent, 27 resulting in a net differential undercount of Hispanics of 2.38 percent. Tr. 56:11-24; PTX- 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 22 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 23 of 126 1 211 at 18. In the 2000 Census, Hispanics were undercounted by .71 percent and non- 2 Hispanic whites were overcounted by 1.13 percent, resulting in a net differential undercount 3 of Hispanics of 1.84 percent. PTX-211 at 18. In the 1990 Census, Hispanics were 4 undercounted by 4.99 percent and non-Hispanic whites were undercounted by .68 percent, 5 resulting in a net differential undercount of Hispanics of 4.31 percent. Id. United States District Court 6 51. Hard-to-count subgroups include low-income persons, persons who do not live 7 in traditional housing, persons who do not speak English fluently or have limited English 8 proficiency, persons who have distrust in the government, racial and ethnic minorities, 9 renters, undocumented immigrants or recent immigrants, and young children. Tr. 1021:19- 10 1023:2 (Abowd); UF 59-60. Census Bureau research shows that there is “substantial 11 overlap” between these hard-to count subgroups and those households most likely not to 12 respond to the 2020 Census because of the citizenship question. Tr. 1023:3-7 (Abowd). 13 52. The Census Bureau has identified four primary obstacles to counting hard-to- 14 count subpopulations: that they are hard to locate, hard to contact, hard to persuade, and hard 15 to interview. Id. at 1023:8-24 (Abowd). For some hard-to-count subgroups, more than one of 16 these obstacles applies. Id. at 1024:7-13 (Abowd). Census Bureau research acknowledges 17 that these obstacles apply to those households most likely not to respond to the 2020 Census 18 because of the citizenship question. Id. at 1023:25-1024:6 (Abowd). b. The Census Bureau’s Partnership and Communications Program 19 20 53. The Census Bureau has developed a range of strategies to address the net 21 differential undercount of “hard-to-count” populations—including targeted marketing and 22 outreach efforts, partnerships with community organizations, deployment of field staff to 23 follow up with individuals who do not respond, and retention of staff with foreign language 24 skills. UF 64. 25 54. In the 2000 and 2010 Censuses, the Census Bureau designed and implemented 26 public advertising campaigns to reach hard-to-count immigrant communities, including 27 using paid media in over a dozen different languages to improve responsiveness, and 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 23 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 24 of 126 1 partnered with local businesses, faith-based groups, community organizations, elected 2 officials, and ethnic organizations to reach these communities and improve the accuracy of 3 the count. UF 65-66. 4 5 campaign, in tandem with the Census Bureau’s NRFU efforts, may mitigate the decline in 6 self-response rates in the 2020 Census. Tr. 798:6-12, 799:21-800:14 (Abowd). Yet there is 7 no evidence in the Administrative Record that Defendants’ planned integrated partnership 8 and communications campaign for the 2020 Census will significantly mitigate such a 9 differential decline in self-response rates. Dr. Abowd agreed that it is “highly unlikely” that 10 the integrated partnership and communications campaign can eliminate the negative effects 11 of adding a citizenship question. Id. at 980:3-11 (Abowd). 12 United States District Court 55. Defendants believe that a similar integrated partnership and communications 56. The Census Bureau also acknowledges that the “trusted partners” that it relies on 13 to convey the importance of participating in the census will have additional challenges 14 communicating that message if the 2020 Census includes the citizenship question. Census 15 Bureau 30(b)(6) Dep. Vol. II 451:21-452:4, 453:2-17; New York Tr. 937:16-23 (Abowd). 16 The CBAMS focus groups of Spanish-speaking respondents found that, “while there were 17 suggestions of trusted voices, there does not seem to be a single trusted voice that could 18 mitigate [respondents’] distrust of the government to uphold the promise of confidentiality.” 19 PTX-153 at 22. Dr. O’Muircheartaigh persuasively testified that this observation shows the 20 citizenship question will “reduce[] the potential impact of the positive input of constituency, 21 community, and association leaders” as these trusted voices attempt to convince their 22 constituents to participate in the 2020 Census. Tr. 153:1-154:9 (O’Muircheartaigh). 23 57. Census Bureau research has noted one messaging strategy that is reassuring to 24 Spanish-speaking respondents is to convey that “[n]one of the questions in this survey will 25 ask about immigration status” and that “[b]y law, [the respondent’s] answers cannot be 26 shared with Immigrations and Customs Enforcement.” PTX-158 at 16. Dr. Barreto similarly 27 observed that, consistent with the findings in Dr. de la Puente’s ethnographic studies, the 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 24 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 25 of 126 1 most effective way— indeed, perhaps the “only way”—to address confidentiality concerns 2 related to the citizenship question is “to assure respondents that no citizenship information is 3 being gathered” in the 2020 Census. Tr. 500:17-501:5 (Barreto). Neither the Census Bureau 4 nor trusted partners can offer such assurances because the citizenship question will be on the 5 2020 Census, unless the Census Bureau is instructed to remove it. Tr. 1052:8-12 (Abowd). 6 58. Moreover, despite the barriers to participation in the 2020 Census associated 7 with the citizenship question, the Census Bureau has not significantly increased its spending 8 on 2020 census outreach relative to that expended in 2010. Tr. 1024:18-1025:9 (Abowd). c. The Census Bureau’s NRFU Operations United States District Court 9 10 59. The Census Bureau’s NRFU workload includes all households that do not 11 initially self-respond to the census. Tr. 851:16-852:2 (Abowd). In the 2010 Census, over 27 12 percent of the persons enumerated were in the NRFU workload. PTX-211 at 32-33 13 (subtracting from the U.S. total population (300,703,000) those persons not in any NRFU 14 universe (219,207,000) and dividing by the total population). The NRFU workload for the 15 2020 Census is expected to rise to between 34.5 and 44.5 percent of the total population. 16 PTX-1 at 172. 17 60. The Census Bureau’s best conservative estimate is that adding a citizenship 18 question to the 2020 Census will increase the NRFU workload by 2.09 million households 19 and 6.5 million persons. PTX-160 at 42. 20 61. Based on his survey data, Dr. Barreto estimated that adding a citizenship 21 question to the 2020 Census will increase the NRFU workload by at least 28 million persons, 22 and that Latinos will be disproportionately represented in that workload. PTX-880; Tr. 23 480:5-14 (Barreto). 24 62. The Bureau’s NRFU operations are designed to obtain an accurate count—and 25 thus, to prevent an undercount—at the national level. Tr. 918:11-16 (Abowd). In recent 26 censuses, however, the Bureau’s NRFU operations have been less effective at counting some 27 subpopulations than others. Tr. 178:7-23 (O’Muircheartaigh). 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 25 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 26 of 126 1 2 that adding a citizenship question will increase the net differential undercount of any 3 subpopulation, after accounting for NRFU operations. Tr. 918:21-24 (Abowd). Dr. Abowd 4 admitted, however, that it is “highly unlikely” that the Census Bureau’s NRFU operations 5 will eliminate a differential undercount in the 2020 Census. Id. at 980:12-981:2. 6 64. The Census Bureau’s NRFU operations for the 2020 Census include in-person 7 follow-up enumeration, proxy enumeration, administrative record enumeration, and 8 imputation by other methods. UF 39-46; Tr. 176:13-177:20 (O’Muircheartaigh). The Census 9 Bureau’s NRFU operations for the 2010 Census included these same processes, with the 10 exception of administrative record enumeration, which was used only on an experimental 11 basis in 2010. Census Bureau 30(b)(6) Dep. Vol. II 400:19-401:21. 12 United States District Court 63. Dr. Abowd testified that he is unaware of any “credible quantitative evidence” 65. The weight of the evidence ultimately shows that these NRFU efforts are 13 unlikely to mitigate significantly the differential decline in self-response caused by the 14 citizenship question and may in fact exacerbate the problem. See Tr. 217:4-218:5 15 (O’Muircheartaigh). i. In-Person Follow-Up Enumeration 16 17 66. The Census Bureau has repeatedly acknowledged that “[t]hose refusing to self- 18 respond due to the citizenship question are particularly likely to refuse to respond in NRFU 19 as well.” PTX-25 at 4; see also PTX-160 at 41, 42 n.59 (“Households deciding not to self- 20 respond because of the citizenship question are likely to refuse to cooperate with 21 enumerators coming to their door . . . .”). 22 67. Although in-person follow-up enumeration is typically more effective than mail 23 solicitation, “in this case for this population, the level of threat embodied by a federal agent 24 arriving at your residence to collect the information is far greater than the threat that might 25 be implied by a piece of paper [] that arrives at your residence.” Tr. 190:2-10 26 (O’Muircheartaigh). 27 68. Given these conditions, the enumeration errors that will result “may not be 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 26 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 27 of 126 1 avoidable simply by spending more money on fieldwork. Once a household decides not to 2 cooperate, it may not be possible to obtain an accurate enumeration no matter how many 3 times an enumerator knocks on their door.” PTX-160 at 43 n.60; see also Tr. 190:20-191:21 4 (O’Muircheartaigh). 5 69. Recent data from ACS in-person follow-up enumeration efforts, specifically the 6 Computer-Assisted Personal Interviewing (CAPI) operation, underscores the challenges that 7 enumerators will face in the 2020 Census if, like the ACS, the census includes a citizenship 8 question. Census Bureau 30(b)(6) Dep. Vol. I 124:19-133:17; Tr. 178:24-185:19 9 (O’Muircheartaigh) (describing PTX-138). The data, which was collected between 2010 10 through 2016, is consistent with the notion that questions on citizenship have become more 11 sensitive since 2010. Census Bureau 30(b)(6) Dep. Vol. I 131:4-11. 12 70. The CAPI data exhibit the following trends: (1) in-person follow-up enumeration 13 has been less effective over time in all census tracts, (2) in-person follow-up enumeration has 14 been differentially less effective in census tracts with a higher proportion of households 15 containing a noncitizen, and (3) the differential between census tracts with a higher 16 proportion of households containing a noncitizen and census tracts with a lower proportion 17 of households containing a noncitizen has grown over time. Census Bureau 30(b)(6) Dep. 18 Vol. I 129:22-130:4, 131:4-18, 133:8-17, Tr. 180:17-181:3 (O’Muircheartaigh). 19 71. The most recent CAPI data—from 2016—for the half of the population with a 20 higher proportion of households containing a noncitizen indicate that in-person follow-up 21 enumeration was 86.63 percent successful. Tr. 183:21-185:9 (O’Muircheartaigh). This rate 22 “is an approximate representation of how . . . such households might behave in the context of 23 the census.” Id. at 185:10-19 (O’Muircheartaigh). Indeed, the success rate was lower (and 24 conversely, the non-interview rate was higher) for in-person follow-up enumeration in the 25 2016 End-to-End Test and the 2018 End-to-End Test. Id. at 186:19-187:14 (describing PTX- 26 482 at 26). 27 72. None of the testing that has been used to plan NRFU staffing levels, the number 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 27 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 28 of 126 1 of field offices, enumerator training, NRFU protocols, or census questionnaire assistance has 2 accounted for a citizenship question on the 2020 Census. Census Bureau 30(b)(6) Dep. Vol. 3 I 198:2-10, 200:9-201:10. Although the Census Bureau’s NRFU operations were used in the 4 2018 End-to End Test, Tr. 819:15-820:9 (Abowd), it did not include a citizenship question, 5 Census Bureau 30(b)(6) Dep. Vol. I 225:13-16; Tr. 820:14-15 (Abowd). 6 7 2018 End-to-End Test. Tr. 820:19-23 (Abowd). But a U.S. Government Accountability 8 (“GAO”) report on NRFU implementation in connection with that test “raises some serious 9 concerns.” Tr. 98:3-8 (O’Muircheartaigh) (describing PTX-482). 10 United States District Court 73. The Census Bureau considers the NRFU operations to have been a success in the 74. That the Census Bureau did not determine the procedures for late-NRFU data 11 collection until after it started work, for example, “seriously undermines the potential of the 12 activity to be successful.” Id. at 98:9-99:7. (O’Muircheartaigh) (describing PTX-482 at 11). 13 This finding, in combination with similar findings that the field workforce was unprepared 14 for certain enumeration challenges, id. at 99:8-100:15 (O’Muircheartaigh), and lacked 15 adequate training, id. at 186:9-18 (O’Muircheartaigh) (describing PTX-482), led Dr. 16 O’Muircheartaigh to conclude that the report was “a little disturbing.” Id. at 101:9-12 17 (O’Muircheartaigh). These findings “cast[] doubt on . . . any projections that the Census 18 Bureau has about how successfully it will operate in 2020, compared, for example, to 2010.” 19 Id. at 101:16-102:4 (O’Muircheartaigh). 20 21 ii. Proxy Enumeration 75. Locating a proxy respondent—a neighbor, landlord, postal worker, or other 22 knowledgeable person who will provide information about another household—is generally 23 not easy. Tr. 195:2-10 (O’Muircheartaigh). The Census Bureau expects that, just as with in- 24 person follow-up enumeration, in census tracts with a higher proportion of households 25 containing a noncitizen, the proxy enumeration rate will be lower than in other tracts. Census 26 Bureau 30(b)(6) Dep. Vol. II 386:2-15; Tr. 196:25-197:6 (O’Muircheartaigh). 27 76. In other words, the challenge of finding willing proxy respondents will be greater 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 28 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 29 of 126 1 in neighborhoods with households that are “fearful of the Administration and fearful of 2 Census.” Tr. 195:13-25 (O’Muircheartaigh). Potential proxy respondents will be “less likely 3 to want to cooperate” if they are concerned about reporting undocumented immigrants. Id. at 4 521:15-522:2 (Barreto). Given that “reference persons are much less likely to answer the 5 citizenship question for nonrelatives in the household than for themselves . . . they may be 6 even less likely to answer it for neighbors.” PTX-160 at 43; Census Bureau 30(b)(6) Dep. 7 Vol. II 386:16-387:10; Tr. 523:3-17 (Barreto). 8 United States District Court 9 77. Even if located and willing to provide a response, proxy respondents generally provide lower quality enumeration data than self-responses. Census Bureau 30(b)(6) Vol. II 10 382:17-21; PTX-22 at 6; Tr. 931:14-24, 951:11-14 (Abowd). For example, in the 2010 11 Census, 97.3 percent of self-responses resulted in a correct enumeration, but the correct 12 enumeration rate for proxy responses was just 70.2 percent. PTX-160 at 42 (citing PTX-211 13 at 33); Tr. 197:14-198:5 (O’Muircheartaigh). 14 78. Proxy responses are particularly inaccurate for persons in tenuous residential 15 arrangements—a subpopulation that is disproportionately made up of Latinos and 16 immigrants. Tr. 198:6-200:4 (O’Muircheartaigh). Because of the nature of these living 17 arrangements—which include, for example, converted garages—proxy respondents “may 18 not actually know how many people live there.” Id. at 522:3-8 (Barreto). 19 79. Census Bureau research has also found that “proxies supply poor quality 20 individual demographic and socioeconomic information about the person on behalf of whom 21 they are responding.” PTX-160 at 41-42; Tr. 200:23-201:8 (O’Muircheartaigh); id. at 937:6- 22 19 (Abowd). Dr. Abowd conceded that the increased use of proxy responses “does impact 23 data quality,” including the quality of characteristic data. Id. at 887:13-24 (Abowd). 24 25 iii. Administrative Record Enumeration 80. Census Bureau research has observed that the quality of administrative records 26 varies depending on the subpopulation. Tr. 204:18-205:3 (O’Muircheartiagh) (describing 27 PTX-288). More specifically, the Bureau is less likely to be able to use administrative 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 29 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 30 of 126 1 records to enumerate hard-to-count subpopulations, including noncitizens and Hispanics. 2 Jarmin Dep. 285:1-286:20; Tr. 948:7-949:12 (Abowd), 205:4-12 (O’Muircheartaigh). 3 Undocumented immigrants are particularly unlikely to be found in administrative records 4 and will be harder to enumerate using such records. Census Bureau 30(b)(6) Dep. Vol. II 5 391:4-19; Tr. 205:13-17 (O’Muircheartaigh). Accordingly, the Census Bureau does not 6 expect administrative record enumeration to be as successful with noncitizens as with 7 citizens. Census Bureau 30(b)(6) Dep. Vol. II 391:21-392:4. 81. Similarly, the Census Bureau will be unable to link Hispanics to administrative 8 9 10 records at as high a rate as it can link non-Hispanic whites. Census Bureau 30(b)(6) Dep. Vol. II 389:12-390:5. 82. Given the inability of the Census Bureau to use administrative records to count United States District Court 11 12 the very subpopulations most likely not respond to the 2020 Census because of the 13 citizenship question, administrative record enumeration will not remediate the differential 14 decline in self-response rates and may indeed exacerbate any differential undercount of 15 noncitizens and Latinos. Tr. 206:4-19 (O’Muircheartaigh). iv. Imputation 16 83. If the Census Bureau is unable to enumerate a household through other NRFU 17 18 operations, it will impute, or model, the number of persons in the household and their 19 characteristics. Tr. 942:17-20 (Abowd). In the decennial census, the Bureau uses “count 20 imputation” to impute the size of the household, and “whole-person imputation” to impute 21 both the size of the household and the characteristics of the people in the household. Id. at 22 892:10-15 (Abowd); PTX-22 at 5. 84. The Census Bureau concedes that whole-person imputations “are not very 23 24 accurate.” Census Bureau 30(b)(6) Dep. Vol. I 253:7-15. 85. The Census Bureau anticipates that there will be 1.477 million more whole- 25 26 person imputations in the 2020 Census because of the citizenship question. PTX-160 at 42- 27 43. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 30 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 31 of 126 1 86. The Census Bureau has not finalized the algorithms it will use for count 2 imputation in the 2020 Census. Tr. 892:16-19 (Abowd). The accuracy of the Census 3 Bureau’s imputation model “is unknown at this time.” PTX-160 at 44. The Census Bureau 4 has recognized that any attempt to use imputation to count nonresponding persons “will be 5 challenging due to the fact that nonresponse is highly correlated with citizenship.” Id. 6 7 imputation model that imputes missing households based on nearby households that the 8 Census Bureau has counted and believes are similar in size, location, and other 9 characteristics. Tr. 892:10-893:11 (Abowd), id at 208:19-209:12 (O’Muircheartaigh). 10 United States District Court 87. As in previous censuses, the Census Bureau expects to use a “hot-deck” 88. Because hot-deck imputation fills in missing data based on data that the Census 11 Bureau has already collected, it is not neutral; by definition, it over-represents the household 12 characteristics of the known population, in which those most likely not to respond to the 13 citizenship question—in particular, noncitizens and Latinos—are underrepresented. Id. at 14 210:14-211:6, 211:24-212:6 (O’Muircheartaigh). Dr. Abowd confirmed that hard-to-count 15 subpopulations will be imputed at a greater rate than the rest of the population. Id. at 981:8- 16 13 (Abowd). 17 89. The Census Bureau’s imputation model also fails to account for the larger 18 household size, on average, of Hispanic households compared to other households. Tr. 19 528:1-24 (Barreto), 1036:25-1037:6 (Abowd). The Census Bureau’s imputation model is 20 built on the assumption that household size is “ignorable” missing data—that it is not 21 correlated with nonresponse. Id. at 525:10-13 (Barreto). Given that those persons most likely 22 not to respond to the 2020 Census because of the citizenship question, however, tend to 23 come from larger households, household size is, in fact, “non-ignorable” data. Id. at 528:1- 24 11 (Barreto); see also Tr. 1036:12-1037:6 (Abowd) (noting that, based on the Census 25 Bureau’s most current data, the average Latino household is larger than the average non- 26 Latino household). The result is bias in the Census Bureau’s imputation model. Id. at 27 985:10-14 (Abowd), 525:20-25 (Barreto). 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 31 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 32 of 126 United States District Court 1 90. Based on his survey data, Dr. Barreto presented quantitative evidence that the 2 Census Bureau’s imputation model will systematically undercount nonresponding 3 households. Id. at 529:14-530:6 (Barreto). The data reveal that, on a national level, 4 households that will not respond to a census with the citizenship question are larger, on 5 average, than households that will respond, and that in California, the gap between these 6 groups expands. Id. at 529:21-25, 530:7-9 (Barreto) (describing PTX-888 and PTX-889). 7 91. Dr. Barreto also constructed an imputation model based on the Census Bureau’s 8 2010 imputation model, as described in PTX-344, the Bureau’s J-12 memorandum. Like the 9 Census Bureau’s imputation model, Dr. Barreto’s version predicted the household size of 10 nonresponding households based on their 20 nearest neighbors, with controls for such factors 11 as housing type, geographic proximity, and household demographics. Tr. 535:3-14 (Barreto). 12 Because Dr. Barreto’s survey data contained the household size of each nonresponding 13 household, he was able to compare the imputation model’s predicted household size to the 14 actual size of these households. Id. at 535:15-19 (Barreto). 15 92. Dr. Barreto’s imputation analysis suggests that the Census Bureau’s imputation 16 model is likely to under-impute the household size of Latinos that do not respond to the 2020 17 Census because of the citizenship question at a rate of three-quarters of a person per 18 household on average, as compared to similarly-situated Latino households that will respond 19 to the census. Id. at 538:22-539:6 (Barreto) (describing PTX-468). 20 93. This evidence, suggests that imputation will not remediate the differential 21 decline in self-response rates. Id. at 540:24-541:8 (Barreto), 212:9-17 (O’Muircheartaigh). In 22 sum, the relative ineffectiveness of the Census Bureau’s NRFU operations with respect to 23 individuals who are unlikely to self-respond as a result of the citizenship question inevitably 24 leads to the conclusion that the NRFU process is unlikely significantly to mitigate the 25 disproportionate effect of the citizenship question on Latino and noncitizen households. 26 3. Inclusion of the Citizenship Question on the 2020 Census Will Result in a Differential Undercount of Noncitizens and Latinos 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 32 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 33 of 126 1 2 that the citizenship question will cause a net differential undercount of noncitizens and 3 Latinos relative to all-citizen households. 4 95. The Census Bureau concedes, based on its own natural experiment, that the 5 citizenship question will cause the self-response rate of noncitizen households to decline at 6 least 5.8 percent. Part III.B.1.c, supra. The Census Bureau has also produced considerable 7 qualitative research suggesting that the citizenship question will cause an even larger 8 differential decline in the self-response rate of noncitizen households, and that these negative 9 effects of the citizenship question will extend to other subpopulations, such as Hispanics. 10 11 United States District Court 94. The weight of the evidence, both qualitative and quantitative, strongly suggests Part III.B.1.d, supra. 96. Dr. Barreto produced quantitative evidence that is consistent with the Census 12 Bureau’s research. Dr. Barreto’s survey results further suggest that the citizenship question 13 will cause a decline in the self-response rate in California that will be greater than the decline 14 in the nation as a whole. Part III.B.1.e, supra. 15 97. In all recent censuses, the Census Bureau has differentially undercounted hard-to 16 count subpopulations, most notably Hispanics, even after implementing all NRFU 17 operations. Part III.B.2.a, supra. 18 98. The persons most likely not to self-respond to the citizenship question are also 19 some of the most unlikely to be counted at every NRFU stage—in-person follow-up 20 enumeration, proxy enumeration, administrative record enumeration, and imputation by 21 other methods. Part III.B.2.c, supra. 22 99. In addition, the Census Bureau’s NRFU operations are not designed to count 23 persons that are missing from the MAF or are left off the roster by a family member or proxy 24 respondent. Part III.B.1.e, supra. If such persons do not self-respond to the 2020 Census 25 because of the citizenship question, they will not be counted. Moreover, noncitizens and 26 immigrants, particularly Mexican immigrants, are more likely to live in housing that is 27 missing from the MAF. Part III.E.1.b. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 33 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 34 of 126 1 2 substantial net differential undercount of noncitizens and Latinos. 3 4. The Citizenship Question Will Harm Data Quality 4 101. Harm to the quality of census data is something the Census Bureau “tr[ies] to 5 avoid.” New York Tr. 953:18-20 (Abowd). It is undisputed, however, that the citizenship 6 question will damage the quality of characteristic data collected through the 2020 Census, 7 separate and apart from the damage to the count. These characteristics include gender, age, 8 race, and ethnicity. Tr. 1001:17-24 (Abowd). The damage to data quality will also cause 9 some people to be counted in the wrong place, including in the wrong area of a municipality, 10 United States District Court 100. In sum, it is more likely than not that the citizenship question will cause a or even in the wrong state. Id. at 1003:5-16 (Abowd). 11 102. In the January 19 Memo, the Census Bureau concluded that adding a 12 citizenship question to the 2020 Census will have an adverse impact on the quality of the 13 data collected by the census. PTX-22 at 4. Because the citizenship question will lower self- 14 response rates, the NRFU workload will increase, which will “degrade data quality because 15 data obtained from NRFU have greater erroneous enumeration and whole-person imputation 16 rates.” Id. at 5. One reason that data quality will suffer is that data collected during NRFU 17 are “much more likely to be collected from a proxy rather than a household member and, 18 when they do come from a household member, that person has less accurate information than 19 self-responders.” Id. at 6. 20 103. In the March 1 Memo, the Census Bureau similarly concluded that a citizenship 21 question will reduce data quality. PTX-25 at 4. The Brown, et al. Memo reached the same 22 conclusion. PTX-160 at 54. Dr. Abowd’s testimony confirms that adding the citizenship 23 question will damage the quality of the data collected in the 2020 Census. New York Tr. 24 885:17-21 (Abowd). Dr. Abowd observed that data produced by lower self-response rates is 25 less accurate than data produced by higher self-response rates. Id. at 881:19-882:5 (Abowd). 26 Likewise, data produced by self-response is much more “reliable” than data produced by 27 NRFU efforts. Id. at 953:2-14 (Abowd); Tr. 942:21-943:2 (Abowd). Therefore, by 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 34 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 35 of 126 1 decreasing self-response rates and thus increasing reliance on NRFU efforts, the citizenship 2 question will reduce the quality and accuracy of data produced during the 2020 Census. New 3 York Tr. 881:19-882:5, 952:23-953:14 (Abowd); Tr. 934:16-935:1, 1001:17-24 (Abowd). 4 5 census would result in fewer persons being linked to administrative records, New York Tr. 6 969:2-23, 979:16-20, 981- 17-19 (Abowd), which would reduce data quality, id. at 981:20- 7 25. In contrast, using administrative records to provide DOJ with block-level CVAP data 8 without adding the citizenship question to the census would not harm the quality of the 9 census data. Id. at 958:5-18 (Abowd). 10 United States District Court 104. Dr. Abowd also acknowledged that inclusion of the citizenship question on the 105. Dr. O’Muircheartaigh, Dr. Barreto, and Dr. Habermann confirmed that adding a 11 citizenship question to the 2020 Census will harm the quality of the census data. Tr. 114:11- 12 15, 217:21-22 (O’Muircheartaigh); id. at 492:16-21 (Barreto); PTX-821 at ¶¶ 47-54, 68. This 13 is the consensus among scientists within and outside the Census Bureau. Tr. 114:11-15 14 (O’Muircheartaigh). 15 106. The increased degradation of data quality that results from adding a citizenship 16 question to the 2020 Census cannot be mitigated. Id. at 935:3-5, 950:6-13, 1001:25-1002:8 17 (Abowd). 18 5. Inclusion of the Citizenship Question on the Census Will Result in a Loss of Federal Funding to Several Plaintiffs 19 a. The California Plaintiffs 20 i. The State of California 21 107. The citizenship question is more likely than not to cause the State of California 22 to lose federal funding. This is because any measurable differential undercount of 23 households containing noncitizens will cause California to lose funding for its state-share 24 programs. Reamer Decl. ¶¶ 20, 74; Tr. at 676:1-2, 677:6-14 (Reamer). 25 108. Dr. Andrew Reamer, who is an expert in the relationship between census data 26 and federal financial assistance, Reamer Decl. ¶¶ 1-8, testified that a significant portion of 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 35 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 36 of 126 1 federal domestic financial assistance is distributed on the basis of statistics derived from the 2 decennial census, id. ¶ 10. 