Promise Arizona v. Petersen

• Court: Court of Appeals for the Ninth Circuit
• Decided: February 25, 2025
• Docket: 24-4029
• Status: Published

Opinion

FOR PUBLICATION

UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT

MI FAMILIA VOTA; VOTO No. 24-3188 LATINO; LIVING UNITED FOR D.C. No. CHANGE IN ARIZONA; LEAGUE 2:22-cv-00509- OF UNITED LATIN AMERICAN SRB CITIZENS ARIZONA; ARIZONA STUDENTS' ASSOCIATION; ADRC ACTION; INTER TRIBAL COUNCIL OF ARIZONA, INC.; OPINION SAN CARLOS APACHE TRIBE; ARIZONA COALITION FOR CHANGE; UNITED STATES OF AMERICA; PODER LATINX; CHICANOS POR LA CAUSA; CHICANOS POR LA CAUSA ACTION FUND; DEMOCRATIC NATIONAL COMMITTEE; ARIZONA DEMOCRATIC PARTY; ARIZONA ASIAN AMERICAN NATIVE HAWAIIAN AND PACIFIC ISLANDER FOR EQUITY COALITION; PROMISE ARIZONA; SOUTHWEST VOTER REGISTRATION EDUCATION PROJECT; TOHONO O'ODHAM NATION; GILA RIVER INDIAN COMMUNITY; KEANU STEVENS; ALANNA SIQUIEROS; LADONNA 2 MI FAMILIA VOTA V. PETERSEN

JACKET,

Plaintiffs - Appellees,

v.

ADRIAN FONTES, in his official capacity as Arizona Secretary of State; KRIS MAYES, in her official capacity as Arizona Attorney General; STATE OF ARIZONA; LARRY NOBLE, Apache County Recorder, in his official capacity; DAVID W. STEVENS, Cochise County Recorder, in his official capacity; PATTY HANSEN, Coconino County Recorder, in her official capacity; SADIE JO BINGHAM, Gila County Recorder, in her official capacity; SHARIE MILHEIRO, Greenlee County Recorder, in her official capacity; RICHARD GARCIA, La Paz County Recorder, in his official capacity; STEPHEN RICHER, Maricopa County Recorder, in his official capacity; KRISTI BLAIR, Mohave County Recorder, in her official capacity; MICHAEL SAMPLE, Navajo County Recorder, in his official capacity; GABRIELLA CAZARES-KELLY, Pima County Recorder, in her official capacity; MI FAMILIA VOTA V. PETERSEN 3

RICHARD COLWELL, Yuma County Recorder, in official capacity; DANA LEWIS, Pinal County Recorder, in official capacity; POLLY MERRIMAN, Graham County Recorder, in her official capacity; JENNIFER TOTH, in her official capacity as Director of the Arizona Department of Transportation; MICHELLE BURCHILL, Yavapai County Recorder, in official capacity; ANITA MORENO, Santa Cruz County Recorder, in her official capacity,

Defendants - Appellees,

WARREN PETERSEN, President of the Arizona Senate; BEN TOMA, Speaker of the Arizona House of Representatives; REPUBLICAN NATIONAL COMMITTEE,

Intervenor-Defendants - Appellants.

MI FAMILIA VOTA; VOTO No. 24-3559 LATINO; LIVING UNITED FOR D.C. No. CHANGE IN ARIZONA; LEAGUE 2:22-cv-00509- OF UNITED LATIN AMERICAN SRB CITIZENS ARIZONA; ARIZONA STUDENTS' ASSOCIATION; 4 MI FAMILIA VOTA V. PETERSEN

ADRC ACTION; INTER TRIBAL COUNCIL OF ARIZONA, INC.; SAN CARLOS APACHE TRIBE; ARIZONA COALITION FOR CHANGE; UNITED STATES OF AMERICA; PODER LATINX; CHICANOS POR LA CAUSA; CHICANOS POR LA CAUSA ACTION FUND; DEMOCRATIC NATIONAL COMMITTEE; ARIZONA DEMOCRATIC PARTY; ARIZONA ASIAN AMERICAN NATIVE HAWAIIAN AND PACIFIC ISLANDER FOR EQUITY COALITION; PROMISE ARIZONA; SOUTHWEST VOTER REGISTRATION EDUCATION PROJECT; TOHONO O'ODHAM NATION; GILA RIVER INDIAN COMMUNITY; KEANU STEVENS; ALANNA SIQUIEROS; LADONNA JACKET,

KRIS MAYES; STATE OF ARIZONA,

Defendants - Appellants. MI FAMILIA VOTA V. PETERSEN 5

PROMISE ARIZONA; SOUTHWEST VOTER No. 24-4029 REGISTRATION EDUCATION D.C. No. PROJECT, 2:22-cv-00509- SRB Plaintiffs - Appellants,

and

MI FAMILIA VOTA, VOTO LATINO, LIVING UNITED FOR CHANGE IN ARIZONA, LEAGUE OF UNITED LATIN AMERICAN CITIZENS ARIZONA, ARIZONA STUDENTS' ASSOCIATION, ADRC ACTION, INTER TRIBAL COUNCIL OF ARIZONA, INC., SAN CARLOS APACHE TRIBE, ARIZONA COALITION FOR CHANGE, UNITED STATES OF AMERICA, PODER LATINX, CHICANOS POR LA CAUSA, CHICANOS POR LA CAUSA ACTION FUND, DEMOCRATIC NATIONAL COMMITTEE, ARIZONA DEMOCRATIC PARTY, ARIZONA ASIAN AMERICAN NATIVE HAWAIIAN AND PACIFIC ISLANDER FOR EQUITY COALITION, TOHONO O'ODHAM NATION, GILA RIVER INDIAN COMMUNITY, KEANU 6 MI FAMILIA VOTA V. PETERSEN

STEVENS, ALANNA SIQUIEROS, LADONNA JACKET,

Plaintiffs,

ADRIAN FONTES, LARRY NOBLE, DAVID W. STEVENS, PATTY HANSEN, SADIE JO BINGHAM, SHARIE MILHEIRO, RICHARD GARCIA, STEPHEN RICHER, KRISTI BLAIR, MICHAEL SAMPLE, GABRIELLA CAZARES-KELLY, SUZANNE SAINZ, RICHARD COLWELL, DANA LEWIS, POLLY MERRIMAN, JENNIFER TOTH, MICHELLE BURCHILL,

Defendants,

WARREN PETERSEN; BEN TOMA; REPUBLICAN NATIONAL COMMITTEE, MI FAMILIA VOTA V. PETERSEN 7

Intervenor-Defendants - Appellees.

Appeal from the United States District Court for the District of Arizona Susan R. Bolton, District Judge, Presiding

Argued and Submitted September 10, 2024 San Francisco, California

Filed February 25, 2025

Before: Kim McLane Wardlaw, Ronald M. Gould, and Patrick J. Bumatay, Circuit Judges.

Opinion by Judge Gould; Dissent by Judge Bumatay

SUMMARY *

Voting Rights

The panel affirmed in part the district court’s rulings on summary judgment and following a bench trial, vacated in part a portion of its factual findings, and remanded, in an action brought by the United States, several nonprofits, the Democratic National Committee, the Arizona Democratic Party, and three federally recognized Tribes who challenged

* This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. 8 MI FAMILIA VOTA V. PETERSEN

two Arizona laws regulating voter registration, H.B. 2492 and H.B. 2243 (together the “Voting Laws”), on the grounds that they were preempted or in violation of the National Voter Registration Act (“NVRA”), the consent decree in League of United Latin Am. Citizens of Ariz. v. Reagan (the “LULAC Consent Decree”), the Civil Rights Act, and the Equal Protection Clause of the United States Constitution. To register to vote in Arizona, an applicant may use the federal form created by the United States Election Assistance Commission or a state form prescribed by Arizona law. The federal form requires applicants to check a box under penalty of perjury indicating they are United States citizens but does not require applicants to disclose their birthplace. The Voting Laws amended provisions regulating voter registration and enabled government officials to require heightened proof of citizenship from federal-form and state-form applications. Specifically, pursuant to H.B. 2492, federal-form applicants without documentary proof of citizenship (“DPOC”) may be registered as federal-only voters but are not eligible to vote for president or to vote by mail. State-form applicants must check a box confirming their citizenship, disclose their birthplace and provide documentary proof of residency (“DPOR”). State-form applications without DPOC must be rejected. Pursuant to H.B. 2243, county recorders must periodically conduct citizenship checks of registered federal-only voters or registered voters who county recorders have “reason to believe” are not citizens and cancel registrations if citizenship is not confirmed. The panel held that the Republican National Committee, the Arizona House Speaker and Senate President (Republican Appellants), and two nonprofit organizations had standing to pursue their appeals. MI FAMILIA VOTA V. PETERSEN 9

The panel affirmed the district court’s rulings on the NVRA claims, the LULAC Consent Decree claim, the Civil Rights Act claims, and the Equal Protection claim. The panel held that although some provisions of the Voting Laws are legitimate and lawful prerequisites to voting, many of the challenged provisions are unlawful measures of voter suppression. Specifically, (1) the requirement that federal form applicants must provide DPOC to vote by mail is preempted by Section 6 of the NVRA and obstacle preemption; (2) the requirement of DPOC to vote in presidential elections is preempted by Section 6 of the NVRA; (3) the requirement that state-form applicants registering for federal elections must provide DPOR violates Sections 6 and 7 of the NRVA; (4) the requirement that county recorders conduct citizen checks of voters that they have “reason to believe” are not citizens violates Section 8(b) of the NVRA; and (5) the periodic cancellation of registrations violates the 90-day Provision of the NVRA to the extent that H.B. 2243 authorizes systematic cancellation of registrations within 90 days before a federal election. The panel held that the requirement that county recorders reject state-form applications without DPOC violates the LULAC Consent Decree. Alternatively, the NVRA does not let states require DPOC from state-form applicants registering for only federal elections. The citizen checkbox requirement relating to Arizona’s state form violates the Civil Rights Act when enforced on a person who has provided DPOC and is otherwise eligible to vote in Arizona. The birthplace disclosure requirement and the requirement that county recorders conduct citizen checks of voters they have reason to believe are not citizens violate the Civil Rights Act. The requirements of DPOC and DPOR for state-form applicants, however, do not violate the Equal 10 MI FAMILIA VOTA V. PETERSEN

Protection Clause under the arbitrary and disparate treatment standard. The panel held that the district court imposed a higher evidentiary standard than required in finding that Arizona enacted H.B. 2243 without intent to discriminate. The panel, therefore, vacated the district court’s factual finding on this issue and remanded with instructions for the district court to apply the proper totality of the circumstances analysis. The panel held that the Republican Appellants’ appeal of the district court’s holding that the Legislative Parties waived legislative privilege was moot. Dissenting, Judge Bumatay stated that while some parts of H.B. 2492 and H.B. 2243 may violate federal law, in no way must they be completely invalidated. Most of the voter- verification laws are consistent with the Constitution and federal law, and the panel should have vacated and substantially narrowed the injunction. Judge Bumatay would reverse the district court’s preliminary injunction enjoining the Voting Law requirements for proof of citizenship to vote for president and to vote by mail and for state voter registration forms. He would also reverse the order enjoining requirements for proof of residence, for the disclosure of birthplace, and for the removal of noncitizens from the voter rolls. Neither the NVRA nor the LULAC consent decree barred enforcement of these requirements. Judge Bumatay disagreed with the majority that the nonprofit organizations had standing to appeal the equal protection claim against H.B. 2243. He also disagreed with the majority’s discriminatory purpose analysis. Given the strong presumption of good faith to legislative enactments, MI FAMILIA VOTA V. PETERSEN 11

there was no basis to overturn the district court’s factual determination. Judge Bumatay joined the majority on three issues. First, he agreed with enjoining the “reason to believe” and the citizenship-checkbox requirements because these requirements violated the Civil Rights Act. He also agreed that the appeal of the district court’s holding that the Legislative Parties waived their legislative privilege was moot.

COUNSEL

Jonathan L. Backer (argued), Bonnie Robin-Vergeer, Matthew N. Drecun, and Margaret Turner, Attorneys, Civil Rights Division, Appellate Section; Kristin Clarke, Assistant Attorney General; United States Department of Justice, Washington, D.C.; Danielle M. Lang (argued), R. Brent Ferguson, Kathryn L. Huddleston, and Jonathan Diaz, Campaign Legal Center, Washington, D.C.; James E. Barton II, Barton Mendez Soto PLLC, Tempe, Arizona; Alexander B. Ritchie, Attorney; Chase A. Velasquez, Assistant Attorney General; Department of Justice San Carlos Apache Tribe, Office of the Attorney General, San Carlos, Arizona; Courtney Hostetler, Free Speech For People, Sharon, Massachusetts; John C. Bonifaz, Free Speech For People, Amherst, Massachusetts; Lee H. Rubin, Mayer Brown LLP, Palo Alto, California; Rachel J. Lamorte, Mayer Brown LLP, Washington, D.C.; Gary A. Isaac, Daniel T. Fenske, Anastasiya K. Lobacheva, and William J. McElhaney III, Mayer Brown LLP, Chicago, Illinois; Ernest I. Herrera (argued), Denise Hulett, and Erika Cervantes, Mexican American Legal Defense and Educational Fund, Los 12 MI FAMILIA VOTA V. PETERSEN

Angeles, California; Daniel R. Ortega Jr., Ortega Law Firm PC, Phoenix, Arizona; Daniel A. Arellano, Roy Herrara, and Jillian L. Andrews, Herrera Arellano LLP, Phoenix, Arizona; Marc E. Elias, Elisabeth C. Frost, Christopher D. Dodge, Daniela Lorenzo, and Qizhou Ge, Elias Law Group LLP, Washington, D.C.; Daniel J. Adelman, Arizona Center For Law In The Public Interest, Phoenix, Arizona; John A. Freedman, Jeremy Karpatkin, Erica McCabe, and Leah Motzkin, Arnold & Porter Kaye Scholer LLP, Washington, D.C.; Leah R. Novak and Andrew Hirschel, Arnold & Porter Kaye Scholer LLP New York, New York; Nina G. Beck, Emily Davis, Jonathan Sherman, Beauregard Patterson, and Michelle K. Cohen, Fair Elections Center, Washington, D.C.; Christopher E. Babbitt, Daniel S. Volchok, Seth P. Waxman, Britany Riley-Swanbeck, and Joseph M. Meyer, Wilmer Cutler Pickering Hale and Dorr LLP, Washington, D.C.; Bruce Samuels, Jennifer Lee-Cota, Papetti Samuels Weiss McKirgan LLP, Scottsdale, Arizona; Sadik H. Huseny, Amit Makker, Evan Omi, and Catherine A. Rizzoni, Latham & Watkins LLP, San Francisco, California; Niyati Shah, Terry A. Minnis, and Noah Baron, Asian Americans Advancing Justice, Washington, D.C.; Andrew M. Federhar, Spencer Fane LLP, Phoenix, Arizona; David B. Rosenbaum, Joshua J. Messer, Osborn Maledon PA, Phoenix, Arizona; Ezra D. Rosenberg, American Civil Liberties Union of New Jersey, Newark, New Jersey; Matthew L. Campbell, Michael S. Carter, Allison A. Neswood, and Jacqueline D. DeLeon, Native American Rights Fund, Boulder, Colorado; Samantha B. Kelty, Native American Rights Fund, Washington D.C.; Marissa L. Sites, Assistant Attorney General; Howard M. Shanker, Attorney General, Tohono O'Odham Nation, Office of the Attorney General, Sells, Arizona; Javier G. Ramos, Senior Counsel, Gila River MI FAMILIA VOTA V. PETERSEN 13

Indian Community, Prima Maricopa Tribe Law Office, Sacaton, Arizona; for Plaintiffs-Appellees. Kory A. Langhofer (argued) and Thomas J. Basile, Statecraft PLLC, Phoenix, Arizona; Tyler R. Green, Gilbert C. Dickey, and Conor D. Woodfin, Consovoy McCarthy PLLC, Arlington, Virginia; for Intervenor-Defendants-Appellants. Joshua M. Whitaker (argued) Joshua D. Bendor, Hayleigh S. Crawford, Attorneys; Kathryn E. Boughton and Timothy E. Durkin Horley, Assistant Attorneys General; Kristin K. Mayes, Arizona Attorney General; Office of the Arizona Attorney General, Phoenix, Arizona; Craig Morgan, Taft Stettinius & Hollister LLP, Phoenix, Arizona; Celeste Robertson, Attorney, Apache County Attorney's Office, St. Johns, Arizona; Christine J. Roberts, Deputy Chief Counsel, Pinal County Attorney's Office, Florence, Arizona; Rose M. Winkeler, Deputy County Attorney, Flagstaff Law Group PLLC, Flagstaff, Arizona; Jefferson R. Dalton, Counsel, Gila County Attorney's Office, Globe, Arizona; Gary Griffith, County Attorney, Greenlee County Attorney’s Office, Clifton, Arizona; Jason W. Mitchell, Trial Attorney, La Paz County Attorney’s Office, Parker, Arizona; Ryan N. Dooley, Attorney, City of St. George, St. George, Utah; Sean M. Moore, Joseph E. La Rue, Jack L. O'Connor III, and Anna Griffin Critz, Deputy County Attorneys; Maricopa County Attorney's Office, Civil Services Division, Phoenix, Arizona; Ryan H. Esplin, Attorney, Mohave County Attorney's Office, Kingman, Arizona; Jason S. Moore, Deputy Assistant Attorney General, Navajo County Attorney's Office, Holbrook, Arizona; Daniel S. Jurkowitz, Pima County Attorney's Office, Tucson, Arizona; William J. Kerekes and Jessica L. Holzer, Deputy County Attorney’s, Office of the Yuma County Attorney, Yuma, Arizona; Craig Cameron, Deputy County Attorney, Pinal County Attorney's 14 MI FAMILIA VOTA V. PETERSEN

Office, Florence, Arizona; Jean A. Roof, Graham County Attorney's Office, Safford, Arizona; Thomas M. Stoxen, Attorney, Yavapai County Attorney's Office, Prescott, Arizona; Christina E. Werther and Justin S. Pierce, Pierce Coleman PLLC, Scottsdale, Arizona; for Defendants- Appellees. Dominic E. Draye, Greenberg Traurig LLP, Phoenix, Arizona; Nick Peterson, Greenberg Traurig LLP, Salt Lake City, Utah; for Amicus Curiae Arizona Free Enterprise Club. Christopher J. Hajec, Immigration Reform Law Institute, Washington, D.C.; Lawrence J. Joseph, Law Office of Lawrence J. Joseph, Washington, D.C.; for Amicus Curiae Immigration Reform Law Institute. Jonathon P. Hauenschild, Center for Election Confidence, Arlington, Virginia, for Amicus Curiae Center for Election Confidence. Michael A. Columbo, Mark P. Meuser, and Harmeet Dhillon, Dhillon Law Group Inc., San Francisco, California; Andrew Gould, Holtzman Vogel Baran Torchinsky & Josefiak PLLC, Phoenix Arizona; for Amicus Curiae Republican Party of Arizona. Elizabeth B. Wydra, Brianne J. Gorod, David H. Gans, and Anna K. Jessurun, Constitutional Accountability Center, Washington, D.C, for Amicus Curiae Constitutional Accountability Center. Justin Levitt, LMU Loyola Law School, Los Angeles, California, for Amicus Curiae Professor Justin Levitt. Patricia J. Yan and Sarah E. Brannon, American Civil Liberties Union Foundation, Washington, D.C.; Sophia L. Lakin, American Civil Liberties Union Foundation, New MI FAMILIA VOTA V. PETERSEN 15

York, New York; R. Adam Lauridsen, Ian Kanig, Imara McMillan, Sara R. Fitzpatrick, and Courtney J. Liss, Keker Van Nest & Peters LLP, San Francisco, California; Jasleen Singh and Sara Carter, Brennan Center for Justice at NYU School of Law, New York, New York; Phi Nguyen and Roni Druks, Demos, New York, New York; for Amici Curiae League of Women Voters, League of Women Voters of Arizona, Secure Families Initiative, and Modern Military Association of America.. 16 MI FAMILIA VOTA V. PETERSEN

OPINION

GOULD, Circuit Judge:

The United States, several nonprofits, the Democratic National Committee, the Arizona Democratic Party, and three federally recognized Tribes (collectively, the “Plaintiff-Appellees”) challenge two Arizona laws regulating voter registration, H.B. 2492 and H.B. 2243 (together the “Voting Laws”), contending these are preempted or in violation of the National Voter Registration Act (“NVRA”), the LULAC consent decree, the Civil Rights Act, and the Equal Protection Clause of the United States Constitution. Consolidating the eight lawsuits challenging the Voting Laws, the district court held that certain provisions of the Voting Laws are preempted by the NVRA, that certain provisions of the Voting Laws violate the NVRA, and that Sections 6 and 9 of the NVRA require county recorders to register state-form applicants without documentary proof of location of residency (“DPOR”) as “federal-only” voters. The district court also held that state- form applicants without documentary proof of citizenship (“DPOC”) must be processed in accordance with the consent decree in League of United Latin Am. Citizens of Ariz. v. Reagan, No. 2:17-cv-4102 (D. Ariz. 2018) (the “LULAC Consent Decree”) or, in the alternative, that the NVRA does not let states require DPOC from state-form applicants registering for only federal elections. Regarding the Civil Rights Act claims, the district court held that two requirements imposed by the Voting Laws violate the “Materiality Provision” of the Civil Rights Act, 52 U.S.C. § 10101(a)(2)(B), and that the requirement that county recorders verify the citizenship status (“citizenship MI FAMILIA VOTA V. PETERSEN 17

checks”) of voters that they have “reason to believe” are not citizens violates the different standards, practices, or procedures provision (“DSPP Provision”) of the Civil Rights Act, 52 U.S.C. § 10101(a)(2)(A). Regarding the Equal Protection claims, the district court held that the requirements of DPOC and DPOR do not violate the Equal Protection Clause of the United States Constitution and found that neither of the Voting Laws was enacted with intent to discriminate. In adjudicating these claims, the district court held that Arizona House Speaker Ben Toma and Arizona Senate President Warren Petersen (together the “Legislative Parties”) waived legislative privilege. The Republican National Committee, Toma, and Petersen (collectively, the “Republican Appellants”) appeal the district court’s holdings about claimed violations of the NVRA, the LULAC Consent Decree, and the Civil Rights Act. The Republican Appellants also appeal the holding that the Legislative Parties waived legislative privilege. Two of the nonprofit Plaintiff-Appellees, Promise Arizona and Southwest Voter Registration Education Project (together the “Promise Cross-Appellants”), cross- appeal the factual finding that H.B. 2243 was not enacted with intent to discriminate. The State of Arizona and the Arizona Attorney General Kris Mayes (in her official capacity) (together “the State”) appeal, contending that the state-form requirement that applicants disclose their birthplace does not violate the Materiality Provision of the Civil Rights Act and that the Promise Cross-Appellants do not have standing to pursue their cross-appeal. Another 18 MI FAMILIA VOTA V. PETERSEN

group of nonprofit entities 1 (collectively, “LUCHA Appellees”) contend that the Republican Appellants do not have standing to appeal and that the DPOC and DPOR requirements violate Equal Protection. The challenges raised in the briefing can be grouped into six general categories: (1) whether certain parties have standing, (2) whether the NVRA preempts provisions of the Voting Laws, (3) whether the Voting Laws violate the LULAC Consent Decree, (4) whether the Voting Laws violate the Civil Rights Act, (5) whether the Voting Laws violate the Equal Protection Clause of the United States Constitution, and (6) whether there was waiver of legislative privilege. There are fourteen specific issues raised in the briefing, namely (1) whether the Republican Appellants have standing to appeal, (2) whether the Promise-Cross Appellants have standing to cross-appeal, (3) whether the DPOC requirement to vote by mail is preempted by the NVRA, (4) whether the DPOC requirement to vote in presidential elections is preempted by the NVRA, (5) whether the DPOR requirement for state-form applicants registering for federal elections is preempted by the NVRA, (6) whether citizenship checks of voters who county recorders have “reason to believe” are not citizens violates the NVRA, (7) whether the periodic cancellation of registrations violates the NVRA, (8) whether the requirement that county recorders reject state-form applications without DPOC violates the LULAC Consent

1 Living United for Change in Arizona; League of United Latina American Citizens; Arizona Students’ Association; ADRC Action; Inter Tribal Council of Arizona, Inc.; San Carlos Apache Tribe, a federally recognized tribe; and Arizona Coalition for Change MI FAMILIA VOTA V. PETERSEN 19

Decree, (9) whether the checkbox requirement violates the Materiality Provision of the Civil Rights Act, (10) whether the birthplace requirement violates the Materiality Provision of the Civil Rights Act, (11) whether the “reason to believe” provision violates the DSPP Provision, (12) whether the district court erred in finding Arizona enacted H.B. 2243 without intent to discriminate, (13) whether the requirements of DPOC and DPOR cause “arbitrary and disparate treatment” violating the Equal Protection Clause, and (14) whether the Legislative Parties waived legislative privilege. We address each issue in turn. Although some provisions of the Voting Laws are legitimate and lawful prerequisites to voting, many of the challenged provisions are unlawful measures of voter suppression. We have jurisdiction under 28 U.S.C. § 1291. We hold that the Republican Appellants and Promise Cross- Appellants have standing to pursue their appeals. We affirm the district court’s rulings on the NVRA claims, the LULAC Consent Decree claim, the Civil Rights Act claims, and the Equal Protection claim. We also vacate the district court’s factual finding that H.B. 2243 was not enacted with intent to discriminate, and we remand for further proceedings consistent with this opinion. We hold that the Republican Appellants’ appeal of the district court’s holding that there was a waiver of legislative privilege is moot. I. FACTS AND PROCEDURAL HISTORY A. Voting and Voter Registration System in Arizona Arizona has a history of discrimination against minorities and of voting discrimination. For example, the Arizona territorial government in 1909 imposed a literacy 20 MI FAMILIA VOTA V. PETERSEN

test prerequisite to voting, with the explicit aim to limit the “ignorant Mexican vote.” After obtaining statehood, Arizona renewed this literacy test in 1912. Next, in the 1970s and 1980s, Arizona conducted voter roll purges of previously-registered voters, which required all previously- registered individuals to re-register to vote and resulted in fewer minority voters re-registering compared to white voters. There is also an example of a Maricopa County election official requesting DPOC around this time, even though it was not yet required by law. To qualify to vote in Arizona, a person must be a United States citizen, a resident of Arizona, at least eighteen years old, and not adjudicated, incapacitated, or convicted of a felony. Ariz. Const. art. VII, § 2. An eligible person can register to vote in Arizona using the “federal form” created by the United States Election Assistance Commission or can register with the state form prescribed by Arizona law. Public assistance agencies in Arizona typically use the state form to register individuals to vote. The NVRA requires states to “accept and use” the federal form to register voters for federal elections, 52 U.S.C. § 20505(a)(1); Arizona v. Inter Tribal Council of Ariz., Inc., 570 U.S. 1, 9 (2013), and the federal form contains:

only such identifying information (including the signature of the applicant) and other information (including data relating to previous registration by the applicant), as is necessary to enable the appropriate State election official to assess the eligibility of the MI FAMILIA VOTA V. PETERSEN 21

applicant and to administer voter registration and other parts of the election process.

52 U.S.C. § 20508(b)(1). The federal form requires applicants to check a box under penalty of perjury indicating that they are citizens of the United States. The federal form does not require applicants to disclose their birthplace. Although Arizona in previous times did not require applicants to disclose their birthplace, Arizona has long collected birthplace information from state-form applicants—including an optional field on the state form for applicants to include their “state or country of birth.” See 1913 Ariz. Rev. Stat. § 2855. Subject to limitations, 2 states may require additional information from applicants seeking to vote in both state and federal elections. See Inter Tribal Council, 570 U.S. at 12. Since 2004, Arizona has required DPOC in its state form for applicants who want to vote in state elections. “[S]atisfactory evidence of citizenship” includes an applicant’s driver’s license, birth certificate, U.S passport, U.S. naturalization documents, the number of the certificate of naturalization, or Bureau of Indian Affairs card number. See Ariz. Rev. Stat. § 16-166(F).

2 Sections 6 and 9 read together permit states to develop “a mail voter registration form” that requires “only such identifying information (including the signature of the applicant) and other information (including data relating to previous registration by the applicant), as is necessary to enable the appropriate State election official to assess the eligibility of the applicant and to administer voter registration and other parts of the election process.” 52 U.S.C. §§ 20505, 20508(b). These state forms “may not include any requirement for notarization or other formal authentication.” 52 U.S.C. § 20508(b)(3). 22 MI FAMILIA VOTA V. PETERSEN

Before the Supreme Court decided Inter Tribal Council, Arizona required DPOC from all applicants regardless of the form used, but we held and the Supreme Court affirmed that the NVRA, 52 U.S.C. § 20505, preempted Arizona’s requirement of DPOC as applied to federal-form applicants. See Gonzalez v. Arizona, 677 F.3d 383, 398–402 (9th Cir. 2012) [hereinafter Gonzalez II] (en banc), aff’d sub nom. Inter Tribal Council, 570 U.S. at 15. Arizona continued to reject state-form applications without DPOC until 2018 when the then-Arizona Secretary of State entered into the LULAC Consent Decree. League of United Latin Am. Citizens of Ariz. v. Reagan, Doc. 37, No. 2:17-cv-4102 (D. Ariz. 2018) [hereinafter LULAC Consent Decree]. The LULAC Consent Decree requires county recorders to register otherwise eligible voters for federal elections regardless whether they provide DPOC. See id. at 8–10, 13. The LULAC Consent Decree mandates that for state-form and federal-form applicants without DPOC, county recorders must search Arizona Department of Transportation (“ADOT”) records to verify citizenship. See id. at 8-10, 13–14. If citizenship is confirmed by the search, the applicant is registered as a full-ballot voter; but if citizenship cannot be confirmed, the applicant is registered as a federal-only voter. See id. Since the LULAC Consent Decree was filed and until the Supreme Court’s order in RNC v. Mi Familia Vota, No. 24A164, 603 U.S. ___, slip. op. (Aug. 22, 2024), Arizona registered both federal-form and state-form applicants without DPOC as federal-only voters eligible to vote in only federal races. As of July 2023, there were 19,439 active federal-only voters in Arizona who were registered without DPOC. These federal-only voters represent less than half a percent of Arizona’s registered voters. About 0.76% of all MI FAMILIA VOTA V. PETERSEN 23

minority voters in Arizona are registered as federal-only voters and 0.35% of white voters are registered as federal- only voters. B. The Voting Laws 1. Legislative History Arizona’s November 2020 presidential election was decided in favor of President Biden by a margin of 10,457 votes. The Arizona Senate established a committee to audit the 2020 election in response to a claim that non-citizens had illegally cast more than 36,000 ballots in the election. This committee found no evidence of voter fraud. Before passing the Voting Laws, the Arizona Legislature (the “Legislature”) did not establish that any non-citizens were registered to vote in Arizona. Neither House Speaker Toma nor Senate President Petersen recalled the Legislature being presented with or considering evidence of non-citizen voter fraud in Arizona. The allegation that persons who were not citizens swayed the election results was apparently fanciful. Nonetheless, the Voting Laws were introduced to the Arizona House of Representatives in 2022. The Arizona Free Enterprise Club (the “Free Enterprise Club”) drafted the Voting Laws. In its initial advocacy for the Voting Laws, the Free Enterprise Club sent lobbying materials to Arizona legislators with the heading “how more illegals started voting in AZ.” In support of H.B. 2492, a state representative asserted during a House Government and Elections Committee meeting that after the LULAC Consent Decree, more than 11,600 individuals had registered without DPOC as federal- only voters. A majority of the House Rules Committee 24 MI FAMILIA VOTA V. PETERSEN

voted in favor of H.B. 2492 despite concerns voiced by the Committee’s legal counsel that the NVRA likely preempted the bill’s DPOC requirement for federal-form applicants. The Legislature persisted in passing the bill, and it was signed into law by the then-Arizona Governor Ducey. As originally drafted, H.B. 2243 amended Ariz. Rev. Stat. § 16-152 to only require a notice on the state form telling voters that their registrations would be cancelled if they moved permanently to a different state. Another bill, H.B. 2617, was introduced the same month and passed by the Legislature in May 2022. Former-Governor Ducey vetoed the bill, however. After this veto, House Speaker Toma decided to include an amended version of H.B. 2617 in H.B. 2243. Senate President Petersen sponsored the amendment in the Arizona Senate and proposed a floor amendment to incorporate H.B. 2617 into H.B. 2243. Senate President Petersen said that the amendments to H.B. 2243 are essentially “identical to” H.B. 2617, except for some “additional notice requirements.” The explanation for these changes in the legislative record is that H.B. 2243 was amended to “address the [Governor’s] veto letter.” In his deposition, House Speaker Toma said that he could not recall another time when a vetoed voting bill was pushed through to passage in this manner. The Legislature passed H.B. 2243, and it was signed into law by former-Governor Ducey. 2. Changes to Arizona Voter Registration Laws The Voting Laws amend provisions regulating voter registration and enable government officials to require heightened proof of citizenship from federal-form and state- form applicants, prescribing consequences if an applicant does not provide such proof. The Voting Laws also provide for monthly comparisons of some registered voters to several MI FAMILIA VOTA V. PETERSEN 25

databases and cancellation of certain registrations after those database comparisons are made. H.B. 2492 made the following specific changes. First, federal-form applicants without DPOC may still be registered as federal-only voters but are not eligible to vote for president or to vote by mail. Ariz. Rev. Stat. §§ 16- 121.01(D)–(E), 16-127(A). Second, state-form applications without DPOC must be rejected, and it is a felony for a county recorder to fail to reject a state-form application without DPOC. Ariz. Rev. Stat. § 16-121.01(C). Finally, state-form applicants must check a box confirming their citizenship (“checkbox requirement”), disclose their birthplace (“birthplace requirement”), and provide DPOR. Ariz. Rev. Stat. §§ 16-121.01(A), 16-123. H.B. 2243 made the following changes. First, county recorders must periodically check available databases to compare the citizenship status of registered federal-only voters and, if they are not confirmed to be citizens, cancel their registrations (“periodic cancellation of registrations”). Ariz. Rev. Stat. §§ 16-165(A)(10), 16-165(G)–(K). The terms of Arizona Revised Statutes §§ 16-165(G)–(K) provide that the county recorder shall research the citizenship status of registered voters by periodically checking available databases including the ADOT, Social Security Administration, Systematic Alien Verification for Entitlements (“SAVE”), National Association for Public Health Statistics and Information Systems (“NAPHSIS”), and city, town, county, state, and federal databases and, if the registrants are not confirmed to be citizens, cancel their registrations. But there is a problem of voter suppression because these provisions may result in actual citizens having their valid voter registrations cancelled if the databases have not been kept up to date. For example, SAVE may not 26 MI FAMILIA VOTA V. PETERSEN

immediately return updated naturalization records if an individual is naturalized before a weekend or a federal holiday. One provision of H.B. 2243 specifically directs that county recorders must each month, or to the extent practicable, conduct citizenship checks of registered federal- only voters or registered voters who county recorders have “reason to believe” are not citizens. Ariz. Rev. Stat. § 16- 165(I). These citizenship checks are to be done through the SAVE program maintained by the U.S. Citizenship and Immigration Services. Ariz. Rev. Stat. § 16-165(I). C. Procedural History The district court consolidated eight lawsuits challenging provisions of the Voting Laws. The district court resolved some claims at summary judgment and others after a 10-day bench trial. Regarding the NVRA claims, the district court specifically held that:

• Section 6 of the NVRA, 52 U.S.C. § 20505(a)(1), preempted H.B. 2492’s provisions prohibiting federal-only voters from voting by mail and in presidential elections; • Sections 6 and 9 of the NVRA require county recorders to register state-form applicants without DPOR as federal-only voters; • The DPOR requirement violates Section 7 of the NVRA, 52 U.S.C. § 20506(a)(6)(A)(ii); MI FAMILIA VOTA V. PETERSEN 27

• Citizenship checks of voters who county recorders have “reason to believe” are not citizens violate Section 8(b) of the NVRA, 52 U.S.C. § 20507(b); and • The periodic cancellation of registrations violates Section 8(c) of the NVRA (the “90-day Provision”), 52 U.S.C. § 20507(c)(2).

The district court also held that state-form applicants without DPOC must be processed in accordance with the LULAC Consent Decree. Alternatively, the district court held that the NVRA does not let states require DPOC from state-form applicants registering for only federal elections. Regarding the Civil Rights Act claims, the district court held that:

• The checkbox requirement violates the Materiality Provision of the Civil Rights Act, 52 U.S.C. § 10101(a)(2)(B), when applicants provide DPOC; • The birthplace requirement violates the Materiality Provision of the Civil Rights Act; and • The “reason to believe” provision of Arizona Revised Statute § 16-165(I) violates the DSPP Provision of the Civil Rights Act, 52 U.S.C. § 10101(a)(2)(A).

Regarding the Equal Protection claims, the district court held that the requirements of DPOC and DPOR do not violate the Equal Protection Clause of the United States 28 MI FAMILIA VOTA V. PETERSEN

Constitution and found that neither of the Voting Laws was enacted with intent to discriminate. In adjudicating these claims, the district court held that the Legislative Parties waived legislative privilege regarding their motives for the Voting Laws. The Legislative Parties complied with the discovery order that they claim violated their legislative privilege. The district court issued its final judgment on May 2, 2024 and permanently enjoined enforcement of the provisions of the Voting Laws inconsistent with its foregoing holdings. II. STANDARD OF REVIEW Summary judgment is reviewed de novo, and we may affirm summary judgment on any ground supported by the record. Campidoglio LLC v. Wells Fargo & Co., 870 F.3d 963, 973 (9th Cir. 2017). After a bench trial, the district court’s legal conclusions are reviewed de novo, and findings of fact are reviewed for clear error. Yu v. Idaho State Univ., 15 F.4th 1236, 1241–42 (9th Cir. 2021); Fed. R. Civ. P. 52(a)(6). III. DISCUSSION A. Standing Because a “question of appellate jurisdiction must always be resolved before the merits of an appeal are examined or addressed,” we first examine the standing issues. In re Application for Exemption from Elec. Pub. Access Fees by Jennifer Gollan & Shane Shifflett, 728 F.3d 1033, 1036 (9th Cir. 2013) (internal quotation marks and citation omitted). MI FAMILIA VOTA V. PETERSEN 29

“[S]tanding must be met by persons seeking appellate review, just as it must be met by persons appearing in courts of first instance.” Hollingsworth v. Perry, 570 U.S. 693, 705 (2013) (internal quotation marks and citation omitted). “All that is needed to entertain an appeal” on an issue, however, “is one party with standing.” Brnovich v. DNC, 594 U.S. 647, 665 (2021). Under Article III of the United States Constitution, a plaintiff has standing if the plaintiff can show (1) an “injury in fact” that is concrete and particularized and actual or imminent, not hypothetical; (2) that the injury is fairly traceable to the challenged action of the defendant; and (3) that it is likely, as opposed to merely speculative, that the injury will be redressed by a favorable decision. Lujan v. Defs. of Wildlife, 504 U.S. 555, 560–61 (1992). Under this general rule, standing requires a showing of injury, causation, and redressability. See id. 1. The Republican Appellants A federal court’s injunction of a state statute’s implementation injures the state. See Abbott v. Perez, 585 U.S. 579, 602 & n.17 (2018) (“[T]he inability to enforce [the State’s] duly enacted plans clearly inflicts irreparable harm on the State.”). “[A] State must be able to designate agents to represent it in federal court.” Hollingsworth, 570 U.S. at 710. “Respect for state sovereignty” considers, however, “the authority of a State to structure its executive branch in a way that empowers multiple officials to defend its sovereign interests in federal court.” Cameron v. EMW Women’s Surgical Ctr., 595 U.S. 267, 277 (2022). The executive branch does not “hold[] a constitutional monopoly on representing [a State’s] practical interests in court.” Berger v. N.C. State Conf. of the NAACP, 597 U.S. 179, 194 30 MI FAMILIA VOTA V. PETERSEN

(2022) (recognizing the authority of the legislative branch to defend state law on behalf of the State because North Carolina has a statute authorizing the House Speaker and Senate President to do so in certain circumstances); see N.C. Gen. Stat. § 120-32.6(b). No party disputes that the district court’s permanent injunction of parts of the Voting Laws causes a clear and obvious injury to the State. See Abbott, 585 U.S. at 602 & n.17. Although Arizona has designated the Attorney General to represent it in federal court, Arizona Revised Statute § 12-1841(A) states that “[i]n any proceeding in which a state statute . . . is alleged to be unconstitutional, the attorney general and the speaker of the house of the representatives and the president of the senate shall be served with” notice “and shall be entitled to be heard.” Like the North Carolina statute in Berger that authorized the North Carolina House Speaker and Senate President to defend North Carolina’s state laws on behalf of the State, Arizona Revised Statute § 12-1841(A) authorizes the Legislative Parties to defend Arizona’s state laws on behalf of the State. Berger, 597 U.S. at 194. A plain reading of the statute’s literal terms shows that the Legislature intended to “reserve[] to itself some authority to defend state law on behalf of the State” and “empowers” the Legislative Parties here to defend Arizona’s sovereign interests in federal court. See id; EMW Women’s Surgical Ctr., 595 U.S. at 277. We hold that the Legislative Parties have standing to bring their appeal. Given that “[a]ll that is needed to entertain” the Republican Appellants’ appeal “is one party with standing,” the Legislative Parties satisfy the standing requirement for Republican Appellants’ appeal. See Brnovich, 594 U.S. at 665. MI FAMILIA VOTA V. PETERSEN 31

2. Promise Cross-Appellants To invoke representational standing, an organization must show that “(a) its members would otherwise have standing to sue in their own right; (b) the interests it seeks to protect are germane to the organization’s purpose; and (c) neither the claim asserted nor the relief requested requires the participation of individual members in the lawsuit.” Students for Fair Admissions, Inc. v. President of Harvard Coll., 600 U.S. 181, 199 (2023). As a general rule of representational standing, when it is clear and not speculative that a member of a group will be adversely affected by a challenged action and a defendant does not need to know the identity of a particular member to defend against an organization’s claims, the organization does not have to identify particular injured members by name. See Nat’l Council of La Raza v. Cegavske, 800 F.3d 1032, 1041 (9th Cir. 2015), overruled on other grounds by Ariz. All. for Retired Americans v. Mayes, 117 F.4th 1165 (9th Cir. 2024). When we analyze injury in fact, “we consider whether the [parties] face a realistic danger of sustaining a direct injury as a result of the statute’s operation or enforcement.” Protectmarriage.com-Yes on 8 v. Bowen, 752 F.3d 827, 839 (9th Cir. 2014). Promise Arizona is a membership organization with 1,043 dues-paying members as of November 2023, and its members include voters who are naturalized citizens. Absent the district court’s injunction, the enforcement of the Voting Laws and H.B. 2243’s citizenship checks would proceed and apply to any registered voter in Arizona if any county recorder has “reason to believe” that the registered voter is not in fact a citizen; from this, Promise Arizona members face an imminent and “realistic danger of sustaining a direct injury.” Bowen, 752 F.3d at 839. Any of 32 MI FAMILIA VOTA V. PETERSEN

Promise Arizona’s members may be subject to a citizenship check if a county recorder has “reason to believe” they are not a citizen. The danger to voting rights here is that properly registered voters, who in fact are citizens, may have their voter registrations cancelled upon mere and potentially arbitrary suspicion of a county recorder, losing their constitutional right to vote. 3 Improper voter suppression here threatens the public because it appears that Promise Arizona’s members include naturalized citizens and “SAVE may not immediately return updated naturalization records if an individual is naturalized prior to a weekend or a federal holiday.” This threat of future injury is traceable to H.B. 2243 and redressable by maintaining the district court’s