109. At least 320 federal domestic assistance programs used census-derived data to United States District Court 3 4 distribute about $900 billion in FY2016. Id. ¶ 10. Of these, there are 24 large federal 5 financial assistance programs with geographic allocation formulas that rely in whole or part 6 on census‐derived data. Id. ¶¶ 10-11, Ex. D (PTX-245); Tr. 668:12-669:9 (Reamer); see also 7 UF 52-56. Eighteen of these 24 programs are “state-share” programs, in that they rely in 8 whole or in part on state share of a U.S. population total. Reamer Decl. ¶¶ 11, 17, Ex. D 9 PTX-245).10 10 110. As previously discussed, the citizenship question will cause a differential 11 undercount of persons living in households containing noncitizens. This, in turn, will lead to 12 a differential undercount of the population of states that have a disproportionate number of 13 such persons, like California. See id. ¶ 17. This will impact a number of federal domestic 14 financial assistance programs with census-tied geographic allocation formulas. Id. ¶¶ 16-18, 15 74. 111. Specifically, a differential undercount in the decennial census among persons 16 17 who live in households containing noncitizens will lead to measurable fiscal losses across 18 numerous federal programs for states with population percentages of households containing 19 noncitizens that are above the national average, including California. Id. ¶¶ 17-18. 112. Dr. Reamer performed calculations using two alternative projections of the 20 21 22 23 24 25 26 27 10 The 18 state-share programs include: Federal Transit Formula Grants, Community Block Development Grants/Entitlement Grants, Crime Victim Assistance, Title I Grants to Local Educational Authorities (LEAs), Special Education Grants, Head Start, Supplemental Nutrition Program for Women, Infants, and Children (WIC), Child Care and Development Block Grants, Supporting Effective Instruction State Grants, Workforce Innovation and Opportunity Act (WIOA) Youth Activities, Rehabilitation Services: Vocational Rehabilitation Grants to the States, Unemployment Insurance administrative costs, Block Grants for Prevention and Treatment of Substance Abuse, Social Services Block Grants, Career and Technical Education—Basic Grants to States, WIOA Disclosed Worker Formula Grants, Special Programs for the Aging, Title III, Part C, Nutrition Services. Reamer Decl. ¶¶ 11, 17, Ex. D (PTX-245). 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 36 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 37 of 126 1 potential undercount of households containing at least one noncitizen resulting from the 2 addition of a citizenship question, which undercount scenarios were applied to projections of 3 the 2020 population by state. Id. ¶ 14. These projections were prepared by Plaintiff expert 4 witness Dr. Bernard Fraga, id. ¶¶ 15, 35, and are discussed in greater detail in Parts III.B.6 5 and III.C.1.b, infra. 113. The two scenarios involve: (1) an undercount of 5.8 percent of households 6 7 containing at least one noncitizen, and (2) using the same starting point but assuming 86.63 8 percent of these households are ultimately counted successfully through NRFU efforts. Id. 9 ¶ 36. United States District Court 10 114. Dr. Reamer calculated the specific financial impact of these projections on three 11 of the 18 state-share programs—Title I grants to local educational agencies, the 12 Supplemental Nutrition Program for Women, Infants, and Children grants, and Social 13 Services Block Grants—to illustrate certain losses that would occur in the event of a 14 differential undercount. Id. ¶¶ 17, 20, 33, 37-40, 43-48, 52-53, 57-63; Tr. 667:8-19 15 (Reamer). 16 115. Under either undercount scenario, California, among other states, would lose 17 funding annually under all three programs. Reamer Decl. ¶¶ 49-50, 54-55, 64-65, and 18 accompanying charts. 19 116. Dr. Reamer’s conclusion that a differential undercount will result in lost 20 funding extends to the other 15 state-share programs he identified, meaning that California, 21 among other states, will lose population share and thus funding under these programs if the 22 citizenship question causes an undercount of individuals living in households containing 23 noncitizens. Id. ¶ 34; Tr. 678:18-679:2 (Reamer). 24 117. Dr. Reamer also opined that the magnitude of the impact varies depending on 25 the extent of the undercount. Id. ¶¶ 15-18. A change in the amount of the differential 26 undercount would impact only the magnitude of the loss to a state-share program, not the 27 existence of a loss. Id. ¶ 20. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 37 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 38 of 126 1 118. Similarly, a change in the funding level or allocation formula would impact 2 only the magnitude of the loss, not the existence of a loss, so long as the allocation formula 3 retains a degree of state-share-based calculation. Id. ¶ 19; Tr. at 669:24-670:11, 675:19-22 4 (Reamer). 5 119. Therefore, because the inclusion of the citizenship question on the 2020 Census 6 is likely to result in a significant differential undercount of households containing 7 noncitizens, the State of California is more likely than not to lose funding for its state-share 8 programs. Reamer Decl. ¶¶ 20, 74; Tr. at 676:1-2, 677:6-14 (Reamer). ii. LAUSD United States District Court 9 10 120. The citizenship question will also cause the Los Angeles Unified School 11 District (“LAUSD”) to lose federal funding. The funding for certain federal assistance 12 programs is distributed among localities within the state according to formulas prescribed by 13 law. Tr. at 677:23-678:10 (Reamer). For example, Title I grants are ultimately distributed to 14 local educational agencies, and grants authorized by the Workforce Innovation and 15 Opportunity Act (WIOA) are distributed to Local Workforce Development (“LWD”) areas. 16 Reamer Decl. ¶¶ 66, 67; see also id. ¶¶ 45, 71 (WIC and Community Development Block 17 Grants); Tr. 677:15-22, 678:11-13 (Reamer) (Community Development Block Grants and 18 WIOA grants). 19 121. Where there is a differential undercount of noncitizens within the locality that is 20 a funding recipient of a state-share program relative to the national population, those 21 localities will experience a loss of federal funding. Tr. 677:23-678:10 (Reamer). 22 122. For example, LAUSD, which has a higher-than average share of households 23 containing noncitizens than the state and national population, would incur a further decrease 24 in Title I funding when the funding received by California is distributed among the local 25 educational agencies within the state. Reamer Decl. ¶ 66 n.2; Escudero Decl. ¶¶ 16, 27; 26 Ryback Decl. ¶ 33. 27 b. San Jose Plaintiffs 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 38 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 39 of 126 123. Among the 18 “state share” programs discussed in paragraph 109, supra, are 1 2 grants authorized under WIOA, including the Youth Activities Program, 29 U.S.C. § 3163, 3 the Adult Activities program, 29 U.S.C. § 3173(b)(2)(A), and the Dislocated Workers 4 Program, 29 U.S.C. § 3173(b)(2)(B). Reamer Decl. ¶ 67; Tr. at 677:19-22; 678:11-13. Also 5 among the 18 “state share” programs are grants distributed via the Community Development 6 Block Grant (“CDBG”) Entitlement Program. Reamer Decl. ¶ 71; Tr. at 677:15-18. 124. Under WIOA, San Jose operates a workforce development program called 7 8 “work2future” that serves a LWD area composed of the cities of San Jose, Campbell, 9 Morgan Hill, Los Altos Hills, Gilroy, Los Gatos, Saratoga, and Monte Sereno, along with the 10 unincorporated areas of Santa Clara County. Melchor Aff. ¶ 2. 125. The cities that comprise work2future’s LWD have a combined population of United States District Court 11 12 1,243,043 residents, of whom at least 197,663, or 16.00%, are noncitizens. Judicially 13 Noticed Facts ¶ 14. 126. Of the 321,004,407 residents of the United States, 22,337,765, or 6.96%, are 14 15 noncitizens. Judicially Noticed Facts ¶ 5.11 127. Of the 38,982,847 residents of California, 5,250,604, or 13.47%, are 16 17 noncitizens. Judicially Noticed Facts ¶ 6. 128. Among the funding programs that use Bureau data are programs administered 18 19 by the Department of Labor under WIOA, which use Bureau data as part of the allocation 20 formulas set forth in 29 U.S.C. §§ 3162(C) and § 3172(C). UF 56. WIOA provides funding 21 to work2future under a two-part formula: first funding is delivered to a state (the “State 22 Allotment”) and the State of California distributes the State Allotment among the LWD’s 23 (the “Sub-State Allotment”). Melchor Aff. ¶ 5. 24 25 26 27 11 A summary of judicial noticed facts pertaining to the San Jose Plaintiffs can be found at ECF 180 in Case No. 18-cv-02279. The order granting their request for judicial notice may be found in ECF 176 of that same case. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 39 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 40 of 126 United States District Court 1 129. Based on his calculations regarding the three example programs, and the fact 2 that WIOA is one of the 18 “state share” programs he identified, Dr. Reamer concluded with 3 a high degree of certainty that California’s State Allotment under WIOA will be lower under 4 each of the scenarios set forth by Dr. Fraga because California’s percentage of noncitizens is 5 higher than the national average. Reamer Decl. ¶ 68. 6 130. Monique Melchor, Director of work2future, Workforce Development Board, 7 Office of Economic Development for the City of San Jose, is tasked with ensuring that the 8 program operates in compliance with federal law and regulations and to ensure that it is 9 properly funded. As part of her duties, she regularly uses the WIOA formula for calculating 10 the Sub-State Allotment for work2future’s LWD to ensure it was properly delivered. 11 Melchor Aff. ¶¶ 5, 8. Melchor uses publicly available Bureau data to make these 12 calculations, which are based on the LWD’s relative share of the total unemployed, the 13 relative share of the excess unemployed, and the local area’s share of disadvantaged adults or 14 youth. Id. ¶ 10. 15 131. Because the LWD, mainly including San Jose, has a higher percentage of 16 noncitizens than California as a whole, a differential undercount of noncitizens, according to 17 Dr. Barreto, will “be particularly severe in San Jose and other plaintiffs' jurisdictions.” Trial 18 Tr. 375:6-7; Judicially Noticed Facts ¶ 14. 19 132. If the population of the LWD (including its disadvantaged adults and 20 disadvantaged youth) is undercounted relative to the State of California, then the LWD’s 21 share of the Sub-State Allocation will decrease. Melchor Decl. ¶¶ 12-13. 22 133. Therefore, if there is a differential undercount of noncitizens in the 2020 23 Census, California will receive a lower State Share of WIOA funding, and the work2future 24 LWD, which includes San Jose, will receive a smaller proportion of the State Share in its 25 Sub-State Allocation, resulting in a double funding loss for the City of San Jose. 26 27 134. The CDBG program, administered by the U.S. Department of Housing and Urban Development (“HUD”), provides funding to eligible “entitlement communities” 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 40 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 41 of 126 1 including the City of San Jose. Reamer Decl. ¶ 71; Judicially Noticed Facts ¶ 16. One of the 2 programs that uses Bureau data is the Home Investment Partnership Program (“HOME”), 3 run by HUD, which uses Bureau data as part of its allocation formula under 42 U.S.C. 4 § 12747(b). UF 54. 5 135. The statutory formula for HOME grants is required to reflect “each 6 jurisdiction’s share of total need among eligible jurisdiction[s] for an increased supply of 7 affordable housing for very low-income and low-income families of different size, as 8 identified by objective measures of inadequate housing supply, substandard housing, the 9 number of low-income families in housing likely to be in need of rehabilitation, the costs of 10 producing housing, poverty, and the relative fiscal incapacity of the jurisdiction to carry out 11 housing activities eligible under section 12742 of this title without Federal assistance. 12 Allocation among units of general local government shall take into account the housing 13 needs of metropolitan cities, urban counties, and approved consortia of units of general local 14 government.” 42 U.S.C. § 12747(b)(1)(A). 136. One of the programs that uses Bureau data is the CDBG, run by HUD, which 15 16 uses Bureau data as part of its allocation formula under 42 U.S.C. § 5306(b). UF 55. This 17 program provides funds to entitlement communities according to a set of formulas prescribed 18 in law and that include data on population, poverty rates, and housing conditions. Reamer 19 Decl. ¶ 71. 137. The statutory formula for CDBG grants considers “the average of the ratios 20 21 between the population of that city and the population of all metropolitan areas; the extent of 22 poverty in that city and the extent of poverty in all metropolitan areas; and the extent of 23 housing overcrowding in that city and the extent of housing overcrowding in all metropolitan 24 areas.” 42 U.S.C. § 5306(b)(1)(A). 25 138. HUD awards the City of San Jose an annual allocation of CDBG and HOME 26 funding; the amount of this funding is directly tied to data from the Census. Clements Aff. 27 ¶ 10. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 41 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 42 of 126 1 139. Of San Jose’s 1,023,031 residents, 176,345, or 17.24%, are noncitizens. 2 Judicially Noticed Facts ¶ 11. Thus, San Jose’s percentage of noncitizens is nearly two-and- 3 a-half times the national percentage of 6.96%. It follows that San Jose is likely to be 4 undercounted relative to the population as a whole if the citizenship question is added to the 5 Decennial Census. Tr. 546:6-17 (Barreto). 6 140. Because CDBG is one of the 18 programs that Dr. Reamer identified as 7 sensitive to changes in population, and because Dr. Reamer concluded that any such program 8 would provide less funding to geographic areas that are undercounted relative to the 9 population as a whole, San Jose will receive less CDBG funding if is undercounted relative 10 to the nation as a whole. Reamer Decl. ¶ 18; Tr. at 677:7-14. Therefore, it is more likely than 11 not that San Jose will receive less CDBG funding if a citizenship question is added to the 12 Census. 13 14 15 141. San Jose’s Office of Emergency Management (“OEM”) also faces a substantial risk of losing funding based on the addition of the citizenship question. 142. In his role as Director of OEM, and in prior positions, Raymond Riordan has 16 applied for funding on behalf of San Jose from the Federal Emergency Management Agency 17 (“FEMA”). Supp. Riordan Aff. ¶ 2. 18 143. When applying for funding from FEMA, Riordan completes a “Preliminary 19 Damage Assessment” in accordance with FEMA guidelines as provided in its Damage 20 Assessment Operations Manual and its Preliminary Damage Assessment for Individual 21 Assistance Operations Manual. These manuals require Riordan to supply census data for 22 areas affected by a disaster. Id. ¶¶ 4-5. 23 144. Riordan has personally worked on funding proposals to FEMA that were denied 24 because FEMA determined not enough people lived in the affected area to qualify for 25 funding. Id. ¶¶ 7-10. Because the total number of individuals affected by a disaster is a key 26 factor in most applications for disaster funding, and because Riordan provides this number 27 based on data from the Census Bureau, a net undercount of San Jose’s population will 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 42 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 43 of 126 1 impede the City of San Jose’s ability to obtain adequate funding when the next disaster 2 occurs. Id. ¶ 14. Indeed, San Jose is in a region prone to natural disasters, including 3 earthquakes, floods, and fires. Id. ¶ 13. 4 6. Inclusion of the Citizenship Question on the 2020 Census Increases the Likelihood that California Will Lose Political Representation 5 145. Adding a citizenship question to the 2020 Census significantly increases the 6 likelihood that California will lose at least one congressional seat. Dr. Bernard Fraga 7 credibly testified that (1) California is expected to maintain its current level of congressional 8 representation (53 seats) if the 2020 Census does not ask a citizenship question, and (2) 9 adding a citizenship question to the 2020 Census increases the probability that California 10 will, contrary to its actual population, lose a congressional seat. 11 146. To estimate the quantitative effect of adding a citizenship question to the 2020 United States District Court 12 Census, Dr. Fraga looked at four scenarios of nonresponse and NRFU—two based on Dr. 13 Barreto’s survey data, and two based on Census Bureau data. Id. ¶ 26. For each scenario, Dr. 14 Fraga estimated how much of each state’s population would not be counted in the 2020 15 Census because of the citizenship question. Id. ¶¶ 57, 58. 16 147. Dr. Fraga estimated that, based on Dr. Barreto’s survey data and assuming 17 NRFU will not remediate the differential in self-response rates (Scenario A), the citizenship 18 question would cause 12.51 percent of Californians not to be reported in the census self19 response. Id. ¶¶ 57-58. 20 148. Dr. Fraga performed the same calculation based on the Census Bureau’s 21 estimate of a decline in nonresponse by 5.8 percent for noncitizen households—again 22 assuming NRFU will not have a mitigating affect (Scenario C). Id. ¶¶ 57, 60. Based on this 23 estimate, the citizenship question would cause 1.68 percent of Californians not to be reported 24 in the census self-response. Because California has a higher proportion of noncitizens than 25 any other state, this was also the highest proportional undercount of all the states. Id. ¶¶ 57, 26 65. Using either the survey results or the Census Bureau’s estimate, California will always 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 43 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 44 of 126 1 have the highest proportional undercount, as long as the Census Bureau’s follow-up efforts 2 are anything less than 100 percent effective. Id. ¶ 65. 149. Dr. Fraga used these different undercount estimates to quantify the impact of 3 4 adding a citizenship question on congressional apportionment, including the probability that 5 apportionment would be affected by the question. Id. ¶ 66. In the baseline scenario with no 6 citizenship question, California is projected to keep its current 53 seats in the House of 7 Representatives. Id. ¶ 75. Based on Dr. Barreto’s estimated self-response rates, however, Dr. 8 Fraga estimates that California is virtually certain to lose three seats if the citizenship 9 question is included in the census, assuming the NRFU process does not mitigate the effects 10 of this differential (Scenario A). Id. ¶¶ 73-74, 76.12 150. Dr. Fraga also concluded that, using the Census Bureau’s 5.8 percent estimate United States District Court 11 12 of nonresponse by noncitizen households and assuming NRFU has no effect on the net 13 differential undercount, the likelihood of California losing at least one seat nearly doubles to 14 fifty percent probability (Scenario C). Id. ¶ 82. Even applying a NRFU success rate of 15 86.63%, Dr. Fraga testified that the likelihood of California losing at least one seat still 16 increases by 15 percent (Scenario D). Id. ¶¶ 48, 82. The 86.63% NRFU success rate derives 17 from Dr. O’Muircheartaigh’s report and reflects the 2016 in person follow-up response rate 18 during the 2016 ACS in census tracts with a higher than average share of noncitizen 19 households. Id. ¶ 49. 20 7. Plaintiffs Have Had to Appropriate Funds to Mitigate the Harm of the Inclusion of the Citizenship Question on the 2020 Census 21 a. California Plaintiffs 22 i. State of California 23 151. The State of California has appropriated and will imminently spend increased 24 25 26 27 12 Defendants argue persuasively that the “NRFU simulation” portion of Dr. Barreto’s survey does not resemble the Census Bureau’s actual NRFU process. Accordingly, the Court declines to rely on Scenario B, which was based in part on this NRFU simulation. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 44 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 45 of 126 1 funds on census-related community outreach due to the citizenship question. Former 2 California Governor Jerry Brown initially proposed to the California Legislature for the FY 3 2018-19 state budget an appropriation of $40.3 million “to be spent over a three-year period 4 for statewide outreach and other activities related to the 2020 Census count.” UF 111. This 5 budget proposal was made prior to Secretary Ross’ issuance of the Decision Memo 6 announcing the addition of the citizenship question. See PTX-502 at 3. 7 8 included an appropriation of $90.3 million “to support the California Complete Count effort, 9 which was established within the Government Operations Agency to perform outreach 10 11 United States District Court 152. The final FY 2018-19 state budget that was enacted in the summer of 2018 focusing on hard-to-count populations for the decennial census.” UF 112. 153. Early on in the budget process, before Secretary Ross issued the Decision 12 Memo, the Legislative Analyst’s Office published an analysis of the census outreach budget 13 item. PTX-502. The analysis observed that the potential introduction of a citizenship 14 question to the 2020 Census could cause an undercount. Id. at 2-5 (noting that changes to the 15 census, including “the potential for a question about citizenship[,] raise the possibility of an 16 undercount in California in 2020”). 17 154. This concern was echoed in legislative committee materials and at least one 18 committee hearing. The legislative history of the FY 2018-19 state budget shows that one of 19 the driving forces behind the increased appropriation was the citizenship question. PTX-504 20 at 140 (summary of FY 2018-19 state budget includes section devoted to census outreach to 21 hard-to-count residents, and states that “[t]he Budget includes $90.3 million for statewide 22 outreach and other efforts related to increasing the participation rate of Californians in the 23 decennial census”); PTX-505 at 1 (description of FY 2018-19 state budget line items 24 references as a “major change” the $90.3 million allocated to “support the California 25 Complete Count effort . . . to perform outreach focusing on hard-to-count populations for the 26 decennial census”); PTX-506 at 8, 76 (Legislative Analyst’s overview of FY 2018-19 state 27 budget includes section describing $90.3 million allocated for outreach activities); PTX-509 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 45 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 46 of 126 1 at 23 (March 15, 2018 Senate Budget and Fiscal Review Subcommittee meeting staff report 2 on the California Complete Count—Census 2020 agenda item notes that concerns about 3 emphasis on Internet self-response, “in combination with the potential for a question about 4 citizenship[,] raise the possibility of an undercount in California in 2020”); PTX-510 at 41- 5 44 (April 24, 2018 Assembly Budget Subcommittee meeting staff report on 2020 Census 6 Outreach agenda item states that one change to the 2020 Census is that “[t]he federal 7 government has decided to include a citizenship question in the census, which is projected to 8 reduce the rate of response,” and identifies the citizenship question as one of the challenges 9 that would justify “additional resources” for outreach); PTX-517 at 45:19-46:12 (statement 10 by Assembly-member David Chiu at the April 24, 2018 Assembly Budget Subcommittee 11 meeting that the citizenship question presents “a different world” that may justify doubling 12 census outreach expenditures); PTX-512 at 24 (May 22, 2018 Senate Committee on Budget 13 and Fiscal Review meeting staff report on California Complete Count—Census 2020 agenda 14 item notes that “[d]ue to the significant changes to the census, providing state funding to 15 target hard-to-count populations is reasonable,” and “[d]ue to both the extreme importance of 16 an accurate census to the state and the high cost of the necessary outreach, additional funding 17 is warranted”); PTX-513 at 30-31 (May 24, 2018 Assembly Budget Subcommittee meeting 18 staff report on the 2020 Census Outreach Funding agenda item proposed $113 million 19 increase in census outreach funding, including $12 million for Los Angeles County’s 20 complete count efforts); PTX-514 at 79 (June 6, 2018 2018-19 Legislative Budget 21 Conference Committee meeting staff report on 2020 Census Outreach cites $153.3 million 22 request from Assembly and $135.3 million request from the Senate for outreach efforts); 23 PTX-515 at 173 (June 8, 2018-19 Legislative Budget Conference Committee meeting staff 24 report on 2020 Census Outreach recommends adopting compromise of $90.3 million 25 allocation for census outreach). 26 27 155. Since the appropriation, the California Complete Count Committee, the body in charge of the outreach efforts, has submitted reports to the Governor and Legislature that 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 46 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 47 of 126 1 underscore the challenge presented by the citizenship question to those outreach efforts. 2 PTX-508 at 4-5, 13, 19, 27-28 (October 2, 2018 report to Governor acknowledges that 3 citizenship question presents challenge to outreach efforts, describes formation of working 4 group on citizenship matters, observes that the question will generate fear, and identifies 5 possible difficulties of hiring trusted messengers); PTX-507 at 9-10 (October 1, 2018 report 6 to Legislature states that in convenings with local partners, “[p]articipants identified the most 7 significant barrier to achieving a complete count to be the Census citizenship question and 8 the current political environment regarding immigrants”). 9 156. The State’s allocation of outreach funding to the County of Los Angeles also 10 confirms increased expenditures due to the citizenship question. Baron Decl. ¶¶ 7-16. The 11 State initially allocated to the County $8.7 million in census outreach funding. Id. ¶ 7, 12. In 12 May 2018, the County requested an additional $3.3 million in funding specifically due to the 13 addition of the citizenship question on the Census. Id. ¶¶ 11-12 & Ex. A. The State met the 14 County’s request, in part. In November 2018, the State announced its County outreach 15 allocations, allocating $9,393,090 to the County of Los Angeles for 2020 Census outreach to 16 hard-to-count populations. UF 113; Baron Decl. ¶¶ 13-15. 17 18 19 b. San Jose Plaintiffs i. City of San Jose 157. San Jose’s population has been undercounted in prior censuses and San Jose is 20 taking steps to mitigate the likely undercount of its population which will be caused by the 21 inclusion of a citizenship question on the 2020 Census. Ruster Aff. ¶¶ 4, 11-19. San Jose, 22 along with other cities, has partnered with the County of Santa Clara to form a “Complete 23 Count Committee” to encourage participation in the Census by hard-to-count communities. 24 Id. ¶ 3, 8. Such partnerships among localities are encouraged by the Bureau to ensure an 25 accurate and complete count. Tr. 799:23-800:14 (Abowd). In fact, San Jose’s preparations 26 for the Census are being conducted in concert with the Bureau’s integrated partnership and 27 communication program. Id. ¶¶ 3-5, 11. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 47 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 48 of 126 1 2 performing outreach and expects to allocate approximately $300,000 more before the Census 3 is conducted. Id. ¶¶ 13, 15. San Jose has dedicated staff resources, including Jeff Ruster, the 4 Assistant Director in the Office of Economic Development, and a full-time consultant, to 5 prepare for the Census. Id. ¶¶ 3, 14, 19. 6 United States District Court 158. San Jose has dedicated approximately $300,000 in resources towards 159. Consistent with recommendations from the Census Bureau and the Department 7 of Commerce, these preparations include targeted outreach that is being performed 8 specifically because Secretary Ross has decided to add a citizenship question to the Census. 9 Id. ¶¶ 10-11, 16; Tr. at 1017:22-1018:17 (Abowd). Indeed, Dr. Abowd agreed that the 10 addition of the citizenship question has made it reasonable for cities to increase their 11 outreach expenditures to encourage participation in the census. Tr. at 979:16-25 (Abowd). 12 160. The targeted outreach being conducted by San Jose is designed specifically to 13 mitigate the impact that adding the citizenship question to the Census will have on hard-to- 14 reach populations in San Jose. Ruster Aff. ¶¶ 16, 17. This diversion of resources to address 15 the citizenship question will limit San Jose’s ability to use its resources to boost participation 16 among other parts of the population. Id. ¶ 19. 17 161. In light of the substantial risk posed by the addition of the citizenship question, 18 it is reasonable for San Jose to spend additional time and money on the outreach to address 19 concerns about the addition of the citizenship question. Tr. at 979:16-25 (Abowd). 20 21 ii. BAJI 162. BAJI’s Executive Director, Opal Tometi has reviewed and received feedback 22 from impacted communities regarding the 2020 Census, formed coalitions with other 23 immigration groups, spoken to members of BAJI, and participated in a number of events, 24 panels, and town halls where concerns about the addition of a citizenship question to the 25 2020 Census were raised. Additionally, Tometi has worked with Black immigrant, refugee, 26 and African American communities since 2010 and is familiar with the ways in which the 27 census count affects the well-being of these communities. Tometi Aff. ¶ 3. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 48 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 49 of 126 United States District Court 1 163. To ensure that BAJI’s members are properly counted in the 2020 Census, BAJI 2 plans to conduct additional outreach to these communities to encourage them to participate 3 in the 2020 Census questionnaire. Supp. Tometi Aff. ¶ 4. BAJI has determined that, due to 4 the citizenship question, such outreach will require the expenditure of additional resources 5 including money, staff time, and operational expenses. Id. To date, BAJI has expended 6 considerable staff time educating constituents, and other community members, about the 7 addition of a citizenship question to the Census. Tometi Decl. ¶ 20 (reaffirmed in Tometi 8 Aff. ¶ 2). 