3 The right to vote is a precious constitutional right. As explained in Reynolds v. Sims, “[u]ndeniably the Constitution of the United States protects the right of all qualified citizens to vote, in state as well as in federal elections. A consistent line of decisions by this Court in cases involving attempts to deny or restrict the right of suffrage has made this indelibly clear.” 377 U.S. 533, 554 (1964) (collecting Supreme Court cases restraining acts of voter suppression). Because the right to vote is fundamental, any deprivation of that right caused by voter suppression measures is of grave concern to the public. Federal circuit judges and district judges have consistently restrained acts of voter suppression. See, e.g., Perkins v. City of West Helena, 675 F.2d 201, 216-17 (8th Cir. 1982); Garza v. Cnty. of Los Angeles, 918 F.2d 763, 774-75 (9th Cir. 1990); Price v. N.Y. State Bd. of Elections, 540 F.3d 101, 110-12 (2d Cir. 2008); Obama for Am. v. Husted, 697 F.3d 423, 428-36 (6th Cir. 2012); Veasey v. Abbott, 830 F.3d 216, 235-43 (5th Cir. 2016); Democratic Exec. Comm. of Fla. v. Detzner, 347 F. Supp. 3d. 1017, 1029-31 (N.D. Fla. 2018); McConchie v. Scholz, 567 F. Supp. 3d 861, 885-89 (N.D. Ill. 2021). Stated another way, the exercise of the fundamental right to vote is a cornerstone premise of democracy; suppression of that right to vote is not only hostile to the right to vote but should also be firmly and unequivocally rejected by the courts that guard that right. MI FAMILIA VOTA V. PETERSEN 33

injunction currently preventing enforcement of H.B. 2243. See Lujan, 504 U.S. at 560. Because the Promise Arizona members satisfy the three prongs for standing required by Lujan, Promise Arizona’s members have standing to sue. See id. at 560–61. Because one or more members of Promise Arizona may be adversely affected by H.B. 2243 and the State does not need to know the identity of a particular member to respond to Promise Arizona’s claim of injury, Promise Arizona need not identify by name its members who would be injured by H.B. 2243 absent the injunction. See Nat’l Council of La Raza, 800 F.3d at 1041. Because Promise Arizona’s “core activities include registering voters, educating voters, and turning out the vote,” protecting the voting rights of its members is germane to Promise Arizona’s purpose. See Students for Fair Admissions, 600 U.S. at 199. Promise Arizona’s cross- appeal and requested relief do not require the participation of its members in this litigation, and the State does not contend otherwise. We hold that Promise Arizona has representational standing, and the Promise Cross-Appellants have standing to pursue their cross-appeal. See Brnovich, 594 U.S. at 665 (“All that is needed to entertain an appeal” on an issue “is one party with standing.”). B. The NVRA “Because the power the Elections Clause confers is none other than the power to pre-empt, the reasonable assumption is that the statutory text accurately communicates the scope of Congress’s preemptive intent . . . . Unlike the States’ historic police powers, the States’ role in regulating 34 MI FAMILIA VOTA V. PETERSEN

congressional elections . . . has always existed subject to the express qualification that it terminates according to federal law.” Inter Tribal Council, 570 U.S. at 14–15 (internal quotation marks and citations omitted); see also Gonzalez II, 677 F.3d at 392 (“[T]he ‘presumption against preemption’ and ‘plain statement rule’ that guide Supremacy Clause analysis are not transferable to the Elections Clause context.” (citation omitted)). State law is preempted when a federal statute expressly preempts state law. Chamber of Com. v. Bonta, 62 F.4th 473, 482 (9th Cir. 2023). State law is also preempted “where it is impossible for a private party to comply with both state and federal requirements, or where state law stands as an obstacle to the accomplishment and execution of the full purposes and objectives of Congress.” English v. Gen. Elec. Co., 496 U.S. 72, 79 (1990) (internal quotation marks and citations omitted). 1. Sections 6, 7, and 9 of the NVRA Under Section 6 of the NVRA, states must “accept and use” the federal form. 52 U.S.C. § 20505(a)(1). The Supreme Court has held that this means that the federal form must “be accepted as sufficient for the requirement it is meant to satisfy.” Inter Tribal Council, 570 U.S. at 10 (emphasis in original). Section 6 of the NVRA permits states to use their own state forms for federal elections. See 52 U.S.C. § 20505(a)(2). But those forms must comply with Section 9 and “require only such identifying information . . . and other information . . . as is necessary to enable the appropriate State election official to assess the eligibility of the applicant and to administer voter registration.” 52 U.S.C. § 20508(b)(1). MI FAMILIA VOTA V. PETERSEN 35

Section 7 of the NVRA provides that any voter registration agency that “provides service or assistance in addition to conducting voter registration shall . . . distribute with each application for such service or assistance” the federal form or an “equivalent” form. 52 U.S.C. § 20506(a)(6)(A)(ii). a. Requirement of DPOC to Vote by Mail The Arizona statutory requirement of DPOC to vote by mail means Arizona’s statute conflicts with its need to “use” the federal form to register federal-form applicants to vote in federal elections by mail, because Arizona would not “accept” the federal form as sufficient without DPOC. Arizona’s statute would require federal-only voters seeking to cast their ballots by mail to provide more information than what the federal form requires. See English, 496 U.S. at 79; Inter Tribal Council, 570 U.S. at 10. Arizona’s statute thereby conflicts with Section 6’s mandate that states “accept and use” the federal form. See Inter Tribal Council, 570 U.S. at 15 (“[A] state-imposed requirement of evidence of citizenship not required by the Federal Form is ‘inconsistent with’ the NVRA’s mandate that States ‘accept and use’ the Federal Form.” (citation omitted)). We conclude that the requirement of DPOC to vote by mail conflicts with Section 6 of the NVRA and so that provision of H.B. 2492 is preempted and cannot stand. The requirement of DPOC to vote by mail is also an obstacle to the NVRA’s purpose and preempted by obstacle preemption as well. The NVRA’s findings state:

the right of citizens of the United States to vote is a fundamental right; it is the duty of the Federal, State, and local governments to 36 MI FAMILIA VOTA V. PETERSEN

promote the exercise of that right; and discriminatory and unfair registration laws and procedures can have a direct and damaging effect on voter participation in elections for Federal office and disproportionately harm voter participation by various groups, including racial minorities.

52 U.S.C. § 20501(a). The NVRA aims to “enhance[] the participation of eligible citizens as voters in elections for Federal office.” 52 U.S.C. § 20501(b)(2). “What is a sufficient obstacle is a matter of judgment, to be informed by examining the federal statute as a whole and identifying its purpose and intended effects.” Crosby v. Nat’l Foreign Trade Council, 530 U.S. 363, 373 (2000). “If the purpose of the act cannot otherwise be accomplished— if its operation within its chosen field else must be frustrated and its provisions be refused their natural effect—the state law must yield to the regulation of Congress within the sphere of its delegated power.” Id. Because the NVRA seeks to “enhance[] the participation of eligible citizens as voters in [federal] elections,” 52 U.S.C. § 20501(b)(2), the requirement of DPOC to vote by mail is a “sufficient obstacle” to the “accomplishment and execution of the [NVRA’s] full purposes” and “must yield to the regulation of Congress” within federal elections. See Crosby, 530 U.S. at 373; see also English, 496 U.S. at 79. By restricting federal-only voters without DPOC to only in- person voting, the DPOC requirement limits federal-only voters’ “fundamental right” to vote, impedes the “duty of the Federal, State, and local governments to promote the exercise of that right,” and frustrates the purpose of the MI FAMILIA VOTA V. PETERSEN 37

NVRA to “enhance[] the participation of . . . voters in [federal] elections.” See 52 U.S.C. §§ 20501(a), 20501(b)(2). Our conclusion is reinforced by the fact that about 89% of Arizona voters cast ballots by mail in 2020. Congress explicitly noted in its findings for the NVRA that “discriminatory and unfair registration laws and procedures can have a direct and damaging effect on voter participation in [federal] elections.” 52 U.S.C. § 20501(a)(3). That finding demonstrates beyond doubt Congress’s intent to increase voter turnout through diminishing barriers to registration laws and procedures. The Republican Appellants contend that the NVRA “governs voter registration—not rules for casting a ballot by mail.” If the NVRA is read, as the Republican Appellants contend, to regulate only “registration” in isolation from the rest of the voting process such as casting a ballot by mail, then states could “accept” the federal form solely to place individuals’ names on the voting rolls but then preclude those who do not provide DPOC from casting vote-by-mail ballots in federal elections. Under such a reading, the federal form would “cease[] to perform any meaningful function, and would be a feeble means of” accomplishing the purpose of “enhanc[ing] the participation of eligible citizens as voters in [federal] elections.” See Inter Tribal Council, 570 U.S. at 13; 52 U.S.C. § 20501(b)(2). Such a narrow view of the NVRA’s purpose is contrary to the text of the NVRA which declares the right “to vote” is a fundamental right and establishes purposes beyond registration. See 52 U.S.C. § 20501. The Republican Appellants’ view also narrows the NVRA’s ability to preempt, contrary to the Supreme Court’s view of Congress’s power to preempt through Elections Clause litigation. Inter Tribal Council, 570 U.S. at 14 (“Because the power the Elections Clause confers is none 38 MI FAMILIA VOTA V. PETERSEN

other than the power to pre-empt, the reasonable assumption is that the statutory text accurately communicates the scope of Congress’s preemptive intent.”) We hold that H.B. 2492’s requirement of DPOC to vote by mail is preempted by Section 6 of the NVRA and by obstacle preemption. b. Requirement of DPOC to vote in presidential elections Requiring DPOC to vote in presidential elections is expressly preempted by the NVRA, which requires states to “accept and use” the federal form “for the registration of voters in elections for Federal office.” See Bonta, 62 F.4th at 482; 52 U.S.C. § 20505(a)(1) (Section 6 of the NVRA); Inter Tribal Council, 570 U.S. at 10. Republican Appellants contend, however, that the NVRA does not apply to presidential elections. They contend that Congress enacted the NVRA under the authority granted to it in U.S. Const. art. I, § 4 (the “Elections Clause”), empowering Congress to preempt only “Manner” regulations for congressional elections. By contrast, U.S. Const. art. II § 1 permits Congress to preempt only “the Time of chusing the Electors, and the Day on which they shall give their Votes” for presidential elections. When analyzing express preemption, we focus on the “plain meaning” of the statute. See Cal. Rest. Ass’n v. City of Berkeley, 89 F.4th 1094, 1101 (9th Cir. 2024). Here, the plain language of the NVRA shows an intent to regulate “voter registration for elections for Federal office” defined to include the “office of President or Vice President.” 52 U.S.C. §§ 20507(a), 30101(3). The NVRA provides that the scope of preemption includes all federal elections, including MI FAMILIA VOTA V. PETERSEN 39

presidential elections. See Inter Tribal Council, 570 U.S. at 14; 52 U.S.C. §§ 20507(a), 30101(3). Aside from the NVRA’s plain language, our precedent also requires us to hold that Congress has the power to control registration for presidential elections. In 1934, the Supreme Court rejected a narrow framing of Congress’s power over presidential elections, like the view argued here by Republican Appellants. The Supreme Court reasoned:

The only point of the constitutional objection necessary to be considered is that the power of appointment of presidential electors and the manner of their appointment are expressly committed by section 1, art. 2, of the Constitution to the states, and that the congressional authority is thereby limited to determining ‘the Time of chusing the Electors, and the Day on which they shall give their Votes; which Day shall be the same throughout the United States.’ So narrow a view of the powers of Congress in respect of the matter is without warrant.

Burroughs v. United States, 290 U.S. 534, 544 (1934). The Court squarely held that Congress had the power to pass legislation to protect the integrity of the federal election process in the presidential election. Id. at 545; see also Buckley v. Valeo, 424 U.S. 1, 13 n.16 (1976) (citing to Burroughs as more generally “recogniz[ing] broad congressional power to legislate in connection with the elections of the President and Vice President”). We have also recognized Congress’s power to regulate all federal elections under the NVRA. See Voting Rts. Coal. 40 MI FAMILIA VOTA V. PETERSEN

v. Wilson, 60 F.3d 1411, 1413–14 (9th Cir. 1995) (rejecting a challenge to the constitutionality of the NVRA in part because “the Supreme Court has read the grant of power to Congress in Article I, section 4 [of the U.S. Constitution] as quite broad” and has endorsed that “[t]he broad power given to Congress over congressional elections has been extended to presidential elections” (citing Burroughs, 290 U.S. at 545)). We hold that H.B. 2492’s requirement of DPOC to vote in presidential elections is preempted by Section 6 of the NVRA. c. Requirement of DPOR for state-form applicants registering for federal elections As former Chief Justice Rehnquist persuasively explained, statutory interpretation requires courts to “presume that the legislature says in a statute what it means . . . [t]hus, our inquiry begins with the statutory text, and ends there as well if the text is unambiguous.” BedRoc Ltd., LLC v. United States, 541 U.S. 176, 183 (2004). “We give the words of a statute their ‘ordinary, contemporary, common meaning,’” absent an indication to the contrary from Congress. Williams v. Taylor, 529 U.S. 420, 431 (2000) (citation omitted). The NVRA allows states to seek only the information “necessary” to assess an applicant’s eligibility, so whether the NVRA lets Arizona require DPOR from state-form applicants registering for only federal elections depends on whether DPOR is necessary for registration. We hold that DPOR is not “necessary” as required by Section 9 of the NVRA. Because Arizona limits voting to residents of the State, an applicant’s location of residence is “necessary to enable the appropriate State election official to assess the eligibility of the applicant” to vote in state MI FAMILIA VOTA V. PETERSEN 41

elections. See 52 U.S.C. § 20508(b)(1); Ariz. Const. art. VII, § 2(A); Ariz. Rev. Stat. § 16-101(A)(3). But DPOR is not “necessary” because voters who obtain an out-of-state license or identification and receive a notice from the county recorder requesting confirmation of residency must only attest “under penalty of perjury” that the voter is still a resident of Arizona. See Ariz. Rev. Stat. § 16-165(F). The ordinary meaning of “necessary” is “essential.” See Williams, 529 U.S. at 431; Necessary, Black’s Law Dictionary (12th ed. 2024); Necessary, Oxford English Dictionary (2d ed. 1989). The requirement of DPOR is not “necessary” for new applicants because attestation sufficiently confirms the eligibility of registered voters. See 52 U.S.C. § 20508(b)(1); Ariz. Rev. Stat. § 16-165(F). Our inquiry ends here because the text of the NVRA is unambiguous. See BedRoc, 541 U.S. at 183. We hold that the DPOR requirement violates Section 6 of the NVRA for state-form applicants registering for federal elections. The district court held that “if the Secretary of State supplies the State Form to public assistance agencies, the State Form must be ‘equivalent’ or ‘virtually identical’ to the Federal Form.” The state form is not equivalent to the federal form because the state form has unnecessary additional requirements of DPOC, DPOR, and birthplace. Compare Ariz. Rev. Stat. §§ 16-121.01(A), 16-121.01(C), 16-123, 16-166(F) with 52 U.S.C. § 20508(b)(1). Because public assistance agencies in Arizona typically use the state form to register individuals to vote, the state form must be “equivalent” to the federal form. See 52 U.S.C. § 20506(a)(6)(A)(ii). The DPOR requirement renders the state form not “equivalent” to the federal form for applicants without DPOR. Applicants who do not include DPOR on the state 42 MI FAMILIA VOTA V. PETERSEN

form will not be registered as federal-only voters, but if the same applicants use the federal form, they will be registered. That difference prevents the forms from being “virtually identical” for applicants without DPOR, and the requirement of DPOR for state-form applicants violates Section 7 of the NVRA. Republican Appellants contend that because Section 9 of the NVRA permits state forms to differ from the federal form, compliance with Section 9 makes a state form equivalent to the federal form for the purposes of Section 7. But “[w]e give the words of a statute their ‘ordinary, contemporary, common meaning,’” absent an indication to the contrary from Congress, and here the ordinary meaning of “equivalent” means “virtually identical.” See Williams, 529 U.S. at 431 (citation omitted); Equivalent, Black’s Law Dictionary (12th ed. 2024); see also Equivalent, Oxford English Dictionary (2d ed. 1989) (defining equivalent as “virtually the same thing; identical in effect”). Also, “[w]hen interpreting the language of a statute, we do not look at individual subsections in isolation” but “read the words in their context and with a view to their place in the overall statutory scheme.” Tovar v. Sessions, 882 F.3d 895, 901 (9th Cir. 2018) (quoting King v. Burwell, 576 U.S. 473, 486 (2015)). While Sections 6 and 9 read together let states develop “a mail voter registration form” that meets the criteria stated in 52 U.S.C. § 20508(b) and let states include information necessary to determine voter eligibility that is not otherwise on the federal form, Section 7 does not do so. Section 7 permits use of only the federal form and “the office’s own form if it is equivalent” to the federal form. Compare 52 U.S.C. §§ 20505(a)(2) (Section 6 of the NVRA), 20508(b) (Section 9 of the NVRA) with 52 U.S.C. § 20506(a)(6)(A) (Section 7 of the NVRA). MI FAMILIA VOTA V. PETERSEN 43

We hold that H.B. 2492’s state-form requirement of DPOR to register for federal elections violates Sections 6 and 7 of the NVRA. 2. Section 8 of the NVRA Section 8(b) of the NVRA provides that “[a]ny State program or activity to protect the integrity of the electoral process by ensuring the maintenance of an accurate and current voter registration roll for elections for Federal office . . . shall be uniform, nondiscriminatory, and in compliance with the Voting Rights Act of 1965.” 52 U.S.C. § 20507(b)(1). In United States v. Florida, the district court held that the Secretary of State’s list maintenance program “probably ran afoul” of Section 8(b) of the NVRA because its “methodology made it likely that the properly registered citizens who would be required to respond and provide documentation would be primarily newly naturalized citizens.” 870 F. Supp. 2d 1346, 1350 (N.D. Fla. 2012). Thus, “[t]he program was likely to have a discriminatory impact on these new citizens.” Id. The 90-day Provision (Section 8(c) of the NVRA) mandates that states “shall complete, not later than 90 days prior to the date of a primary or general election for Federal office, any program the purpose of which is to systematically remove the names of ineligible voters from the official lists of eligible voters.” 52 U.S.C. § 20507(c)(2)(A). It also lists exceptions to the 90-day Provision. See 52 U.S.C. § 20507(c)(2)(B). These exceptions are removals “at the request of the registrant,” or “by reason of criminal conviction or mental incapacity,” “the death of the registrant,” “a change in the residence of the registrant,” or “correction of registration records pursuant to this chapter.” 52 U.S.C. §§ 20507(a)(3)–(4), 20507(c)(2)(B)(ii). 44 MI FAMILIA VOTA V. PETERSEN

a. Citizenship checks of voters who county recorders have “reason to believe” are not citizens Under H.B. 2243, county recorders must conduct citizenship checks of registered federal-only voters or registered voters who county recorders have “reason to believe” are not citizens using the SAVE program maintained by the U.S. Citizenship and Immigration Services. Ariz. Rev. Stat. § 16-165(I). The citizenship checks are non-uniform and are discriminatory in effect because it is “likely that the properly registered citizens who would be required to respond and provide documentation would be” naturalized citizens. See Florida, 870 F. Supp. 2d at 1350. Although the Voting Laws are written as if they confirm the citizenship status of all voters, running a citizenship check through SAVE requires an immigration number. See Ariz. Rev. Stat. § 16-165(I). As a result, county recorders can only conduct SAVE checks on naturalized citizens and non-citizens. Absent injunction, naturalized citizens would be at risk of county recorders’ subjective decisions to investigate their citizenship status because of the “reason to believe” provision, which will not apply to U.S.-born citizens. The citizenship checks are “likely to have a discriminatory impact on [naturalized] citizens,” and on its face, the “reason to believe” provision would have a non-uniform and discriminatory impact. See id; Florida, 870 F. Supp. 2d at 1350. We hold that H.B. 2243’s citizenship checks violate Section 8(b) of the NVRA. b. Periodic cancellation of registrations The Republican Appellants contend that because “[t]he NVRA does not discuss . . . a State’s authority to remove noncitizens from the voter rolls,” the NVRA does not MI FAMILIA VOTA V. PETERSEN 45

regulate the periodic cancellation of registrations and does not forbid removal of noncitizens from voter rolls. But that contention mischaracterizes the district court’s holding, which never said that the NVRA forbids removal of noncitizens from voter rolls. Rather, the district court held that the periodic cancellation of registrations violates the 90- day Provision of the NVRA to the extent it “allow[s] systematic cancellation of registrations within 90 days of a[] [federal] election.” The Republican Appellants also contend that the periodic cancellation of registrations is not subject to the 90- day Provision because the 90-day Provision is limited to “general program[s]” to remove ineligible voters who are no longer eligible because of conviction, death, or change in residence. See 52 U.S.C. §§ 20507(a)(3)–(4). “We give the words of a statute their ‘ordinary, contemporary, common meaning,’” absent an indication to the contrary from Congress. See Williams, 529 U.S. at 431 (citation omitted). “Where Congress explicitly enumerates certain exceptions to a general prohibition, additional exceptions are not to be implied, in the absence of evidence of a contrary legislative intent.” Andrus v. Glover Constr. Co., 446 U.S. 608, 616–17 (1980). The 90-day Provision requires that states “shall complete, not later than 90 days prior to [a federal election] . . . any program” that “systematically remove[s] the names of ineligible voters from the official lists of eligible voters.” 52 U.S.C. § 20507(c)(2)(A) (emphasis added). Based on the ordinary meaning of “any,” “program” should be construed to have an expansive meaning. Any, Oxford English Dictionary (rev. ed. 2024) (defining any “[w]ith singular noun in affirmative contexts” as being “used to refer to a 46 MI FAMILIA VOTA V. PETERSEN

member of a particular group or class without distinction or limitation”), available at https://doi.org/10.1093/OED/4481770737. The Supreme Court has commented that “the word ‘any’ has an expansive meaning,” namely, “one or some indiscriminately of whatever kind.” United States v. Gonzales, 520 U.S. 1, 5 (1997). The prior provision, 52 U.S.C. § 20507(c)(1), limits the applicable program to 52 U.S.C. § 20507(a)(4) by saying that “[a] State may meet the requirement of subsection (a)(4) by establishing a program.” By contrast, the 90-day Provision does not limit the applicable programs to a specific provision and instead enumerates exceptions. See 52 U.S.C. §§ 20507(c)(2)(A)–(B). That the 90-day Provision does not contain a similar limiting provision to describe the programs to which it applies suggests that Congress intended “any program” in the 90-day Provision to have an expansive meaning. Similarly, Congress’s enumerated exceptions to the 90-day Provision suggest that Congress intended for “any program” to have a broad meaning absent an exception. See 52 U.S.C. § 20507(c)(2)(B). Holding that the 90-day Provision does not apply to the periodic cancellation of registrations would create a new exception, and “[w]here Congress explicitly enumerates certain exceptions . . . additional exceptions are not to be implied, in the absence of evidence of a contrary legislative intent.” See Andrus, 446 U.S. at 616–17; Williams, 529 U.S. at 431. We conclude that the 90-day Provision applies to the periodic cancellation of registrations. The plain language of the 90-day Provision lets states continue any non-systematic cancellation of registrations within the 90-day window. 52 U.S.C. § 20507(c)(2)(A). A non-systematic or “individualized” removal program relies MI FAMILIA VOTA V. PETERSEN 47

on “individualized information or investigation” to determine removal of ineligible voters from voting rolls rather than cancelling batches of registrations based on a set procedure such as “us[ing] a mass computerized data- matching process to compare the voter rolls with other state and federal databases, followed by the mailing of notices.” See Arcia v. Fla. Sec’y of State, 772 F.3d 1335, 1344 (11th Cir. 2014). The periodic cancellation of registrations is required by H.B. 2243. But that statute’s language does not limit cancellation to at least 90 days before a federal election. See Ariz. Rev. Stat. §§ 16-165(A)(10), 16-165(G)–(K). And here, none of the NVRA’s enumerated exceptions to the 90- day Provision applies. Compare Ariz. Rev. Stat. §§ 16- 165(A)(10), 16-165(G)–(K) with 52 U.S.C. §§ 20507(a)(3)– (4), 20507(c)(2)(B). Whether the periodic cancellation of registrations required by Arizona’s law violates the 90-day Provision depends on whether it is a “systematic” or an “individualized” removal program. Arizona Revised Statute § 16-165(A)(10) provides that “[t]he county recorder shall cancel a registration: . . . [w]hen the county recorder obtains information pursuant to this section and confirms that the person registered is not a United States citizen” and before cancelling the registration, the “county recorder shall send the person notice by forwardable mail that the person’s registration will be cancelled in thirty-five days unless the person provides satisfactory evidence within thirty-five days.” Arizona Revised Statutes §§ 16-165(G)–(K) provides that the county recorder shall obtain such information by periodically checking available databases including the ADOT, Social Security Administration, SAVE, NAPHSIS, and city, town, county, state, and federal databases to research the 48 MI FAMILIA VOTA V. PETERSEN

citizenship status of registered voters 4 and, if they are not confirmed to be citizens, cancel their registrations. This periodic cancellation of registrations does not rely on “individualized information or investigation” but rather comparisons to databases. It is a systematic removal program and violates the 90-day Provision because it permits systematic cancellation of registrations within 90 days preceding a federal election. Like the program that violated the 90-day Provision in Arcia, H.B. 2243 uses “a mass computerized data-matching process to compare the voter rolls with other state and federal databases, followed by the mailing of notices.” 772 F.3d at 1344. Cancellation of batches of registered voters based on a set procedure is systematic as opposed to individualized, and like the program in Arcia, one database that H.B. 2243 uses is SAVE: the “Systematic Alien Verification for Entitlements.” See id. (emphasis in original). The Republican Appellants contend that such periodic cancellation is individualized because Arizona Revised Statute § 16-165(A)(10) provides a person with mail notice and opportunity to respond after information is obtained “pursuant to this section . . . that the person registered is not a United States citizen.” That argument does not persuade us because the statute details how such information is obtained: through the systematic comparison of all—or