9 164. Given the nature of the census taking process, BAJI is reserving the majority of 10 the expenditure it will use to address the addition of the citizenship question—resources that 11 will likely be diverted from its other essential services—for its efforts to bolster census 12 participation among its members and other underrepresented minority communities who are 13 fearful about responding to the citizenship question. Id. (reaffirmed in Tometi Aff. ¶ 2). 14 Accordingly, BAJI expects to allocate at least an additional $200,000 in the next two years 15 to addressing the addition of a citizenship question to the Census and attempting to mitigate 16 its harmful effects. Id. (reaffirmed in Tometi Aff. ¶ 2). 17 165. BAJI has taken these actions because an undercount of noncitizens would 18 disproportionately affect BAJI members in that BAJI’s membership has a high proportion of 19 immigrants who live in immigrant-rich metropolitan areas. Id. ¶¶ 9-10 (reaffirmed in Tometi 20 Aff. ¶ 2). The impact of the addition of a citizenship question to the Census, and BAJI’s 21 diversion of its resources to address the same, has therefore impaired BAJI’s ability to carry out 22 its mission of fostering racial, economic, and social equality for Black immigrants and other 23 historically underrepresented communities. Id. ¶ 19 (reaffirmed in Tometi Aff. ¶ 2). 24 C. Conclusions of Law Related to Standing 25 The California Plaintiffs and the San Jose Plaintiffs ultimately succeed in 26 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 49 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 50 of 126 1 establishing standing with respect to the APA and the Enumeration Clause Claims.13 Once 2 the Court determines that at least one of the San Jose Plaintiffs and one of the California 3 Plaintiffs has standing, it need not consider the standing of the remaining parties. See 4 Leonard v. Clark, 12 F.3d 885, 888 (9th Cir. 1994) (“The general rule applicable to federal 5 court suits with multiple plaintiffs is that once [a] court determines that one of the plaintiffs 6 has standing, it need not decide the standing of the others”). Here, several plaintiffs across 7 both actions have standing. 8 1. Injury-in-Fact 9 To establish injury-in-fact, a plaintiff must demonstrate it “has sustained or is 10 immediately in danger of sustaining a direct injury” as a result of the challenged action. 11 Spokeo, 136 S. Ct. at 1552 (Thomas, J., concurring) (quoting Ex parte Levitt, 302 U.S. 633, 12 634 (1937) (per curiam)). The injury or threat of injury must be “concrete and particularized” 13 rather than conjectural or hypothetical. Lujan, 504 U.S. at 560. Thus, where standing is 14 based upon an alleged future injury, the plaintiff must demonstrate either that the harm is 15 certainly impending, or that there is a substantial risk the harm will occur. Susan B. Anthony 16 List v. Driehaus, 573 U.S. 149, 158 (2014). Injury-in-fact also exists where there is a 17 “substantial risk” that harm will occur, which prompts plaintiffs reasonably to incur costs to 18 mitigate or avoid that harm. Clapper v. Amnesty Int’l USA, 568 U.S. 398, 414 n.5 (2013). “For standing purposes, a loss of even a small amount of money is ordinarily an 19 20 ‘injury.’” Czyzewski v. Jevic Holding Corp., 137 S. Ct. 973, 983 (2017); see also Carpenters 21 Indus. Council v. Zinke, 854 F.3d 1, 5 (D.C. Cir. 2017) (“A dollar of economic harm is still 22 an injury-in-fact for standing purposes.”); Council of Ins. Agents & Brokers v. Molasky- 23 24 25 26 27 13 As previously explained, the San Jose Plaintiffs also advance a claim for relief under the Apportionment Clause. They do not, however, adequately establish standing with respect to the Apportionment Clause Claim as they have not shown how, even were California to lose congressional representation, that would impact San Jose in particular. This finding is of little practical consequence, however, given that the San Jose Plaintiffs successfully establish standing with respect to the Enumeration Clause Claim. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 50 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 51 of 126 1 Arman, 522 F.3d 925, 932 (9th Cir. 2008) (noting that Supreme Court has found injury-in- 2 fact even where magnitude of harm was only a few dollars). 3 4 injuries include, (1) the loss of federal funding, (2) an increased risk of losing political 5 representation, and (3) the expenditure of funds for census outreach to mitigate the 6 substantial risk of harm flowing from the citizenship question. 7 8 9 United States District Court The evidence establishes that Plaintiffs will be injured in at least three ways. These a. Loss of Federal Funding A plaintiff who is likely to lose federal funding due to a differential undercount of the population has suffered an injury-in-fact for the purposes of Article III standing. Carey v. 10 Klutznick, 637 F.2d 834, 838 (2d Cir. 1980); City of Detroit v. Franklin, 4 F.3d 1367, 1373- 11 1375 (6th Cir. 1993) (standing established where “census undercount will result in a loss of 12 federal funds” to plaintiffs’ city). Lost federal funding, no matter the magnitude, qualifies as 13 an injury-in-fact. Czyzewski, 137 S. Ct. at 983. 14 A significant portion of federal domestic financial assistance is distributed based on 15 census-derived data, including from 24 large federal financial assistance programs with 16 geographic allocation formulas that rely in whole or in part on census-derived data. If, as 17 Plaintiffs have shown, there is any measurable differential undercount of households 18 containing noncitizens, California will lose federal funding, because California has a larger 19 proportion of noncitizens relative to other states. 20 Similarly, some federal domestic financial assistance that is based on census-derived 21 data is distributed among localities within the state. If, as Plaintiffs have shown, there is any 22 measurable differential undercount of households containing noncitizens, LAUSD and San 23 Jose will also lose federal funding, because these localities have a larger proportion of 24 noncitizens relative to other localities. Having shown a “substantial risk” that the citizenship 25 question will cause them to lose federal funding, the State of California and LAUSD (of the 26 California Plaintiffs), and the City of San Jose (of the San Jose Plaintiffs) have each 27 established injury-in-fact. See Susan B. Anthony List, 573 U.S. at 158. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 51 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 52 of 126 1 2 A plaintiff’s “expected loss of a Representative to the United States Congress 3 undoubtedly satisfies the injury-in-fact requirement of Article III standing.” Dep’t of 4 Commerce v. U.S. House of Representatives, 525 U.S. 316, 331-332 (1999); Carey, 637 F.2d 5 at 838 (holding that a disproportionate undercount resulting in the loss of congressional 6 representation confers standing); City of New York v. U.S. Dep’t of Commerce, 713 F. Supp. 7 48, 50 (E.D.N.Y. 1989) (holding that a likely undercount of populations disproportionately 8 represented in plaintiff states conferred standing). 9 United States District Court b. Loss of Political Representation (State of California Only) The California Plaintiffs have shown that adding a citizenship question to the 2020 10 Census will cause a differential undercount of the State of California’s population relative to 11 other states, creating a substantial risk that California will lose its fair share of political 12 representation in Congress, and by extension, the Electoral College. Even using the most 13 conservative data and assumptions (Scenario D), Dr. Fraga concluded that the addition of the 14 citizenship question would increase the likelihood by 15 percent that California will lose a 15 congressional seat. See Part III.B.6. 16 This estimate is conservative. As previously discussed, Scenario D is based on the 17 Census Bureau’s estimate that the self-response rate in noncitizen households will decline by 18 5.8 percent relative to all-citizen households. The Census Bureau has acknowledged, and the 19 Plaintiff experts agree, that this is a conservative estimate. Accordingly, it is safe to conclude 20 that the true differential decline in self-response rates will in fact be higher than 5.8 percent. 21 Furthermore, the Census Bureau’s NRFU efforts are unlikely to mitigate this differential 22 decline in self-response rates. See Part III.B.2, supra. 23 In light of the considerable evidence in the record that the Census Bureau’s NRFU 24 will be differentially less effective in counting noncitizens and Latinos and therefore will not 25 significantly mitigate the effects of the citizenship question, Dr. Fraga’s estimates under 26 Scenario C are the most probative. Under Scenario C, Dr. Fraga once again adopted the 27 Census Bureau’s conservative estimate that the differential decline in self-response rates for 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 52 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 53 of 126 1 noncitizen households relative to all-citizen households will be 5.8 percent. Unlike Scenario 2 D, however, Scenario C relies upon the assumption that the Census Bureau’s NRFU 3 processes will not remediate the differential in self-response rate caused by the citizenship 4 question. Under this scenario, the likelihood that the state will lose a congressional seat nearly United States District Court 5 6 doubles from 26.1 percent to a 49.9 percent, creating a near 50-50 chance that California will 7 lose representation. Fraga Aff. ¶¶ 81-82. This evidence establishes that California faces a 8 substantial risk of losing at least one seat as a result of the citizenship question. Such 9 malapportionment of the State of California’s congressional representation is a “threat of 10 vote dilution” that “is ‘concrete’ and ‘actual or imminent, not conjectural or hypothetical.” 11 Dep’t of Commerce, 525 U.S. at 332 (quoting Whitmore v. Arkansas, 495 U.S. 149, 155 12 (1990). 13 14 15 c. Expenditure of Funds to Mitigate Harm i. The California Plaintiffs The State of California has reasonably increased its expenditures on census outreach 16 to attempt to mitigate the decline in self-response rates and the resulting differential 17 undercount of Plaintiffs’ residents caused by the citizenship question. These additional 18 expenditures, which constitute a direct injury to the State of California, are sufficient to 19 establish injury-in-fact for standing purposes. Clapper, 568 U.S. at 414 n.5 (standing may be 20 based on “reasonably incur[red] costs to mitigate or avoid” a “substantial risk” of harm); 21 Monsanto Co. v. Geertson Seed Farms, 561 U.S. 139, 153-155 (2010) (finding injury-in-fact 22 where alfalfa growers increased administrative costs to minimize likelihood of potential 23 contamination of their crops from genetically-altered alfalfa plant). 24 The California Plaintiffs’ expenditures were reasonably incurred because they face a 25 substantial risk of a differential undercount of their residents. These plaintiffs have proven 26 that the citizenship question will cause them to be differentially undercounted because they 27 have a disproportionate share of noncitizens, immigrants, and Latino residents and that this 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 53 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 54 of 126 1 differential undercount will cause them to lose federal funding and create a substantial risk 2 that the State of California will lose political representation. United States District Court 3 The legislative history of California’s FY 2018-19 state budget and follow-up reports 4 to the Governor and Legislature show that the State appropriated, and the California 5 Plaintiffs are spending, additional funds on census outreach to attempt to mitigate these 6 negative impacts caused by the citizenship question. Although it is not possible to pinpoint 7 precisely how much the citizenship question drove the increase in the state budget’s census 8 outreach line item, the California Plaintiffs have shown that the Legislature’s decision to 9 boost the Governor’s initial allocation for census outreach ($40.3 million) to a higher 10 allocation in the enacted budget ($90.3 million) is due at least in part to the citizenship 11 question. UF 111-112. 12 The California Plaintiffs also provided evidence that the citizenship question 13 prompted the County of Los Angeles to request $3.3 million in additional funds to meet the 14 County’s need for funding for census outreach to the hard-to-count populations most likely 15 not to respond to the 2020 Census because of the citizenship question. The State partially 16 met this request by allocating hundreds of thousands of dollars of additional funding to the 17 County for census outreach. Because any amount of costs incurred to mitigate harm is 18 sufficient to confer standing, as long as such costs were reasonably incurred, the State of 19 California’s expenditures on census outreach constitute a direct injury to the state’s budgets 20 and resources that is sufficient to establish injury-in-fact for standing purposes. 21 ii. The San Jose Plaintiffs 22 The City of San Jose has also spent, and will continue to spend, additional money on 23 outreach specifically because Secretary Ross directed that a citizenship question be added to 24 the Census. Because there is a substantial risk that adding the citizenship question will cause 25 a net differential undercount of noncitizens, which in turn would cause San Jose to lose 26 federal funding, the City’s diversion of resources to prevent that harm is reasonable. 27 Accordingly, the City of San Jose has suffered injury-in-fact based on its additional outreach 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 54 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 55 of 126 1 spending. Because BAJI has similarly diverted resources to encourage its constituents to 2 3 participate in the census and to counteract the chilling effects of the citizenship question, it 4 too has suffered an injury-in-fact. An injury-in-fact exists where a nonprofit organization 5 shows “a drain on its resources from both a diversion of its resources and frustration of its 6 mission.” Fair Hous. of Marin v. Combs, 285 F.3d 899, 905 (9th Cir. 2002); Havens Realty 7 Corp. v. Coleman, 455 U.S. 363, 378-380 (1982). BAJI’s Executive Director, Opal Tometi, testifies that BAJI has already expended 8 United States District Court 9 additional staff time and related financial resources to educate its constituents about the 10 addition of a citizenship question to the Census and encourage participation. BAJI intends to 11 spend an additional $200,000 to counteract the chilling effect of the citizenship question on 12 its members. BAJI’s membership has a high proportion of immigrants and is concentrated in 13 immigrant-rich metropolitan areas. Accordingly, an undercount of immigrant populations, 14 and any resulting loss of funding or political representation, would impair BAJI’s ability to 15 carry out its mission of fostering racial, economic, and social equality for Black immigrants and 16 other historically underrepresented communities. In sum, BAJI’s diversion of resources to 17 mitigate the negative impact of the citizenship question on its organizational mission 18 qualifies as an injury in fact.14 19 2. Traceability 20 The second standing element requires a plaintiff to demonstrate that the injury is 21 “fairly traceable to the challenged action of the defendant, and not the result of the 22 independent action of some third party not before the court.” Mendia v. Garcia, 768 F.3d 23 1009, 1012 (9th Cir. 2014) (quoting Bennett v. Spear, 520 U.S. 154, 167 (1997)). This 24 25 26 27 14 BAJI also advances two separate theories of associational standing. Hunt v. Washington State Apple Advertising Com’n, 432 U.S. 333, 343 (1977) (setting for the requirements for associational standing). There is, however, no need to reach these theories in light of the several bases for standing for the San Jose Plaintiffs that have already been discussed. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 55 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 56 of 126 1 requirement is less demanding than a proximate cause standard. Lexmark Intern., Inc. v. 2 Static Control Components, Inc., 572 U.S. 118, 134 n.6 (2014) (“Proximate causation is not 3 a requirement of Article III standing, which requires only that the plaintiff's injury be fairly 4 traceable to the defendant’s conduct.”). Indeed, “[c]ausation may be found even if there are 5 multiple links in the chain connecting the defendant’s unlawful conduct to the plaintiff’s 6 injury, and there’s no requirement that the defendant’s conduct comprise the last link in the 7 chain.” Mendia, 768 F.3d at 1012. When harm is caused by multiple actions, “what matters 8 is not the length of the chain of causation, but rather the plausibility of the links that 9 comprise the chain.” Id. at 1012-13 (quotation and citation omitted). United States District Court 10 While an injury is not fairly traceable it if is “‘th[e] result [of] the independent action 11 of some third party not before the court,’ . . . that does not exclude injury produced by 12 determinative or coercive effect upon the action of someone else.” Bennett, 520 U.S. at 169 13 (emphasis in original) (quoting Lujan, 504 U.S. at 560). The key question is whether the 14 “government’s unlawful conduct is at least a substantial factor motivating the third parties’ 15 actions.” Mendia, 768 F.3d at 1013 (citations and quotations omitted). Even when some 16 links in the chain of causation are illegal acts by third parties, the injury may still be “fairly 17 traceable” to the original challenged action. See Attias v. Carefirst, Inc., 865 F.3d 620, 629 18 (D.C. Cir. 2017) (harm caused by a data breach is “fairly traceable” to company’s inadequate 19 security standards even though data was stolen by third party hackers). 20 Plaintiffs have established that there will be a drop in self-response to the 2020 21 Census caused by the addition of the citizenship question. That nonrespondents have a legal 22 duty to respond to the census does not alter this conclusion because the citizenship question 23 is a “substantial factor” contributing to the nonresponse. Mendia, 768 F.3d at 1013. The 24 Bureau and its top officials have concretely affirmed the predictable impact of adding a 25 citizenship question to the 2020 Census on self-response rates. The harms Plaintiffs will 26 suffer—namely the loss of federal funding, the risk to California of losing political 27 representation, and the expenditure of resources to mitigate these harms—inevitably flow 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 56 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 57 of 126 1 from the disproportionate undercount of particular demographic groups that is likely to result 2 from Secretary Ross’s decision to include the citizenship question on the 2020 Census. 3 Defendants argue that any differential decline in self-response rates among 4 noncitizens and Latinos is attributable to the hostile macro-environment surrounding issues 5 of immigration, and that Plaintiffs fail to show that the citizenship question in particular 6 caused this decline. Dr. Barreto testified persuasively, however, that a hostile macro- 7 environment combined with the inclusion of sensitive questions on a survey can have a 8 cumulative effect that is greater than either of these factors would have on their own. Tr. 9 386:21-25, 411:5-14 (Barreto). More to the point, both Dr. Barreto’s survey and the Census 10 Bureau’s natural experiment attempted to isolate the effect of the citizenship question within 11 the same macro-environment. Dr. Barreto’s survey asked respondents about their willingness 12 to participate in the 2020 Census both with and without the citizenship question. Tr. 568:5- 13 569:1. The Census Bureau’s natural experiment, and subsequent statistical analysis, 14 compared response rates to the 2010 Census (no citizenship question) to those of the 2010 15 ACS (which included a citizenship question). PTX-103 at 6-7; PTX-148 at 6-7. Accordingly, 16 the differential decline in response rates found in these experiments cannot be exclusively 17 attributed to the current macro-environment surrounding immigration. 18 19 In sum, the injuries Plaintiffs have identified are fairly traceable to the Secretary’s decision to include the citizenship question in the 2020 Census. 20 3. Redressability 21 Finally, to meet the third standing requirement, Plaintiffs must show that it is likely 22 that a favorable decision will redress their injuries. Lujan, 504 U.S. at 561. 48. A plaintiff 23 need show only that “an injury” be redressed by a favorable decision. Larson v. Valente, 456 24 U.S. 228, 243 n.5 (1982) (emphasis in original). A plaintiff “need not show that a favorable 25 decision will relieve [] every injury.” Id. (emphasis in original). A favorable decision 26 vacating or enjoining the decision to add a citizenship question to the 2020 Census would 27 redress Plaintiffs’ injuries by diminishing the funds that they would need to be expended on 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 57 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 58 of 126 1 census outreach and by ensuring that Plaintiffs do not lose federal funding or, in the case of 2 California, political representation because of the citizenship question. 3 4 In sum, both the San Jose Plaintiffs and the California Plaintiffs have carried the burden of establishing all three of the standing requirements. United States District Court 5 IV. APA CLAIM 6 A. Legal Standard 7 Under section 706 of the APA, a reviewing court must “hold unlawful and set aside 8 agency action, findings, and conclusions found to be . . . arbitrary, capricious, an abuse of 9 discretion, or otherwise not in accordance with law; contrary to constitutional right, power, 10 privilege, or immunity; in excess of statutory jurisdiction, authority, or limitations, or short 11 of statutory right; [or] without observance of procedure required by law.” 5 U.S.C. 12 § 706(2)(A)-(D). Accordingly, the decision-making process that ultimately leads to the 13 agency action must be “logical and rational.” Allentown Mack Sales & Serv., Inc. v. NLRB, 14 522 U.S. 359, 374 (1998). Courts should be careful, however, not to substitute their own 15 judgment for that of the agency. Suffolk Cty. v. Sec’y of Interior, 562 F.2d 1368, 1383 (2d 16 Cir. 1977). Ultimately, a reviewing court may uphold agency action “only on the grounds 17 that the agency invoked when it took the action.” Michigan v. EPA, 135 S. Ct. 2699, 2710 18 (2015). Post hoc rationalizations may not be considered. American Textile Mfrs. Inst., Inc. v. 19 Donovan, 452 U.S. 490, 539 (1981). 20 B. Scope of Review 21 In evaluating an APA claim, courts “typically” limit their review to the 22 Administrative Record existing at the time of the decision. Sw Ctr. for Biological Diversity 23 v. U.S. Forest Service, 100 F.3d 1443, 1450 (9th Cir. 1996); accord Ranchers Cattlemen 24 Action Legal Fund United Stockgrowers of Am. v. U.S. Dep't of Agric., 499 F.3d 1108, 1117 25 (9th Cir. 2007). “Under limited circumstances, however, extra-record evidence can be 26 admitted and considered.” Ranchers Cattlemen, 499 F.3d at 1117. These exceptions include: 27 (1) when plaintiffs make a showing of agency bad faith, or (2) when the agency failed to 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 58 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 59 of 126 1 2 1. Use of Extra-Record Evidence to Evaluate Plaintiffs’ Pretext Argument 3 A court may consider extra-record evidence that is relevant to the reason for an 4 agency action where there has been a strong showing of bad faith or improper behavior by 5 the decision-makers. Public Power Council v. Johnson, 674 F.2d 791, 795 (9th Cir. 1982); 6 Ranchers Cattlemen, 499 F.3d at 1117. In such circumstances, consideration of extra-record 7 evidence is “necessary to a meaningful judicial review” of the agency’s actual decision- 8 making process. Tummino v. Torti, 603 F. Supp. 2d 519, 543 (E.D.N.Y. 2009). 9 United States District Court consider “all relevant factors” of the decision. Id. As discussed in greater detail in Part IV.C.12, infra, the Administrative Record alone 10 provides a “strong showing” of bad faith which authorizes this Court to look outside the 11 record for further evidence of pretext. That is particularly true here, where the 12 Administrative Record includes pre-decision communications between Secretary Ross and 13 his “point person” on the citizenship question issue expressing caution about what the 14 Administrative Record would include. PTX-96, PTX-362. 2. Defendants proceeded to 15 submit an initial Administrative Record that mischaracterized the Secretary’s decision- 16 making process and concealed important circumstances surrounding the DOJ’s request for 17 the addition of the citizenship question. Defendants supplemented the record only after being 18 ordered to do so in the New York matter. For these reasons, and the reasons set forth in Part 19 IV.C.12, infra, consideration of extra-record evidence is appropriate, but not necessary, for 20 the purpose of determining whether Secretary Ross’s decision was pretextual. 21 2. Use of Extra-Record Evidence to Evaluate Whether Defendants Considered All Relevant Factors 22 An agency’s decision is arbitrary and capricious under the APA if, among other 23 24 25 26 27 things, the agency failed to consider all “relevant factors,” Ranchers Cattlemen, 499 F.3d at 1115; ignored “an important aspect of the problem,” Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43-44 (1983); or made “an irrational departure from [settled] policy,” INS v. Yueh-Shaio Yang, 519 U.S. 26, 32 (1996). To apply these standards, a court must—as a threshold matter—understand what is 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 59 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 60 of 126 1 “relevant,” “important,” or “settled policy” in the field where the challenged agency decision 2 was made. In many cases, the Administrative Record will provide the relevant benchmarks. 3 But evidence outside the “bare record” may be required to determine “the applicable 4 standard” to apply in evaluating the completeness of the agency’s reasoning and in 5 determining whether the agency ignored critical factors or information. See Citizens to 6 Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 420 (1971); see also Nat’l Audubon 7 Soc. v. U.S. Forest Service, 46 F.3d 1437, 1447 (9th Cir. 1993). “It will often be impossible, especially when highly technical matters are involved, 8 United States District Court 9 for the court to determine whether the agency took into consideration all relevant factors 10 unless it looks outside the record to determine what matters the agency should have 11 considered but did not.” Asarco, Inc. v. EPA, 616 F.2d 1153, 1160 (9th Cir. 1980). The court 12 cannot adequately discharge its duty to engage in a “substantial inquiry” if it is required to 13 take the agency’s word that it considered all relevant matters. Id.; see also Nat’l Audubon 14 Soc. v. Hoffman, 132 F.3d 7, 15 (2d Cir. 1997) (“[t]he omission of technical scientific 15 information is often not obvious from the record itself”). Because this case involves complex technical issues related to survey methodology 16 17 and census-related practices, meaningfully evaluating whether Defendants considered all 18 relevant factors or irrationally departed from settled policy would be difficult on the 19 Administrative Record alone. The Court cannot simply accept Defendants’ assurances that 20 they considered all relevant factors. Accordingly, extra-record evidence will be admitted for 21 this limited purpose as well. In the interest of facilitating review of this decision by higher courts, the findings of 22 23 fact and conclusions of law that are based exclusively on the administrative record will be 24 discussed separately from those that rely on extra-record evidence. C. Findings of Fact Based Exclusively on the Administrative Record15 25 26 27 15 The numbering of the findings of fact picks up where it left off in Part III.B, supra. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 60 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 61 of 126 1. Compilation of the Administrative Record 1 166. Defendants produced an initial Administrative Record, along with a certification 2 3 4 5 6 and index on June 8, 2018, consisting of only 1,320 pages. See PTX-1 (AR 1-1320). These materials contain little documentation of internal discussions that took place before December 2017 and communications between the Departments of Commerce and Justice about the citizenship question. Id. 167. On June 21, 2018, Defendants filed a supplemental memorandum composed by 7 8 9 Secretary Ross that revised the narrative of how a citizenship question came to be placed on the decennial census. PTX-2. 168. On July 3, 2018, and memorialized in a July 5, 2018 order, Defendants were 10 11 United States District Court 12 13 14 15 ordered to supplement the Administrative Record in New York et al. v. Department of Commerce, et al., No. 18-cv-2921 (S.D.N.Y.).16 In response to that order, Defendants produced extensive supplemental Administrative Record documents. The parties agreed that all documents bearing prefix-less Bates stamps between 000001 and 0013024 are part of the Administrative Record. Joint Pretrial Statement 11-13. 2. Secretary Ross’s Deliberations Prior to Receiving the DOJ Letter 16 169. In March 2017, the Bureau reported to Congress the five “topics” that would be 17 18 19 20 included on the 2020 Census, including gender, age, race, ethnicity, and homeownership status. The subjects did not include citizenship or immigration status. PTX-264 at 5-15. This report also listed topics limited to the ACS, including citizenship. Id. at 55. 170. That same month, Secretary Ross exchanged emails with his Deputy Chief of 21 22 23 24 25 26 27 Staff and Director of Policy, Earl Comstock, regarding whether noncitizens are included in the census count for the purposes of congressional apportionment. PTX-30; PTX-55. Comstock specifically emailed Ross an answer to “Your Question on the Census,” to confirm that noncitizens are indeed counted on the census. PTX-55. 16 Judicial notice is taken of the order in New York et al. v. Department of Commerce, et al., No. 18-cv-2921 (S.D.N.Y.) requiring Defendants to supplement the record. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 61 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 62 of 126 171. On April 5, 2017, Ross’s executive assistant wrote an email directed to Ross 1 2 indicating that “Steve Bannon has asked the Secretary to talk to someone about the census.” 