4 Some provisions are limited to specific types of registered voters. While most provisions apply to all registered voters, Arizona Revised Statute § 16-165(I) specifies citizenship checks against SAVE will be for persons “who the county recorder has reason to believe are not United States citizens and persons who are registered to vote without satisfactory [DPOC].” Arizona Revised Statute § 16-165(J) similarly limits checks against NAPHSIS to persons registered to vote without DPOC. MI FAMILIA VOTA V. PETERSEN 49

groups of—registered voters to various databases. See Ariz. Rev. Stat. §§ 16-165(G)–(K). The mailing of notices is to individuals, but this is only after the systematic comparison prompts the mailing, as opposed to it being prompted by an individualized investigation. Our holding is consistent with the purposes of the 90-day Provision and of the NVRA generally. The NVRA’s purposes include “protect[ing] the integrity of the electoral process,” “ensur[ing] that accurate and current voter registration rolls are maintained,” and “establish[ing] procedures that will increase the number of eligible citizens who register to vote in elections for Federal office.” 52 U.S.C. § 20501(b). As the Eleventh Circuit has recognized, the 90-day Provision is designed to balance with care the NVRA’s purposes by acting “cautious[ly]” with respect to systematic cancellation programs in the lead up to an election because such programs can cause inaccurate removal and “[e]ligible voters removed days or weeks before Election Day will likely not be able to correct the State’s errors in time to vote.” Arcia, 772 F.3d at 1346. In sharp contrast, individualized removals that are not prohibited by the 90-day Provision are based on more “rigorous individualized inquir[ies], leading to a smaller chance for mistakes.” Id. In light of the purposes of the 90-day Provision and the NVRA, the periodic cancellation of registrations required by Arizona’s law is precisely the type of systematic cancellation program that the 90-day Provision was meant to preclude. The periodic cancellation of registrations is based on the systematic comparison of registered voters to various databases, see Ariz. Rev. Stat. §§ 16-165(G)–(K), which will likely cause inaccurate removals. Mailing notices to individuals does not change that because if the affected voter 50 MI FAMILIA VOTA V. PETERSEN

does not respond to the notice with “satisfactory evidence within thirty-five days,” their voter registration will still be cancelled. Ariz. Rev. Stat. § 16-165(A)(10). Because of that short period for response to be given, there is an unduly high risk that voter registrations will be inaccurately cancelled because of the systematic comparisons and eligible voters “will likely not be able to correct the State’s errors in time to vote,” depriving them of their fundamental right to vote. See Arcia, 772 F.3d at 1346. Such a voter suppression measure should not be tolerated by the law, which protects the constitutional right of citizens to vote. We hold that H.B. 2243’s periodic cancellation of registrations violates the 90-day Provision of the NVRA to the extent that H.B. 2243 authorizes systematic cancellation of registrations within 90 days before a federal election. C. The LULAC Consent Decree A consent decree approved by a court is an enforceable, final judgment with the force of res judicata. SEC v. Randolph, 736 F.2d 525, 528 (9th Cir. 1984); see also Rufo v. Inmates of Suffolk Cnty. Jail, 502 U.S. 367, 391 (1992) (“[A] consent decree is a final judgment that may be reopened only to the extent that equity requires.”). For this reason, “the equitable decree based on the [parties’] agreement ‘is subject to the rules generally applicable to other judgments and decrees.’” Gates v. Shinn, 98 F.3d 463, 468 (9th Cir. 1996) (quoting Rufo, 502 U.S. at 378). Because it is a final judgment, a consent decree “may not lawfully be revised, overturned or refused faith and credit by another Department of Government.” Taylor v. United States, 181 F.3d 1017, 1024 (9th Cir. 1999) (en banc) (quoting Chicago & S. Air Lines, Inc. v. Waterman S.S. Corp., 333 U.S. 103, 113 (1948)). Consent decrees are binding final judgments MI FAMILIA VOTA V. PETERSEN 51

that remain in force permanently even if the entering court explicitly retains jurisdiction only for a limited period of time. See id. at 1024–26; see, e.g., Thompson v. U.S. Dep’t of Hous. & Urb. Dev., 404 F.3d 821, 828, 833 (4th Cir. 2005) (court retained authority to enforce terms of decree beyond seven-year period during which it retained jurisdiction); Roberts v. St. Regis Paper Co., 653 F.2d 166, 171–72 (5th Cir. 1981) (clause retaining jurisdiction for five years did not “refer[] to the life of the decree itself,” and decree’s injunction was permanent). Although the district court entering the LULAC Consent Decree retained jurisdiction only until December 21, 2020, the consent decree has never been set aside. See Taylor, 181 F.3d at 1024. That the court retained jurisdiction for a limited period of time supports that the LULAC Consent Decree is a final judgment under Taylor and does not suggest that the preclusive effect of the final judgment expired after the docket was closed. See id. at 1023. The LULAC Consent Decree remains an enforceable, binding final judgment. Contrary to the LULAC Consent Decree requirement that Arizona county recorders accept state-form applications without DPOC and register those applicants as federal-only voters, H.B. 2492 would require county recorders to do the opposite and reject state-form applications without DPOC. Compare LULAC Consent Decree at 8–10 with Ariz. Rev. Stat. § 16-121.01(C). Because H.B. 2492 requires county recorders to violate the LULAC Consent Decree’s requirements, the LULAC Consent Decree bars enforcement of this provision of H.B. 2492. Republican Appellants contend that the Secretary of State cannot “via a private contract divest the Legislature of 52 MI FAMILIA VOTA V. PETERSEN

any portion of its sovereign authority.” See State v. Prentiss, 786 P.2d 932, 936 (Ariz. 1989) (“The legislature has the exclusive power to declare what the law shall be [in Arizona].”). But the LULAC Consent Decree does not divest the Legislature of its sovereign authority. Instead, it cabins the authority of parties to the decree, specifically the Secretary of State of Arizona and the Maricopa County Recorder, and limits the ability of executive officers in Arizona to enforce legislation contrary to the final judgment of the federal decree. See LULAC Consent Decree at 1. Sitting en banc in Taylor v. United States, we recognized that “[t]he Constitution’s separation of legislative and judicial powers denies [Congress] the authority” to “enact[] retroactive legislation requiring an Article III court to set aside a final judgment.” 181 F.3d at 1026; see also id. at 1024 (“Congress may change the law and, in light of changes in the law or facts, a court may decide in its discretion to reopen and set aside a consent decree . . . but Congress may not direct a court to do so with respect to a final judgment (whether or not based on consent) without running afoul of the separation of powers doctrine.”) (emphasis in original). The Republican Appellants present no authority suggesting that Arizona’s state legislature may permissibly nullify a final judgment entered by an Article III court. The principle stated in our en banc panel decision in Taylor applies with equal force here. As Chief Justice Marshall explained: “If the legislatures of the several states may, at will, annul the judgments of the courts of the United States, and destroy the rights acquired under those judgments, the constitution itself becomes a solemn mockery.” United States v. Peters, 9 U.S. (5 Cranch) 115, 136 (1809); see Cooper v. Aaron, 358 U.S. 1, 18 (1958) (noting that “Chief Justice Marshall spoke for a unanimous Court” in Peters). We decline Arizona’s MI FAMILIA VOTA V. PETERSEN 53

invitation for us to reject the law established by Chief Justice John Marshall and a unanimous court in 1809. That law has never been in doubt. We hold that the LULAC Consent Decree bars Arizona election officials from enforcing H.B. 2492’s mandate to reject state-form applications without DPOC. 1. Alternatively, the NVRA does not let Arizona require DPOC from applicants registering for only federal elections. As discussed in Section III.B.1.c, although Section 6 of the NVRA lets states use their own state forms for federal elections, those forms must comply with Section 9, under which states may seek only information “necessary” to assess an applicant’s eligibility to vote. 52 U.S.C. §§ 20505(a)(2), 20508(b)(1); see supra pp. 40–41. The NVRA does not let Arizona require DPOC from state-form applicants registering for only federal elections because DPOC is not legitimately necessary for registration. To elaborate, DPOC is not “necessary” as required by Section 9 of the NVRA because, although citizenship is “necessary to enable the appropriate State election official to assess the eligibility of the applicant” to vote in federal elections, see 52 U.S.C. § 20508(b)(1), the state form’s checkbox requirement supplies proof of citizenship by an attestation. Ariz. Rev. Stat. § 16-121.01(A). The ordinary meaning of “necessary” is “essential,” and the challenged requirement of DPOC for state-form applicants registering to vote in only federal elections is not “essential” because the checkbox requirement already gives proof of citizenship. See Williams, 529 U.S. at 431; Necessary, Black’s Law Dictionary (12th ed. 2024); Necessary, Oxford English Dictionary (2d ed. 1989); 52 U.S.C. § 20508(b)(1). Our 54 MI FAMILIA VOTA V. PETERSEN

inquiry ends here because the text of the NVRA is unambiguous. See BedRoc, 541 U.S. at 183. Republican Appellants urge that we have held that Section 9 “plainly allow[s] states, at least to some extent, to require their citizens to present evidence of citizenship when registering to vote.” See Gonzalez v. Arizona, 485 F.3d 1041, 1050–51 (9th Cir. 2007) [hereinafter Gonzalez I]). Although Gonzalez I holds that “[t]he language of the statute does not prohibit documentation requirements,” the Gonzalez I case was decided at the preliminary injunction stage, addressing only whether plaintiffs showed a likelihood of succeeding on the merits of this claim. See 485 F.3d at 1050–51. We have not decided whether and to what extent states may “require their citizens to present evidence of citizenship when registering to vote.” See id. at 1051. And because we on en banc review did not decide that question in Gonzalez II, the quoted language from Gonzalez I is not persuasive here. The issue presented in this case was not decided in our en banc decision in Gonzalez II. See 677 F.3d at 400 (“Even assuming, without deciding, that Arizona is correct in its interpretation of [Section 9 of the NVRA] . . .”). Similarly, Section 7 of the NVRA requires that state forms supplied to public assistance agencies be “‘equivalent’ or ‘virtually identical’” to the federal form. 52 U.S.C. § 20506(a)(6)(A)(ii); see supra pp. 41–42. Because public assistance agencies in Arizona typically use the state form to register individuals to vote, the state form must be “equivalent” to the federal form. See 52 U.S.C. § 20506(a)(6)(A)(ii). Here, the state form is not equivalent to the federal form because the state form has unnecessary additional requirements of DPOC, DPOR, and birthplace. MI FAMILIA VOTA V. PETERSEN 55

Compare Ariz. Rev. Stat. §§ 16-121.01(A), 16-121.01(C), 16-123, 16-166(F) with 52 U.S.C. § 20508(b)(1). The DPOC requirement renders the state form not “equivalent” to the federal form for applicants without DPOC. If applicants who do not include DPOC use the state form, they will not be registered as federal-only voters but if they use the federal form, they will be registered. That difference prevents the forms from being “virtually identical” for applicants without DPOC, and the requirement of DPOC for state-form applicants violates Section 7 of the NVRA. We hold that the NVRA does not let states require DPOC from applicants registering for only federal elections. D. The Civil Rights Act 1. The Materiality Provision The Materiality Provision prohibits states from denying an individual the right to vote “because of an error or omission on any record or paper relating to any application, registration, or other act requisite to voting, if such error or omission is not material in determining whether such individual is qualified under State law to vote in such election.” 52 U.S.C. § 10101(a)(2)(B). Normal principles of statutory interpretation, as explained by the Supreme Court, require courts to “presume that the legislature says in a statute what it means . . . [t]hus, our inquiry begins with the statutory text, and ends there as well if the text is unambiguous.” BedRoc, 541 U.S. at 183. “We give the words of a statute their ‘ordinary, contemporary, common meaning,’” absent an indication to the contrary from Congress. See Williams, 529 U.S. at 431 (citation omitted). 56 MI FAMILIA VOTA V. PETERSEN

Arizona cannot deny an individual the right to vote because of an “error or omission [that] is not material in determining” an applicant’s eligibility to vote. See 52 U.S.C. § 10101(a)(2)(B). The Materiality Provision requires invalidation of any voting prerequisite that does not convey “material” information that has a probability of affecting an election official’s eligibility determination. See Williams, 529 U.S. at 431; see also Material, Black’s Law Dictionary (12th ed. 2024); Material, Oxford English Dictionary (2d ed. 1989). 5 The erroneous or omitted information need not be absolutely essential to determine if a person is eligible to vote, but it must have probable impact on eligibility to vote. a. The checkbox requirement In light of our holding on the meaning of “material,” the state form’s checkbox requirement violates the Materiality Provision because confirming citizenship via the checkbox “is not material in determining” an applicant’s eligibility to vote when they have already provided DPOC. See 52 U.S.C. § 10101(a)(2)(B). DPOC is sufficient to show citizenship— a requirement to vote in Arizona—so the state form’s checkbox requirement has no probable impact in determining applicant’s eligibility to vote when DPOC has been provided. See Ariz. Const. art. VII, § 2; Ariz. Rev. Stat. §§ 16-121.01(A), 16-121.01(C).

5 Black’s Law Dictionary defines material as “having some logical connection with the consequential facts” or “[o]f such a nature that knowledge of the item would affect a person’s decision-making.” Material, Black’s Law Dictionary (12th ed. 2024). The Oxford English Dictionary defines material as “of such significance as to be likely to influence the determination of a cause.” Material, Oxford English Dictionary (2d ed. 1989). MI FAMILIA VOTA V. PETERSEN 57

Our holding is consistent with the purpose of the Materiality Provision. The Materiality Provision was “intended to address the practice of requiring unnecessary information for voter registration with the intent that such requirements would increase the number of errors or omissions on the application forms, thus providing an excuse to disqualify potential voters.” Schwier v. Cox, 340 F.3d 1284, 1294 (11th Cir. 2003). In League of Women Voters of Arkansas v. Thurston, the district court held that a voting law violated the Materiality Provision because it required absentee voters to provide information about their eligibility to vote “several times,” and voters had their ballots “rejected on the basis of a mismatch or omission in one of the multiple documents they ha[d] provided” even when they “correctly provided th[e] information at least once.” No. 5:20-cv- 05174, 2021 WL 5312640, at *4 (W.D. Ark. Nov. 15, 2021). The checkbox requirement similarly creates the danger that Arizona may reject a state-form application based on a “mismatch” between documents, such as an incomplete checkbox on a state form, notwithstanding that a voter registration applicant had already given DPOC. See Thurston, 2021 WL 5312640, at *4; Ariz. Rev. Stat. §§ 16- 121.01(A), 16-121.01(C). By requiring voters to provide information about their citizenship status “several times,” Arizona “increase[s] the number of errors or omissions” on the application forms “and provide an excuse to disenfranchise otherwise qualified voters.” See Thurston, 2021 WL 5312640, at *4; Schwier, 340 F.3d at 1294. The checkbox requirement contradicts the purpose of and violates the Materiality Provision. We hold that H.B. 2492’s checkbox requirement relating to Arizona’s state form violates the Materiality Provision of 58 MI FAMILIA VOTA V. PETERSEN

the Civil Rights Act when enforced on a person who has provided DPOC and is otherwise eligible to vote in Arizona. b. The birthplace requirement Given our holding on the meaning of “material,” the state form’s birthplace requirement also violates the Materiality Provision because disclosing one’s birthplace has no probable impact on and “is not material in determining” an applicant’s eligibility to vote. See 52 U.S.C. § 10101(a)(2)(B). To vote in Arizona, a person must be a United States citizen, a resident of Arizona, at least eighteen years old, and not adjudicated, incapacitated, or convicted of a felony. Ariz. Const. art. VII, § 2. At no place in Arizona law is birthplace location a prerequisite to vote in Arizona. An individual’s birthplace does not directly verify an individual’s citizenship or place of residence. But the State nonetheless asserts without basis that the birthplace requirement can be used to verify an individual’s identity. The district court found that county recorders “do not use birthplace information to determine an applicant’s eligibility to vote, nor do county recorders need birthplace to verify an applicant’s identity.” Although Arizona has collected birthplace information from state-form applicants and included a field in the state form for applicants to include their “state or country of birth” since 1979, Arizona did not require birthplace information for voter registration until 2022 and has determined prior voters qualified to vote despite the absence of birthplace information. That fact strongly indicates that birthplace has no probable impact in determining eligibility to vote. Indeed, an expert at trial, Dr. Eitan Hersh, testified that about one-third of currently registered voters in Arizona had not MI FAMILIA VOTA V. PETERSEN 59

provided birthplace information when they registered to vote. The Voting Laws do not require county recorders to verify an individual’s birthplace or to reject state-form applications with an incorrect birthplace. See Ariz. Rev. Stat. § 121.01(A). Dr. Hersh also testified at trial that about 200,000 voter registrations in Arizona merely list “the United States” as the voter’s birthplace, and county recorders manually enter an applicant’s birthplace (when provided) “exactly as it appears on the state-form,” resulting in non-uniform birthplace information for existing registered voters. Moreover, some birthplace designations are unclear such as “CA,” which could refer to either California or Canada. And many applicants write only their city or county (which can refer to multiple locations) despite the state form’s request that applicants include “state or country of birth.” “If the substance of the [birthplace field] does not matter, then it is hard to understand how . . . this requirement has any use in determining a voter’s qualifications.” Migliori v. Cohen, 36 F.4th 153, 164 (3d Cir. 2022) (holding that omitting the date on a ballot was immaterial because ballots were only to be set aside if the date was missing— not incorrect), vacated on other grounds by Ritter v. Migliori, 143 S. Ct. 297 (2022). We hold that H.B. 2492’s birthplace requirement violates the Materiality Provision of the Civil Rights Act. 2. Different Standards, Practices, and Procedures Provision The DSPP Provision of the Civil Rights Act states “[n]o person acting under color of law shall in determining whether any individual is qualified under State law or laws to vote in any election, apply any standard, practice, or 60 MI FAMILIA VOTA V. PETERSEN

procedure different from the standards, practices, or procedures applied under such law or laws to other individuals within the same county, parish, or similar political subdivision who have been found by State officials to be qualified to vote.” 52 U.S.C. § 10101(a)(2)(A). Case authorities from extra-circuit cases decided by district courts illustrate the type of fact patterns that district courts have said violate the DSPP Provision. For example, in the case of U.S. Student Ass’n Foundation v. Land, the district court held that the DSPP Provision “requires that if Michigan wishes to impose unique procedural requirements on the basis of a registrant’s original voter ID being returned as undeliverable, it must impose those requirements on everyone whose original ID is returned as undeliverable.” 585 F. Supp. 2d 925, 949–50 (E.D. Mich. 2008) (emphasis in original). As another example, in Frazier v. Callicutt, the district court held different standards and procedures existed where the registrar summarily denied and referred the registration of every Black student whose registration listed a previous address outside of the county, potentially indicating lack of residency, to the board of election commissioners, but the registrar approved nearly all non-students whose registrations similarly listed a previous address outside of the county. 383 F. Supp. 15, 18–19 (N.D. Miss. 1974). Also, in Shivelhood v. Davis, the district court held that the Board of Civil Authority, in charge of examining voter applications, “must use its best efforts to insure that any questionnaire [concerning domicile] is equally relevant to all applicants and not designed only to apply to student applicants” to comply with the DSPP Provision. 336 F. Supp. 1111, 1115 (D. Vt. 1971). MI FAMILIA VOTA V. PETERSEN 61

H.B. 2243’s “reason to believe” provision in effect encourages county recorders to apply different standards, practices, and procedures to naturalized citizens than those standards, practices, and procedures they apply to U.S.-born citizens. See Ariz. Rev. Stat. § 16-165(I); 52 U.S.C. § 10101(a)(2)(A). Although a county recorder may in some cases have a reason to think that a person seeking to register to vote is not a citizen, county recorders can only conduct SAVE checks on naturalized citizens and non-citizens because running a citizenship check through SAVE requires an immigration number. See Ariz. Rev. Stat. § 16-165(I). Absent injunction, naturalized citizens would be at risk of county recorders’ subjective decisions to further investigate their citizenship status because of the open-ended “reason to believe” provision, and that provision will not apply to U.S.- born citizens. See id. Because the “reason to believe” provision “determine[s] whether any individual is qualified under State law . . . to vote in any election” and “appl[ies] a[] standard, practice, or procedure” for naturalized citizens “different from the standards, practices, or procedures applied under such law” to U.S.-born citizens, the “reason to believe” provision violates the DSPP Provision. See 52 U.S.C. § 10101(a)(2)(A); Ariz. Rev. Stat. § 16-165(I); U.S. Student, 585 F. Supp. 2d at 949–50; Frazier, 383 F. Supp. at 18–19; Shivelhood, 336 F. Supp. at 1115. It need hardly be added that the “reason to believe” provision invites county recorders to pose a barrier to registration for any disfavored individual. The Republican Appellants contend that the “reason to believe” provision is not discriminatory because a county recorder must run a citizenship check through SAVE on any voter the recorder has “reason to believe” is not a citizen. 62 MI FAMILIA VOTA V. PETERSEN

These citizenship checks will not have utility for U.S.-born citizens because the system cannot yield substantive information without an inputted alien registration number. See Ariz. Rev. Stat. § 16-165(I). Because SAVE contains no information on U.S.-born citizens, however, the district court found that the “reason to believe” provision “solely” impacts naturalized citizens and cannot be used if the subject of the inquiry is a U.S.-born citizen. By requiring the use of SAVE to check citizenship status whenever the county recorder is suspicious about citizenship, rather than a method that could be applied to both naturalized and U.S.-born citizens, Arizona Revised Statute § 16-165(I) limits the “reason to believe” provision to a subset of the electorate: persons with immigration numbers. It is not merely a matter of “utility” then, as the Republican Appellants contend; a query cannot start without an immigration number so county recorders cannot run a citizenship check through SAVE for U.S.-born citizens. For this reason, we conclude that the “reason to believe” provision applies different standards, practices, or procedures to naturalized citizens compared to U.S.-born citizens. As Republican Appellants contend, Arizona can investigate the citizenship status of registered voters to ensure that only qualified individuals are registered to vote. For example, county recorders must check the ADOT, Social Security Administration, and city, town, county, state, and federal databases for all registered voters. See Ariz. Rev. Stat. §§ 16-165(G)–(H), 16-165(K). That does not violate the DSPP Provision. The Supreme Court has alluded that holding otherwise “would raise serious constitutional doubts” regarding the DSPP Provision. See Inter Tribal Council, 570 U.S. at 17. But because the “reason to believe” MI FAMILIA VOTA V. PETERSEN 63

provision subjects only naturalized citizens to database checks, this provision violates the DSPP Provision. We hold that H.B. 2243’s “reason to believe” provision violates the DSPP Provision of the Civil Rights Act. E. Factual Finding Regarding Discriminatory Intent Although the clear error standard for reviewing factual findings is deferential, “it is not a rubber stamp.” Alexander v. S.C. State Conf. of the NAACP, 602 U.S. 1, 18 (2024). We must ensure that the applicable law or standard is properly applied. See Masayesva v. Zah, 65 F.3d 1445, 1453 (9th Cir. 1995), as amended on denial of reh’g and reh’g en banc (Dec. 5, 1995) (“[W]e review the district court’s application of law to facts for clear error where it is ‘strictly factual,’ but de novo where application of law to fact requires ‘consideration of legal principles.’”). The Supreme Court in Village of Arlington Heights v. Metropolitan Housing Development Corp. set out a non- exhaustive list of factors for courts to consider in evaluating whether a law was enacted with discriminatory intent: (1) historical background, (2) the relevant legislative history, (3) the sequence of events leading up to the enactment, including departures from the normal legislative process, and (4) whether the law has a disparate impact on a specific racial group. 429 U.S. 252, 266–68 (1977). Under Arlington Heights, a plaintiff must “‘simply produce direct or circumstantial evidence demonstrating that a discriminatory reason more likely than not motivated’ the defendant and that the defendant’s actions adversely affected the plaintiff in some way.” Pac. Shores Props., LLC v. City of Newport Beach, 730 F.3d 1142, 1158 (9th Cir. 2013) (quoting McGinest v. GTE Serv. Corp., 360 F.3d 1103, 1122 (9th Cir. 2004)). “A plaintiff does not have to prove that the 64 MI FAMILIA VOTA V. PETERSEN

discriminatory purpose was the sole purpose of the challenged action, but only that it was a ‘motivating factor.’” Arce v. Douglas, 793 F.3d 968, 977 (9th Cir. 2015) (quoting Arlington Heights, 429 U.S. at 266). “Necessarily, an invidious discriminatory purpose may often be inferred from the totality of the relevant facts,” Washington v. Davis, 426 U.S. 229, 242 (1976), in large part because “discriminatory intent is rarely susceptible to direct proof,” Mhany Mgmt., Inc. v. Cnty. of Nassau, 819 F.3d 581, 606 (2d Cir. 2016). The Supreme Court’s decision in Desert Palace, Inc. v. Costa supports the principle that a plaintiff may rely successfully on either circumstantial or direct evidence to demonstrate that a law was enacted with discriminatory intent. See 539 U.S. 90 (2003); see also Cornwell v. Electra Cent. Credit Union, 439 F.3d 1018, 1029–30 (9th Cir. 2006) (recognizing because of Costa that plaintiffs may rely on circumstantial evidence in the Title VII context). In Costa, the Supreme Court explained that “[t]he reason for treating circumstantial and direct evidence alike is both clear and deep rooted: ‘Circumstantial evidence is not only sufficient, but may also be more certain, satisfying and persuasive than direct evidence.’” 539 U.S. at 100 (quoting Rogers v. Mo. Pac. R.R. Co., 352 U.S. 500, 508 n.17 (1957)). Here, the district court applied a heightened version of the Arlington Heights analysis to the facts—insisting that Plaintiff-Appellees directly link the motive of the Legislature to every piece of evidence offered under each prong of the Arlington Heights framework. Because the district court’s reasoning imposed a higher evidentiary standard than that required by the Arlington Heights test MI FAMILIA VOTA V. PETERSEN 65

analyzing the “totality of circumstances,” the district court clearly erred. We address each Arlington Heights prong: 1. Historical background First, the district court acknowledged that “Arizona does have a long history of discriminating against people of color” and gave examples of the state’s past discrimination. But the district court then failed to meaningfully address the significance of that history in its analysis of whether Arizona acted with discriminatory intent in enacting the Voting Laws. Rather, the district court dismissed Arizona’s history as too old to be determinative, and insisted that Plaintiff- Appellees show “a nexus between Arizona’s history of animosity toward marginalized communities and the Legislature’s enactment of the voting laws.” The district court’s “nexus” requirement could not be satisfied, absent an unambiguous admission from the Legislature that the purpose of the Voting Laws was to perpetuate Arizona’s “well-documented history of voting discrimination.” That of course was not likely ever to happen. Such evidence is rare because legislators “seldom, if ever, announce on the record that they are pursuing a particular course of action because of their desire to discriminate against a racial minority.” Arce, 793 F.3d at 978 (quoting Smith v. Town of Clarkton, 682 F.2d 1055, 1064 (4th Cir. 1982)). In Cornwell, we recognized in the Title VII context that “[a]lthough some plaintiffs might discover direct evidence that a defendant’s nondiscriminatory justification is pretext, most will not.” 439 F.3d at 1029. Consequently, plaintiffs may rely on circumstantial evidence. Id. While the context here is different, the reasoning in Cornwell applies with equal force because direct evidence of legislators’ 66 MI FAMILIA VOTA V. PETERSEN

discriminatory purpose is similarly rare, and consequently most plaintiffs will not be able to show direct evidence of a discriminatory legislative purpose. See Arce, 793 F.3d at 978. In light of the Supreme Court’s recognition in Costa that circumstantial evidence may be “more certain, satisfying and persuasive than direct evidence,” the district court should not have required plaintiffs to produce direct evidence of discriminatory purpose. See 539 U.S. at 100. In creating its onerous “nexus” requirement, the district court misapplied the Arlington Heights framework by requiring Plaintiff-Appellees to provide direct evidence of racial animus for every prong of the test, rather than applying a totality of the circumstances analysis that also took into account circumstantial evidence. If the district court had viewed the evidence in its totality, a different conclusion may have been reached. A historical pattern of discriminatory behavior from a legislative body, particularly as it pertains to voting laws, gives context as to whether the same legislative body has acted with discriminatory purpose in enacting new voting laws. The district court erred in its analysis of the first prong of the Arlington Heights framework. 2. Legislative history Second, the district court found that “[n]othing in the legislative hearings [on the Voting Laws] evince a motive to discriminate against voters based on race or national origin,” and concluded that the legislators were instead motivated by a desire to control the increase in federal-only voters in Arizona who had not provided DPOC. The district court did not properly analyze the evidence in its totality, however, as required by the Arlington Heights test. See United States v. Carrillo-Lopez, 68 F.4th 1133, 1140 (9th Cir. 2023), cert. MI FAMILIA VOTA V. PETERSEN 67

denied, 144 S. Ct. 703 (2024) (“Courts must consider the totality of the evidence presented by the plaintiff” when conducting an Arlington Heights analysis). The political climate in Arizona leading to enactment of the Voting Laws provides circumstantial evidence of discriminatory intent. After the November 2020 presidential election, there were claims that non-citizens had illegally cast more than 36,000 votes in the election. The Arizona Senate then established a committee to audit the 2020 election results. The audit did not reveal any evidence of voter fraud, yet the Legislature proceeded to enact legislation aimed at remedying the voter fraud issue that was contradicted by its own findings. 6 When considering both the charged political climate and the events leading to the passage of the Voting Laws, see infra, the Legislature’s insistence on pressing forward with the Voting Laws despite its own audit revealing no voter fraud is circumstantial evidence “demonstrating that a discriminatory reason more likely than not motivated” the Legislature in enacting the Voting Laws. 7 Pac. Shores, 730 F.3d at 1158 (quoting McGinest, 360 F.3d at 1122).

6 A state has a legitimate interest in “preserving the integrity of its election process,” regardless whether there is actual evidence of fraud. See Eu v. San Francisco Cnty. Democratic Cent. Comm., 489 U.S. 214, 231 (1989). But the absence of evidence of voter fraud can still be considered when assessing the motivations of the Legislature as is specifically required by the holistic Arlington Heights standard. 7 This conclusion is bolstered by the evidence demonstrating that the claim there was illegal voting by non-citizens was repeated on many occasions throughout the legislative process, even though the Legislature’s own audit contradicted his claim. For example, Senate President Petersen repeated the illegal-voter accusation when discussing 68 MI FAMILIA VOTA V. PETERSEN

Despite the Legislature’s failed audit and the charged political climate leading to the passage of the Voting Laws, the district court did not infer that there was discriminatory intent, instead concluding that the Plaintiff-Appellees failed to “adduce evidence challenging the sincerity” of the Legislature’s belief that non-citizens were voting in Arizona elections. But in addressing an issue of voter suppression, we are not bound by questions of sincerity of legislators, but rather must look to what was actually done, and the purported reasons for and the effects of legislative action, which cannot be determined by legislative say-so but requires a demonstration through a presentation of facts. The Legislature’s failure to show evidence of voter fraud in its audit calls into question the sincerity of its belief in the existence of voter fraud. But more importantly, this “sincerity” requirement imposed by the district court exists nowhere in the Arlington Heights framework established by the United States Supreme Court. Rather, Arlington Heights asks that courts make a “sensitive inquiry into [] circumstantial and direct evidence” of discriminatory intent, because “discriminatory intent is rarely susceptible to direct proof.” Mhany Mgmt., 819 F.3d at 606. By requiring direct evidence that the Legislature was not acting out of sincerely held beliefs, the district court misapplied Arlington Heights. Next, the Free Enterprise Club played a vital role in enacting the Voting Laws. As the district court acknowledged, the “Free Enterprise Club helped author the Voting Laws.” And in his deposition, Senate President

the Voting Laws in an Arizona Senate Judiciary Committee meeting on March 10, 2022. And Greg Blackie of the Free Enterprise Club also repeated the claim that there was illegal voting by non-citizens in an email to Republican members of the Arizona Senate Judiciary Committee. MI FAMILIA VOTA V. PETERSEN 69

Petersen said that the Free Enterprise Club drafted “most of [the Voting Laws.]” 8 But in its findings, the district court excluded evidence demonstrating how deep the Free Enterprise Club’s involvement ran. For example, House Speaker Toma, referring to the Free Enterprise Club, called H.B. 2243 “their” bill. And Greg Blackie of the Free Enterprise Club testified to the details of the bill as the Senate Government Committee’s expert witness on March 14, 2022. Also, the bill’s sponsor, state Representative Jacob Hoffman, deferred to Blackie when asked questions about the bill in a committee hearing. Representative Hoffman emphasized the role of the Free Enterprise Club, telling the same committee that he had been “working with the Free Enterprise Club on this bill, and they’ve spent hundreds of hours digging into this.” The Free Enterprise Club, in its advocacy for the Voting Laws, sent lobbying materials to Arizona legislators with the heading “how more illegals started voting in AZ.” “[T]he use of ‘code words’ may demonstrate discriminatory intent,” Ave. 6E Invs., LLC v. City of Yuma, 818 F.3d 493, 505 (9th Cir. 2016) (citation omitted), and the term “illegals” can evidence racial animus for members of the Latino community in Arizona. This suggests that the Free Enterprise Club—an architect and advocate of the Voting Laws—was motivated by a discriminatory purpose in drafting and advocating for the Voting Laws, which, in turn, supports a conclusion that the Voting Laws were the product of intentional discrimination. See Ave. 6E Ins., 818 F.3d at 504 (“The presence of community animus can support a

8 In its amicus brief in this case, the Free Enterprise Club also claims that it was “instrumental in the drafting and adoption of the statutes at issue in this case.” 70 MI FAMILIA VOTA V. PETERSEN

finding of discriminatory motives by government officials, even if the officials do not personally hold such views.”). The Free Enterprise Club’s involvement sets this case apart from Brnovich v. DNC. In Brnovich, the Supreme Court reversed our decision and held that the district court did not clearly err in finding that a different Arizona voting law was not enacted with discriminatory intent. See 594 U.S. at 687–88. There, the main evidence of discriminatory animus in the legislative process was a former senator’s “unfounded and far-fetched allegations of ballot collection fraud” and a “‘racially-tinged’ video created by a private party,” both of which led to what the district court concluded was “a serious legislative debate on the wisdom of early mail-in voting.” Id. at 688. Here, in sharp contrast, discriminatory animus permeated each and every step of the legislative process because the Free Enterprise Club was involved with the Voting Laws’ enactment from start to finish, from conception to passage. Although we may accept the district court’s conclusion that some members of the Legislature may have been sincerely motivated by a desire to control the increase in federal-only voters for a non- discriminatory purpose, the sincerity of some legislators’ actions does not change the totality of the circumstances— starting with assertions that non-citizens had voted in the 2020 election and continuing with discriminatory animus of the Free Enterprise Club in drafting and lobbying for the Voting Laws. We conclude that the totality of the circumstances suggests the Voting Laws were the product of intentional discrimination. The district court did not view the evidence in its totality, instead concluding that “Plaintiff[-Appellees] presented no persuasive evidence that the Legislature relied on the Free Enterprise Club’s coded appeals, nor that the Legislature MI FAMILIA VOTA V. PETERSEN 71

enacted the Voting Laws to prevent anyone other than non- citizens from voting,” and that “[t]he legislative record lacks any indicia of a nefarious motive.” We conclude that these conclusions are not supported by the record, as we view it. And the district court imposed a higher evidentiary burden than is mandated by the Supreme Court’s precedent in Arlington Heights, which expressly permits “circumstantial evidence demonstrating that a discriminatory reason more likely than not motivated the defendant.” Pac. Shores, 730 F.3d at 1158 (internal quotation marks and citation omitted). Plaintiff-Appellees did not need to provide direct evidence showing that every member of the Legislature relied upon the Free Enterprise Club’s coded discriminatory appeal. But the district court should have done what Arlington Heights requires and should have evaluated the political climate leading to the Voting Laws and the Free Enterprise Club’s involvement within their context—a context that in the totality of the circumstances supports an inference of discriminatory intent. See Davis, 426 U.S. at 242. 3. Departures from the normal legislative process Third, there were departures from ordinary procedure throughout the legislative process. Such departures “might afford evidence that improper purposes are playing a role.” Arlington Heights, 429 U.S. at 267. Consider H.B. 2243’s frenzied passage on the final day of the 2022 legislative session. After the initial version of H.B. 2243 was vetoed by former-Governor Ducey, an amended version of the bill was distributed to the legislators only minutes before it was to be debated and brought to a final vote, giving the legislators little time to review the substantial amendment. In his deposition, House Speaker Toma admitted that he could not 72 MI FAMILIA VOTA V. PETERSEN

recall another time when a vetoed voting bill was pushed through to passage this way. And testimony revealed that amendments that “change everything that was in a prior version of a bill” in the final stages of the legislative process, as the amendment did here, are not a common occurrence. Despite these departures from the usual legislative procedure, the district court found that “[t]he speed with which the Legislature passed H.B. 2243 as amended was not so abrupt as to infer an improper motive, considering the Legislature had previously passed H.B. 2617 through the ordinary legislative process.” But this is not probative because the amended bill contained many substantive changes from its previous version that even supportive legislators had not previously considered. 9 The abrupt passage of this bill occurred in the final moments of the legislative session. The district court should have viewed those departures from typical legislative procedure in the context of the totality of the circumstances when determining whether an improper motive should be inferred. If it had done so, the district court may have drawn a different conclusion. These departures from ordinary legislative procedure, considered with the evidence supporting the other Arlington Heights factors, could indicate discriminatory intent.

9 For example, House Speaker Toma himself was not aware of many changes made by the bill. He was not aware that the notice period to cure for those suspected to be not citizens had been reduced from 90 days to 35 days. He learned about this change for the first time when he was deposed on November 28, 2023. MI FAMILIA VOTA V. PETERSEN 73

4. Impact on a minority group Finally, we focus on one troubling aspect of the district court’s decision: its finding that “Plaintiff[-Appellees] did not show the Arizona Legislature enacted the Voting Laws because of any impact on minority voters or naturalized citizens.” In so finding, the district court said that “[e]vidence of a law’s disparate impact is generally insufficient alone to evidence a legislature’s discriminatory motive.” But Plaintiff-Appellees did not ask the district court to view evidence of the Voting Laws’ disparate impact alone, nor contend that disparate impact should be dispositive. The district court’s narrow view of the evidence was clear error. The district court, by requiring direct evidence of legislators’ motive on this prong, imposed a stricter test than held by Arlington Heights, which required district courts to consider evidence of disproportionate impact along with other direct and circumstantial evidence offered for each of the Arlington Heights prongs. The district court clearly erred by viewing each piece of evidence in isolation and expecting Plaintiff-Appellees to proffer direct evidence of animus for each prong of the Arlington Heights framework, rather than examining the circumstantial evidence as part of a larger totality of the circumstances analysis. See Carrillo-Lopez, 68 F.4th at 1140. The contentious political climate arising from claims of illegal voting may seem innocuous standing alone. So might the Free Enterprise Club’s use of the term “illegals” in lobbying materials, if standing alone. So might H.B. 2243’s hasty passage departing from legislative norms, if standing alone. But viewed in context these discrete pieces of evidence take on a different meaning and support an 74 MI FAMILIA VOTA V. PETERSEN

inference of discriminatory intent. Factfinders considering whether a law was passed with discriminatory intent must analyze the totality of the circumstances. See Davis, 426 U.S. at 242. Because the district court erred by misapplying Arlington Heights and did not show that it was viewing the evidence in context, we vacate and remand the issue of whether H.B. 2243 was enacted with discriminatory intent, with instructions for the district court to apply the proper totality of the circumstances analysis that is required by the Supreme Court’s precedent of Arlington Heights. F. Equal Protection Clause The Fourteenth Amendment provides that “[n]o State shall . . . deny to any person within its jurisdiction the equal protection of the laws.” U.S. Const. amend. XIV, § 1. The Fifteenth Amendment provides that “[t]he right of citizens of the United States to vote shall not be denied or abridged by the United States or by any State on account of race, color, or previous condition of servitude.” U.S. Const. amend. XV, § 1. Bush v. Gore relied on the principle that in the voting context, “arbitrary and disparate treatment” that does not meet “the rudimentary requirements of equal treatment and fundamental fairness” will not survive constitutional scrutiny under the Equal Protection Clause. 531 U.S. 98, 104–05, 109 (2000) (per curiam). Bush v. Gore held that the Equal Protection Clause has a “minimum requirement for nonarbitrary treatment of voters.” Id. at 105; see also Election Integrity Project Cal., Inc. v. Weber, 113 F.4th 1072, 1089 (9th Cir. 2024). MI FAMILIA VOTA V. PETERSEN 75

Bush v. Gore famously stated that its “consideration [wa]s limited to the present circumstances.” 531 U.S. at 109. That statement was not believed by many commentators. 10 What the Supreme Court says in its decisions normally affects future cases raising the same issues. 11 And in most cases in which we have applied the “arbitrary and disparate treatment” standard, we have like Bush v. Gore focused on the one-person, one-vote principle that was first laid down in Reynolds v. Sims. 377 U.S. 533 (1964); see, e.g., Idaho Coal. United for Bears v. Cenarrusa, 342 F.3d 1073, 1076– 77, 1077 n.7 (9th Cir. 2003); Sw. Voter Registration Educ. Project v. Shelley, 344 F.3d 882, 894–95 (9th Cir. 2003), rev’d on other grounds en banc, 344 F.3d 914 (9th Cir. 2003). “The general principle that Bush applied—that ‘the rudimentary requirements of equal treatment and fundamental fairness’ prohibits states from engaging in wholly ‘arbitrary and disparate treatment’ of members of the