3 PTX-58.17 In their subsequent communications, Stephen Bannon connected Secretary Ross 4 with Kris Kobach, former Vice Chair of the Presidential Commission on Election Integrity 5 and Kansas Secretary of State, to discuss adding a citizenship question to the census. PTX- 6 19; PTX-58. Kobach told Secretary Ross by phone that a citizenship question was necessary 7 to address the “problem that aliens who do not actually ‘reside’ in the United States are still 8 counted for congressional apportionment purposes.” PTX-19 at 2. 172. A few days later, Mark Neuman, an outside advisor to Secretary Ross, emailed United States District Court 9 10 Comstock with the subject line “One of the Supreme Court cases that informs planning for 11 the 2020 Census. . . .” PTX-182. The email contained only a link to the Supreme Court’s 12 decision in LULAC v. Perry, id., which considered citizen voting age population in assessing 13 claims under Section 2 of the VRA. On April 13, Comstock emailed Neuman, asking when 14 the Census Bureau would need to notify Congress of the questions that would appear on the 15 2020 Census. PTX-88; PTX-181. Neuman responded to Comstock on April 14, 2017, that 16 the notification deadline for topics had already passed, and that “[t]here would be another 17 opportunity next year.” PTX-88. 18 173. On May 2, 2017, Secretary Ross emailed Comstock: “I am mystified why 19 nothing [has] been done in response to my months old request that we include the citizenship 20 question. Why not?” PTX-89. In reference to the statutory requirement that the topics for the 21 Census be submitted by March 2017, Secretary Ross wrote to Comstock: “Worst of all they 22 emphasize that they have settled with congress on the questions to be asked.” Id. Comstock 23 wrote back: “On the citizenship question we will get that in place . . . . We need to work with 24 Justice to get them to request that citizenship be added back as a census question.” Id. 25 3. Comstock Seeks an Agency Willing to Request Addition of the Citizenship 26 27 17 At the time, Bannon was the White House Chief Strategist. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 62 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 63 of 126 1 2 3 4 5 6 7 8 9 10 11 United States District Court 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Question 174. Secretary Ross’s May 2, 2017 email led to immediate action by Comstock. On May 3, 2017, Eric Branstad, a Senior White House Advisor, reached out on Comstock’s behalf to Matthew J. Flynn, Senior Director of Cabinet Affairs at the Executive Office of the President, to find “the best counterpart to reach out to at DOJ regarding Census and Legislative issue?” PTX-85. Branstad subsequently referred Comstock to Mary Blanche Hankey who was the White House liaison within DOJ. PTX-370. 175. Hankey directed Comstock to speak with James McHenry, the director of the Executive Office of Immigration Review at DOJ. Id. Comstock spoke “several times” with McHenry of DOJ about adding a citizenship question to the census. Id. McHenry ultimately informed Comstock that the DOJ did not want to request the citizenship question “given the difficulties Justice was encountering in the press at the time (the whole Comey matter).” Id. 176. McHenry therefore referred Comstock to the Department of Homeland Security. Id. DHS, however, likewise declined to request the citizenship question. Id. Following his failed discussions with DHS, Comstock asked James Uthmeier, of the Office of General Counsel at the Department of Commerce, to investigate “how Commerce could add the question to the Census itself.” Id. 177. On May 24, 2017, Secretary Ross and David Langdon, a policy advisor who reported to Comstock, met “all afternoon.” PTX-151 at 2. During that meeting, Secretary Ross asked questions about the content of the decennial Census and “seemed . . . puzzled why citizenship is not included in the 2020” census. PTX-86 at 1. Later that afternoon, Burton Reist, Chief of Decennial Communications and Stakeholder Relations at the Census Bureau, emailed Langdon a 1988 internal DOJ memorandum that opined the Constitution does not mandate the counting of undocumented U.S. residents in the census apportionment count. PTX-448, PTX-449 at 1-2. That evening, Langdon requested further information from Census Bureau staff including Reist regarding “the criteria used to pick topics for 2020 versus ACS. Say, citizenship.” PTX-151. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 63 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 64 of 126 1 2 immigrants,” which stated: “the counting of illegal immigrants (or of the larger group of 3 noncitizens) has a solid and fairly long legal history . . . [there is] a Bush 41 era DOJ opinion 4 that proposed legislation to exclude illegal aliens from the decennial census was illegal.” 5 PTX-397. Comstock responded to Langdon that day, asking for further information on the 6 selection of questions for the census versus the ACS, and passing along the Supreme Court 7 decision of LULAC v. Perry that Neuman had previously provided for the proposition that 8 the government might have a use for citizenship data. Id.; PTX-182. 9 United States District Court 178. That same day, Langdon sent an email to Comstock entitled “Counting of illegal 179. On July 14, 2017, Kobach emailed Secretary Ross to remind him of their prior 10 telephone discussion “a few months ago.” PTX-19 at 2. Kobach wrote that during their 11 earlier discussion, he and Secretary Ross “talked about the fact that the US census does not 12 currently ask respondents about their citizenship,” and further advised Secretary Ross that 13 the absence of such a question “leads to the problem that aliens who do not actually ‘reside’ 14 in the United States are still counted for congressional apportionment purposes.” Id. Kobach 15 further wrote that “it was essential that one simple question be added to the upcoming 2020 16 census” and that a variant of the question that appears on the ACS “needs to be added to the 17 census.” Id. 18 180. On July 21, 2017, Kobach called Wendy Teramoto, a Senior Advisor and Chief 19 of Staff to Secretary Ross. Id. at 1. He also emailed her, forwarding his July 14 email to 20 Secretary Ross stating that he had spoken to Secretary Ross about adding a citizenship 21 question to the 2020 Census at the direction of Steve Bannon. Id. Teramoto subsequently 22 arranged a call between Kobach and Secretary Ross for a few days later. Id. 23 181. On August 8, 2017, Secretary Ross emailed Comstock, asking “where is DOJ in 24 their analysis” of whether to add a citizenship question to the 2020 Census, and advising 25 “[i]f they still have not come to a conclusion please let me know your contact person and I 26 will call the AG.” PTX-96; PTX-98. The very next day, Comstock emailed Secretary Ross, 27 stating “we are preparing a memo and full briefing for you on a citizenship question. The 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 64 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 65 of 126 1 memo will be ready by Friday . . . Since this issue will go to the Supreme Court we need to 2 be diligent in preparing the administrative record.” PTX-96, PTX-362. Secretary Ross 3 responded to Comstock the next day, “I would like to be briefed on Friday by phone . . . we 4 should be very careful about everything, whether or not it is likely to end up in the SC.” 5 PTX-96, PTX-362. 6 182. On August 11, Comstock and Uthmeier exchanged edits on briefing materials 7 for Secretary Ross related to a citizenship question. During this exchange, Uthmeier wrote 8 that he had “recommendations on execution,” stating that he thought “our hook” was 9 “ultimately, we do not make decisions on how the [citizenship] data will be used for 10 apportionment, that is for Congress (or possibly the President) to decide.” PTX-437 at 2. 11 That same day, Comstock emailed Secretary Ross and Teramoto “a draft memo on the 12 citizenship question” prepared by James Uthmeier. PTX-3 at 1140; PTX-147. The 13 memorandum has not been produced. 14 183. On September 1, 2017, Secretary Ross complained to Comstock and Teramoto 15 about a number of issues, including that he had “received no update [on] the issue of the 16 census question” and Comstock responded, “Understood. Wendy and I are working on it.” 17 PTX-45; PTX-97. 18 184. On September 6, Secretary Ross and his senior staff had a meeting to discuss 19 adding a citizenship question to the decennial census. PTX-31; PTX-35; PTX-36; PTX-46; 20 PTX-47. The next day, Secretary Ross requested from his staff an update on “progress since 21 the discussion yesterday regarding the citizenship question.” PTX-37, PTX-49. A few days 22 later, Uthmeier contacted Neuman to discuss “some Census legal questions for the 23 Secretary.” PTX-38. That same day, Comstock sent Secretary Ross a memo reporting on his 24 efforts to identify someone who would request that a citizenship question be added to the 25 2020 Census, and advising that, as of that date, he had not been successful in locating any 26 such agency. PTX-48; PTX-134. He also informed the Secretary that, once his discussions 27 with DOJ and DHS fell through, he had enlisted Uthmeier’s help in advancing the 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 65 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 66 of 126 1 United States District Court 2 citizenship question. PTX-48; PTX-134. 185. In mid-September, John Gore of DOJ contacted and later spoke on the 3 telephone with Teramoto “about a DOJ-DOC” issue. PTX-59; PTX-60. Following that 4 conversation, Gore worked with an aide to Attorney General Sessions to set up a phone call 5 between Secretary Ross and then Attorney General Sessions. PTX-63, PTX-67, PTX-68. On 6 September 17, 2017, Danielle Cutrona in the Office of the Attorney General wrote to Wendy 7 Teramoto, Ross’s Chief of Staff, to say that “[t]he Attorney General is available on his cell,” 8 and that “it sounds like we can do whatever you all need us to do and the delay was due to a 9 miscommunication . . . . The AG is eager to assist.” PTX-62 (emphasis added); see also 10 PTX-67; PTX-68. Secretary Ross and Attorney General Sessions proceeded to speak on the 11 phone regarding the subject of Gore and Teramoto’s earlier conversation, presumably about 12 adding the citizenship question to the census. PTX-57; PTX-61; PTX-62. 13 186. The most reasonable inference to be drawn from these facts is that, on or before 14 September 18, 2017, Secretary Ross spoke with Attorney General Sessions and asked that he 15 instruct his subordinates at DOJ to request a question on citizenship be added to the Census. 16 187. On Sunday, October 8, Secretary Ross sent an email to staffer Peter Davidson 17 of the Department of Commerce with the subject line, “Letter from DOJ” asking “what is its 18 status.” PTX-52. Davidson responded, “I’m on the phone with Mark Neuman right now . . . 19 he is giving me a readout of his meeting last week.” Id. On the evening of November 27, 20 2017, Secretary Ross emailed Davidson again, stating, “Census is about to begin translating 21 the questions into multiple languages and has set the printing contract. We are out of time. 22 Please set up a call for me tomorrow with whoever is the responsible person at Justice. We 23 must get this resolved.” PTX-144. 24 188. The Administrative Record shows that Secretary Ross issued a “request that we 25 include the citizenship question,” PTX-89, that Comstock assured him that “we will get that 26 in place,” id, that Secretary Ross’s staff attempted to find a way for the Department of 27 Commerce to add the citizenship question without an outside agency request when it 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 66 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 67 of 126 1 appeared unlikely that any such request would materialize, and that Secretary Ross 2 intervened with the Attorney General when career staff at DOJ and DHS refused to request 3 the addition of the question. The reasonable inference to be drawn from these actions is that 4 Secretary Ross had already made up his mind to include the citizenship question on the 2020 5 Census before he received the DOJ request letter. United States District Court 6 189. There is no writing of any kind in the Administrative Record authored by the 7 Secretary or anyone at the Commerce Department (or anyone else) that directly describes the 8 reasons why the Secretary wanted to add a citizenship question prior to December 12, 2017. 9 190. The Administrative Record reveals, however, that at the same time the 10 Secretary was discussing adding a citizenship question to the census in the spring and 11 summer of 2017, he was asking Comstock questions about whether congressional 12 apportionment based on the census included noncitizens, he was informed that including 13 noncitizens in congressional apportionment was legally required, and he was advised by 14 Kobach that the inclusion of noncitizens in the census was a “problem.” PTX-19; PTX-55; 15 PTX-58; PTX-86; PTX-89; PTX-151; PTX-397; PTX-437; PTX-444. Nowhere in the record 16 does Secretary Ross explicitly state whether he subscribed to the same rationale for 17 including the citizenship question as Kobach. 18 191. Nevertheless, it is clear that Secretary Ross’s decision prior to December 12, 19 2017 to include the citizenship question was unrelated to the enforcement of the VRA. 20 Secretary Ross’s direction to his staff to find a way to include the citizenship question on the 21 census, the fact that his staff solicited DHS as well as DOJ for a request, and the fact that 22 Comstock’s attempts to solicit a request were not limited to the Division of Civil Rights 23 belie any notion that Secretary Ross’s motivation was to meet DOJ’s VRA enforcement 24 needs. Rather, it was DOJ that was meeting his preferences. PTX-62; PTX-67; PTX-68. 25 192. The Administrative Record strongly supports the following inferences 26 regarding the period from March 2017 through December 2017: (1) that before Secretary 27 Ross received DOJ’s request to add the citizenship question, he had already decided to 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 67 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 68 of 126 1 include the citizenship question on the 2020 Census; (2) that his reasons for wanting the 2 question asked were unrelated to DOJ’s alleged need for citizenship data at the census block 3 level to enforce the VRA; (3) that he directed his staff to find a way to get the citizenship 4 question added to the Census; (4) that, when he was told that he needed to have an agency 5 ask him to add the question, he directed his staff to find an agency to ask him; and (5) that he 6 did not care who asked for the question or the reasons for asking the question, so long as 7 some agency asked. 8 United States District Court 9 4. DOJ Letter Requesting the Addition of the Citizenship Question a. DOJ’s Previous Decision Not to Request the Citizenship Question 10 193. In June 2014, Arthur Gary of DOJ wrote to the General Counsel of the 11 Department of Commerce to indicate what census-derived data the DOJ uses and to confirm 12 that it continued to use such data. PTX-1 at 278-83. Gary indicated that DOJ used citizenship 13 data collected by the ACS to enforce the VRA, and that the “lowest geography” for which 14 DOJ needed citizenship data was the “Census block group” level. Id. at 280. Gary did not 15 indicate that DOJ needed citizenship data at the census block level, or that it needed a 16 citizenship question on the 2020 Decennial Census. 17 194. In July 2016, Gary wrote to the Census Bureau to confirm that the DOJ “had no 18 needs to amend the current content and uses or to request new content” for the 2020 Census. 19 PTX-1 at 311. In November of 2016, Gary supplemented that letter to “formally request[] 20 that the Census Bureau consider a new topic in the ACS relating to LGBT populations.” Id. 21 At no time prior to December 2017 did Gary express a need for more granular citizenship 22 data or a citizenship question on the Census. 23 b. DOJ Changes Positions and Requests the Citizenship Question 24 195. On December 12, 2017, Arthur Gary of DOJ sent a formal letter to Dr. Ron 25 Jarmin, Acting Director of the Census Bureau, requesting that a citizenship question be 26 added to the 2020 Census (December 12 Letter). PTX-32. The December 12 Letter requests 27 the addition of a citizenship question for purposes of VRA enforcement. Id. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 68 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 69 of 126 1 2 citizenship data is strictly necessary for the purposes of VRA enforcement. Rather, the letter 3 contends, “the Department [of Justice] believes that decennial census questionnaire data 4 regarding citizenship, if available, would be more appropriate for use in redistricting and in 5 Section 2 litigation than the ACS citizenship estimates.” Id. at 3. It further cites numerous 6 published cases for the proposition that, “in order to assess and enforce compliance with 7 Section 2’s protection against discrimination in voting, the Department needs to be able to 8 obtain citizen voting-age population data for census blocks…where potential Section 2 9 violations are alleged or suspected.” Id. at 2. 10 United States District Court 196. The letter does not state that a citizenship question and/or block-level 197. The December 12 Letter states that one of the reasons decennial census data on 11 citizenship would be preferable to ACS data concerns the margin of error: “The ACS 12 estimates are reported at a ninety percent confidence level, and the margin of error increases 13 as the sample size—and, thus, the geographic area—decreases . . . By contrast, decennial 14 census data is a full count of the population.” Id. at 3. 15 198. The letter does not state that any plaintiffs had lost any Section 2 enforcement 16 actions due to insufficient CVAP data from the ACS. Nor does the letter suggest that DOJ 17 has declined to bring any Section 2 enforcement actions due to insufficient CVAP data from 18 the ACS. 19 5. The Census Bureau’s Analyses and Recommendations 20 199. Soon after the Census Bureau received the December 12 Letter, Dr. Abowd 21 directed the SWAT Team, to formulate a response to the suggestion that a citizenship 22 question be added, which Dr. Abowd managed and reviewed. PTX-4D at 284; PTX-75; 23 PTX-148; UF 19. In a series of technical reports, responses to questions posed by Secretary 24 Ross, and other briefing documents, the Census Bureau consistently recommended against 25 adding a citizenship question to the 2020 Census. PTX-22; PTX-101; PTX-133; PTX-148. 26 27 200. The Census Bureau concluded that the stated goals of DOJ with respect to enforcement of the VRA could be accomplished, in a less costly and more effective manner, 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 69 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 70 of 126 1 by linking Census responses to other administrative data sets available to the federal 2 government. PTX-22; PTX-101; PTX-133; PTX-148. In other words, the Census Bureau 3 determined that including a citizenship question on the decennial census was not necessary 4 to provide complete and accurate data in response to DOJ's request. PTX-22; PTX-101; 5 PTX-148. In fact, the Census Bureau advised Secretary Ross that relying exclusively on 6 administrative records would result in more accurate citizenship data than adding a 7 citizenship question to the census. PTX-22; PTX-101; PTX-148. The Bureau further 8 concluded that relying exclusively on administrative records would produce more accurate 9 citizenship data than attempting to use both administrative records and citizenship responses. 10 United States District Court 11 PTX-24; PTX-25. 201. As part of its analysis, the Census Bureau examined the initial drop in self- 12 response due to the citizenship question, as well as the potential costs of NRFU as a result of 13 that drop. PTX-22; PTX-101; PTX-133; PTX-148. The Administrative Record does not, 14 however, contain any information showing that the Census Bureau analyzed whether NRFU 15 could or would mitigate the decrease in self-response rates, or whether the citizenship 16 question would ultimately result in an undercount. 17 18 a. December 22 Census Bureau Memo 202. On or about December 15, 2017, Dr. Abowd directed senior executives and 19 expert employees of the Census Bureau to evaluate alternative methods of providing 20 estimates of the CVAP to support redistricting under Public Law 94 -171 (P.L. 94-171) and 21 Section 2 of the VRA. PTX-4D at 284; PTX-148 at 1. This evaluation is reflected in a 22 memorandum dated December 22, 2017 and was provided to the Commerce Department 23 (December 22 Memo). PTX-148. 24 203. In the December 22 Memo, the Bureau reported on the differential decline in 25 self-response rates from the 2010 Decennial Census to the 2010 ACS in households with at 26 least one noncitizen and found that there was a 5.1% greater decline among such households, 27 which is “consistent with citizenship questions being more sensitive for households with 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 70 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 71 of 126 1 2 204. Additionally, the Bureau found that “there is a tendency for noncitizen ACS 3 respondents to report being U.S. citizens,” and that “roughly 40 percent” of those who are 4 legal resident noncitizens reported being citizens on the ACS. Id. at 7. 5 205. The memo recommended that that the citizenship data for DOJ VRA 6 enforcement be obtained through the use of administrative records rather than the addition of 7 a question to the decennial census instrument. Id. at 11. The Bureau based this 8 recommendation on several factors, including its conclusion that the question would create a 9 differential increase in nonresponse rates of at least 5.1 percent in households with at least 10 one noncitizen, and the fact that citizenship status is often inaccurately reported by 11 noncitizens. 12 United States District Court noncitizens.” Id. at 6-7. 206. The Census Bureau identified eight administrative sources of citizenship 13 information either already in use by the Census Bureau or available for acquisition by the 14 Census Bureau. Id. at 3. Accordingly, the December 22 Memo advised that the best way to 15 meet DOJ’s stated need was to provide it with citizenship data from administrative records. 16 Id. at 11. 17 207. The memo observed that, if the recommendation were followed, “[t]he 2020 18 Census questionnaire would not be altered, and the field operations would not have to be 19 expanded to compensate for the lower rate of voluntary compliance predicted for a census 20 that asks the citizenship question directly.” Id. at 12. 21 22 b. January 3 Memo 208. Following the December 22 Memo, the Census Bureau further memorialized its 23 research and analysis of potential sources of citizenship data in a January 3, 2018 24 memorandum from Dr. Abowd to Dr. Jarmin (January 3 Memo). PTX-101. 25 209. This document explained that the Census Bureau produces CVAP data in “two 26 related data products: the P.L. 94-171 redistricting data produced by April 1st of the year 27 following a decennial census under the authority of 13 U.S.C. Section 141, and the Citizen 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 71 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 72 of 126 1 Voting Age Population by Race and Ethnicity (CVAP) tables produced every February from 2 the most recent five-year ACS data.” PTX-101 at 1. While the P.L. 94-171 data is released at 3 the census block level, the CVAP data is released at the census block group level. Id. 4 210. The memo proceeded to analyze the cost and data quality implications of three 5 alternative methods of meeting DOJ’s request for census block-level estimates of CVAP. Id. 6 211. Alternative A was to “[m]aintain the status quo for data collection, preparation 7 and publication,” but then prepare a special product for DOJ that combines the P.L. 94-171 8 and CVAP tables to produce the Bureau’s best estimate of block-level CVAP data. Id. 9 Alternative A was estimated to cost $200,000. Id. at 2. The Census Bureau concluded this United States District Court 10 option was “not very costly and does not harm the quality of the census count.” Id. at 3. 11 212. Alternative B was to “[a]dd a citizenship question to the 2020 Census 12 questionnaire.” The memo estimated that Alternative B would increase census nonresponse 13 rates by at least 5.1 percent of all households with one or more noncitizens, or 700,000 14 households. Id. at 2. This would increase the NRFU workload and increase the cost of the 15 2020 Census by “at least $27.5 million.” Id. While the memo stated that Alternative B suited 16 DOJ’s stated uses better than Alternative A, it noted that this option would be “very costly” 17 and—because NRFU is less accurate than self-responses—would harm the accuracy of the 18 census. Id. at 2, 3. 19 213. Alternative C would involve creating block-level citizenship data using 20 administrative records without asking respondents about their citizenship status. Id. at 1. 21 This option was estimated to cost less than $1 million. Id. at 2. The Bureau concluded that 22 relying exclusively on administrative records would deliver higher quality citizenship data 23 than Alternative B because the administrative records provide “very accurate” citizenship 24 information. Id. at 3. The memo explained that this is because the administrative record data 25 required proof of citizenship, whereas citizenship information is self-reported less 26 accurately, and proxies report citizenship even less accurately than self-responding 27 individuals. Id. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 72 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 73 of 126 1 2 January 3 memo, reasoning that, compared to Alternatives A and B, Alternative C “even 3 better meets DoJ’s stated uses, is comparatively far less costly than Alternative B, and does 4 not harm the quality of the census count.” Id. On January 4, 2018, Dr. Abowd wrote to 5 various census officials, including Dr. Jarmin, “Ron reports that he has discussed this with 6 the Under Secretary, and she agrees with the recommendation of Alternative C, but 7 Alternative A remains a possibility as well.” PTX-122 at 1. 8 c. January 19 Memo 9 United States District Court 214. The Census Bureau therefore expressly recommended Alternative C in the 215. On January 19, 2018, Dr. Abowd sent another memorandum to Secretary Ross 10 on the “Technical Review of the Department of Justice Request to Add Citizenship Question 11 to the 2020 Census” (January 19 Memo). PTX-22. This memo presents the view of Dr. 12 Abowd and the technical team that evaluated Alternatives A, B and C. Id. It contains the 13 same recommendation and rationale as in the January 3 Memo, along with some additional 14 details of their analysis. Id. 15 216. The memo specifically examined the issue of item nonresponse to the 16 citizenship question, i.e. nonresponse to the particular question, rather than the whole 17 questionnaire. Id. at 4. It noted that item nonresponse rates for the citizenship question “are 18 much greater than the comparable rates for other demographic variables like sex, 19 birthdate/age, and race/ethnicity.” Id. Moreover, between 2013 and 2016, the item 20 nonresponse rate of Hispanics to the citizenship question was approximately double that of 21 non-Hispanic whites. Id. 22 217. The memo also examined the break-off rate for internet responses to the 2016 23 ACS, i.e. at what question people stopped taking the survey. Id. at 5. It found that “[b]ecause 24 Hispanics and non-Hispanic non-whites breakoff much more often than non-Hispanic 25 whites, especially on the citizenship-related questions, their survey response quality is 26 differentially affected.” Id. 27 218. The Census Bureau reiterated its estimate, based on the 2010 census and ACS, 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 73 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 74 of 126 1 that Alternative B would cause a 5.1 percent drop in self response from households 2 containing at least one noncitizen. Id. at 4-5. It explained that, while both citizen and 3 noncitizen households responded to the ACS (which had a citizenship question) at lower 4 rates than to the census (which did not), the decline between the two surveys was 5.1 percent 5 greater for noncitizen households. Id. at 4. 219. The January 19 Memo further explained that lowered self-response rates would 6 7 lower census data quality because data obtained in NRFU have greater rates of erroneous 8 enumeration and whole-person imputation. Id. at 5-6. Erroneous enumerations are 9 enumerations of a person who should not have been counted and whole-person imputations 10 are enumerations of all characteristics of a person. Id. at 5. 220. The Census Bureau provided additional information about the shortcomings of United States District Court 11 12 attempting to ascertain citizenship status based on direct inquiries. It specifically explained 13 that Alternative C would yield more accurate citizenship data than Alternative B because, 14 based on historical census and ACS data, noncitizens misreport themselves as citizens “for 15 no less than 23.8% of the cases, and often more than 30%.” Id. at 7. Specifically, “[i]n 2010 16 and 2016, individuals for whom the administrative data indicate noncitizen respond citizen 17 in 32.7% and 34.7% of the ACS questionnaires, respectively.” Id. at 8. By contrast, 18 Alternative C would provide highly accurate citizenship data because administrative record 19 citizenship data is “verified” based on proof of citizenship or legal resident alien status. Id. at 20 7. 21 221. The memo also addressed the increased burden on respondents associated with 22 the addition of the citizenship question. The Census Bureau explained: “Survey 23 methodologists consider burden to include both the direct time costs of responding and the 24 indirect costs arising from nonresponse due to perceived sensitivity of the topic.” Id. at 5. 25 Thus, a citizenship question “would make the 2020 Census modestly more burdensome in 26 the direct sense, and potentially much more burdensome in the indirect sense that it would 27 lead to a larger decline in self-response for noncitizen households.” Id. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 74 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 75 of 126 United States District Court 1 222. Finally, the Census Bureau explained the $27.5 million increase in costs due to 2 the addition of the citizenship question “is a conservative estimate because the other 3 evidence cited in this report suggests that the differences between citizen and noncitizen 4 response rates and data quality will be amplified during the 2020 Census compared to 5 historic levels. Hence, the decrease in self-response for [non]citizen households in 2020 6 could be much greater than the 5.1 percentage points we observed during the 2010 Census.” 