10 See, e.g., Laurence H. Tribe, Bush v. Gore and Its Disguises: Freeing Bush v. Gore from its Hall of Mirrors, 115 HARV. L. REV. 170, 271 (November 2001) (“Many see the Court’s attempt to limit the case to whatever ‘the present circumstances’ might be as profoundly illegitimate. These critics argue that the Court was in essence trying to free itself from the discipline of stare decisis, which forces a court either to eat its own words in future cases or else give good reasons for spitting them out.”). 11 See id. (“Indeed, whenever an Article III court renders a decision, these commentators argue, that decision must have precedential effect.”); Planned Parenthood v. Casey, 505 U.S. 833, 866 (1992) (“[T]he Court’s legitimacy depends on making legally principled decisions under circumstances in which their principled character is sufficiently plausible to be accepted by the Nation.”); Frederick Schauer, Precedent, 39 STAN. L. REV. 571, 589 (1987) (“[T]he conscientious decisionmaker must recognize that future conscientious decisionmakers will treat her decision as precedent, a realization that will constrain the range of possible decisions about the case at hand.”). 76 MI FAMILIA VOTA V. PETERSEN

public—is not unique to that case,” and we should not hesitate to apply it when relevant. See Election Integrity, 113 F.4th at 1090 n.15 (citing 531 U.S. at 107, 109). We apply Bush v. Gore, because despite its disclaimer, it is relevant precedent. Here, the requirements of DPOC and DPOR do not match the “varying” and complete lack of specific standards which violate Equal Protection under the “arbitrary and disparate treatment” standard. See Bush v. Gore, 531 U.S. at 106–07. In Bush v. Gore, the Florida Supreme Court had directed election officials to discern the intent of voters whose “punchcard” ballots were not registering perforation, but the attempted recount resulted in disparate treatment among similarly situated voters because there were no standards by which to determine voter “intent.” Id. at 105–06. Each of the counties involved had used “varying standards” to determine what was a legal vote, and the Supreme Court held that “[t]he problem inheres in the absence of specific standards to ensure its equal application.” Id. at 106–07. In contrast, we held in Election Integrity that California’s vote counting rules satisfied the minimum requirement for nonarbitrary treatment of voters because California’s voting rules were “more than sufficiently detailed and uniform” than “the standardless vote counting order considered in Bush” and California’s “vote counting standard applies uniformly to the counting of all ballots and votes regardless of the vote tabulation method used.” 113 F.4th at 1095 (internal quotation marks and citation omitted). Here, the requirements of DPOC and DPOR apply uniformly, and consequently do not violate Equal Protection under the “arbitrary and disparate treatment” standard. Unlike Bush v. Gore, in which each of the Florida counties MI FAMILIA VOTA V. PETERSEN 77

involved in the votes to be tabulated had used “varying standards” to determine what was a legal vote, here the requirements of DPOC and DPOR are “more than sufficiently detailed and uniform.” See 531 U.S. at 107; Election Integrity, 113 F.4th at 1095. Arizona Revised Statute § 16-121.01(C) requires county recorders to reject state-form applications without DPOC and Arizona Revised Statute § 16-123 requires state-form applicants to provide DPOR. A failure to provide either will result in rejection of the state-form application to vote, and this standard applies to all applicants using the state-form application. The district court also found that there was no evidence that county recorders will act arbitrarily when confirming an individual’s citizenship status. That county recorders will not act arbitrarily is reinforced by the permanent injunction prohibiting enforcement of Arizona Revised Statute § 16- 165(I)’s “reason to believe” provision. The periodic cancellation of registrations, relevant here because Arizona Revised Statutes §§ 16-165(I)–(J) specify citizenship checks against SAVE and NAPHSIS for “persons who are registered to vote without satisfactory [DPOC],” is a systematic removal program with cancellation of batches of registered voters based on the set procedure of routine comparison to certain databases. See supra pp. 47– 50. Unlike the “absence of specific standards to ensure its equal application” in Bush v. Gore, here the standards are specific, clearly defined, and based on an established procedure. See 531 U.S. at 106. Because the DPOC and DPOR requirements and the procedures implementing these requirements are uniform, they are consistent with the minimum requirement for nonarbitrary treatment of voters set forth in Bush v. Gore and they do not violate the Equal Protection Clause. We conclude that there have been 78 MI FAMILIA VOTA V. PETERSEN

statutory violations under the NVRA and the Civil Rights Act, but no constitutional violations under the Equal Protection Clause. We hold that H.B. 2492’s requirements of DPOC and DPOR for state-form applicants do not violate the Equal Protection Clause. G. Legislative Privilege The district court held that the Legislative Parties had waived legislative privilege. We need not decide that issue for the reasons that follow. The doctrine of legislative immunity protects state legislators “from criminal, civil, or evidentiary process that interferes with their ‘legitimate legislative activity.’” Puente Ariz. v. Arpaio, 314 F.R.D. 664, 669 (D. Ariz. 2016) (quoting Tenney v. Brandhove, 341 U.S. 367, 376 (1951)). Legislative privilege is a corollary to legislative immunity and is a qualified privilege that generally shields legislators from compulsory evidentiary process. Lee v. City of Los Angeles, 908 F.3d 1175, 1187–88 (9th Cir. 2018). The Legislative Parties here complied with the discovery order that they contend violated their legislative privilege. Because “[c]ompliance with a discovery order renders moot an appeal of that order,” this issue of whether legislative privilege was waived is moot. See Richmark Corp. v. Timber Falling Consultants, 959 F.2d 1468, 1479 (9th Cir. 1992). 12

12 Although the Supreme Court has held that compliance with administrative summons and subpoenas does not moot challenges to those requests, that holding is inapposite here. See Church of MI FAMILIA VOTA V. PETERSEN 79

IV. CONCLUSION We hold that Republican Appellants and Promise Cross- Appellants have standing to pursue their appeals. We AFFIRM the district court’s rulings regarding the NVRA claims, the LULAC Consent Decree, the Civil Rights Act claims, and the Equal Protection claim. We VACATE the district court’s factual finding that H.B. 2243 was not enacted with intent to discriminate, and we REMAND for further proceedings consistent with this opinion based on the record that the district court previously developed in its bench trial. We hold that the Republican Appellant’s appeal regarding the district court’s holding that there was a waiver of legislative privilege is moot.

BUMATAY, Circuit Judge, dissenting:

In the wake of the 2020 election, Arizona enacted two sets of voter-verification laws: House Bill (“H.B.”) 2492 and H.B. 2243. Arizona sought to amend its voting laws to improve verification of those registered to vote in the State. These voter-verification amendments made several changes:

• H.B. 2492 prohibits applicants who have not provided “satisfactory evidence of citizenship” from voting in presidential

Scientology v. United States, 506 U.S. 9, 12–13 (1992). In Church of Scientology, the issue was not moot because the “[t]axpayers have an obvious possessory interest in their records . . . and a court can effectuate relief by ordering the Government to return the records.” Id. at 13. Here, the district court’s discovery order is not an administrative summons or subpoena, and the court cannot order for the Legislative Parties’ depositions to be undone, let alone returned. 80 MI FAMILIA VOTA V. PETERSEN

elections. Ariz. Rev. Stat. § 16- 127(A)(1). • H.B. 2492 prohibits applicants who have not provided “satisfactory evidence of citizenship” from voting by mail. Id. § 16-127(A)(2). • H.B. 2492 requires voter-registration applicants using the state-created voter- registration form to provide “satisfactory evidence of citizenship.” Id. § 16- 121.01(C). • H.B. 2492 requires voter-registration applicants using the state-created form to provide satisfactory proof of residence. Id. §§ 16-121.01(A), 16-123. • H.B. 2243 requires county recorders to periodically check available databases to verify the citizenship of registered voters and cancel registrations of foreign citizens. Id. § 16-165(A)(10), (G), (H), (J), (K). • H.B. 2492 requires applicants using the state voter-registration form to provide their birthplace and check a “box” confirming U.S. citizenship. Id. § 16- 121.01(A). • H.B. 2243 requires county recorders to verify citizenship in the Systematic Alien Verification for Entitlements (“SAVE”) database maintained by the U.S. MI FAMILIA VOTA V. PETERSEN 81

Citizenship and Immigration Services (“USCIS”) if the county recorder has “reason to believe” a registered voter is not a citizen. Id. § 16-165(I).

Before these voter-verification amendments went into effect, the Democratic National Committee (“DNC”), the Arizona Democratic Party, the Biden Administration’s Department of Justice Civil Rights Division, and various aligned groups (collectively, “Voting Law Opponents”) sought to stop the voter-verification laws in their tracks. They sued alleging violations of the National Voting Rights Act (“NVRA”), the Civil Rights Act of 1964, a consent decree, and the Constitution. In an unprecedented ruling, the district court granted the Voting Law Opponents virtually everything they wanted, except for finding that H.B. 2243 was enacted with discriminatory intent. The district court enjoined enforcement of most of H.B. 2492 and H.B. 2243—just months before the 2024 election. In an emergency appeal, the Republican National Committee (“RNC”) and two Arizona legislators (collectively, “Voting Law Proponents”) sought to lift the injunction on the three proof-of-citizenship requirements. 1 A motions panel of our court granted a partial stay of the injunction—allowing the proof-of-citizenship requirement for the state-voter registration forms—but otherwise declined to upset the injunction. In an extraordinary move, a divided merits panel reconsidered the motions panel order

1 At least the Arizona legislators have standing to bring this appeal. See Mi Familia Vota v. Fontes (“Mi Familia Vota III”), 111 F.4th 976, 994 (9th Cir. 2024) (Bumatay, J., dissenting). 82 MI FAMILIA VOTA V. PETERSEN

and vacated the partial stay a mere two weeks later. The Supreme Court quickly reversed the merits-panel majority and allowed the proof-of-citizenship requirement to be enforced. Now, the majority tries again. This time, ignoring the Supreme Court’s direction on at least the state voter-form issue, it again affirms the injunction wholesale. But even more, the majority thinks that the district court didn’t go far enough in overturning Arizona’s voter-verification laws. While following the district court’s legal rulings on the NVRA, Civil Rights Act, and the consent decree, the majority reverses the district court’s factual findings and all but declares H.B. 2243 the product of discrimination. Unprecedented yet again. When courts are forced to enter the political realm—as challenges to voting laws require—we must be our most deliberate, careful, and thoughtful. Our robes are not blue or red but black. Sweeping rulings setting aside a State’s laws don’t help. While some parts of H.B. 2492 and H.B. 2243 may violate federal law, in no way must they be completely invalidated. Most of the voter-verification laws are consistent with the Constitution and federal law, and we should have vacated and substantially narrowed the injunction. I respectfully dissent. I. Proof of Citizenship to Vote in Presidential Elections H.B. 2492 prohibits registered voters who do not provide “satisfactory evidence of citizenship” from voting in presidential elections. Ariz. Rev. Stat. § 16-127(A)(1). The district court ruled that Section 6 of the NVRA preempts this MI FAMILIA VOTA V. PETERSEN 83

provision. Under that section of the NVRA, States “shall accept and use” federally created voter-registration forms “for the registration of voters in elections for Federal office.” 52 U.S.C. § 20505(a)(1). The district court interpreted this NVRA provision to require States to allow any individual who submits the federal form to vote in presidential elections—regardless of proof of citizenship—and enjoined the Arizona law. But because the Constitution doesn’t grant Congress the power to regulate who may vote in presidential elections, we should have reversed this ruling. A. The NVRA gives citizens who want to vote in federal elections two options for registration. First, citizens may register to vote through a federal voter-registration form issued by the Election Assistance Commission. 52 U.S.C. § 20505(a). Second, citizens may also register through state voter-registration forms—forms designed by each State for that State’s elections. Id. The NVRA mandates that “[e]ach State . . . accept and use” the federal voter-registration form “for the registration of voters in elections for Federal office.” Id. § 20505(a)(1). The NVRA defines “Federal office” to include the “office of President or Vice President.” Id. §§ 20502(2), 30101(3). Congress derived its authority to enact the NVRA from the Elections Clause of the Constitution. See Arizona v. Inter Tribal Council of Arizona, Inc. (“ITCA”), 570 U.S. 1, 8–9 (2013): see also id. at 40 (Alito, J., dissenting) (“[T]he NVRA was the first significant federal regulation of voter registration enacted under the Elections Clause since Reconstruction[.]”). But, as a matter of constitutional text, the Elections Clause doesn’t govern presidential elections. The Elections Clause of Article I provides that “[t]he Times, Places and 84 MI FAMILIA VOTA V. PETERSEN

Manner of holding Elections for Senators and Representatives, shall be prescribed in each State by the Legislature thereof; but the Congress may at any time by Law make or alter such Regulations, except as to the Places of chusing Senators.” U.S. Const. art. I, § 4, cl. 1 (emphasis added). Under that Clause, States have the “duty” to set the time, place, and manner of holding congressional elections, but Congress has the power to “alter” those regulations or “supplant them altogether.” See ITCA, 570 U.S. at 8. The Court has held that the “Times, Places, and Manner” of holding elections “embrace authority to provide a complete code for congressional elections,” including regulation of voter registration. Id. at 8–9. 2 But the Clause is expressly limited to “Elections for Senators and Representatives.” Thus, while the Elections Clause may give Congress power over registration in congressional elections, it doesn’t extend that authority over presidential elections.

2 As a matter of original understanding, this conclusion may not provide the full picture. Both the Voter Qualifications Clause and the Seventeenth Amendment direct that States set the “qualifications” for electors for the House of Representatives and Senate. U.S. Const. art. I, § 2, cl. 1 (“the Electors in each State shall have the Qualifications requisite for Electors of the most numerous Branch of the State Legislature”); id. amend. XVII (“The electors in each State shall have the qualifications requisite for electors of the most numerous branch of the State legislatures”). “Taken together, these provisions suggest that the United States Constitution commits wholly to the states decisions about who may vote in federal elections[.]” James A. Gardner, Liberty, Community and the Constitutional Structure of Political Influence: A Reconsideration of the Right to Vote, 145 U. Pa. L. Rev. 893, 964 (1997); see also ITCA, 570 U.S. at 26 (Thomas, J., dissenting) (“Congress has no role in setting voter qualifications, or determining whether they are satisfied[.]”). Even so, as an inferior court, we are bound by ITCA’s holding. MI FAMILIA VOTA V. PETERSEN 85

Other Clauses of Article II cover presidential elections. First, the Electors Clause lays out much of the groundwork—granting nearly all authority to the States. It provides that “[e]ach State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors[.]” U.S. Const. art. II, § 1, cl. 2. Unlike the grant of a revisory power to Congress in the Elections Clause, the Electors Clause gives the States sole power over the “Manner” of appointing electors to the electoral college. See U.S. Term Limits, Inc. v. Thornton, 514 U.S. 779, 805 (1995) (describing the Electors Clause as the sort of “express delegation[] of power to the States” by the Constitution necessary for them “to act with respect to federal elections”). Second, the Time of Chusing Clause provides a narrow role for Congress in presidential elections. The Time of Chusing Clause says that “Congress may determine the Time of chusing the Electors, and the Day on which they shall give their Votes; which Day shall be the same throughout the United States.” U.S. Const. art. II, § 1, cl. 4. So rather than having any power over the “Manner” of holding congressional elections, Congress merely has authority to choose the date of the presidential election and date of the electoral college vote. “Any shadow of a justification for congressional power with respect to congressional elections therefore disappears utterly in presidential elections.” Oregon v. Mitchell, 400 U.S. 112, 212 (1970) (Harlan, J., concurring in part); see also Nicholas O. Stephanopoulos, The Sweep of the Electoral Power, 36 Const. Comment. 1, 54 (2021) (“As a textual matter, the [Time of Chusing] Clause is plainly narrower than the Elections Clause. It only authorizes Congress to set the time of presidential elections.”). 86 MI FAMILIA VOTA V. PETERSEN

Together, these Clauses form a cohesive structure governing federal elections—States and Congress share authority over congressional elections, but States retain near- exclusive power over presidential elections. Thus, the Constitution forecloses congressional authority to control voter-registration requirements for presidential elections. Under the Electors Clause, that power falls within the province of the States alone. And congressional authority under the Elections Clause can’t be twisted to encompass presidential elections. See ITCA, 570 U.S. at 16 (“[O]ne cannot read the Elections Clause as treating implicitly what . . . other constitutional provisions regulate explicitly.”). Giving Congress a narrow role over presidential elections makes sense for the separation of powers. As Hamilton explained, a central concern at the Founding was that “the Executive should be independent for his continuance in office on all but the people themselves. He might otherwise be tempted to sacrifice his duty to his complaisance for those whose favor was necessary to the duration of his official consequence.” The Federalist No. 68 (Alexander Hamilton). Imagine then a Congress with power to regulate presidential elections—the Executive may fear retaliation from Congress in the form of unfavorable election laws. State ratification debates echoed this concern. As James Wilson put it in Pennsylvania’s debates: “Was the President to be appointed by the legislature? . . . To have the executive officers dependent upon the legislative, would certainly be a violation of that principle, so necessary to preserve the freedom of republics, that the legislative and executive powers should be separate and independent.” See Debates in the Several State Conventions on the Adoption of the Federal Constitution, as Recommended by the General MI FAMILIA VOTA V. PETERSEN 87

Convention at Philadelphia in 1787 511–12 (J. Elliot 2d ed. 1836). In its briefing, the Civil Rights Division waves this all away—claiming that the Necessary and Proper Clause, along with Congress’s more limited electoral duties, instead supports Congress’s broad authority over presidential elections. The Civil Rights Division vaguely lists three clauses as support for this authority. See, e.g., U.S. Const. amend. XII (vesting in Congress powers and duties in connection with the election of the President and Vice President); id. amend. XIV, § 2 (setting forth a process for penalizing States for denial of “the right to vote at any election for the choice of electors for President and Vice- President of the United States” and other federal offices); id. amend. XXIV, § 1 (prohibiting denial or abridgment of the right to vote in any “election for President or Vice President” and other federal offices based on failure to pay a poll tax). The Civil Rights Division cites no authority for its broad view of federal power. And the Necessary and Proper Clause may not serve as a workaround to the Constitution’s express provisions. Regardless of that Clause’s scope, a “federal statute . . . must . . . not be prohibited by the Constitution.” United States v. Comstock, 560 U.S. 126, 135 (2010) (simplified). And the Constitution “could [not] be clearer in stating what Congress can control and what it cannot control” when it comes to presidential elections. ITCA, 570 U.S. at 16 (simplified). Thus, the NVRA can’t preempt state laws governing presidential elections. See id. at 35 n.2 (Thomas, J., dissenting) (While “the NVRA purports to regulate presidential elections,” that is “an area over which the Constitution gives Congress no authority whatsoever.”). 88 MI FAMILIA VOTA V. PETERSEN

B. The opponents of the proof-of-citizenship requirement seemingly acknowledge the States’ role over the “Manner” of appointing electors under the Electors Clause. But, citing McPherson v. Blacker, 146 U.S. 1 (1892), they argue that “Manner” refers only to a narrow right to select the mode of choosing electors—either by popular election, appointment, or some other mechanism. But once a State chooses a mode, they contend that Congress has a free hand to regulate presidential elections as it pleases. Four reasons prove this argument unconvincing. First, their argument would contradict ITCA. If “Manner” in the Electors Clause only means the mode of an election, then Congress too would not have authority to enact voter-registration regulations under the Elections Clause, which also refers to the “Manner of holding elections.” U.S. Const. art. I, § 4, cl. 1. But ITCA directly held that Congress has such power. 570 U.S. at 8–9. Indeed, the phrasing of the Elections Clause is narrower than the Electors Clause. The Elections Clause refers only to the “Manner of holding elections,” compared to the broadly worded Electors Clause allowing States to decide the “Manner” of appointing electors “as the Legislature thereof may direct.” Compare U.S. Const., art. I, § 4, cl.1 (emphasis added) with id., art. II, § 1, cl. 2. It would be inconsistent to read the Electors Clause more narrowly than the Elections Clause. Second, as a matter of common sense, if States may let no one vote for presidential electors (by letting legislatures pick them), then they may decide to let only some vote for electors. In other words, subject to other constitutional constraints like the Fourteenth and Fifteenth Amendments, the power to disenfranchise all its citizens suggests the MI FAMILIA VOTA V. PETERSEN 89

power to franchise only some of its citizens—those meeting certain registration requirements. Indeed, at the Founding, the States had different requirements for voting—for example, some had race, property, religious, or literacy tests. Akhil Reed Amar, The Words That Made Us: America’s Constitutional Conversation, 1760–1840 226 (2021). So it’s wrong to think of choosing “popular election” as an all-or- nothing option. States could choose a “popular election” with varying levels of enfranchisement. Third, McPherson doesn’t support this overly narrow role for States. McPherson determined that Michigan could establish district-level elections for the selection of presidential electors under the Electors Clause. 146 U.S. at 24. The Court remarked that, historically, the Electors Clause meant that States may “appoint [electors] in any mode its legislature saw fit to adopt”—meaning through legislative vote, general popular vote, district-level vote, or other “mode.” Id. at 29. In that case, the Court reasoned that “Manner” of appointment included “mode” of appointment. But McPherson didn’t establish the definitive scope of “Manner” in the Electors Clause or determine that “Manner” only meant the “mode” of choosing. Rather, McPherson reinforced the narrow role the federal government plays in presidential elections compared to the “plenary power” state legislatures enjoy “in the matter of the appointment of electors.” Id. at 35 (emphasis added). While “Congress is empowered to determine the time of choosing the electors and the day,” “otherwise the power and jurisdiction of the state is exclusive.” Id. (emphasis added). Indeed, McPherson confirmed that “[t]he right to vote in the states comes from the states.” Id. at 38. So McPherson teaches us that States have plenary and exclusive power to plan the 90 MI FAMILIA VOTA V. PETERSEN

administration of presidential elections and Congress can’t encroach on that power. Fourth and most importantly, this narrow view of the scope of “Manner” contravenes the original understanding of the Electors Clause. At the Founding, the “Manner” of appointing electors was broad enough to encompass regulating voter-registration requirements. At the time, “Manner” meant “Way; mode”; “Custom; habit; fashion”; or “Form; method.” Samuel Johnson, A Dictionary of the English Language (1773); see also Noah Webster, An American Dictionary of the English Language (1828) (defining “Manner” as “Form; method; way of performing or executing”; “Custom; habitual practice”; and “Way; mode.”). These definitions establish that “Manner” included a broad range of election regulations—not just a choice between popular vote and legislative appointment. See Robert G. Natelson, The Original Scope of the Congressional Power to Regulate Elections, 13 U. Pa. J. Const. L. 1, 20 (2010) (The word “Manner” in the Electors Clause “was an acknowledgment of state power to fix the qualifications (or identity) of the person or persons appointing the presidential electors[.]”). Before the Founding, sources from England and elsewhere used the phrase “manner of election,” and its synonyms, in various ways: “the times, places, and mechanics of voting; legislative districting; provisions for registration lists; the qualifications of electors and elected; . . . and the rules of decisions.” Id. at 20. For instance, rules setting out the “manner of election” in London dealt with the election of candidates from districts, the qualifications of the electorate, the choice of candidate, and methods of certification. Id. at 10 (citing 1 Philip Morant, The History and Antiquities of the County of Essex 98 (London, 1768)). MI FAMILIA VOTA V. PETERSEN 91

Parliamentary legislation governing the “manner of election” to the House of Commons prescribed the creation and maintenance of a list of qualified and disqualified voters, public notice and proclamations, times and places of voting, the duties of supervising officers, viva voce voting, adjudication of disputed elections, and punishment for vote- selling. Id. at 11 (citing, e.g., Determinations of the Honourable House of Commons, Concerning Elections, and All Their Incidents 42–79 (London 1774); 4 John Comyns, A Digest of the Laws of England 330–32, 557 (1780)). The main limit on the use of “manner of election” in these sources was that it did not include the governance of campaigns. Id. at 12. And “Americans ascribed the same general content to the phrase ‘manner of election’ as the English . . . did.” Id. at 12–13. Take a 1721 South Carolina election code that referred to oaths and enrollment of electors, the choice of election managers, and the conduct of voter assemblies, as part of “the Manner and Form of electing Members” to the colonial assembly. Id. at 13 (citing S.C. Stat. 113–15 (1721) (“An Act to ascertain the Manner and Form of electing members . . . in the Commons House of Assembly.”)). Likewise, a 1787 New York statute treated inspection of the poll lists, voters’ receipt of their ballots in the presence of inspectors, the administration of oaths to voters of questionable loyalty, and the qualifications of voters as part of the “Mode” of conducting an election. Id. at 16 (citing An Act for Regulating Elections (Feb. 13, 1787), § VI, reprinted in 2 Laws of the State of New York 27, 29–30 (1789)). And a 1781 Maryland law included the administration of oaths to voters in the “manner” in which special elections were conducted. Id. (citing An Act for Holding Special Elections in Caecil County, 1781 Md. 92 MI FAMILIA VOTA V. PETERSEN

Laws, ch. IX). Similar examples abound. See id. at 12–16 (collecting sources). Thus, without more, the historical understanding of “Manner” in the context of elections included within its meaning voter-registration regulations. 3 Compare too the ratification-era debates over congressional and presidential elections. First, how congressional elections would work under the Elections Clause generated heated debate. Across the country, Federalists had to refute predictions that the federal government would entrench itself by exploiting power over voting qualifications in congressional elections. See ITCA, 27, 31–34 (Thomas, J., dissenting) (collecting sources). “Madison explained that ‘reduc[ing] the different qualifications in the different States to one uniform rule would probably have been as dissatisfactory to some of the States as it would have been difficult to the convention.’” Id. (quoting The Federalist No. 52). Put another way, “setting voter qualifications in the constitution could have jeopardized ratification, because it would have been difficult to convince States to give up their right to set voting qualifications.” Id. (citing Joseph Story, Commentaries on the Constitution of the United States 216, 218–19 (abridged ed. 1833)). Thus, federal government power over who may

3 While “manner of elections” is broad enough to encompass voter- registration regulations, the Constitution may have carved away congressional regulation of voter qualifications in congressional elections through Article I, § 2, cl. 1 and the Seventeenth Amendment. See note 2 above. What’s more, congressional authority under the Elections Clause is narrower than “manner of elections”—it only applies to the “Manner of holding elections.” U.S. Const. art. I, § 4, cl. 1 (emphasis added). This textual difference may further limit congressional power over voter qualifications and registrations. But, once again, ITCA governs this question. See 570 U.S. at 8–9, 17–18. MI FAMILIA VOTA V. PETERSEN 93

vote in congressional elections was a point of serious contention. In contrast, the Electors Clause sparked little concern over federal government interference with presidential elections. Hamilton observed that “[t]he mode of appointment of the Chief Magistrate of the United States is almost the only part of the [Constitution], of any consequence, which has escaped without severe censure, or which has received the slightest mark of approbation from its opponents.” The Federalist No. 68 (Alexander Hamilton); see also The Federalist No. 45 (James Madison) (“Without the intervention of the State legislatures, the President of the United States cannot be elected at all. They must in all cases have a great share in his appointment, and will, perhaps, in most cases, of themselves determine it.”). Thus, the ratification debates suggest that the Founders left regulation of presidential elections (apart from the narrow “Time of chusing”) wholly to the States—otherwise, we would expect the same tension as raised over congressional elections. In sum, “Manner” in the Electors Clause is broad. It sweeps in modern voter-registration requirements. And it leaves States with the exclusive right to regulate voter registration for presidential elections. C. And no controlling precedent alters the States’ exclusive power over presidential elections. Citing Ex parte Yarbrough, 110 U.S. 651 (1884), and Burroughs v. United States, 290 U.S. 534 (1934), the district court claimed that the Court has recognized Congress’s power to regulate presidential elections. But that’s wrong. If anything, these precedents reaffirm the principle that Congress’s role in 94 MI FAMILIA VOTA V. PETERSEN

presidential elections is limited, and that the manner of appointing presidential electors is within the “exclusive” “power and jurisdiction of the state[s].” See McPherson, 146 U.S. at 35. Even in the modern era, the Court has continued to express that “the state legislature’s power to select the manner for appointing [presidential] electors is plenary[.]” Bush v. Gore, 531 U.S. 98, 104 (2000) (per curiam). In Ex parte Yarbrough, several men severely beat a Black citizen to prevent him from voting in a congressional election and were convicted under two federal statutes criminalizing the violent intimidation of citizens attempting to vote in a federal election. 110 U.S. at 657. They sought the writ of habeas corpus on the ground that those statutes exceeded Congress’s constitutional authority. Id. In denying the petition, the Court affirmed the power of Congress to protect all voters in federal elections—it is “the duty of that government to see that [a voter] may exercise this right freely, and to protect him from violence while so doing, or on account of so doing.” Id. at 662. According to the Court, this duty comes “from the necessity of the government itself.” Id. Thus, “its service shall be free from the adverse influence of force and fraud practiced on its agents, and that the votes by which its members of congress and its president are elected shall be the free votes of the electors.” Id. Rather than broadly proclaiming an atextual and expansive role for Congress in presidential elections, Yarbrough simply recognized the federal government’s power to enact laws to secure “election[s] from the influence of violence, of corruption, and of fraud.” Id. at 657. This authority to guard against violence is distinct from the authority to establish voter qualifications or organize voter MI FAMILIA VOTA V. PETERSEN 95

registration. Indeed, Yarbrough itself separated the protection of voters to vote “free from force and fraud” from the power to establish the “qualification of the voter[, which is] determined by the law of the state where he votes.” Id. at 663. In other words, there is a difference between a federal law that operates on third parties involved in presidential elections and a federal law that operates directly on the States to mandate certain rules and requirements for presidential elections. While the Court understood the necessity of federal power over the former, Yarbrough had nothing to say about federal power over the latter. So Yarbrough doesn’t support congressional power to override the States’ exclusive power to establish the “Manner” of presidential elections, including over voter-registration requirements. Nor did Burroughs confer broad power over presidential elections on Congress. That case involved the indictment of a political committee treasurer and chairman for failing to disclose contributions and expenditures in a presidential election. 290 U.S. at 543. The defendants challenged the indictment claiming that Congress lacked authority to enact a campaign finance law for presidential elections under the Electors Clause. Id. at 544. Once again, the Court recognized the difference between regulating third parties involved in presidential elections and regulating the States’ administration of presidential elections. Because the campaign finance law did not cross into the States’ exclusive authority to decide the procedures and requirements for a presidential election, it was constitutional. As the Court said,

Neither in purpose nor in effect does [the law] interfere with the power of a state to appoint 96 MI FAMILIA VOTA V. PETERSEN

electors or the manner in which their appointment shall be made. It deals with political committees organized for the purpose of influencing elections in two or more states, and with branches or subsidiaries of national committees, and excludes from its operation state or local committees. Its operation, therefore, is confined to situations which, if not beyond the power of the state to deal with at all, are beyond its power to deal with adequately. It in no sense invades any exclusive state power.

Id. at 544–45 (emphasis added). The Court thus contrasted authority over the rules and requirements for presidential elections with the power to protect the federal government from “impairment or destruction, whether . . . by force or by corruption.” Id. at 545. While the federal government could legislate against the actions of third parties seeking to impair elections, the Court has never recognized the power to directly legislate the States’ choices in appointing electors. See also Mitchell, 400 U.S. at 291 (Stewart, J., concurring in part) (observing that “the qualifications that voters must have when . . . selecti[ng] electors” is “left to the States” and that Burroughs only acknowledges “Federal Government . . . power to assure that such elections are orderly and free from corruption”). Indeed, the Court never suggested that voter registration is “beyond [a State’s] power to deal with adequately.” Burroughs, 290 U.S. at 544–45. This distinction also flows from the original public meaning of “Manner,” which appears not to extend to the governance of campaigns. See Natelson, Original Scope, at 12. MI FAMILIA VOTA V. PETERSEN 97

So, much like Yarbrough, Burroughs recognized the federal government’s power to regulate third parties who seek to corrupt a federal election—whether by dollars or by fists. While Congress can bar third parties from disrupting federal elections, it cannot establish or regulate the registration process for a presidential election. Thus, the Court’s later characterization of Burroughs in another campaign finance case as recognizing “broad congressional power to legislate in connection with the election[] of the President” is also beside the point. See Buckley v. Valeo, 424 U.S. 1, 13 n.16 (1976) (per curiam). And the Ninth Circuit hasn’t recognized broad federal power over voter registration either. In Voting Rights Coalition v. Wilson, 60 F.3d 1411, 1413 (9th Cir. 1995), California challenged the “motor voter” provisions of the NVRA. While acknowledging Congress’s role over congressional elections under the Elections Clause, California argued that the NVRA provisions interfered with its sovereign authority because they “will have a significant impact on its registration procedures applicable to elections of state and local officials.” Id. at 1415–16. We respected California’s concern for its sovereignty. Id. But, as a facial challenge, we observed that “at this point we cannot determine the extent to which, if at all, these [NVRA] changes impinge on the legitimate retained sovereignty of the states.” Id. at 1416. We directed California to comply with the NVRA but “[w]e fores[aw] the possibility in which the district court will be asked to determine whether a certain implementation of the statute sought by the United States . . . is properly resisted by the state on substantial grounds related to its sovereignty.” Id. We also admonished that “our opinion is not intended to foreclose future judicial review of any [constitutional] issues” and that our opinion 98 MI FAMILIA VOTA V. PETERSEN

spoke “only with respect to an as yet unapplied statute.” Id. at 1413. Thus, Wilson was a limited ruling that had nothing to do with the Electors Clause or presidential elections, and we cautioned against overreading its precedential value. Yet the opponents of the proof-of-citizenship requirement rely on Wilson for a single, throwaway line from the opinion. That line says that “[t]he broad power given to Congress over congressional elections has been extended to presidential elections.” Id. at 1414 (citing Burroughs, 290 U.S. at 545). This single statement, which misreads Burroughs, doesn’t alter the constitutional design. First, as Wilson itself warned, the opinion was not meant to answer complex constitutional questions for the circuit and didn’t “foreclose future judicial review” of these issues. Id. at 1413. Second, while the Ninth Circuit adheres to the “binding dicta” rule, even this odd rule has its limits. “Where a panel confronts an issue germane to the eventual resolution of the case, and resolves it after reasoned consideration in a published opinion, that ruling becomes the law of the circuit, regardless of whether doing so is necessary in some strict logical sense.” United States v. McAdory, 935 F.3d 838, 843 (9th Cir. 2019) (simplified). But “we are not bound by a prior panel’s comments made casually and without analysis, . . . uttered in passing without due consideration of the alternatives, or . . . done as a prelude to another legal issue that commands the panel’s full attention.” Id. (simplified). Thus, Wilson’s unreasoned musing on Burroughs is not binding on our court. Rather than invent a surprising new balance of power between the States and the federal government divorced from constitutional text out of a single line of dicta, we should look to the historical understanding of the Constitution’s meaning. MI FAMILIA VOTA V. PETERSEN 99

Thus, no precedent alters the original public meaning of the Electors Clause and the plenary authority of the States to decide the requirements for voting in presidential elections. D. Finally, the opponents of the proof-of-citizenship requirement also argue that the NVRA is a proper exercise of Congress’s powers under the Fourteenth and Fifteenth Amendments. The district court did not reach this question. See Mi Familia Vota v. Fontes (“Mi Familia Vota I”), 691 F. Supp. 3d 1077, 1090 n.7 (D. Ariz. 2023). Because we are a court of “review, not first view,” I would remand to the district court to consider this question in the first instance. See Roth v. Foris Ventures, LLC, 86 F.4th 832, 838 (9th Cir. 2023). * * * Given all this, we should have reversed the district court’s injunction of § 16-127(A)(1). II. Proof of Citizenship to Vote by Mail in Federal Elections H.B. 2492 prohibits voters registered to vote in only federal elections from voting by mail if they do not provide “satisfactory evidence of citizenship.” Ariz. Rev. Stat. § 16- 127(A)(2) (“A person who has not provided satisfactory evidence of citizenship . . . and who is eligible to vote only for federal offices is not eligible to receive an early ballot by mail.”). The district court likewise ruled that Section 6 of the NVRA preempts this provision. Recall that section of the NVRA commands States to “accept and use” federally created voter registration forms “for the registration of voters 100 MI FAMILIA VOTA V. PETERSEN

in elections for Federal office.” 52 U.S.C. § 20505(a)(1). While the NVRA’s text refers only to “registration” and not to “voting,” the district court read this provision to prevent States from imposing any other requirement on mail-in voting, like proof of citizenship. It interpreted the NVRA’s provision permitting States to “require” first-time voters “to vote in person” to mean that States may not add any other mail-in voting requirements. Mi Familia Vota I, 691 F. Supp. 3d at 1090–91 (citing 52 U.S.C. § 20505(c)(1)). The district court also ruled that NVRA’s “purpose” to “enhance[] participation of eligible citizens as voters” preempted Arizona’s mail-in provision. Id. at 1091–92 (citing 52 U.S.C. § 20501(b)(2)). But because the text of the NVRA doesn’t preempt States’ mail-voting rules, we should have reversed this ruling. As background, the “default” rule is that States hold “responsibility for the mechanics of congressional elections.” Foster v. Love, 522 U.S. 67, 69 (1997). Of course, under the Elections Clause, Congress may override State regulations for congressional elections. Id. Because Congress’s regulations are “paramount” to those of the States, if state and federal law “conflict,” then state law “so far as the conflict extends, ceases to be operative.” Ex parte Siebold, 100 U.S. 371, 384 (1879). To show preemption, a party must point to “a constitutional text or a federal statute t[hat] assert[s]” preemptive force. See P.R. Dep’t of Consumer Affs. v. Isla Petroleum Corp., 485 U.S. 495, 503 (1988). “Invoking some brooding federal interest or appealing to a judicial policy preference should never be enough to win preemption of a state law.” Va. Uranium, Inc. v. Warren, 587 U.S. 761, 767 (2019) (lead opinion of Gorsuch, J.). Thus, we must look to the NVRA’s text to see if a conflict exists. MI FAMILIA VOTA V. PETERSEN 101

A. First, the NVRA’s text does not support preempting Arizona’s mail-voting requirements. The NVRA only mandates that States “accept and use” federal voter- registration forms “for the registration of voters in elections for Federal office[.]” 52 U.S.C. § 20505(a)(1) (emphasis added). As a matter of plain text, this provision about voter registration doesn’t conflict with state-specific rules for voting by mail in federal elections. Here, it’s not impossible for Arizona to both “accept and use” the federal form for registering voters and require proof of citizenship for mail voting. See Whistler Investments, Inc. v. Depository Tr. and Clearing Corp., 539 F.3d 1159, 1164 (9th Cir. 2008) (Preemption occurs only when “a party’s compliance with both federal and state requirements is impossible[.]”). At most, the NVRA may require States to allow eligible federal-form applicants to vote in congressional elections. See ITCA, 570 U.S. at 12 (“[T]he Federal Form guarantees that a simple means of registering to vote in federal elections will be available.”) (emphasis added). But the NVRA doesn’t prescribe the way in which those voters must cast their vote—either in person, by mail, or other method. Once a State has complied with its obligation to register the federal-form applicants to vote, nothing prevents the State from prohibiting registered voters from voting by mail unless they meet certain conditions. In other words, while the NVRA may require that the federal form be “accepted as sufficient” to be eligible to vote in congressional elections, it doesn’t require the federal form to be sufficient for all purposes—like satisfying heightened mail-voting requirements. Id. at 10. Thus, the NVRA doesn’t bar States from imposing added safeguards before allowing voters to cast a ballot outside of traditional in-person voting. 102 MI FAMILIA VOTA V. PETERSEN

Indeed, aside from military or overseas voters, no federal law requires States to allow all its citizens to vote by mail. After all, when it comes to state mail-in voting rules, “[i]t is . . . not the right to vote that is at stake . . . but a claimed right to receive absentee ballots.” McDonald v. Bd. of Election Com’rs of Chicago, 394 U.S. 802, 807 (1969). And States may have different approaches to mail balloting. Cf. id. at 809 (“[A] legislature traditionally has been allowed to take reform one step at a time, addressing itself to the phase of the problem which seems most acute to the legislative mind.”) (simplified). Some States offer broader access to mail ballots than others. See Nat’l Conf. of State Legislatures, Table 2: Excuses to Vote Absentee (Jan. 3, 2024). 4 Some States let all voters vote by mail. Others demand voters clear certain hurdles to vote by mail. Those States demand an excuse, such as absence from the locality, illness, or disability. Id. So many States have required more than what’s required to vote in person. None of the opponents of the proof-of-citizenship requirement argue that the NVRA displaces all these requirements. Instead, the Civil Rights Division conceded at oral argument that the NVRA did no such thing. But how can they draw such an arbitrary distinction? Imagine a State with one of these mandates. The hypothetical law provides that “a person who has not provided satisfactory evidence of a disability is not eligible to receive an early ballot by mail.” But what’s the functional difference between this hypothetical law and Arizona’s statute? Arizona’s statute establishes that “[a] person who has not provided satisfactory evidence of citizenship . . . is not eligible to receive an early ballot by mail.” Ariz. Rev. Stat. § 16-

4 Available at: perma.cc/B4ML-L6KJ. MI FAMILIA VOTA V. PETERSEN 103

127(A)(2). Thus, nothing in the text of the NVRA reflects Congress’s intent to require all federal-form applicants to be allowed to vote by mail—regardless of these individual state mandates. B. That the NVRA expressly permits States to require first- time voters to vote in person doesn’t foreclose States from imposing other qualifications on mail voting. See 52 U.S.C. § 20505(c)(1). The NVRA provides that “a State may by law require a person to vote in person if—(A) the person was registered to vote in a jurisdiction by mail; and (B) the person has not previously voted in that jurisdiction.” Id. The district court took the negative implication of this anti-fraud provision to affirmatively bar States from imposing any other requirements for mail-in voting. The district court surmised, “[h]ad Congress intended to permit states . . . to require in-person voting under additional circumstances[,] . . . it could have said so in the NVRA.” Mi Familia Vota I, 691 F. Supp. 3d at 1091. But this logic makes little sense. First, as discussed above, the “default” position is that States decide the mechanism of elections. See Foster, 522 U.S. at 69. States create election law and state law governs unless it conflicts with federal law. It would be odd for Congress to displace the whole field of mail-in voting rules through such an opaque provision. Reading this narrow provision to establish a new status quo and to preempt a broad swath of state mail-in voting laws would violate the principle that Congress does not “hide elephants in mouseholes.” Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 468 (2001). After all, negative inferences from statutory text only work if it is “fair to suppose that Congress 104 MI FAMILIA VOTA V. PETERSEN

considered the unnamed possibility and meant to say no to it.” Marx v. Gen. Revenue Corp., 568 U.S. 371, 381 (2013) (simplified). Second, Congress enacted this provision as an anti-fraud provision—not a broad preemption clause. As we have held, this provision is one of “numerous fraud protections” in the NVRA. See Gonzalez v. Arizona, 677 F.3d 383, 403 (9th Cir. 2012) (en banc). The “NVRA allows states to require first-time voters who register by mail to vote in person at the polling place, where the voter’s identity can be confirmed.” Id. at 403 n.28. Thus, Congress didn’t work to create a major upheaval in mail-in voting laws and preclude States from adopting other anti-fraud measures through a provision to empower States to weed out voter fraud. And third, this argument proves too much. The district court’s logic would mean that all state limitations on absentee and mail voting would be preempted. But no one argues that the NVRA goes this far. Indeed, this would be too thin a reed to support implied preemption of a field historically and constitutionally left to the States. C. Lastly, the NVRA’s purpose doesn’t get us to preemption. The district court relied on one of the NVRA’s statutory purposes to read a broad preemptive intent to occupy the field of mail voting. Looking to the NVRA’s purpose to “enhance[] participation of eligible citizens as voters,” see 52 U.S.C. § 20501(b)(2), the district court saw the law as preempting States’ mail-voting requirements. But there are dangers in using supposed purpose rather than statutory text to interpret the law. See generally Rojas v. FAA, 989 F.3d 666, 693 (9th Cir. 2021) (Bumatay, J., MI FAMILIA VOTA V. PETERSEN 105

dissenting in part). And reading a broad preemption regime from the NVRA’s purpose falls into these traps. First, this reading ignores that “[l]egislation . . . is often about the art of compromise.” Id. at 695. Legislation encompasses “the clash of purposes, interests, and ideas,” and its text “may reflect hard-fought compromises.” Id. (simplified). And “no legislation pursues its purposes at all costs, so it frustrates rather than effectuates legislative intent simplistically to assume that whatever furthers the statute’s primary objective must be the law.” Id. (simplified). This case is a perfect example of this principle. The NVRA had multiple statutory purposes—which the district court ignored. Besides broadening the franchise, the NVRA’s purpose was also “to protect the integrity of the electoral process” and “to ensure that accurate and current voter registration rolls are maintained.” 52 U.S.C. § 20501(b)(3)– (4). Thus, both expanding voting and preventing voter fraud were at the heart of the NVRA. If we are to govern by purpose rather than by text, which purpose must prevail here? While some legislators may have felt that letting as many people as possible vote by mail was paramount, others may have believed that combatting voter fraud was more critical. Permitting States to require proof of citizenship to ensure the integrity of the mail-voting system furthers that latter purpose. As judges, we are not well situated to step into the shoes of our elected representatives and select which purpose should guide our interpretation. So it was a mistake to let one singular purpose guide the preemption analysis here without any express textual command. 106 MI FAMILIA VOTA V. PETERSEN