7 Id. at 6. Ultimately, the memo concluded that adding a citizenship question to the decennial 8 census, “is very costly, harms the quality of the census count, and would use substantially 9 less accurate citizenship status data than are available from administrative sources.” Id. at 1. 10 223. Bureau staff had a single meeting with Secretary Ross on February 12, 2018, to 11 discuss the January 19 Memo. The only record of the meeting in the Administrative Record 12 is a February 12, 2018 email from James Treat of the Census Bureau to other Bureau 13 employees, including Dr. Jarmin and Dr. Abowd, identifying five “actions from yesterday’s 14 meeting with the Secretary.” PTX-128. 15 d. March 1 Memo 16 224. At some point after the February 12 meeting, Secretary Ross requested analysis 17 of a fourth option, Alternative D, which would involve “combining Alternative B (asking the 18 citizenship question of every household on the 2020 Census) and Alternative C (do not ask 19 the question, link reliable administrative data on citizenship status instead).” PTX-133 at 2. 20 On March 1, 2018, Dr. Abowd sent an additional recommendation memorandum to Secretary 21 Ross performing an analysis of this fourth alternative. PTX-133. In this memo, the Census 22 Bureau concluded that, “Alternative D would result in poorer quality citizenship data than 23 Alternative C. It would still have all the negative cost and quality implications of Alternative 24 B outlined in the draft January 19, 2018 memo to the Department of Commerce.” Id. at 5. 25 225. The Census Bureau also identified additional problems with Alternative D in the 26 March 1 Memo. First, census responses would be unreliable for filling in the data gaps for 27 those who do not match to administrative records, because undocumented immigrants “have 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 75 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 76 of 126 1 a strong incentive to provide an incorrect [citizenship] answer, if they answer at all.” Id. at 4. 2 226. Second, lowered self-response rates due to the citizenship question would 3 decrease the number of people who can be linked to administrative records, because NRFU 4 personal identifying information is of lower quality than personal identifying information 5 that is self-reported. Id. 6 227. Further, the March 1 Memorandum explains: 7 Under Alternative C, there will be error in the administrative records, but we believe these to be relatively limited due to the procedure following by SSA, USCIS and State. In both Alternatives, the modeled cases will be subject to prediction error . . . . Alternative D has an additional source or error, response error. This is where 2020 respondent give the incorrect status. Statisticians often hope these errors are random and cancel out. However, we know from prior research that citizenship status responses are systematically biased for a subset of noncitizens. Response error is only an issue in alternative D. 8 9 10 11 United States District Court 12 PTX-133 at 8. 13 228. The memo concludes that, under Alternative D, for the group of 22 million 14 people for which the Bureau has both a census response and administrative records, but 15 where they do not match, the citizenship data will be less accurate due to response errors. Id. 16 229. The Bureau also concluded that under Alternative D, it would not receive a 17 response to the citizenship question from 35.4 million people and would likely be able to 18 observe citizenship status for 21.5 million people and impute citizenship status for 13.8 19 million people, concluding that “there will be a need for imputing many cases across either 20 alternative.” Id. at 7-8. 21 22 230. As in previous Census Bureau memoranda on the subject, the March 1 Memo recommended against adding a citizenship question to the 2020 Census. Id. at 5. 23 6. The Set of 35 Questions Answered by the Census Bureau 24 231. Following the January 19 Memo, Comstock and Uthmeier developed and sent 25 to the Census Bureau a set of 35 questions for the Census Bureau to answer about the 26 analysis in the January 19 Memo. PTX-377. The Census Bureau’s responses to the questions 27 were submitted to the Commerce Department on March 1, 2018, along with Dr. Abowd’s 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 76 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 77 of 126 1 2 232. Question 31 asked, “What was the process that was used in the past to get 3 questions added to the decennial Census or do we have something similar where a precedent 4 was established?” Id. at 21. The Census Bureau provided the following response: 5 6 7 8 9 10 United States District Court March 1 memorandum to Secretary Ross. PTX-133. The Census Bureau follows a well-established process when adding or changing content on the census or ACS to ensure the data fulfill legal and regulatory requirements established by Congress. Adding a question or making a change to the Decennial Census or the ACS involves extensive testing, review, and evaluation. This process ensures the change is necessary and will produce quality, useful information for the nation. The Census Bureau and the Office of Management and Budget (OMB) have laid out a formal process for making content changes. 11  12  13  14 16    17 Id. at 21-22. 15 18 First, federal agencies evaluate their data needs and propose additions or changes to current questions through OMB. In order to be included, proposals must demonstrate a clear statutory or regulatory need for data at small geographies or for small populations. Final proposed questions result from extensive cognitive and field testing to ensure they result in proper data, with an integrity that meets the Census Bureau’s high standards. This process includes several opportunities for public comment. The final decision is made in consultation with OMB. If approved, the Census Bureau implements the change. 233. The description of the “well-established” process in the Census Bureau’s 19 response to Question 31 is consistent with other Census Bureau documents describing the 20 process to add a question or change the content of the decennial census. PTX-4 at AR 3890, 21 3560, 9867; PTX-135; PTX-141. 22 234. The Administrative Record shows that, despite the fact that the questions were 23 directed to the Census Bureau, Commerce Department Deputy General Counsel Michael 24 Walsh drafted a different answer to Question 31. PTX-14. That answer states: 25 26 27 No new questions were added to the 2010 Decennial Census, so there is no recent precedent for considering a request to add questions to a Decennial Census. Consistent with longstanding practice for adding new questions to the ACS survey, the Census Bureau is working with relevant stakeholders to ensure that legal and regulatory requirements are fulfilled and that the 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 77 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 78 of 126 1 2 3 4 5 6 7 8 9 10 11 United States District Court 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 question would produce quality, useful information for the nation. As you are aware, that process is ongoing. Upon its conclusion, you will have all of the relevant data at your disposal to make an informed decision about the pending request from the Department of Justice. PTX-1 at 1296. 235. In Defendants’ first production of documents in the Administrative Record, which they represented at the time constituted the complete administrative record, they included a version of the 35 questions and answers that included only Walsh’s answer to Question 31. See PTX-1 at 1296. The Census Bureau’s March 1 response to Question 31 was produced later and as a result of the New York court’s order to supplement the record. See PTX-133 at 21-22. 236. Question 1 of the 35 questions asked, “With respect to Alternatives B and C, what is the difference, if any, between the time when the data collected under each alternative would be available to the public?” PTX-133 at 11. The Census Bureau answered Question 1 by stating that, between Alternatives B and C, there was no difference in the timing in which the citizenship data could be offered to the public. Id. 7. Memo Re: “Key Differences Between Alternative C and Alternative D” 237. The Administrative Record also includes a memorandum entitled “Summary Analysis of the Key Differences Between Alternative C and Alternative D.” PTX-24. Like the March 1 Memo, this memorandum recommends using administrative data alone (Alternative C) and not adding a citizenship question. Id. at 1-2. 238. The Census Bureau explained that while both Alternative C and D will require the citizenship of a portion of the population to be imputed, or “modeled,” Alternative D will suffer from accuracy issues because many noncitizens self-report as citizens, which will systematically bias the modeling in Alternative D. Id. at 2. 239. Neither this memorandum, or any of the memoranda previously discussed, analyzed whether NRFU would fully mitigate the nonresponse or whether it would ultimately result in an undercount of certain populations. See PTX-22; PTX-101; PTX-133; PTX-148. 8. DOJ’s Refusal to Discuss Its Request with the Census Bureau 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 78 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 79 of 126 1 2 the Census Bureau sought to meet with DOJ to discuss its stated need for block-level 3 citizenship data. PTX-4C at 1971; PTX-72. Dr. Jarmin emailed an initial response to the 4 December 12 Letter on December 22. PTX-72. In that letter, he advised Gary that “the best 5 way to provide P.L. 94-171 block-level data with citizen voting population by race and 6 ethnicity would be through utilizing a linked file of administrative and survey data the 7 Census Bureau already possesses. This would result in higher quality data produced at a 8 lower cost.” Id. In Dr. Jarmin’s December 22 letter, he suggested to Gary a “meeting of 9 Census and DOJ technical experts to discuss the details of this proposal.” Id. 10 United States District Court 240. Following the Census Bureau’s receipt of the December 12 Letter from DOJ, 241. On January 2, 2018, Dr. Jarmin and Gary exchanged emails about meeting the 11 following week. PTX-102. Although no date was set, Gary promised to “get back to” Dr. 12 Jarmin about the timing. Id. 13 242. However, on February 6, 2018, Dr. Jarmin reported to Undersecretary Karen 14 Dunn Kelley: “I spoke with Art Gary. He has spoken with DOJ leadership. They believe the 15 letter requesting citizenship to be added to the 2020 Census fully describes their request. 16 They do not want to meet.” PTX-3 at 2136. 17 243. The Administrative Record contains no evidence that before Secretary Ross 18 issued his March 26, 2018 Decision Memo, DOJ ever met with either the Census Bureau or 19 the Commerce Department to discuss the request in DOJ’s December 12 Letter. 20 9. Outside Stakeholders Weigh In 21 244. In early 2018, prior to issuing his Decision Memo, Secretary Ross conferred 22 with multiple external stakeholders, including academics and representatives of interest 23 groups, regarding the addition of a citizenship question to the Census. This input included 24 communications from: (1) Six former Directors of the Census Bureau, PTX-1 at 1057 25 (“There is a well-proven multi-year process to suggest and test new questions. We strongly 26 believe that adding an untested question on citizenship status at this late point in the 27 decennial planning process would put the accuracy of the enumeration and success of the 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 79 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 80 of 126 1 Census in all communities at grave risk.”); (2) Members of the Census Bureau’s Census 2 Scientific Advisory Committee, PTX-1 at 794 (“We hold the strong opinion that including 3 citizenship in the 2020 Census would be a serious mistake which would result in a substantial 4 lowering of the response rate.”); and (3) various state officials, professional associations, and 5 other organizations expressing concern about the impact of the citizenship question on the 6 integrity of the 2020 Census. See e.g., PTX-1 at 787, 798, 1053, 1073, 1082, 1090, 1122, 7 1150, 1222, 1235, 1239, 1269; PTX-3 at 2281, 2284. 8 9 necessary for Section 2 VRA enforcement. See e.g., PTX-1 at 799 (letter from The 10 Leadership Conference on Civil and Human Rights); id. at 1122 (letter from national Jewish 11 organizations); PTX-3 at 2281 (letter from Constitutional Accountability Center). 12 United States District Court 245. Numerous stakeholder letters also advised that a citizenship question was not 246. In January and February 2018, before Secretary Ross issued the March 26 13 decision memo, the Leadership Conference on Civil and Human Rights, the Constitutional 14 Accountability Center, and more than 15 other external stakeholders and voting rights 15 experts, submitted letters to Secretary Ross explaining that that enforcement of Section 2 of 16 the VRA has never once previously depended upon having enumerated citizenship data since 17 the statute’s enactment in 1965. See, e.g., PTX-1 at 799, 1122; PTX-3 at 2281. 18 247. Defendants attempted to enlist stakeholders to express support for the 19 citizenship question, but had trouble doing so. PTX-71; PTX-4B at 137 (“We are trying to 20 find someone who can give a professional expression of support for the proposal in contrast 21 to the many folks we can find to give professional statements against the proposal”). 22 248. On February 13, 2018, Dr. Jarmin wrote to an individual at the American 23 Enterprise Institute (AEI): “We are trying to set up some meetings for Secretary Ross to 24 discuss the proposed citizenship question on the 2020 Census with interested stakeholders. 25 Most stakeholders will speak against the proposal. We’re looking for someone thoughtful 26 who can speak to the pros of adding such a question . . . .” PTX-4B at 138. That same day, 27 Michael Strain of the AEI responded: “None of my colleagues at AEI would speak favorably 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 80 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 81 of 126 1 about the proposal.” Id. Dr. Jarmin then wrote to Under Secretary Kelley, “Please see the 2 thread below. Appears no one at AEI is willing to speak in favor of putting question on the 3 2020.” Id. at 137. 4 249. Although they were in the minority, some groups and individuals expressed 5 support for Secretary Ross’s proposal. PTX-1 at 1198 (letter from Senator Tom Cotton); 6 PTX-1 at 1199 (letter from Senator Ted Cruz); see also PTX-1 at 1203, 1206, 1261. Director 7 Jarmin reported that certain individuals associated with the Center for Immigration Studies 8 and the Heritage Foundation endorsed the addition of the citizenship question. PTX-1 at 9 1206, 1261; PTX-4C at 1645. 10 10. Secretary Ross’s March 26 Decision Memorandum 11 250. On March 26, 2018, Secretary Ross issued his formal decision memorandum 12 (Decision Memo) announcing and explaining his decision to adopt “Option D” (known in the 13 Census Bureau memoranda as “Alternative D”) and add a citizenship question to the 14 decennial census. PTX-26. The Decision Memo states that Secretary Ross “set out to take a 15 hard look” at the citizenship question “[f]ollowing receipt of the DOJ request” for it. PTX-26 16 at 1. Secretary Ross states that DOJ has requested CVAP data for census blocks “where 17 potential Section 2 violations are alleged or suspected, and DOJ states that the current data 18 collected under the ACS are insufficient in scope, detail, and certainty to meet its purpose 19 under the VRA.” Id. 20 251. The Decision Memo includes several statements that are inconsistent with the 21 Census Bureau’s unrefuted estimate that the citizenship question would cause a differential 22 drop in self-response rates of at least 5.1 percent between households with at least one 23 noncitizen and all-citizen households. PTX-26 at 3; see also PTX-101. These include: 24  The statement that, with respect to “Option B” (the option of adding a citizenship 25 question, referred to in the Census Bureau memoranda as “Alternative B”) neither the 26 Census Bureau nor the concerned stakeholders could document that the response rates 27 would decline materially. PTX-26 at 3. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 81 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 82 of 126 1 The statement that a former Chief Operating Officer of the Census Bureau confirmed 2 that to the best of his knowledge, “no empirical data existed on the impact of a 3 citizenship question on responses.” Id. 4  Secretary Ross’s conclusion that “while there is widespread belief among many 5 parties that adding a citizenship question could reduce response rates, the Census 6 Bureau’s analysis did not provide definitive, empirical support for that belief.” Id. at 7 4. 8 9 United States District Court   The statement that “there is no information available to determine the number of people who would in fact not respond due to a citizenship question being added, and 10 no one has identified any mechanism for making such a determination.” Id. at 5. 11 252. The Decision Memo sought to justify the selection of Option D by citing 12 purported “[a]dditional empirical evidence about the impact of sensitive questions on survey 13 response rates… from the SVP of Data Science at Nielsen.” Id. at 6. The only evidence in the 14 Administrative Record from this Nielsen representative are notes from a telephone call three 15 days earlier between her (Christine Pierce) and Secretary Ross. Those notes indicate that 16 “Ms. Pierce stated that her biggest concerns [sic] was that the reinstatement of a citizenship 17 question could lead to a lower response rate.” PTX-1 at 1276. During that call, Pierce also 18 “noted the importance of testing questions.” Id. There is no “empirical evidence” in the 19 Administrative Record at all from Pierce regarding the citizenship question. 20 253. Secretary Ross also asserts in his memo that the citizenship question has been 21 “well tested” because it has been on the ACS since 2005. PTX-26 at 2. The Secretary does 22 not discuss the Census Bureau’s usual process of testing new questions before placing them 23 on the census. Nor does the memo identify any other testing standards that apply to addition 24 of the citizenship question or discuss whether those standards have been met. 25 254. The Decision Memo asserts that asking the citizenship question of all people, 26 “may eliminate the need for the Census Bureau to have to impute an answer for millions of 27 people.” Id. at 5. The Memo does not address the Census Bureau’s estimate that, with a 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 82 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 83 of 126 1 citizenship question on the census, it will have to impute the citizenship data of 13.8 million 2 people. PTX-24 at 2; see also PTX-133 at 7-10. 3 255. Secretary Ross also states that Option D “would maximize the Census Bureau’s 4 ability to match the decennial census responses with administrative records,” PTX-26 at 4, so 5 as to allow for “more complete” citizenship data. The memo does not address the Census 6 Bureau’s analysis showing that adding a citizenship question will drive down the self- 7 response rate and put more households into NRFU operations, thereby reducing the Census 8 Bureau’s ability to match survey responses with administrative records and resulting in less 9 accurate citizenship data. PTX-25 at 4. United States District Court 10 256. The Decision Memo ultimately states that the Department of Commerce 11 prioritizes the goal of “obtaining complete and accurate [CVAP] data” over all else, and 12 concludes that adding the citizenship question is the best way to achieve this goal. PTX-26 at 13 1, 7. The memo, however, points to no evidence to refute the Census Bureau’s analysis 14 which showed that use of a citizenship question would result in less accurate data than 15 administrative records alone. PTX-22, PTX-25. 16 257. The Decision Memo does, however, make the following assertion: 17 [P]lacing the question on the decennial census and directing the Census Bureau to determine the best means to compare the decennial census responses to administrative records will permit the Census Bureau to determine the inaccurate response rate for citizens and noncitizens 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 noncitizen responses to impute for that small percentage of cases where it is necessary to do so. 18 19 20 21 22 23 24 25 26 27 PTX-26 at 5. 258. Nowhere in the Administrative Record does the Census Bureau state that adding a citizenship question would increase the accuracy of its estimate of inaccurate citizenship responses. Nor is it apparent why the inaccuracy rate of responders would help impute the citizenship data of nonresponders. If actual citizenship is benchmarked to administrative records, and the Bureau would use those records in any event, then adding a citizenship question to the decennial census would not assist in the imputation. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 83 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 84 of 126 United States District Court 1 259. Finally, the Decision Memo does not discuss whether the Census Bureau’s 2 NRFU and imputation processes are likely substantially to mitigate any drop in self-response 3 caused by the citizenship question. 4 11. Testing of the Citizenship Question 5 260. The Administrative Record contains no evidence that the cognitive and field 6 testing that are part of the usual process for adding new questions to the census, as described 7 in the Census Bureau’s answer to Question 31 and in other documents in the Administrative 8 Record, were performed. Moreover, the Administrative Record contains no evidence that 9 Defendants considered any testing requirements from the Office of Management and Budget. 10 261. The Administrative Record does not include any information about what testing 11 was performed, or how the citizenship question performed on that testing, before it was 12 added to the ACS. 13 262. There is also no indication that Defendants considered obtaining any kind of 14 waiver of any applicable agency guidelines regarding testing, or that the Census Bureau 15 publicly noticed and provided a period for public comment about the citizenship question 16 before Secretary Ross made the decision to add it to the Census, as required by the Census 17 Bureau’s “well-established” process described with the Census Bureau’s answer to Question 18 31, and in other documents in the Administrative Record. See PTX-3 at 2236-37; PTX-4A at 19 155-56; PTX-4D at 812; PTX-135; PTX-141. 20 12. The Purpose and Timing of Secretary Ross’s Decision 21 263. The weight of the evidence clearly shows that Secretary Ross made the decision 22 to add a citizenship question before knowing whether DOJ had any need or even desire to 23 add the question. In other words, Secretary Ross did not decide to add the citizenship 24 question to the decennial census to aid in enforcement of Section 2 of the VRA. 25 264. There is no writing of any kind either in the Administrative Record authored by 26 the Secretary or anyone at the Commerce Department (or anyone else) that expressly and 27 directly describes the reasons why the Secretary wanted to add a citizenship question as early 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 84 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 85 of 126 1 as the first quarter of 2017. The record reflects, however, that Secretary Ross was urged to 2 include the citizenship question by Kobach, among others, to facilitate the exclusion of 3 noncitizens from the population count for congressional apportionment. See, e.g., PTX-19 at 4 2, PTX-55, PTX-437 at 2, PTX-448, PTX-449. United States District Court 5 265. The following facts are particularly relevant: (1) Secretary Ross admits that he 6 discussed a citizenship question with “senior administration officials” before he became 7 Secretary of Commerce; (2) Steve Bannon, the Chief Strategist and Senior Counselor to the 8 President, then asked him to speak with Kobach, who wanted a citizenship question on the 9 census to exclude noncitizens from the apportionment count rather than for VRA 10 enforcement purposes; (3) on May 24, 2017, following Secretary Ross’s meeting with David 11 Langdon regarding the citizenship question, Langdon exchanged emails with the Census 12 Bureau and received from them a 1988 DOJ memo about excluding “illegal immigrants” 13 from the census count, and; (4) by August of 2017, when Secretary Ross’s staff was 14 preparing “a memo and full briefing…on a citizenship question” (which was not disclosed in 15 the initially submitted administrative record), Commerce Department legal counsel emailed 16 that their “hook” was that the Department “do[es] not make decisions on how the 17 [citizenship] data will be used for apportionment,” PTX-437 at 2. 18 266. By contrast, prior to the December 12 Letter, the Administrative Record 19 includes no communications at all from Secretary Ross or any other Commerce Department 20 officials indicating that they wanted to add the citizenship question for the purpose of aiding 21 VRA enforcement. The only implicit reference to the VRA in the Administrative Record 22 prior to December 12 is the LULAC v. Perry court opinion that Neuman emailed to Secretary 23 Ross, and that Comstock emailed to Langdon in May 2017 as the Commerce Department 24 was investigating whether it could request the citizenship question itself. There is no 25 discussion in the record about the meaning of this case. 26 27 267. The fact that the Commerce Department began exploring ways to add the citizenship question without getting a request from another agency when it appeared no 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 85 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 86 of 126 1 request was forthcoming strongly suggests that aiding DOJ’s Section 2 enforcement was not 2 the real reason behind the decision to add the question. PTX-370. The fact that Comstock 3 attempted to solicit a request for the citizenship question from DHS after initially being 4 turned away by DOJ further demonstrates that the Department of Commerce’s reasons for 5 wanting to add the citizenship question had nothing to do with VRA enforcement. Id. 6 268. The finding that Secretary Ross did not add the citizenship question for Section 7 2 enforcement is also supported by a dearth of evidence explaining why he would go to such 8 lengths to persuade DOJ to request the question despite the agency’s initial refusals. United States District Court 9 269. Defendants’ decision-making process further supports the finding that Secretary 10 Ross was not motivated by VRA enforcement. The Administrative Record contains no 11 disclosure by the Commerce Department to the Census Bureau about the intent to add a 12 citizenship question until after the December 12 Letter was delivered, despite the fact that 13 Secretary Ross and his staff spent the better part of 2017 communicating about the 14 citizenship question and strategizing about how to elicit a request for it. While Secretary 15 Ross is not obligated to inform the Census Bureau about all policy discussions he has with 16 his staff, one would expect an open-minded decision-maker to consult with the experts at the 17 Census Bureau early on to learn more about the consequences of adding the citizenship 18 question to the census and to allow for the requisite testing to take place. 19 270. It is also highly implausible that Secretary Ross’s true purpose was Section 2 20 enforcement because, as set forth in Part IV.C.10, supra, and discussed in greater detail in 21 Part IV.E.2.a, infra, his Decision Memo includes statements plainly at odds with the evidence 22 in the Administrative Record. 23 271. Finally, Defendants’ initial failure to disclose the full administrative record 24 suggests bad faith in disclosing the true basis of Secretary Ross’s decision. Defendants now 25 concede that the Administrative Record consists of over 13,000 pages of documents, even 26 though their initial submission contained only 1,320 pages. See Joint Pretrial Statement 11- 27 13. Furthermore, the initial submission of the Administrative Record mischaracterized 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 86 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 87 of 126 1 Secretary Ross’s decision-making process and omitted critical information about the 2 circumstances surrounding the DOJ request letter. It is also noteworthy that Defendants 3 initially failed to produce the Census Bureau’s answer to Question 31 (and produced only the 4 Commerce Department’s more favorable version regarding question testing), and that the 5 Administrative Record includes correspondence between Ross and Comstock expressing 6 caution about what ends up in the Administrative Record. PTX-362. 272. Together, this evidence establishes that Defendants intended to use the VRA United States District Court 7 8 enforcement as a pretext for adding the citizenship question when VRA enforcement was not, 9 in fact, their true purpose. In sum, Plaintiffs have made a clear showing that (1) Secretary 10 Ross acted in bad faith in disclosing the basis of his decision, and (2) Defendants acted in bad 11 faith in compiling the Administrative Record. 12 D. Findings of Fact Based on Extra-Record Evidence 13 1. Secretary Ross’s Decision-Making Process a. Secretary Ross’s Early Interest in the Citizenship Question 14 273. Extra-record evidence confirms that Secretary Ross talked to Steve Bannon 15 16 about the citizenship question in the spring of 2017. Bannon asked Secretary Ross to speak 17 with Kobach about adding a citizenship question to the decennial census. RFA response 18 (ECF 146-6 at 3); 18 RFA response (ECF 146-3 at 31). Secretary Ross’s conversation with 19 Kobach about adding a citizenship question to the 2020 Census came before any request 20 from DOJ to add a citizenship question to the decennial census. RFA response (ECF 146-3 at 21 39); RFA Response (ECF 146-6 at 2-3). 22 274. Around that same time, Comstock set out to come up with a “legal rationale” to 23 support the Secretary’s request to add a citizenship question. Comstock Dep. 266:4-12 (ECF 24 175-3). Comstock believed that he needed another agency to request to add the question 25 because OMB and the Paperwork Reduction Act required the Commerce Department to 26 27 18 Where only one ECF docket number is listed, the docket refers to Case No. 18-cv-01865. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 87 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 88 of 126 1 “justify” why a citizenship question was needed, and Comstock understood simply saying 2 that the Secretary wanted to include the question would not “clear [the] legal thresholds.” Id. 3 at 153:6-154:11. He testified that it was his job to “help [the Secretary] find the best 4 rationale” for adding the question, because “[t]hat’s what a policy person does.” Id. at 266:4- 5 267:6. 6 275. According to Comstock, he did not “need to know what [the Secretary’s] 7 rationale might be, because it may or may not be one that is . . . legally-valid.” Id. at 267:10- 8 14. In fact, Comstock testified that the Secretary never told him why he wanted to add a 9 citizenship question to the census. Id. at 251:1-254:17. Teramoto and Department of 10 Commerce Undersecretary Kelley similarly testified that they had no knowledge of why the 11 Secretary wanted to add a citizenship question. Teramoto Dep. 32:5-23 (ECF 175-9); Kelley 12 Dep. 39:3-40:1 (ECF 175-6). 13 276. Even outside of the Administrative Record, no writing produced in discovery 14 authored by the Secretary or anyone else describes the reasons why the Secretary wanted to 15 add a citizenship question as early as the first quarter of 2017. 16 277. When Comstock contacted DOJ on May 4, 2017 for the purpose of adding a 17 citizenship question to the 2020 Census, he was not specifically seeking to promote 18 enforcement of the VRA. See Comstock Dep. 167:8-172:5. 19 278. On July 25, 2017, Secretary Ross had a further telephone conversation with 20 Kobach concerning the addition of the citizenship question to the 2020 Census. RFA 21 Response (ECF 146-3 at 40). 22 279. Comstock, Teramoto, and Undersecretary Kelley have all denied having any 23 recollection of their September 5 meeting with Secretary Ross to discuss adding the 24 citizenship question. Comstock Dep. 221:1-16; Teramoto Dep. 60:6-61:11; Kelley Dep. 25 105:5-107:15. 26 27 280. Secretary Ross has publicly claimed that “what triggered the investigation, the real study, what triggered the process that led to the determination to [add the citizenship 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 88 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 89 of 126 1 question] was the letter from the Department of Justice.” PTX-472. Secretary Ross testified 2 before the House Committee of Ways and Means on March 22, 2018. PTX-346. When asked 3 whether the Department of Commerce planned to add a citizenship question to the 2020 4 Census, Secretary Ross began his response by stating, “Department of Justice, as you know, 5 initiated the request for inclusion of the citizenship question.” Id. 6 United States District Court 7 b. Role of DOJ in Requesting the Citizenship Question 281. The Census Bureau informed federal agencies that they were to submit requests 8 for 2020 Census content by July 1, 2016. New York Tr. 995:18-996:3 (Abowd); see also 9 PTX-214. DOJ, however, did not contact the Census Bureau about adding a citizenship 10 question until December 2017. New York Tr. 996:4-10 (Abowd). Indeed, prior to the 11 December 2017 request, DOJ had never communicated to the Census Bureau that ACS 12 CVAP data was not ideal for DOJ’s VRA enforcement purposes. Id. at 996:19-23 (Abowd). 13 282. John Gore, who was responsible for drafting the DOJ request letter, was the 14 Acting Assistant Attorney General (AAAG) for Civil Rights at DOJ at the time. Gore Dep. 15 18:20-19:3 (ECF 175-4). One of the sections within the DOJ Civil Rights Division is the 16 Voting Section, which enforced Section 2 of the VRA. Id. at 19:7-14. 17 283. Prior to becoming AAAG, Gore was previously an attorney in private practice. 18 Id. at 14:20-22. As an attorney in private practice, he litigated numerous cases under the 19 VRA on behalf of defendants. Id. at 14:20-16:1. The issue of the adequacy of CVAP data 20 never came up in any of the VRA cases litigated by Gore and he never took the position a 21 plaintiff’s block-level CVAP data was insufficient because it was based on sample survey 22 data rather than a “hard count” from the decennial census. Id. at 16:19-17:10. 23 284. Gore testified he does not personally believe that it is necessary for DOJ’s VRA 24 enforcement efforts to collect CVAP data through the census questionnaire. Id. at 300:8-11. 25 He also stated that he has no experience drawing districts for the purposes of complying with 26 the preconditions for VRA liability set forth in Thornburg v. Gingles, 478 U.S. 30 (1986), or 27 using block-level data about the characteristics of populations. Gore Dep. 17:21-18:15. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 89 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 90 of 126 1 2 conversation with Hankey and then Attorney General Sessions. Id. at 73:2-75:14. On or 3 around Labor Day of 2017, Gore had a conversation with Sessions regarding a citizenship 4 question. Id. at 83:16-84:1. 5 286. Beginning roughly in mid-September, the Commerce Department initiated 6 direct conversation with Gore. Id. at 91:18-92:6, 94:17-95:5. Throughout the autumn of 7 2017, Gore spoke with three individuals from the Commerce Department about a citizenship 8 question: Peter Davidson, James Uthmeier, and Wendy Teramoto. Id. at 92:18-94:3. None of 9 these conversations were initiated by Gore or anyone in DOJ’s Civil Rights Division. Id. at 10 United States District Court 285. Gore became involved in the issue of the citizenship question through 94:17-95:5. 11 287. Gore was first contacted by Davidson in mid-September, who called Gore to 12 discuss adding the citizenship question to the 2020 Census. Id. at 92:18-93:6, 97:19-98:3. 13 Davidson asked Gore to reach out to Teramoto. Id. at 96:4-97:14. The “DOJ-DOC” issue 14 referred to in Gore’s email to Teramoto was the addition of a citizenship question to the 15 2020 Census. Id.; see also PTX-59. Gore and Teramoto spoke about the citizenship question 16 on or about September 15, 2017, Gore Dep. 102:2-103:4, however Teramoto testified that 17 she had no recollection of this conversation, Teramoto Dep. 74:22-77:20. 18 288. On or about September 22, 2017, Gore spoke on the phone with Uthmeier about 19 adding a citizenship question to the 2020 Census. Gore Dep. 118:15-17. After their call, 20 Gore was also provided with Uthmeier’s August 11 memorandum discussing the addition of 21 a citizenship question. Id. at 118:15-119:4. With the August 11 memorandum, Gore also 22 received a handwritten note from Uthmeier. Id. This handwritten note from Uthmeier 23 contained information that DOJ considered in drafting the final letter requesting a citizenship 24 question. Id. at 123:16-124:2. That note was not produced in the Administrative Record and 25 Defendants withheld it on the basis of privilege. 26 289. On or about November 1, 2017, Gore wrote the initial draft of the December 12 27 Letter. Id. at 126:2-127:2, 127:12-17. The only career staffer in the Voting Rights Section to 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 90 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 91 of 126 1 provide input at any stage of drafting the December 12 Letter was Chris Herren, Chief of 2 DOJ’s Voting Section. Id. at 151:21-153:22. In early November, Gore emailed Herren, 3 copying Ben Aguiñaga. Id. at 126:2-15. Gore attached the first draft of the letter that he had 4 written requesting a citizenship question be added to the 2020 Census questionnaire and 5 requested Herren’s input. Id. at 126:13-127:2. 6 7 first draft. Id. at 136:21-137:8. At the time, Aguiñaga and Pickett were political appointees in 8 the front office of the Civil Rights Division who began working there in 2017, having 9 graduated from law school around 2015. Id. at 133:9-134:11. 10 United States District Court 290. Herren, Aguiñaga, and Bethany Pickett provided substantive feedback on this 291. Gore does not recall sharing subsequent drafts of the December 12 Letter with 11 Herren or receiving any additional feedback from him. Id. at 444:17-445:1. Gore has no 12 recollection of receiving any other input from career Civil Rights Division staff. Id. at 13 152:12-153:5. The only other people in DOJ’s Civil Rights Division from whom Gore can 14 recall soliciting or receiving input on the draft December 12 Letter were Pickett and 15 Aguiñaga. Id. at 136:21-137:8. 16 292. In mid-November Gore discussed adding a citizenship question with Rachael 17 Tucker, then counsel in the Office of the Attorney General, and Robert Troester, then 18 Associate Deputy Attorney General. Id. at 141:6-14. Neither Tucker nor Troester had 19 experience as counsel in VRA cases, litigating Section 2 redistricting cases involving the use 20 of CVAP data, or otherwise assessing the reliability of CVAP data used in VRA litigation. 21 Id. at 140:1-17. 22 293. In late November 2017, again Gore solicited and received edits on the draft 23 letter requesting a citizenship question from Tucker and Troester. Id. at 138:12-142:17. Gore 24 did not receive substantive edits from anyone besides Tucker and Troester in the last few 25 days before the December 12 Letter was sent. Id. at 146:7-11. 26 27 294. Final authorization to send the letter came from either Tucker or Troester on behalf of Attorney General Sessions in mid-December. Id. at 159:18-160:18. Attorney 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 91 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 92 of 126 1 General Sessions ultimately made the decision for DOJ to request that the Census Bureau 2 ask a citizenship question on the Census. Id. at 442:13-19. United States District Court 3 295. Attorney General Sessions subsequently directed DOJ to refuse to meet with 4 the Census Bureau to discuss DOJ’s request in the December 12 Letter. Id. at 271:21-272:13. 5 296. Dr. Hermann Habermann described meetings with a requesting agency, such as 6 the meeting Director Jarmin requested with DOJ on December 22, 2017, as “normal Census 7 Bureau procedure. [Such a meeting] allows the technical experts to better understand how 8 the Census Bureau can meet the needs of the proposers.” PTX-821 at ¶¶ 28-29. Dr. Jarmin 9 similarly testified that it was typical for the Census Bureau to meet with federal agencies 10 requesting data in order to understand their needs and come up with the best way to meet 11 those needs. Jarmin Dep. 33:1-15, 36:14-20 (ECF No. 175-5). Dr. Jarmin did not agree with 12 DOJ’s reasoning for refusing to meet because the Census Bureau would have liked 13 additional information on how DOJ used CVAP data. Id. at 101:9-20. 14 297. Dr. Abowd also observed that it is “very unusual” for an agency to make a 15 request to the Census Bureau to collect data through the census but then refuse to meet to 16 discuss the technical aspect of that data request. Tr. 1055:8-12 (Abowd). He also stated that 17 it is “very unusual” for the head of a cabinet agency personally to direct staff not to meet 18 with the Census Bureau to discuss the Census Bureau’s ideas for producing better quality 19 data for that agency at a lower cost. Id. at 1055:13-17 (Abowd). 20 298. It is unknown whether the block-level CVAP data collected with a citizenship 21 question on the 2020 Census will have a margin of error any more precise than the CVAP 22 data on which the Department of Justice currently relies. New York Tr. 1045:19-1046:8 23 (Abowd). Gore is not aware of any communications between DOJ and the Census Bureau 24 about whether or not adding a citizenship question to the census would in fact produce data 25 that has smaller margins of error than the citizenship data currently used by DOJ, due to 26 required disclosure avoidance techniques. Gore Dep. 228:11-20, 233:4-234:2. 27 c. Role of the Census Bureau in Evaluating the Citizenship Question 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 92 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 93 of 126 1 2 email on December 15, 2017. New York Tr. 879:9-17 (Abowd). Following receipt of the 3 December 12 Letter, Dr. Jarmin asked Dr. Abowd to assemble a team of technical experts, 4 nicknamed the SWAT team, to discuss how to respond to the DOJ request. Id. at 878:23- 5 880:5 (Abowd). The SWAT team specifically looked into using administrative records in 6 lieu of a citizenship question on the census because Dr. Jarmin believed that, under Title 13, 7 the Bureau is supposed to use administrative records in lieu of direct collection when 8 possible. Jarmin Dep. 59:9-60:7. A group of Census Bureau decision-makers in collaboration 9 with Department of Commerce Undersecretary Kelley ultimately decided not to conduct a 10 randomized controlled trial of the content of the citizenship question. New York Tr. 925:19- 11 22 (Abowd). 12 United States District Court 299. Dr. Abowd first learned about DOJ’s request to add a citizenship question via 13 d. January 19 Memo 300. The January 19 Memo to Secretary Ross summarized the opinions of the 14 Census Bureau senior executive staff and was based on the SWAT team’s work and other 15 Census Bureau research. Id. at 880:10-18 (Abowd). The January 19 Memo memorializes the 16 Census Bureau’s credible, quantitative evidence, as well as its analysis, that adding a 17 citizenship question to the 2020 Census could be expected to lower the self-response rate in 18 households that may contain noncitizens. Id. at 881:4-10 (Abowd). 19 301. The views in the January 19 Memo are a summary of the technical work 20 performed by the SWAT team and includes contributions made by other senior executives at 21 the Census Bureau. Id. at 883:4-6 (Abowd). Dr. Abowd agrees with the conclusions in the 22 January 19 Memo, id. at 883:7-9 (Abowd), which were ultimately reviewed and approved by 23 Dr. Jarmin, id. at 883:10-12 (Abowd). 24 302. The analyses in section B(1), (2), and (3) of the memo all support the 25 conclusion that the citizenship question would cause a lower self-response rate to the 2020 26 Census. Id. at 890:3-891:3 (Abowd). At the time the memo was released, 5.1 percent was the 27 Census Bureau’s best estimate of the effect of adding a citizenship question in terms of the 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 93 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 94 of 126 1 question’s differential impact of self-responses of noncitizen households as compared to 2 citizen households. Id. at 893:12-22 (Abowd). A reduction in self response of 5.1 of 3 noncitizen households would send more than a million additional people into NRFU. Id. at 4 894:1-16 (Abowd). 5 303. The lower self-response rates resulting from adding a citizenship question will 6 increase the cost of conducting the 2020 Census. Id. at 950:10-13 (Abowd). This is because 7 more people will have to be enumerated through NRFU, which costs money. Id. at 950: 15- 8 20 (Abowd). The Census Bureau’s estimate that NRFU costs will increase by $27.5 million 9 is conservative because, among other things, the differences in self-response rates to the 10 2020 Census between citizen and noncitizen households is likely to be even greater than 11 estimated in the memo. Id. at 951:11-19 (Abowd). Another reason this figure represents a 12 lower-bound cost estimate is that it may take more NRFU visits to enumerate households 13 that do not respond due to the citizenship question than assumed. Id. at 952:2-6 (Abowd). 14 Moreover, this cost estimate does not incorporate any estimate about the effect of a 15 citizenship question on reducing self-response rates from all-citizen households. Id. at 952:7- 16 11 (Abowd). Finally, this figure does not capture increased communication campaign costs 17 that may be needed as a result of the citizenship question. Id. at 952:12-16 (Abowd). 18 304. Citizenship status is a characteristic where administrative records tend to be 19 more accurate than survey responses. Id. at 955:21-24 (Abowd). In other words, when an 20 ACS response says that a person is a citizen, but the administrative records says the person is 21 not a citizen, the most likely conclusion is that the person is, in fact, a noncitizen. Id. at 22 955:6-20 (Abowd). In fact, for more than 30 percent of noncitizens who provide a response 23 to the ACS citizenship question, the response is incorrect. Id. at 956:16-21 (Abowd). 24 305. The Bureau has no empirical basis to believe that noncitizens for whom a 25 response is provided to the citizenship question on the census will have more accurate 26 responses than they do to the citizenship question on the ACS. Id. at 956:22-957:2 (Abowd). 27 Instead, the Census Bureau found strong indications that responses by noncitizens to a 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 94 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 95 of 126 1 citizenship question on the 2020 Census will be even less accurate than they have 2 historically been on the ACS. Id. at 957:3-7 (Abowd). United States District Court 3 306. The Census Bureau still hasn't made any determination about how it will 4 address disagreement between survey responses and the administrative records when 5 producing block-level CVAP data used by the Department of Justice after the 2020 Census. 6 Id. at 957:8-13 (Abowd). The Bureau has concluded, however, that using administrative 7 records would deliver higher quality block-level CVAP data by race and ethnicity than 8 including a citizenship question on the census. Id. at 958:19-22 (Abowd). The Census 9 Bureau's proposal to generate such block-level CVAP data using administrative records 10 rather than a citizenship question had the backing of the Census Bureau's redistricting office. 11 Id. at 958:24-959:3 (Abowd). 12 307. In the January 19 Memo, the Census Bureau concluded that a citizenship 13 question on the 2020 Census would be a sensitive one for Hispanics. Id. at 917:4-7 (Abowd). 14 This memo also provided empirical support for the conclusion that adding a citizenship 15 question will reduce self-response rates to the 2020 Census. Id. at 922:4-10 (Abowd). Dr. 16 Jarmin agrees with the findings in the January 19 Memo, including that using administrative 17 records would provide higher quality CVAP data at a lower cost than a citizenship question 18 on the 2020 Census. Jarmin Dep. 65:22-67:2, 115:20-117:15. 19 20 e. The Set of 35 Questions Answered by the Census Bureau 308. After the Census Bureau communicated its views to Secretary Ross, the 21 Commerce Department sent a list of 35 follow-up questions to the Census Bureau. New 22 York Tr. 1004:19-25 (Abowd). Dr. Abowd was charged with making sure that the responses 23 to the 35 questions were accurate. Id. at 1005:23-1006:1 (Abowd). 24 309. As of March 1, 2018, it was Dr. Abowd’s understanding that adding a new 25 question to the decennial census involves extensive testing, review, and evaluation. Id. at 26 1007:7- 9 (Abowd). The answer to Question 31 found at AR 10900 summarized the Census 27 Bureau and OMB’s formal process for making content changes to the census. Id. at 1006:2- 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 95 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 96 of 126 1 1008:8 (Abowd); Jarmin Dep. 137:6-138:21; see also PTX-4D at 1845-46; PTX-133 at 21- 2 22. 3 4 administrative record, he does not know who wrote it, and it does not appear in the last 5 version of the document in the possession of the Census Bureau. New York Tr. 1010:12- 6 1011:3 (Abowd). The text in that document is not the text that the Census Bureau transmitted 7 to the Commerce Department, id. at 1014:18-23 (Abowd), and does not represent the final 8 Census Bureau version. Jarmin Dep. 137:6-16. Dr. Jarmin also does not know who wrote the 9 answer to Question 31 that appears in Defendants’ initial administrative record submission. 10 11 United States District Court 310. Dr. Abowd did not write the answer to Question 31 that appeared in the initial Id. at 211:19-21. f. February 12 Meeting Between Census Bureau and Secretary Ross 12 311. Dr. Abowd met with Secretary Ross to discuss the January 19 Memo on 13 February 12, 2018. New York Tr. 883:17-19 (Abowd). In addition to Dr. Abowd, the Census 14 Bureau staff attending the meeting were Dr. Jarmin, Dr. Llamas, Associate Director for the 15 2020 Census Al Fontenot, Assistant Director for the 2020 Census Jim Treat, and Special 16 Assistant to the Director Krista Jones. Tr. 824:10-18 (Abowd). The February 12 meeting also 17 included several members of Secretary Ross’s staff. Id. 18 312. The February 12 meeting was the only meeting Dr. Abowd had with Secretary 19 Ross to discuss the citizenship question before he issued the Decision Memo. New York Tr. 20 884:11-14 (Abowd). Prior to the meeting with Secretary Ross, Dr. Abowd had a pre-meeting 21 with Undersecretary Kelley to discuss the memo. Id. at 883:20-884:1 (Abowd). During that 22 meeting, she did not express any disagreements with the analysis in the January 19 Memo. 23 Id. at 884:2-5 (Abowd). 24 25 26 27 313. At the February 12 meeting, Ross quickly dismissed Alternative A (not collecting block-level CVAP data) as a possibility. Tr. 826:20-25 (Abowd). 314. Dr. Abowd informed Secretary Ross during the February 12 meeting that the Census Bureau thought that the difference in self-response rates on the ACS and the census, 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 96 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 97 of 126 1 when comparing citizen and noncitizen households, was related to the citizenship question 2 on the ACS. New York Tr. 922:11-17 (Abowd). He also explained the Census Bureau’s 3 conclusion that using administrative records would better meet DOJ’s stated uses than 4 relying on a citizenship question on the census. Id. at 959:12-19 (Abowd). 5 6 and Undersecretary Kelley wanted Abowd to evaluate Alternative D (using both 7 administrative records and citizenship question responses). Id. at 965:25-966:5 (Abowd). 8 9 United States District Court 315. After the February 12 meeting, Dr. Jarmin told Dr. Abowd that Secretary Ross g. March 1 Memo 316. The views in the March 1 memo are those of the senior executive staff at the 10 Census Bureau. New York Tr. 966:23-25 (Abowd). The Census Bureau did not recommend 11 Alternative D and still does not recommend Alternative D. Id. at 967:16-21 (Abowd). 12 Instead, the memo concluded that using administrative records alone would be more accurate 13 than attempting to combine administrative records and survey responses under Alternative 14 D. Id. at 988:12-16 (Abowd). 15 317. As the memo explains, survey-collected citizenship data would not be reliable 16 for many of the people falling in the gaps in the administrative records. Id. at 973:23-974:3 17 (Abowd). The memo specifically states that citizenship survey data gathered under 18 Alternative D would be of “suspect quality” whereas administrative data on citizenship is 19 “high quality.” Id. at 974:12-20 (Abowd). 20 318. Accordingly, for the portion of the population that cannot be linked to 21 administrative records, it would be more accurate to impute/model their citizenship status 22 based on administrative records (Alternative C) than to obtain the information through their 23 survey responses (Alternative D). Id. at 974:22-975:9 (Abowd). This is the view of both Dr. 24 Abowd and the Census Bureau. Id. at 974:22-975:15. 25 319. Moreover, under Alternative D, due to the lower quality personal data on 26 census responses from an increased number of households going through NRFU, there will 27 be a reduction in the number of individuals whom the Census Bureau can link to 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 97 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 98 of 126 1 administrative records. Id. at 969:2- 23 (Abowd). In other words, the number of individuals 2 that could be matched to administrative records that contain citizenship information would 3 be higher under Alternative D than under Alternative C. Id. at 975:17-21 (Abowd). 4 5 6 United States District Court 7 320. The Census Bureau also concluded in this memo that Alternative C is cheaper than Alternative D. Id. at 988:9-11 (Abowd). 321. These analyses and conclusions were communicated to Secretary Ross before he issued his Decision Memo. Id. at 988:17-19 (Abowd). 8 h. Secretary Ross’s March 26 Decision Memorandum 9 322. Although Secretary Ross repeatedly claims in the Decision Memo that there 10 was a lack of evidence that the citizenship question would lower self-response rates, Dr. 11 Abowd provided Secretary Ross with credible, quantitative evidence that doing so would 12 lower the self-response rate for households that contain a noncitizen. Tr. 1059:16-21 13 (Abowd). 14 323. The Decision Memo also states that Option D will provide more “complete and 15 accurate” citizenship data than using administrative records alone. PTX-26 at 5, 7. However, 16 the Census Bureau’s analysis concluded that adding a citizenship question is not necessary to 17 provide complete and accurate data in response to the Department of Justice’s request. Tr. 18 1063:18-22 (Abowd). 19 324. The Decision Memo stated: 20 Finally, placing the question on the decennial census and directing the Census Bureau to determine the best means to compare the decennial census responses to administrative records will permit the Census Bureau to determine the inaccurate response rate for citizens and noncitizens 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 noncitizen responses to impute for that small percentage of cases where it is necessary to do so. 21 22 23 24 25 26 27 PTX-26 at 5. However, as of March 26, 2018, the Census Bureau had not analyzed these presumptions. New York Tr. 977:25-978:7 (Abowd). The presumptions were never discussed with Dr. Abowd and the Census Bureau does not agree with them. Id. at 976:18977:24 (Abowd). In fact, adding a citizenship question will make it more difficult for the 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 98 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 99 of 126 1 Census Bureau to establish the accurate ratio of citizen to noncitizen responses to impute. Tr. 2 1061:8-11 (Abowd). 3 325. The March 26 Decision Memo states that “no one has identified any 4 mechanism” for determining whether the citizenship question would cause a drop in self- 5 response. PTX-26 at 5. There were, however, several mechanisms available to determine 6 whether the citizenship question would cause people not to participate in the census. Tr. 7 1061:23-1062:11 (Abowd). One such mechanism was the statistical analysis performed by 8 the Census Bureau. Id. at 1061:23-1062:3 (Abowd). Another mechanism would have been 9 an RCT, which the Census Bureau could have conducted, but did not. Id. at 1062:4-11 10 (Abowd). A group-of decision-makers in collaboration with Undersecretary Kelley decided 11 not to conduct a randomized controlled trial of the content of the citizenship question. New 12 York Tr. 925:19-22 (Abowd). 13 2. Testing Requirements for New Questions a. The Census Bureau’s Statistical Quality Standards 14 15 326. The Census Bureau’s Statistical Quality Standards set forth the Bureau’s 16 internal standards, guidelines, and requirements on pretesting questionnaires and data 17 collection instruments. PTX-205; Tr. 82:6-19 (O’Muircheartaigh), 832:1-833:8 (Abowd). 18 The Statistical Quality Standards apply when new questions are added to a data collection 19 instrument or existing questions are revised. PTX-205 at 18. 20 21 22 23 24 327. Sub-Requirement A2-3.3 of the Statistical Quality 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 18; Tr. 82:9-84:13 (O’Muircheartaigh), 833:11-18 (Abowd). 25 328. Under Sub-Requirements A2-3.3-1c and A2-3.3-1d, pretesting must be 26 performed when “Review by cognitive experts reveals that adding pretested questions to an 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 99 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 100 of 126 1 existing instrument may cause potential context effects” and when “An existing data 2 collection instrument has substantive modifications (e.g., existing questions are revised or 3 new questions added).” PTX-205 at 18. 4 5 6 7 8 9 10 329. One exception to the pretesting requirement of the Statistical Quality Standards is that, “On 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.; Tr. 833:19834:7, 1046:18- 1047:19 (Abowd). 330. Another exception is that, “Pretesting is not required for questions that 11 performed adequately in another survey.” PTX-205 at 18; Tr. 833:19-834:7, 1047:12-19 United States District Court 12 (Abowd). Those and other similar pretesting standards are used in the survey methodology 13 14 15 16 17 and data collection profession more generally. Tr. 84:19-21 (O’Muircheartaigh). b. OMB’s Standards and Guidelines for Statistical Surveys 331. Under Congress’s direction, the Office of Management and Budget (OMB) has also issued standards for designing, developing, and pretesting survey content. PTX-821 at ¶¶ 55-56; PTX-262; PTX-266; PTX-267; PTX-612. The OMB-promulgated standards for 18 pretesting content on data collection instruments can be found in the OMB Standards and 19 Guidelines for Statistical Surveys. PTX-266. 20 332. The Census Bureau must follow the OMB Standards and Guidelines for 21 Statistical Surveys when preparing for and implementing the decennial census. Census 22 Bureau 30(b)(6) Dep. Vol. I 321:14-17; New York Tr. 989:15-17 (Abowd); PTX-821 at 23 ¶¶ 55-56; Tr. 88:22-89:12 (O’Muircheartaigh). These guidelines require that agencies 24 conduct a pretest of all components of a survey, including by conducting a field test and full 25 “dress rehearsal” for “highly influential surveys.” PTX-266 at 14. 26 333. OMB Standard 1.4 requires that agencies “ensure that all components of a 27 survey function as intended when implemented in the full-scale survey and that measurement 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 100 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 101 of 126 1 error is controlled” prior to implementing the data collection instrument. Id. at 14; PTX-821 2 at ¶ 56. This is done either by “conducting a pretest of the survey components” or by “having 3 successfully fielded the survey components on a previous occasion.” PTX-266 at 14. 4 5 6 7 8 9 10 334. OMB 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.” Id. at 16. OMB Guideline 2.3.1 similarly demands that agencies “[d]esign the data collection instrument in a manner that minimizes respondent burden, while maximizing data quality.” Id. at 16. 335. The Census Bureau has conceded that—within the meaning of OMB Standard 2.3—Alternative D would result in lower data quality, higher cost, and higher respondent 11 burden than the Census Bureau’s recommended Alternative C. Census Bureau 30(b)(6) Dep. United States District Court 12 Vol. I 321:18-322:19; New York Tr. 989:6-990:6 (Abowd). While Alternative C would 13 14 15 16 17 comport with OMB Guideline 2.3.1, New York Tr. 990:7-991:1 (Abowd), Alternative D would not, given the degradation to data quality that would result. c. Census Bureau Process for Adding or Modifying Census Content 336. In addition to abiding by the standards set forth above, the Census Bureau follows a well-established process for adding or modifying questions on the decennial 18 census. This decade-long process involves multiple tests, including various randomized 19 control trials. New York Tr. 994:18-22 (Abowd). It also involves extensive cognitive and 20 field testing, ongoing research, and input from advisory committees. Id. at 996:24-997:14 21 (Abowd); PTX-214 at 4; Jarmin Dep. 47:13-48:17, 52:5-11, 138:16-139:19. 22 337. Another standard pretesting method is the randomized control trial (RCT), 23 which tests an operation with an added element and compares that to a test of the operation 24 without the element. Tr. 102:7-23 (O’Muircheartaigh). According to Dr. Abowd, the RCT is 25 the “gold standard” for testing a proposed question’s effect on the census count and data 26 collection. Id. at 874:10-23, 1039:10-17 (Abowd); New York Tr. 923:16-924:9 (Abowd); 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 101 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 102 of 126 1 2 Census Bureau 30(b)(6) Dep. Vol. II 426-430. 338. Based on the result of pretesting, the Census Bureau must finalize the actual 3 2020 Census questionnaires and then must submit them for OMB approval of the 2020 4 Census information collection. PTX-821 at ¶ 62. 5 6 7 8 9 10 11 d. Past Practices for Testing the Decennial Census Questionnaire 339. Past decennial census questionnaires—the complete 2010 Census questionnaire, for example—were subject to extensive cognitive testing and field testing. New York Tr. 997:11- 23 (Abowd). After the 1990 Decennial Census, the Census Bureau investigated the possibility of adding a question concerning respondents’ Social Security numbers on the decennial census short form questionnaire. Id. at 998:25-999:4 (Abowd). 340. To test that potential Social Security number question, the Census Bureau conducted an RCT comparing a version of the short form with the Social Security number United States District Court 12 question and one without. Id. at 999:5-8 (Abowd). That RCT allowed the Bureau to assess 13 the impact on self-response rates of a Social Security number question. Id. at 999:9-11. 14 15 16 17 341. In that RCT, the self-response rate fell off in the group that had the Social Security number question by 3.4 percent. Id. at 999:12-15. The conclusion drawn from that RCT was that asking for a Social Security number would be sensitive. Id. at 999:16-18 (Abowd). As a result, the Census Bureau decided not to include a Social Security number 18 question on the decennial census questionnaire. Id. at 999:19-24 (Abowd). The Census has 19 never requested Social Security numbers on the census questionnaire, and one of the reasons 20 is the effect of the question on self-response rates, as revealed by the RCT. Id. 21 342. The RCT to assess the impact of a Social Security number question was 22 conducted before any decision was made about whether to include a Social Security number 23 question on the decennial census. Id. at 1000:8-13 (Abowd). 24 3. Whether ACS Data is Sufficient for VRA Enforcement Purposes 25 343. Plaintiffs offer the expert testimony of Dr. Lisa Handley and Professor Pamela 26 S. Karlan regarding whether the inclusion of a question about citizenship status in the 27 decennial census would assist in the enforcement Section 2 of the VRA. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 102 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 103 of 126 344. Dr. Handley is a consultant in redistricting and in electoral district design. She United States District Court 1 2 has over thirty years of experience as an expert in redistricting, minority voting rights, and 3 the use of census data for voting rights enforcement purposes, advising governments, non- 4 profits, and NGOs on minority voting rights and redistricting-related issues and serving as an 5 expert in dozens of voting rights cases, including five Section 2 redistricting cases on behalf 6 of DOJ. PTX-819 at 788:22-796:3 (Handley). Based on her education, experience, and 7 knowledge, Dr. Handley is well-qualified to offer reliable and credible opinions on Section 2 8 of the VRA, and the use of census data in Section 2 litigation and enforcement proceedings. 9 345. Dr. Handley testified that in her professional opinion “currently available 10 census data has proven perfectly sufficient to ascertain whether an electoral system or 11 redistricting plan dilutes minority votes.” Id. at 796:22-797:12, 819:19-23 (Handley); PTX- 12 650. 13 346. Citizenship data is most relevant to the first Gingles precondition, which is that 14 the minority group in question be sufficiently large and geographically compact to constitute 15 a majority in a single-member district. Id. at 798:6-799:19 (Handley); see also Karlan Trial 16 Dep. 30:25-32:14 (ECF 145). Dr. Handley testified that, in her experience, CVAP estimates 17 at the census tract or block group level are generally sufficient to satisfy the first Gingles 18 precondition in Voting Rights Act cases. PTX-819 at 807:24-811:6 (Handley). 19 347. Moreover, Dr. Handley explained, where it would be helpful to present CVAP 20 data at the block level, this information can be reliably and accurately estimated by applying 21 the CVAP ratios from the census tract level to the block-level figures for total voting-age 22 population. Id. at 808:10-815:5 (Handley). In Dr. Handley’s expert opinion, block-level 23 CVAP estimates derived from ACS data at the census tract or block group levels are reliable 24 and accurate. Id. at 815:8-819:23, 855:23-856:10 (Handley). 25 348. Dr. Handley further explained that, in the district-specific functional analysis 26 that she employs in VRA analysis, which is also used by DOJ, the outcome does not depend 27 on a precise measurement of CVAP at the individual block level, but rather on an analysis of 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 103 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 104 of 126 1 turnout rates and voting patterns within a district. The exact number of minority citizens of 2 voting age at the block level is “essentially irrelevant” to the analysis. Id. at 820:2-823:11 3 (Handley). 4 5 year ACS CVAP data, and she is also not aware of any VRA claim that failed due to a 6 plaintiff’s reliance upon 5-year ACS CVAP data. Id. at 832:14-21 (Handley). 7 United States District Court 349. Dr. Handley’s work as a VRA expert has never been impeded by her use of 5- 350. Professor Karlan has served as Deputy Assistant Attorney General in DOJ’s 8 Civil Rights Division from January 2014 through September 2015, overseeing the work of 9 the Voting Section, which enforces the VRA. She has also served as assistant counsel to the 10 NAACP Legal Defense Fund, litigating voting rights cases on behalf of plaintiffs and amici 11 curiae, including numerous cases brought under Section 2 of the VRA before the U.S. 12 Supreme Court. Karlan Trial Dep. 10:9-21, 12:18-14:7, 22:3-15. She has co-authored two 13 casebooks which covers the VRA, among other topics, and has written approximately one 14 dozen academic articles about the VRA. Id. at 18:2-20:3; 24:22-25:21. 15 351. Based on her education, experience, and knowledge, Professor Karlan is well- 16 qualified to offer reliable and credible opinions on Section 2 of the VRA, and the use of 17 census data in Section 2 litigation and enforcement proceedings. 18 352. Professor Karlan testified that, in her professional opinion, existing data sources 19 from the ACS are sufficient for plaintiffs to bring and prevail in cases brought under Section 20 2 of the VRA. Id. at 29:14-23, 66:15-23. She specifically observed that no reported Section 2 21 case has ever failed on account of the purported inadequacy of ACS data (or, prior to the 22 advent of the ACS, data from the longform census questionnaire) as a measure of CVAP. Id. 23 at 52:14-53:18. 24 353. Nothing in the DOJ’s December 12 Letter altered Professor Karlan’s 25 professional opinion that existing data sources are sufficient for plaintiffs to bring and 26 prevail in litigation under Section 2 of the Voting Rights Act. Id. at 66:24-67:20. Indeed, the 27 December 12 Letter did not identify any cases in which the inaccuracy or inadequacy of 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 104 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 105 of 126 United States District Court 1 ACS data caused a plaintiff to lose a Section 2 case. Id. at 54:5-15. 2 E. Conclusions of Law 3 “A person suffering legal wrong because of agency action . . . is entitled to judicial 4 review thereof.” 5 U.S.C. § 702. Under the APA, “[t]he reviewing court shall . . . hold 5 unlawful and set aside agency action, findings, and conclusions found to be,” among other 6 things, “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with 7 law.” 5 U.S.C. § 706(2)(A). 8 1. “Contrary to Law” Review 9 An agency decision may also violate the APA if the agency action is contrary to law. 10 5 U.S.C. § 706(2)(A); Nat’l Wildlife Fed’n v. U.S. Army Corps of Eng’rs, 384 F.3d 1163, 11 1166 (9th Cir. 2004). Based on the Administrative Record alone, the decision to add the 12 citizenship question violates the APA as contrary to law. 13 a. Section 6(c) 14 Section 6(c) of the Census Act requires the Secretary to use administrative records to 15 address DOJ’s data request rather than adding a citizenship question on the census. Although 16 Congress delegated to the Secretary a degree of discretion in conducting the census, section 17 6(c), among other provisions, limits that discretion. Title 13, section 6 states in full: 18 (a) The Secretary, whenever he considers it advisable, may call upon any other department, agency, or establishment of the Federal Government, or of the government of the District of Columbia, for information pertinent to the work provided for in this title. 19 20 21 22 23 24 (b) The Secretary may acquire, by purchase or otherwise, from States, counties, cities, or other units of government, or their instrumentalities, or from private persons and agencies, such copies of records, reports, and other material as may be required for the efficient and economical conduct of the censuses and surveys provided for in this title. 26 (c) To 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. 27 13 U.S.C. § 6. Subdivision (c) of section 6 was added to the statute in the 1976 Census Act. 25 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 105 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 106 of 126 1 See 1976 Census Act § 5(a), 90 Stat. at 2460. So while the statute previously merely 2 authorized the use of government records for census-related purposes, the amendment made 3 the use of those records a mandatory alternative to “direct inquiries” in certain circumstances. 4 Subdivision (c) of section 6 serves “the dual interests of economizing and reducing 5 respondent burden.” H.R. CONF. REP. No. 94-1719, at 10 (1976), reprinted in 1976 6 U.S.C.C.A.N. 5476, 5478. United States District Court 7 The Administrative Record here demonstrates that it was “possible” to “acquire and 8 use” administrative records from other government agencies that would produce data 9 “consistent with the kind, timeliness, quality and scope of the statistics required.” 13 U.S.C. 10 § 6(c). The “kind” of data here would be the same regardless of whether it is gathered by 11 administrative records or a census question—in both cases, the relevant data is simply block- 12 level data on citizens versus noncitizens. Moreover, the Census Bureau advised Secretary 13 Ross in the March Memo that regardless of whether administrative data or a citizenship 14 question were used, there was no difference in timing on when the citizenship data would be 15 available, PTX-133 at 11, and there is no evidence in the Administrative Record to the 16 contrary. 17 Finally, the Census Bureau repeatedly advised Secretary Ross that the quality of the 18 citizenship data would be higher from administrative records than from a citizenship question 19 due to noncitizens’ propensity to self-report as citizens. See Parts IV.C.5.d, IV.C.7, supra. 20 The scope of the data would be the same, regardless of source, because data would be 21 obtained for residents of the entire country (with a minority requiring imputation, regardless 22 of the data source). 23 Thus, under every criterion set forth in section 6(c), with no evidence in the record to 24 the contrary, using administrative records alone was superior to adding a citizenship question 25 to the decennial census. Secretary Ross was therefore legally required to obtain citizenship 26 data through administrative records rather than a citizenship question on the census. His 27 decision to do otherwise violates the APA. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 106 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 107 of 126 b. Section 141(f) 1 Adding a citizenship question to the 2020 Census is also contrary to law because 2 3 Secretary Ross changed the subjects to be included on the Census after submitting his 4 Section 141(f)(1) report despite the absence of any new circumstances justifying such a 5 change. Section 141(f) provides: 6 (f) With respect to each decennial and mid-decade census conducted under subsection (a) or (d) of this section, the Secretary shall submit to the committees of Congress having legislative jurisdiction over the census— 7 8 (1) not later than 3 years before the appropriate census date, a report containing the Secretary’s determination of the subjects proposed to be included, and the types of information to be compiled, in such census; 9 10 (2) not later than 2 years before the appropriate census date, a report containing the Secretary’s determination of the questions proposed to be included in such census; and 11 United States District Court 12 (3) after submission of a report under paragraph (1) or (2) of this subsection and before the appropriate census date, if the Secretary finds new circumstances exist which necessitate that the subjects, types of information, or questions contained in reports so submitted be modified, a report containing the Secretary’s determination of the subjects, types of information, or questions as proposed to be modified. 13 14 15 16 17 13 U.S.C. § 141(f). Secretary Ross submitted his section 141(f)(1) report in March of 2017. 18 PTX-264. That report did not include the subject of citizenship as a topic for the 2020 19 Census. Id. In March of the following year, Secretary Ross submitted his section 141(f)(2) 20 report.19 Consistent with his Decision Memo, that report states a citizenship question will be 21 included on the 2020 Census. Secretary Ross has not, however, submitted a report pursuant 22 to subdivision (f)(3), nor did he identify the “new circumstances” that justified the addition 23 of this new topic. 24 Defendants argue that only Congress can enforce section 141(f) because courts 25 26 27 19 Judicial notice is taken of Secretary Ross’s 141(f)(2) report, Questions Planned for the 2020 Census and American Community Survey, available at https://www.census.gov/library/publications/2018/dec/planned-questions-2020-acs.html. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 107 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 108 of 126 1 cannot redress any injury resulting from an inadequate report. Guerrero v. Clinton, 157 F.3d 2 1190, 1191 (9th Cir. 1998); Renee v. Duncan, 686 F.3d 1002, 1016-17 (9th Cir. 2012). In 3 both Guerrero and Renee the Ninth Circuit held that the courts could not police the quality 4 of congressional reports that were “purely informational” and from which no legal 5 consequences flowed. Guerrero, 157 F.3d at 1194-1195; Renee, 686 F.3d at 1016-17. United States District Court 6 While Defendants are correct that the primary function of section 141(f) is to impose 7 reporting requirements, it also imposes substantive limitations on the Secretary’s ability to 8 modify the census. This section specifically prevents the Secretary of Commerce from 9 changing the topics or questions to be included in the census after submission of the (f)(1) or 10 (f)(2) reports absent “new circumstances” justifying the change. Therefore, while Defendants 11 present a credible argument that the reporting requirement is generally not subject to judicial 12 review, they do not address section 141(f)(3)’s substantive restrictions on the Secretary’s 13 authority to make last minute changes to the census without good cause, which is judicially 14 reviewable. 15 Section 141(f)(3) requires that, “if the Secretary finds new circumstances exist which 16 necessitate that the subjects, types of information, or questions” included in the census be 17 modified, he submit a new report detailing these modifications. 13 U.S.C § 141(f)(3). The 18 plain meaning of this provision is that the Secretary may not deviate from the subjects and 19 questions outlined in his or her reports to Congress absent new circumstances. Construing 20 the statute in any other manner would render the “new circumstances” clause utterly 21 meaningless. This runs afoul of one of the most basic canons of construction, that a “statute 22 should be construed so that effect is given to all its provisions, so that no part will be 23 inoperative or superfluous, void or insignificant.” Corley v. United States, 556 U.S. 303, 314 24 (2009) (quoting Hibbs v. Winn, 542 U.S. 88, 101 (2004)). 25 The only interpretation of the statute that gives full effect to all provisions is that, 26 once the Secretary has submitted either the (f)(1) or (f)(2) reports, he or she may not deviate 27 from the contents set forth in those reports absent “new circumstances” that necessitate a 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 108 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 109 of 126 1 late-stage change to the census. Here, this inquiry ultimately collapses with the pretext 2 question discussed above. Indeed, the only new circumstances identified by Defendants is 3 the submission of the DOJ request letter. As previously explained, the VRA enforcement 4 rationale is nothing more than a pretext designed to provide cover for the Secretary’s 5 unexplained desire to add the citizenship question to the census. Just as the DOJ letter does 6 not provide the true basis for Secretary Ross’s decision, neither may it qualify as a new 7 circumstance that requires a change to the census. Accordingly, the Secretary’s decision to 8 add the citizenship question was contrary to law for this reason as well. 9 United States District Court 10 2. Arbitrary and Capricious Review The standard for evaluating whether an agency’s decision was arbitrary and 11 capricious is whether the decision “was the product of reasoned decisionmaking.” State 12 Farm, 463 U.S. at 52. This standard is deferential, Pac. Dawn LLC v. Pritzker, 831 F.3d 13 1166, 1173 (9th Cir. 2016), and does not permit the Court to “substitute its judgment for that 14 of the agency,” Ctr. for Bio. Diversity v. Zinke, 868 F.3d 1054, 1057 (9th Cir. 2017). The 15 focus at all times must remain on whether the agency “considered the relevant factors and 16 articulated a rational connection between the facts found and the choices made.” Nw. Ecosys. 17 All. v. U.S. Fish & Wildlife Serv., 475 F.3d 1136, 1140, 1145 (9th Cir. 2007) (quotation 18 omitted). 19 20 21 a. Review Based on the Administrative Record Alone i. Pretext The APA requires an agency decision-maker to “disclose the basis of its” decision 22 and to “give clear indication that it has exercised the discretion with which Congress has 23 empowered it.” Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168 (1962); 24 accord Federal Power Comm’n v. Texaco Inc., 417 U.S. 380, 396 (1974). Where the agency 25 decision-maker fails to disclose the substance of relevant information that has been presented 26 to it, the court “must treat the agency’s justifications as a fictional account of the actual 27 decisionmaking process and must perforce find its actions arbitrary.” See Home Box Office, 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 109 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 110 of 126 1 Inc. v. F.C.C., 567 F.2d 9, 54-55 (D.C. Cir. 1977); see also U.S. Lines, Inc. v. Federal 2 Maritime Comm’n, 584 F.2d 519, 534-535 (D.C. Cir. 1978) (explaining that the basis of an 3 agency’s decision must be disclosed, at the very latest, in the final decision to permit 4 meaningful judicial review). 5 Secretary Ross violated the APA by failing to disclose the basis for his decision to 6 add a citizenship question to the 2020 Census. As explained in the Findings of Fact Based 7 Exclusively on the Administrative Record, the evidence overwhelmingly shows that 8 Secretary Ross decided to add the citizenship question well before DOJ made the request in 9 December of 2017 and that his reason for doing so was not to improve enforcement of 10 United States District Court 11 Section 2 of the VRA. This purported purpose was a mere pretext. While there is some evidence in the Administrative Record that Secretary Ross’s 12 interest in the citizenship question was related to the inclusion of noncitizens in the 13 apportionment count, there is no need definitively to ascertain the Secretary’s true purpose. 14 For the purposes of the APA, it is relevant only that Section 2 enforcement did not supply 15 the true basis of Secretary Ross’s decision and that he disclosed no other basis. Secretary 16 Ross has therefore violated the APA by failing to disclose the actual basis for his decision to 17 add a citizenship question to the 2020 Census. 18 19 ii. Whether Secretary Ross Considered All Relevant Factors An agency action must also be set aside as arbitrary and capricious if the agency fails 20 “to consider an important aspect of the problem.” State Farm, 463 U.S. at 43; see also 21 SecurityPoint Holdings, Inc. v. Transp. Sec. Admin., 769 F.3d 1184, 1188 (D.C. Cir. 2014) 22 (vacating agency order where agency failed to consider potential harms of its changes to an 23 airport advertising program); Stewart v. Azar, 313 F. Supp. 3d 237, 263 (D.D.C. 2018) 24 (vacating HHS Secretary’s waiver of several requirements of expanded Medicaid because 25 “[f]or starters, the Secretary never once mentions the estimated 95,000 people who would 26 lose coverage, which gives the Court little reason to think that he seriously grappled with the 27 bottom-line impact on healthcare” (emphasis in original)). 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 110 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 111 of 126 United States District Court 1 Defendants failed to consider the potential harms the citizenship question could cause 2 to the accuracy of the Census Bureau’s final enumeration, and therefore to the allocation of 3 federal funding and apportionment of congressional representation. The Decision Memo 4 states that “[t]he citizenship data provided to DOJ will be more accurate with the question 5 than without it, which is of greater importance than any adverse effect that may result from 6 people violating their legal duty to respond.” PTX-26 at 7. Secretary Ross does not, 7 however, articulate the potential harms that he is allegedly weighing. In particular, he does 8 not grapple in any meaningful way with the possibility of that a severe differential decline in 9 self-response rates could ultimately affect congressional apportionment. Instead, the 10 Secretary based his decision to add the citizenship question upon a purported lack of 11 evidence about the effects of the citizenship question on the census—specifically self- 12 response rates. Part IV.C.10. 13 The Secretary attempts to justify barreling ahead in the face of this alleged 14 uncertainty by stating that “no one has identified any mechanism” for evaluating the impact 15 of the citizenship question on the census. id. This statement is, frankly, perplexing. It is 16 notable that Secretary Ross does not claim that the Census Bureau lacks the tools to test the 17 citizenship question. Nor does Secretary Ross claim the Census Bureau advised him that 18 testing the effects of the citizenship question was beyond the Bureau’s ability. Ultimately, it 19 is highly implausible that Secretary Ross was unaware that other testing mechanisms, 20 beyond the Census Bureau’s natural experiment, existed. 21 Secretary Ross’s failure to investigate and consider the likely effects of the 22 citizenship question on the accuracy of the Census Bureau’s enumeration—and therefore on 23 congressional apportionment and the allocation of federal funding—was an abdication of his 24 duty to consider all relevant factors before making his decision. Therefore, the Secretary’s 25 decision to prioritize the inclusion of the citizenship question on the census over any harm 26 that might result was necessarily arbitrary and capricious. 27 iii. Whether the Decision Ran Counter to the Evidence 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 111 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 112 of 126 United States District Court 1 An agency action is also arbitrary and capricious under the APA if the agency offers 2 “an explanation for its decision that runs counter to the evidence before the agency.” State 3 Farm, 463 U.S. at 43. An agency must “articulate a satisfactory explanation for its action 4 including a ‘rational connection between the facts found and the choice made.’” Id. (quoting 5 Burlington Truck Lines, 371 U.S. at 168. Where a decision-maker adopts a “plainly inferior” 6 course of action, that decision is arbitrary and capricious. Pub. Citizen, Inc. v. Mineta, 340 7 F.3d 39, 56 (2d Cir. 2003). 8 Defendants’ decision to add the citizenship question to the 2020 Census is contrary to 9 the evidence. This decision is predicated on the assertion in the Decision Memo that adding a 10 citizenship question on the 2020 Census will result in the “most complete and accurate” 11 citizenship data for DOJ’s stated purpose of VRA enforcement. PTX-26 at 1 (“The 12 Department and Census Bureau’s review of the DOJ request—as with all significant Census 13 assessments— prioritized the goal of obtaining complete and accurate data” (emphasis in 14 original); see also id. at 5 (“It is my judgment that Option D will provide DOJ with the most 15 complete and accurate CVAP data in response to its request”), 7 (“[h]owever, even if there is 16 some impact on responses, the value of more complete and accurate data derived from 17 surveying the entire population outweighs such concerns”), 8 (“To conclude, after a 18 thorough review of the legal, program, and policy considerations, as well as numerous 19 discussions with the Census Bureau leadership and interested stakeholders, I have 20 determined that reinstatement of a citizenship question on the 2020 Decennial Census is 21 necessary to provide complete and accurate data in response to the DOJ request.”). 22 In light of this stated goal, Defendants’ decision to add a citizenship question to the 23 2020 Census runs counter to the evidence in the Administrative Record that shows adding a 24 citizenship question will result in citizenship data less accurate and no more complete than 25 citizenship data gathered through administrative records alone (Alternative C). This was the 26 conclusion of every scientific analysis in the Administrative Record that addressed the issue, 27 and this conclusion was repeatedly communicated to Secretary Ross. See, e.g., PTX-22 at 1 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 112 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 113 of 126 1 (January 19 Memo) (explaining that adding the citizenship question would result in 2 “substantially less accurate citizenship status data than are available from administrative 3 sources”); PTX-25 at 5 (March 1 Memo) (“Alternative D would result in poorer quality 4 citizenship data than Alternative C.”). 5 6 accuracy between a census response and administrative records data. First, the citizenship 7 data in administrative records is “very accurate” because that data point requires people to 8 have provided proof of citizenship or legal resident alien status. PTX-101 at 3; PTX-22 at 7. 9 Second, citizenship data that is self-reported in surveys is inaccurate for noncitizens. 10 Historical census and ACS data show that noncitizens misreport themselves as citizens “for 11 no less than 23.8% of the cases, and often more than 30%.” PTX-22 at 7. 12 United States District Court The Census Bureau’s analyses offered several explanations for the difference in Third, the Census Bureau found that lowered self-response rates due to the 13 citizenship question will decrease the number of people who can be linked to administrative 14 records (which contain citizenship information), because the personal identifying 15 information gathered in NRFU is lower quality than personal identifying information 16 gathered through self-response. PTX-22 at 2; PTX-25 at 4. Fourth, imputation will be less 17 accurate if based in part on self-reported citizenship data rather than administrative records 18 alone. This is because many noncitizens inaccurately report that they are citizens, therefore 19 the imputation model will be biased under Option D. In contrast, the imputation model under 20 Alternative C would be benchmarked to accurate administrative records and therefore would 21 not suffer from this bias. 22 Notably, there is no evidence in the Administrative Record supporting Secretary 23 Ross’s assertion that self-reported citizenship data is more accurate than citizenship data 24 from administrative records. Thus, all of the evidence shows that citizenship information 25 gathered through a citizenship question on the Census (Alternative D) will be less accurate 26 than citizenship information gathered through administrative records (Alternative C). 27 All of the evidence in the Administrative Record also shows that Option D will not 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 113 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 114 of 126 1 yield more “complete” data than Alternative C. Secretary Ross implies in the Decision 2 Memo that citizenship data from administrative records would be incomplete because using 3 administrative records alone would require imputation of citizenship status for 10 percent of 4 the population, whereas a citizenship question on the 2020 Census “may eliminate the need 5 for the Census Bureau to have to impute an answer for millions of people.” PTX-26 at 4, 5. 6 In fact, the Census Bureau estimates that under both alternatives millions of people 7 would need their citizenship status imputed and the total number of people assigned a 8 citizenship status would be approximately the same (330 million). PTX-24 at 1-4. In sum, all 9 of the evidence in the Administrative Record shows that adding a citizenship question to the 10 2020 Census would yield citizenship data that is less accurate and no more complete than 11 gathering that data using administrative records alone. The decision to add the citizenship 12 question was therefore arbitrary and capricious in violation of the APA. 13 The decision was also counter to the evidence because it was replete with flawed 14 assertions that are either not based on any evidence or contrary to the evidence in the 15 Administrative Record. First, the Decision Memo repeatedly claimed that there was no 16 evidence the citizenship question would cause a drop in self response, and that “no one has 17 identified any mechanism” for obtaining such evidence. PTX-26 at 3-5. This assertion is 18 contrary to the evidence in the Administrative Record. The Census Bureau performed a 19 scientific analysis leading to an estimate that 5.1 percent of households with at least one 20 noncitizen would not respond to the census due to the citizenship question. The Bureau 21 repeatedly communicated that estimate to Secretary Ross. PTX-22 at 4; PTX-148 at 6-7. 22 Nothing in the Administrative Record supports a contrary conclusion. 23 Second, the Decision Memo states that asking the citizenship question of all people 24 “may eliminate the need for the Census Bureau to have to impute an answer for millions of 25 people.” PTX-26 at 5. However, the Census Bureau had estimated that with a citizenship 26 question on the census, it would still have to impute the citizenship data of 13.8 million 27 people. PTX-24 at 2. Nothing in the Administrative Record supports a contrary conclusion. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 114 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 115 of 126 1 Third, the Decision Memo states that Option D “would maximize the Census 2 Bureau’s ability to match the decennial census responses with administrative records,” PTX- 3 26 at 4, so as to allow for “more complete” citizenship data. However, the Administrative 4 Record reflects that because adding a citizenship question would drive down the self- 5 response rate and put more households into NRFU operations, Option D actually reduces the 6 Census Bureau’s ability to match survey responses with administrative records. PTX-25 at 4. United States District Court 7 Fourth, the Decision Memo also attempts to justify Option D by stating that adding 8 the citizenship question to the census “will permit the Census Bureau to determine the 9 inaccurate response rate” with respect to the citizenship question, and suggests that this 10 would improve the accuracy of the imputation model. PTX-26 at 5. However, nowhere in the 11 Administrative Record, including in the Abowd memoranda, does the Census Bureau state 12 that adding a citizenship question would increase the accuracy of its estimate of inaccurate 13 citizenship responses. Nor is it apparent from the Administrative Record why the inaccuracy 14 rate of responders would help impute the citizenship data of nonrespondents. If citizenship is 15 benchmarked to administrative records, and the Bureau would be using those records in any 16 event, then adding a census question would not assist in the imputation. Accordingly, 17 Defendants’ decision is arbitrary and capricious because these key statements in the Decision 18 Memo, purportedly justifying the choice of “Option D,” were counter to the evidence. 19 b. Extra-Record Evidence Confirms the Secretary’s Decision was Arbitrary and Capricious 20 i. Pretext 21 Extra-record evidence confirms that Secretary Ross’s justification for his decision to 22 add the citizenship question was pretextual. Several facts are particularly striking. First, 23 Secretary Ross’s senior officials at the Commerce Department all claim, rather implausibly, 24 to be ignorant of why Secretary Ross wanted the citizenship question on the 2020 Census. 25 Even Comstock, the Director of Policy in charge of soliciting the DOJ request for the 26 question, claims that he never asked Secretary Ross to explain his reasoning. Comstock Dep. 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 115 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 116 of 126 1 171-72. This suggests either that, despite several months of discussion, Secretary Ross kept 2 his senior staff in the dark about his reason for wanting to include the citizenship question or 3 that his staff are dissembling in order to avoid revealing Secretary Ross’s true purpose. Both 4 of these conclusions suggest that Secretary Ross does not wish his reason for requesting the 5 inclusion of the citizenship question on the 2020 Census to come to light. 6 Second, DOJ’s December 12 Letter requesting the citizenship question was drafted 7 with limited involvement from VRA career staff. Gore initially prepared the letter and sent 8 an early draft to Chris Herren, the Chief of the Voting Rights Section, on November 1, 9 requesting feedback. This was the only involvement by any career staff in the Voting Rights 10 Section, even though the letter continued to be reviewed and revised by political appointees 11 for more than a month afterwards. Id. The limited involvement of VRA staff in this process 12 casts doubts on the true need for this data to support VRA enforcement. This bolster’s the 13 conclusion that the December 12 Letter was a mere pretext orchestrated by Secretary Ross, 14 rather than a sincere request upon which the Secretary reasonably relied. 15 Third, the Administrative Record reveals that DOJ refused to meet with Census 16 Bureau staff to discuss its request that a citizenship question be added to the 2020 Census. 17 Extra-record evidence further establishes that DOJ’s refusal to meet and discuss its data 18 needs was both “very unusual” for a requesting agency and directed by Attorney General 19 Sessions himself. Tr. 1055:5-9 (Abowd); Gore Dep. 271:21-272:13. That DOJ would choose 20 not to pursue an offered alternative that the Census Bureau believed would better suit DOJ’s 21 stated needs is extraordinary—and all the more so given that the Attorney General himself 22 made the decision. This evidence strongly suggests that VRA enforcement was not the true 23 goal of either the DOJ officials who prepared the request or Secretary Ross, who solicited 24 the request in the first place. 25 26 27 ii. Whether Secretary Ross Considered All Relevant Factors Extra-record evidence also confirms that Defendants failed to consider several important factors when deciding whether to include the citizenship question on the census. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 116 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 117 of 126 1 First, Defendants failed to consider whether the question had been adequately tested under 2 the applicable agency standards. While the Decision Memo states that the citizenship 3 question is “well tested” because it has been on the ACS since 2005, PTX-26 at 2, 4 Defendants point to no evidence in the Administrative Record demonstrating that the 5 citizenship question was “performing adequately” on the ACS for the purposes of pretesting. 6 See Part IV.D.2.a. Defendants also failed to consider the effect of documented changes in the 7 macro-environment around issues of immigration on respondents’ willingness to fill out a 8 questionnaire that includes a citizenship question. United States District Court 9 Second, Defendants failed to consider the effect of the Census Bureau’s 10 confidentiality obligations and disclosure avoidance practices on the fitness of citizenship 11 data for DOJ’s stated purpose, enforcement of Section 2 of the VRA. Under its disclosure 12 avoidance protocols, the Census Bureau will apply disclosure avoidance techniques to data 13 collected from every census block, meaning that even after adding a citizenship question, 14 there will not be a single census block for which the reported citizenship data directly 15 reflects the responses of the census block’s inhabitants to the 2020 Census questionnaire, 16 unless by random chance. New York Tr. 1033:16-21 (Abowd); Census Bureau 30(b)(6) Dep. 17 Vol. I 53:12-17, 69:6-71:12. Therefore, even if the citizenship data is obtained through 18 enumeration, some margin of error will unavoidably exist. Id. The Census Bureau does not 19 know how that margin of error will compare to the margin of error for the ACS citizenship 20 data currently in use. Id. The Bureau has not determined if, after disclosure avoidance, the 21 error margins for block-level CVAP data based on information collected through the 22 decennial enumeration will allow redistricting offices and the Department of Justice to use 23 the data effectively. Census 30(b)(6) Dep. Vol. I 100:21-101:15. 24 This complete absence of certainty, confirmed by Dr. Abowd, of whether the Census 25 Bureau’s disclosure avoidance protocols will result in greater error margins than the ones 26 associated with currently available CVAP data, or will generate a product that is “effective” 27 for its intended use, directly controverts the Secretary’s argument that “hard-count” 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 117 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 118 of 126 1 citizenship data from the decennial enumeration will provide DOJ with the “most complete 2 and accurate” data for its VRA enforcement efforts. PTX-26 at 5. 3 Third, extra-record evidence reveals that Defendants failed to analyze and consider 4 whether hard-count citizenship data would aid in VRA enforcement. The unrebutted expert 5 testimony of Plaintiffs’ voting rights expert Dr. Lisa Handley confirms that a citizenship 6 question on the census is not necessary to enforce the VRA. Dr. Handley testified that, in her 7 experience, CVAP estimates at the census tract or block group level are generally sufficient 8 to satisfy the first Gingles precondition in VRA cases. New York Tr. 807:24-811:6 9 (Handley). Dr. Handley further explained that, where it would be helpful to present CVAP 10 data at the block level, this information can be reliably and accurately estimated using block- 11 level CVAP data by applying the CVAP ratios from the census tract level to the block-level 12 figures for total voting-age population. Id. at 808:10-815:5. 13 The unrebutted expert testimony of Plaintiffs’ voting rights expert Professor Pamela 14 Karlan also confirms that a citizenship question on the census is not “necessary” to enforce 15 the VRA. Professor Karlan explained that no Section 2 case has ever failed on account of the 16 purported inadequacy of ACS data (or, prior to the advent of the ACS, data from the long- 17 form census questionnaire) as a measure of CVAP. Karlan Trial Dep. 52:14-53:18. 18 Defendants point to no evidence in the Administrative Record that suggests a 19 contrary conclusion. Moreover, there is no indication that Secretary Ross independently 20 weighed the value of the requested block level CVAP data, even after DOJ refused to meet 21 with the Census Bureau to clarify its data needs. Ultimately, Secretary Ross’s failure to 22 evaluate for himself the value of the requested data renders his decision to prioritize the 23 acquisition of this information over all else arbitrary and capricious. 24 In short, Defendants’ failure to consider these additional factors further supports a 25 finding that the Secretary’ decision was arbitrary and capricious. 26 V. CONSTITUTIONAL CLAIMS 27 As previously explained, both sets of plaintiffs advance a claim for violation of the 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 118 United States District Court Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 119 of 126 1 Enumeration Clause. The San Jose Plaintiffs also advance a claim under the Appropriations 2 Clause; however, as discussed in Part III.C, supra, they do not adequately establish standing 3 with respect to that claim. Accordingly, only the Enumeration Clause claim will be discussed 4 here. 5 A. Legal Standard 6 The United States Constitution mandates the “actual Enumeration” of the population 7 every ten years for the purpose of apportioning congressional representatives among the 8 states. U.S. Const. art. I, § 2, cl. 3. Congress has delegated the duty of taking the census to 9 the Secretary of Commerce. 13 U.S.C. § 141(a). Although Congress and the states use 10 census data for many purposes, including for allocating federal funding, the only 11 constitutional purpose of the census is to apportion congressional representatives based on 12 the “actual Enumeration” of the population of each state. U.S. Const. art. I, § 2, cl. 3; id. 13 amend. XIV, § 2; see also Franklin v. Massachusetts, 505 U.S. 788, 806 (1992) (reasoning 14 that Secretary of Commerce’s decision to include overseas federal employees in the 15 apportionment count did not violate Enumeration Clause because the decision “does not 16 hamper the underlying constitutional goal of equal representation”); Utah v. Evans, 536 U.S. 17 452, 500 (2002) (Thomas, J., concurring in part and dissenting in part) (observing that the 18 Framers’ “[d]ebate about apportionment and the census . . . focused for the most part on 19 creating a standard that would limit political chicanery”). 20 The Census Bureau is not constitutionally required to perform an absolutely accurate 21 count of the population. Wisconsin v. Cty. of New York, 517 U.S. 1, 6 (1996). Nevertheless, 22 there is a “strong constitutional interest in accuracy” of the census. Evans, 536 U.S. at 478. 23 The type of accuracy which most directly implicates the constitutional purpose of the census 24 is distributive accuracy, as opposed to numerical accuracy. See Wisconsin, 517 U.S. at 20 25 (explaining that “a preference for distributive accuracy . . . would seem to follow from the 26 constitutional purpose of the census, viz., to determine the apportionment of the 27 Representatives among the States”). Numerical accuracy refers to the accuracy of the overall 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 119 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 120 of 126 1 count, whereas distributive accuracy refers to the accuracy of the proportions in which 2 residents are counted in their proper locations. See id. at 11 n.6. 3 4 actions to bear “a reasonable relationship to the accomplishment of an actual enumeration of 5 the population, keeping in mind the constitutional purpose of the census,” which is to 6 determine the apportionment of the Representatives among the States. Id. at 20. While each 7 and every question on the census need not be related to the goal of actual enumeration, a 8 decision to alter the census in a way that affirmatively interferes with the actual enumeration, 9 and does not fulfill any other reasonable governmental purpose, is subject to a challenge 10 United States District Court To promote distributive accuracy, the Enumeration Clause requires the Secretary’s under the Enumeration Clause. 11 B. Scope of Review 12 Defendants argue that the Enumeration Clause claims must be decided on the basis of 13 the Administrative Record alone. See, e.g., Bellion Spirits, LLC v. United States, 335 F. 14 Supp. 3d 32, 43 (D.D.C. 2018) (“[W]hen a constitutional challenge to agency action requires 15 evaluating the substance of an agency’s decision made on an administrative record, that 16 challenge must be judged on the record before the agency”); Chiayu Chang v. U.S. 17 Citizenship & Immigration Servs., 254 F. Supp. 3d 160, 161 (D.D.C. 2017) (summarizing 18 cases holding that the assertion of constitutional claims does not remove a matter from the 19 APA’s procedural strictures); Jarita Mesa Livestock Grazing Ass’n v. U.S. Forest Serv., 58 20 F. Supp. 3d 1191, 1237 (D.N.M. 2014) (holding that the fact that a case “alleges 21 constitutional violations as well as statutory ones does not take it outside of the APA”). 22 As the district court in Bellion recognized, “caselaw on a plaintiff’s ability to 23 supplement an administrative record to support a constitutional cause of action is sparse and 24 in some tension.” 335 F. Supp. 3d at 41. Some courts have held that such claims are subject 25 to the strictures of the APA because section 706 specifically provides for review of agency 26 action that is “contrary to constitutional right.” See, e.g., Jarita Mesa, 58 F. Supp. 3d at 27 1237-38 (quoting 5 U.S.C. § 706(2)(B)); see also Chiayu Chang, 254 F. Supp. 3d at 161-162 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 120 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 121 of 126 1 (collecting cases). Other courts have reached a similar conclusion based on more practical 2 concerns. See, e.g., Alabama-Tombigbee Rivers Coal. v. Norton, No. CV-01-S-0194-S, 2002 3 WL 227032, at *3-6 (N.D. Ala. Jan. 29, 2002). Yet another group of courts has concluded 4 that extra-record evidence may sometimes be considered in resolving such claims. See, e.g., 5 Puerto Rico Pub. Hous. Admin. v. U.S. Dep’t of Hous. & Urban Dev., 59 F. Supp. 2d 310, 6 327-28 (D.P.R. 1999); Rydeen v. Quigg, 748 F. Supp. 900, 906 (D.D.C. 1990); see also 7 Carlsson v. U.S. Citizenship & Immigration Servs., No. 12-cv-07893, 2015 WL 1467174, at 8 *13 (C.D. Cal. Mar. 23 2015). United States District Court 9 Fortunately, there is no need to wade into the morass. Even assuming the usual APA 10 strictures apply, Secretary Ross’s failure to consider the impact of the citizenship question on 11 the accuracy of the enumeration, and therefore on the apportionment of congressional 12 representation, see Part E(1)(a)(ii), supra, empowers this Court to look beyond the 13 Administrative Record to determine whether the Secretary’s decision to include the 14 citizenship question is consistent with his Constitutional obligation to carry out an accurate 15 enumeration of the public. See Ranchers Cattlemen, 499 F.3d 1108 at 1117. To find 16 otherwise would allow government agencies to insulate themselves from judicial review of 17 the constitutionality of their actions by simply refusing to investigate pertinent factors during 18 the decision-making process. Accordingly, the Findings of Fact Related to Standing set forth 19 in Part III.B, supra, form the basis of the Enumeration Clause claim, in addition to the 20 findings of fact associated with the APA claim laid out in Part IV.C-D, supra. 21 C. Conclusions of Law 22 The Secretary’s decision to add a citizenship question to the 2020 Census violates the 23 Enumeration Clause of the Constitution because its inclusion will materially harm the 24 accuracy of the census without advancing any legitimate governmental interest. This is no 25 ordinary demographic inquiry. The record reveals that the inclusion of the citizenship 26 question on the upcoming census will have a unique impact on the Census Bureau’s ability 27 to count the public, to the point where the inclusion of this question is akin to a mechanics- 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 121 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 122 of 126 1 of-counting-type issue. In short, Secretary Ross’s decision to add the citizenship question to 2 the 2020 Census undermines the “strong constitutional interest in [the] accuracy” of the 3 census, Utah, 536 U.S. at 478, and does so despite the fact that adding this question does not 4 advance any identifiable government purpose. United States District Court 5 First, as previously discussed, the citizenship question will significantly impair the 6 distributive accuracy of the census because it will uniquely and substantially impact specific 7 demographic groups. Part III.B.1-3, supra. Specifically, the citizenship question will cause 8 an undercount of noncitizens and Latinos and, by extension, localities where many such 9 persons reside. Part III.B.3, supra. This, in turn, substantially increases the risk that 10 California will lose a seat in the House of Representatives. Indeed, the likelihood that 11 California will lose a congressional seat increases from 26.1 percent to at least 49.9 percent 12 with the addition of the citizenship question. Parts III.B.6, III.C.1.b, supra. There is also 13 credible evidence, based on Dr. Barreto’s survey, that California could lose up to three 14 congressional seats due to the citizenship question. Part III.B.6 Accordingly, the addition of 15 the citizenship question implicates the “strong constitutional interest in accuracy.” Evans, 16 536 U.S. at 478; see also Wisconsin, 517 U.S. at 20. 17 Second, Defendants fail to identify any countervailing governmental interest that 18 could justify this harm to the census. It is clear from the Administrative Record that 19 Secretary Ross’s purported reliance on the DOJ letter was nothing more than a pretext 20 designed to provide cover for Secretary Ross’s unexplained desire to add the citizenship 21 question to the census. See Part IV.C.12, supra. Extra-record evidence bolsters this 22 conclusion. Evidence produced through discovery reveals that the DOJ letter was drafted at 23 the Attorney General’s direction, with minimal involvement from the Civil Rights Division’s 24 voting rights staff. Attorney General Sessions subsequently prevented DOJ officials from 25 engaging in follow-up discussions with Census Bureau staff to clarify the request. Moreover, 26 aside from DOJ’s December 12 Letter, which was produced under curious circumstances and 27 represented an abrupt change in DOJ’s position, the record is devoid of any evidence to 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 122 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 123 of 126 1 support the VRA enforcement rationale. Plaintiffs, on the other hand, adduce persuasive 2 evidence that block-level CVAP data would have little effect on the prosecution of VRA 3 actions. United States District Court 4 Even assuming DOJ’s VRA enforcement efforts would benefit from block-level 5 citizenship data, there is no evidence to support the conclusion that surveying residents about 6 their citizenship would increase the accuracy of available block-level citizenship data. 7 Indeed, all evidence before this Court points to the opposite conclusion—that the inclusion 8 of the citizenship question would ultimately yield less citizenship accurate data than relying 9 on administrative records and imputing the citizenship status of persons for whom such data 10 does not exist. Part IV.C.5.d. In sum, the evidence shows that this question would impede 11 the stated goal of obtaining more accurate block-level CVAP. Accordingly, Defendants have 12 failed to point to any way in which including the citizenship question on the 2020 census 13 will advance a legitimate governmental interest. 14 This is not to say that the Census Bureau may never ask about citizenship on future 15 census questionnaires. It simply means that, where the inclusion of a particular question will 16 degrade the accuracy of the Census to the point where the proper apportionment of 17 representatives among the states is at risk, the government must identify a legitimate 18 governmental purpose that is sufficiently weighty to justify this significant harm to the 19 census. Defendants falls well short of this standard. 20 Moreover, finding that adding a citizenship question is unconstitutional does not 21 automatically render all demographic questions on the census unconstitutional. Indeed, it is 22 well established that each and every question on the census need not relate to the goal of 23 enumeration. See Baldridge v. Shapiro, 455 U.S. 345, 353 (1982) (acknowledging that the 24 census “fulfills many important and valuable functions”). There is, however, no evidence 25 that any other demographic question included on the census is likely to result in distributive 26 27 inaccuracy that threatens to distort the apportionment of representatives, or that these questions were introduced despite the absence of any legitimate governmental interest in 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 123 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 124 of 126 1 2 their being asked. Defendants argue against an Enumeration Clause finding because it would suggest 3 the constitutionality of a particular question could vary over time depending upon the social 4 and political context. This argument is unpersuasive. As the Supreme Court has recognized, 5 the constitutionality of a particular governmental action may depend on the larger social 6 7 8 9 10 context in which that action occurs. See Shelby Cty., Ala. v. Holder, 570 U.S. 529, 553-557 (2013) (striking down Section 4 of the Voting Rights Act because, inter alia, the preclearance formula set forth in that section was based on “decades-old data relevant to decades-old problems, rather than current data reflecting current needs”); Grutter v. Bollinger, 539 U.S. 306, 342-43 (2003) (explaining that “race-conscious admissions policies must be limited in time” and predicting that 25 years later affirmative action would no longer 11 be necessary or constitutionally justified). Ultimately, Secretary Ross’s primary obligation United States District Court 12 under the Constitution is to ensure a reasonably accurate enumeration of the public, and to 13 14 15 16 17 18 19 20 attempt to design a survey that will achieve that goal in the year 2020. The fact that the citizenship question may have been perfectly harmless in 1950, or that may be harmless again in the year 2050 is of little consequence to the Secretary’s constitutional obligations with respect to the accuracy of the 2020 Census. The addition of a citizenship question to the 2020 Census cannot be said to bear a “reasonable relationship to the accomplishment of an actual enumeration of the population.” Wisconsin, 517 U.S. at 20. Secretary Ross’s decision thus violates the Enumeration Clause of the Constitution. 21 VI. REMEDIES 22 A. Vacatur and Remand 23 In light of this Court’s ruling that Secretary Ross’s decision to include the citizenship 24 question on the 2020 Census violated the APA, the Secretary’s decision is hereby vacated 25 and remanded to the Department of Commerce. 5 U.S.C. § 706(2) (“[A] reviewing court 26 shall . . . hold unlawful and set aside agency action found to be” in violation of the APA). 27 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 124 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 125 of 126 B. Injunctive Relief20 1 A plaintiff seeking a permanent injunction must satisfy four requirements: “(1) that it 2 has suffered an irreparable injury; (2) that remedies available at law, such as monetary 3 damages, are inadequate to compensate for that injury; (3) that, considering the balance of 4 5 6 United States District Court 7 hardships between the plaintiff and defendant, a remedy in equity is warranted; and (4) that the public interest would not be disserved by a permanent injunction.” eBay Inc. v. MercExchange, L.L.C., 547 U.S. 388, 391 (2006). The decision whether to grant injunctive relief ultimately falls within a court’s equitable discretion. Id. 8 Here, each of the four elements is satisfied. It is simply beyond dispute that a loss of 9 political representation to the state of California, among other harms identified by Plaintiffs, 10 qualifies as irreparable harm that cannot be remedied at law. Moreover, given the likely 11 harm to the quality of the census and the absence of any valid basis for Secretary Ross’s 12 decision, the balance of the hardships and the public interest both favor barring Defendants 13 from including the citizenship question in the census questionnaire. Accordingly, an 14 injunction is appropriate here. The proper scope of the injunction, however, requires 15 additional analysis. As previously noted, this Court is mindful of concerns regarding the authority and 16 17 propriety of an individual district judge in one judicial district issuing an injunction of 18 nationwide application. In light of the unitary nature of the question at issue, by definition no 19 injunctive relief could be limited to only one geographic area or to only certain litigants. Put 20 simply, the citizenship question is either on or off the 2020 Census, thereby practically 21 requiring an injunction nationwide in scope. The appropriate terms of this injunction depend on which claim forms the basis for 22 23 24 25 26 27 relief. To the extent that the injunction relates solely to the APA claim, the reasoning in New York et al. v. Department of Commerce, et al., No. 18-cv-2921 (S.D.N.Y.) is persuasive. As 20 Plaintiffs also request declaratory relief. In light of the vacatur of Secretary Ross’s decision and the injunction barring Defendants from including the citizenship question on the 2020 Census, providing declaratory relief would serve little useful purpose and is therefore denied. 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 125 Case 3:18-cv-01865-RS Document 205 Filed 03/06/19 Page 126 of 126 1 explained in that order, “an injunction is necessary to make the Court’s vacatur effective, as 2 it prevents Secretary Ross from arriving at the same decision without curing the problems 3 identified in this Opinion.” Id. at 272. Accordingly, with respect to the APA violation, 4 Defendants are enjoined from including the citizenship question on the 2020 Census unless: 5 (1) they establish that direct inquiries regarding citizenship are necessary given the “kind, 6 timeliness, quality and scope of the statistics required” and that administrative records will 7 not suffice, 13 U.S.C. § 6(c); (2) they identify new circumstances that necessitate the last 8 minute addition of the citizenship question to the census, id §141(f)(3); and (3) Secretary 9 Ross considers all relevant factors and evidence, and sets forth the actual basis for his United States District Court 10 decision. 11 The Enumeration Clause violation, however, requires a more expansive injunction. 12 The record in this case has clearly established that including the citizenship question on the 13 2020 Census is fundamentally counterproductive to the goal of obtaining accurate 14 citizenship data about the public. This question is, however, quite effective at depressing 15 self-response rates among immigrants and noncitizens, and poses a significant risk of 16 distorting the apportionment of congressional representation among the states. In short, the 17 inclusion of the citizenship question on the 2020 Census threatens the very foundation of our 18 democratic system—and does so based on a self-defeating rationale. In light of these 19 findings, Defendants do not get another bite at the apple. Defendants are hereby enjoined 20 from including the citizenship question on the 2020 Census, regardless of any technical 21 compliance with the APA. 22 23 IT IS SO ORDERED. 24 25 26 27 Dated: March 6, 2019 ____________________________________ RICHARD SEEBORG United States District Judge 28 CASE NO. 18-cv-01865-RS; 18-cv-02279-RS 126