* * * Thus, nothing in the text of the NVRA precludes Arizona from requesting proof of citizenship before allowing voters to vote by mail. We should have reversed the district court order enjoining enforcement of § 16-127(A)(2). III. Proof of Citizenship to Register to Vote Using State Forms H.B. 2492 requires voters who register to vote through Arizona’s state voter-registration form to provide “satisfactory evidence of citizenship” and requires state election officials to “reject any application for registration that is not accompanied by satisfactory evidence of citizenship.” Ariz. Rev. Stat. § 16-121.01(C). The district court held that this provision was barred by the terms of a consent decree signed by Arizona’s Secretary of State and that the NVRA preempts it. The Supreme Court stayed the district court’s injunction on this matter and allowed the law to take effect. We should have taken the hint and ruled that neither the consent decree nor the NVRA bars enforcement of this provision. A. The LULAC Consent Decree Doesn’t Bar Proof of Citizenship In 2018, the former Arizona Secretary of State and former Maricopa County Recorder entered a consent decree with the League of United Latin American Citizens of Arizona (“LULAC”). See LULAC v. Reagan, Doc. 37, No. 2:17-cv-4102 (D. Ariz. 2018). The LULAC Consent Decree bars Arizona county recorders from categorically rejecting MI FAMILIA VOTA V. PETERSEN 107

the registration of applicants who use the state voter- registration form but provide no proof of citizenship. Under this regime, applicants who did not provide proof of citizenship and whose citizenship could not be verified in state databases would be registered to vote only in federal elections. The district court held that the LULAC Consent Decree precludes Arizona from rejecting state-form registrations lacking proof of citizenship. Because this holding raises alarming separation-of-powers concerns, I would reverse. Even if § 16-121.01(C) conflicts with the LULAC Consent Decree, Arizona’s law must prevail. The view that a settlement by a single state executive-branch official may forever curtail the state legislature’s lawmaking power presents disturbing separation-of-powers concerns. Under that view, state executive-branch officials can permanently circumvent legislative authority by entering whatever arrangements they want with private parties. The opportunity for abuse is clear. A state official could collude with like-minded parties to “sue and settle” to prevent a legislature from enacting contrary policies. As the Supreme Court has recognized, consent decrees have the potential to “improperly deprive future officials of their designated legislative and executive powers.” Horne v. Flores, 557 U.S. 433, 449–50 (2009) (simplified). While these separation-of-powers concerns would apply to any restriction of a state legislature’s lawmaking power, they’re particularly acute in the election-law context, where state legislatures enjoy express constitutional authority to act. As discussed above, the Constitution leaves it to state legislatures to set the mechanisms for elections. See Moore v. Harper, 600 U.S. 1, 10 (2023) (observing that the “state legislatures” have the “duty to prescribe rules governing 108 MI FAMILIA VOTA V. PETERSEN

federal elections”) (simplified); see also Carson v. Simon, 978 F.3d 1051, 1060 (8th Cir. 2020) (“[T]he Secretary [of State] has no power to override the Minnesota Legislature” by stipulating to the tabulation of absentee ballots received after Election Day.). These separation-of-powers concerns animate the many cases signifying that legislative acts must trump consent decrees, not the other way around. After all, consent decrees cannot be used to handcuff governments in perpetuity. Thus, consent decrees may need to give way to intervening changes in law, including legislative enactments. See, e.g., Horne, 557 U.S. at 450 (“[C]ourts must . . . ensure that [the] responsibility for discharging the State’s obligations is returned promptly to the State and its officials when the circumstances warrant.”) (simplified); Rufo v. Inmates of Suffolk Cnty. Jail, 502 U.S. 367, 388 (1992) (“[A] consent decree must of course be modified if . . . one or more of the obligations placed upon the parties has become impermissible under federal law,” and that modification may also be warranted “when the statutory or decisional law has changed to make legal what the decree was designed to prevent.”); Agostini v. Felton, 521 U.S. 203, 215 (1997) (“[T]he court cannot be required to disregard significant changes in law . . . if it is satisfied that what it has been doing has been turned through changed circumstances into an instrument of wrong[.]”) (simplified); Miller v. French, 530 U.S. 327, 347 (2000) (“[W]hen Congress changes the law underlying a judgment awarding prospective relief, that relief is no longer enforceable to the extent it is inconsistent with the new law.”); Keith v. Volpe, 118 F.3d 1386, 1393 (9th Cir. 1997) (Parties to a consent decree “c[annot] agree to terms which would exceed their authority and supplant state law.”); League of Residential Neighborhood Advocates MI FAMILIA VOTA V. PETERSEN 109

v. City of Los Angeles, 498 F.3d 1052, 1055 (9th Cir. 2007) (A consent decree “cannot be a means for state officials to evade state law.”); Imprisoned Citizens Union v. Ridge, 169 F.3d 178, 189 (3d Cir. 1999) (opinion of Alito, J.) (When a consent decree conflicts with later legislative action, absent a finding of a “current and ongoing violation of federal law, the law demands nothing less than the immediate termination of the consent decree.”); Biodiversity Assocs. v. Cables, 357 F.3d 1152, 1169–70 (10th Cir. 2004) (A consent decree “does not freeze the provisions of the statute into place. If the statute changes, the parties’ rights change, and enforcement of their agreement must also change. Any other conclusion would allow the parties, by exchange of consideration, to bind not only themselves but Congress and the courts as well.”). So when a change in statutory law conflicts with a consent decree, it’s the statute that governs. Of course, state laws must yield to federal constitutional rights. So a consent decree guarding a federal right is a different matter. But “[w]ithout . . . finding[]” that a “remedy is necessary to rectify a violation of federal law,” federal courts have no authority to “override[] state law provisions” and “parties can only agree to that which they have the power to do outside of litigation.” League of Residential Neighborhood Advocates, 498 F.3d at 1058 (simplified). At no point did the district court that entered the LULAC Consent Decree hold that the requirement of proof-of-citizenship violates federal law. In fact, the LULAC Consent Decree notes the Secretary of State’s continued assertion of the law’s constitutionality, despite the compromise. So the LULAC Consent Decree is not a judicial remedy necessary to enforce federal law. Rather, the basis for the decree hides in plain sight—consent alone. And the consent of a single state executive-branch official is no 110 MI FAMILIA VOTA V. PETERSEN

basis to upset the balance of power among the branches of state government or the balance of power between the state and federal governments. Opponents of the proof-of-citizenship requirement frame this issue as one of federal supremacy and judicial finality— that a state legislature cannot reverse the binding effect of a federal court’s final judgment. True, consent decrees “are essentially contractual agreements that are given the status of a judicial decree.” Hook v. State of Ariz., Dep’t of Corr., 972 F.2d 1012, 1014 (9th Cir. 1992). But “finality” isn’t the end all and be all in the law. No doubt, “[h]aving achieved finality, . . . a judicial decision becomes the last word of the judicial department with regard to a particular case or controversy, and Congress may not declare by retroactive legislation that the law applicable to that very case was something other than what the courts said it was.” Plaut v. Spendthrift Farm, Inc., 514 U.S. 211, 227 (1995). But that principle does not “call[] into question” a legislature’s ability to pass legislation that “alter[s] the prospective effect of injunctions entered by Article III courts.” Id. at 232. Regardless of whether a prospective remedy is an injunction or a consent decree, “a court does not abdicate its power to revoke or modify its mandate, if satisfied that what it has been doing has been turned through changing circumstances into an instrument of wrong.” Sys. Fed. No. 91 v. Wright, 364 U.S. 642, 650–51 (1961) (simplified). And so a consent decree—though blessed by a federal court—doesn’t forever foreclose legislative change. Indeed, it would detract—rather than augment—respect for federal law to claim that federal courts are powerless to stop a state executive official from teaming up with like- minded private litigants to tie the hands of future state legislatures. It’s this picture that turns federal supremacy on MI FAMILIA VOTA V. PETERSEN 111

its head at the expense of the separation of powers in the States. In no way are federal courts forced to “bind state and local officials to the policy preferences of their predecessors” and erode state legislative powers. See Horne, 557 U.S. at 449 (simplified). After all, “[a] State, in the ordinary course, depends upon successor officials, both appointed and elected, to bring new insights and solutions” to its government. Frew ex rel. Frew v. Hawkins, 540 U.S. 431, 442 (2004). Curiously, the majority argues that enjoining § 16- 121.01(C) poses no threat to the Arizona “Legislature[’s] sovereign authority” because it does not bar the legislature from enacting the law—it only bars executive officials from enforcing the law. See Maj. Op. at 52. That is no solace for the Arizona Legislature. Instead, “the inability to enforce its duly enacted plans clearly inflicts irreparable harm on the State.” Abbott v. Perez, 585 U.S. 579, 602 n.17 (2018). Of course, “completely nullif[ying] any vote by the Legislature” flouts the separation of powers. Ariz. State Legislature v. Ariz. Indep. Redistricting Com’n, 576 U.S. 787, 804 (2015). After all, the heart of the legislative power is to transform the words of proposed legislation into enforceable statutes. We can’t turn a blind eye to neutering the Arizona Legislature by sophistry. Finally, it is claimed § 16-121.01(C) can’t be enforced because no party has moved to modify the consent decree under Federal Rule of Civil Procedure 60(b). But “the general rule” is that “only a party to the action” can move under Rule 60. Wright & Miller 21A Fed. Proc., L. Ed. § 51:170 (2024). And no one here was a party to the LULAC Consent Decree. Courts have “emphasize[d] the fundamental nature of the general rule that a litigant is not bound by a judgment to which she was not a party.” Taylor 112 MI FAMILIA VOTA V. PETERSEN

v. Sturgell, 553 U.S. 880, 898 (2008). Simply, the LULAC Consent Decree “does not conclude the rights of strangers” and “collateral attack” is proper when, as here, the decree “affects [a stranger’s] legal rights.” Martin v. Wilks, 490 U.S. 755, 762–63 (1989) (simplified); see also Sys. Fed. No. 91, 364 U.S. at 650–51. After all, “[a] court that invokes equity’s power to remedy a constitutional violation by an injunction mandating systemic changes to an institution has the continuing duty and responsibility to assess the efficacy and consequences of its order.” Brown v. Plata, 563 U.S. 493, 542 (2011) (emphasis added). Given the profound effect of the LULAC Consent Decree on the structure of Arizona’s government, the fundamental instruction to federal courts to continually reassess prospective relief applies here too. So no procedural obstacle prevents enforcement of § 16-121.01(C). B. The NVRA Doesn’t Preempt the Proof-of-Citizenship Requirement Nor does the NVRA preempt Arizona’s requirement for proof of citizenship. Opponents of the requirement make two arguments under the NVRA. First, they assert that the requirement violates § 20508(b)(1)’s “necessary” information rule. Second, they contend that § 20506(a)(6)(A)’s “public assistance agencies” provision bars enforcement of § 16-121.01(C). Both arguments are wrong. 1. NVRA’s Necessary Information Provision Because the district court ruled based on the LULAC Consent Decree, it relegated its NVRA analysis to a mere MI FAMILIA VOTA V. PETERSEN 113

footnote. See Mi Familia Vota I, 691 F. Supp. 3d at 1096 n.13. The district court tersely reasoned that the NVRA preempts § 16-121.01(C) because the statute “precludes states from requiring [documentary proof of citizenship] to register applicants for federal elections.” Id. As the following shows, that’s wrong. Once again, the NVRA creates two paths for citizens to register to vote. They may register using a federally created voter-registration form or they may register with a state- created voter-registration form. See 52 U.S.C. § 20505(a)(1)–(2). The NVRA places different constraints on the design and use of both forms, though States have leeway to design their state form. The NVRA directs that a State may “develop and use” a state form so long as it “meets all of the criteria stated in section 20508(b) of this title for the registration of voters in elections for Federal office.” Id. § 20505(a)(2). The NVRA then establishes the substantive rules that the state form must follow. Id. § 20508(b). It provides that the state form “may require only such identifying information . . . and other information . . . as is necessary to enable the appropriate State election official to assess the eligibility of the applicant and to administer voter registration and other parts of the election process.” Id. § 20508(b)(1). It also mandates that the state form “include a statement that”: (A) “specifies each eligibility requirement (including citizenship);” (B) “contains an attestation that the applicant meets each such requirement; and” (C) “requires the signature of the applicant, under penalty of perjury.” Id. § 20508(b)(2)(A)–(C). Despite these set requirements, § 20508(b) is no straitjacket on the States. In the end, “state-developed forms 114 MI FAMILIA VOTA V. PETERSEN

may require information the Federal Form does not.” ITCA, 570 U.S. at 12. At all times, “States retain the flexibility to design and use their own registration forms[.]” Id. The key word here is “flexibility.” After all, why would Congress want to micromanage what information can be included on a state form when they already obligated States to “accept and use” the federal form? The NVRA thus confirms the States’ plenary authority to design state election forms— subject to a few mandatory requirements. So we should largely defer to the States to develop their own forms with the sole constraint that the State must only request information it finds “necessary.” Id. § 20508(b)(1). And there’s no reason to read “necessary” information as meaning only the bare minimum amount of information. While § 20508(b)(1) permits the States to ask for “necessary” information, elsewhere the NVRA limits States to asking for “only the minimum amount of information necessary to . . . enable State election officials to assess the eligibility of the applicant and to administer voter registration and other parts of the election process.” Id. § 20504(c)(2)(B) (providing the standard for “motor voter” forms). So Congress distinguished between “information” that was “necessary” in the eyes of state officials and “information” that was the “minimum amount . . . necessary” for state officials. See Fish v. Kobach, 840 F.3d 710, 734 (10th Cir. 2016) (holding that § 20504(c)(2)(B) imposes a “stricter principle” than § 20508(b)(1)). And “when the legislature uses certain language in one part of the statute and different language in another, the court assumes different meanings were intended.” Cheneau v. Garland, 997 F.3d 916, 920 (9th Cir. 2021) (en banc) (simplified). While it would be fair to strictly enforce necessity in § 20504(c)(2)(B), § 20508(b)(1) still gives States flexibility. MI FAMILIA VOTA V. PETERSEN 115

So “necessary” in § 20508(b)(1) doesn’t impose a least- restrictive-means test on state forms. Here, we have no basis to overrule Arizona’s determination that documentary proof of citizenship is “necessary to enable [its] election official[s] to assess the eligibility of the applicant.” 52 U.S.C. § 20508(b)(1). Such a requirement obviously would ensure the citizenship of the voter—a necessary qualification. And precedent already supports States’ authority to request proof of citizenship. As the Court said, “[s]ince the power to establish voting requirements is of little value without the power to enforce those requirements, . . . it would raise serious constitutional doubts if a federal statute precluded a State from obtaining the information necessary to enforce its voter qualifications.” ITCA, 570 U.S. at 17. The Court even used Arizona’s proof-of-citizenship requirement as the example of the type of information that “state-developed forms may require” that “the Federal Form does not.” Id. at 12. And our own court has remarked that the NVRA “plainly allow[s] states, at least to some extent, to require their citizens to present evidence of citizenship when registering to vote.” Gonzalez v. Arizona, 485 F.3d 1041, 1050–51 (9th Cir. 2007) (observing that “[t]he language of the [NVRA] does not prohibit documentation requirements” and refusing to enjoin Arizona’s documentary proof-of-citizenship requirement). Given this overwhelming support for Arizona’s law, opponents of the law must climb a steep hill to support the injunction—a burden they do not meet. First, they primarily rely on an out-of-circuit interpretation of a different provision of the NVRA. Citing Fish, they argue that mere attestation of citizenship is all that States may request and documentary proof is too far. True, Fish held that attestation 116 MI FAMILIA VOTA V. PETERSEN

“is the presumptive minimum amount of information necessary for state election officials to carry out their [duties].” 840 F.3d at 717. But Fish was applying § 20504(c)(2)(B)’s “motor voter” stricter standard, which only permits the “minimum amount of information necessary.” Id. It had nothing to do with § 20508(b)(1)— the issue here. Given their different standards, it’s more appropriate to use Fish to show why Arizona’s law meets § 20508(b)(1)’s more permissive standard. Their next out-of-circuit authority fares no better. Kobach v. U.S. Election Assistance Comm’n, 772 F.3d 1183 (10th Cir. 2014), is an Administrative Procedure Act case deferentially reviewing the EAC’s determination of “necessity” for the federal voter-registration form. Kobach applied “very deferential” review to that question. Id. at 1187–88, 1197. There’s no similar agency action here. More to the point, EAC’s determinations about what’s necessary for the federal form don’t govern what’s necessary for the state form. See ITCA, 570 U.S. at 12. Finally, they point to the district court’s factual finding that “non-citizens voting in Arizona is quite rare” and so they argue Arizona’s law is unnecessary. See Mi Familia Vota v. Fontes (“Mi Familia Vota II”), 719 F. Supp. 3d 929, 967 (D. Ariz. 2024). But this ignores that the district court found that non-citizen voting does occur—even if it isn’t widespread. Id. And Arizona’s elected officials—not federal judges— get to determine what level of voter fraud the State may tolerate. Indeed, even if no voter fraud were proven, state officials may still decide that the concern for voter fraud warrants legislative action. Cf. Brnovich v. Democratic Nat’l Comm., 594 U.S. 647, 686 (2021) (noting a State “may take action to prevent election fraud without waiting for it to occur and be detected within its own borders”). MI FAMILIA VOTA V. PETERSEN 117

2. NVRA’s Public Assistance Agencies Provision Opponents of Arizona’s proof-of-citizenship requirement make a final argument under the NVRA. Relying on the district court’s holding that the NVRA preempts the state form because of its proof-of-residency requirement under the “public assistance agencies” provision, they contend that the proof-of-citizenship requirement is also preempted. See Mi Familia Vota II, 719 F. Supp. 3d at 997. This provision establishes that States must designate “public assistance agencies” that will provide to all applicants for services either the federal voter- registration form or “the office’s own form if it is equivalent to the [federal] form.” 52 U.S. § 20506(a)(6)(A)(i)–(ii). In the district court’s view, because the proof-of-residency (and proof-of-citizenship) requirements make Arizona’s state form not “equivalent” to the federal form, those requirements must give way. Instead, the district court ruled that any state form provided by a public assistance agency must be “virtually identical to the Federal Form.” Mi Familia Vota II, 719 F. Supp. 3d at 997 (simplified). First, “equivalent” doesn’t always mean “identical.” Common definitions show that “equivalent” can fall short of meaning the “exact same”—especially when two different things have the same function or cause similar effects. See Equivalent, American Heritage Dictionary 291 (4th ed. 2000) (“Similar or identical in function or effect”); Equivalent, Oxford English Dictionary 358 (2d ed. 1989) (Equal in value, power, efficacy, or import”; “That is virtually the same thing; identical in effect; tantamount”; “Having the same relative position or function; corresponding.”); Equivalent, Webster’s Third New Int’l 118 MI FAMILIA VOTA V. PETERSEN

Dictionary 769 (1981) (“like in signification or import”; “corresponding or virtually identical esp. in effect or function”). So this provision doesn’t demand that state public assistance agencies use a form that is identical to the federal form. Rather, like the state form, an “equivalent” form need only have the same “effect” for purposes of registration. And demanding that the federal form and the state form be identical would render § 20505(a) void and contravene ITCA. Allowing some variation between the federal form and the public assistance agencies’ “own form” best accounts for the NVRA’s “context” and “overall statutory scheme.” King v. Burwell, 576 U.S. 473, 486 (2015) (simplified). As discussed above, the NVRA creates a two-track approach for voter registration: applicants may use either the federally created voter-registration form or a state-created form. See 52 U.S.C. § 20505(a). States have some freedom in designing the state form if they follow the permissive requirements of § 20505(a)(2). The upshot of this statutory framework is that voters can pick a “simple means of registering to vote in federal elections” through the federal form or they can choose the state form, which can “require information the Federal Form does not.” ITCA, 570 U.S. at 12. It is an elegant scheme that respects the balance of power between the federal government and the States. It would thus be odd if Congress gave States the flexibility to create their own form in § 20505(a) but then took away all that freedom through the “public assistance agencies” provision of § 20506(a)(6)(A). It’s doubtful that Congress expected a third form—a public agency’s “own form” that must be identical to the federal form. Thus, the best way to harmonize all these provisions is to consider a compliant state form—one “that meets all of the criteria stated in MI FAMILIA VOTA V. PETERSEN 119

section 20508(b)”—as “equivalent” to the federal form. See 52 U.S.C. §§ 20505(a)(2), 20506(a)(6)(A). At the very least, even if the district court were right that the state form is not “equivalent” to the federal form, the remedy isn’t to redesign Arizona’s chosen form. The proper remedy would have been to have Arizona’s “public assistance agencies” distribute the federal form. Such a narrowly tailored remedy would respect the State’s sovereignty and fulfill the commands of the NVRA. C. Finally, opponents of the proof-of-citizenship requirement assert an equal protection challenge to the law. Even the majority agrees this argument was a stretch. See Maj. Op. at 74–78. * * * For all these reasons, we should have reversed the district court order enjoining enforcement of § 16-121.01(C). IV. Requiring Proof of Residence to Register to Vote H.B. 2492 requires a person who registers to vote to provide “an identifying document that establishes proof of location of residence.” Ariz. Rev. Stat. § 16-123; see also id. § 16-121.01(A). A “valid and unexpired Arizona driver license” constitutes “satisfactory proof of location of residence.” Id. § 16-123. If a person fails to provide proof of residence, then the person will be registered to vote in only federal elections. The district court held that the NVRA’s “public assistance agencies” provision barred enforcement of this provision, see 52 U.S.C. § 20506(a)(6)(A)(i)–(ii), for the same reasons as the proof- 120 MI FAMILIA VOTA V. PETERSEN

of-citizenship requirement. For the reasons discussed above, the district court’s analysis was wrong, and we should have reversed it. Opponents of the proof-of-residence requirement also make an equal protection argument against it. The majority properly dismisses that contention. See Maj. Op. at 74–78. The district court also ruled that the proof-of-residency requirement violated the necessity provision of § 20508(b)(1). Recall that § 20508(b)(1) requires that state- created voter registration forms “may require only such identifying information . . . and other information . . ., as is necessary to enable the appropriate State election official to assess the eligibility of the applicant and to administer voter registration and other parts of the election process.” 52 U.S.C. § 20508(b)(1). Once again, we have no basis to overrule what Arizona thought was “necessary” for state voter-registration forms. See id. The district court’s ruling that proof-of-residence isn’t “necessary” hinged on what it perceived to be an inconsistency in Arizona’s registration requirements. Under the law, new voter-registration applicants must provide proof-of-residence, Ariz. Rev. Stat. §§ 16-121.01(A), 16- 123, but existing registered voters who obtain an out-of-state license or identification must only provide a signed statement under the penalty of perjury that they are still a resident of Arizona, id. § 16-165(F). “The Court cannot reconcile why [documentary proof of residence] would be necessary for new applicants when an attestation is sufficient to determine the eligibility of registered voters who subsequently obtain an out-of-state identification.” Mi Familia Vota II, 719 F. Supp. 3d at 996. Respectfully, the district court could have tried harder to reconcile the two provisions. There is a clear difference between an existing MI FAMILIA VOTA V. PETERSEN 121

registered voter who has previously been verified as a legitimate voter and a new applicant who has not yet gone through the State’s vetting process. It makes sense to require heightened proof for the unverified applicant. That Arizona permits existing voters with a known track record to provide less proof of residence than unknown, new applicants doesn’t make proof of residence unnecessary. In other words, what may be “necessary” in some cases may not be “necessary” in all cases. Further, § 20508(b)(1) doesn’t impose a least-restrictive- means test on what sort of documentation a state form can require. The State has no duty to do just the bare minimum of vetting. If the State finds it “necessary,” it may request more thorough proof of eligibility. Otherwise, we impose a non-existent narrow-tailoring test onto § 20508(b)(1). And no one disputes that residency is a valid eligibility requirement to vote in Arizona. See Ariz. Const. art. VII, § 2(A); Ariz. Rev. Stat. § 16-101(A)(3). Without convincing proof that information serves no function, we have no basis to second-guess Arizona’s determination of necessity. For these reasons, we should have reversed the district court order enjoining enforcement of §§ 16-121.01(A) and 16-123. V. Removal of Noncitizens Within 90 Days of an Election H.B. 2243 directs state officials to conduct periodic, often monthly, inspections of Arizona’s voter roll to determine whether any person is ineligible to vote or not a U.S. citizen. See Ariz. Rev. Stat. § 16-165(G)–(K). If election officials “obtain[] information” from these 122 MI FAMILIA VOTA V. PETERSEN

inspections and “confirm” that a “person registered is not a United States citizen,” they “shall cancel the registration.” Id. § 16-165(A)(10). The district court held that the cancellation of an improperly registered foreign citizen’s registration violates the NVRA’s “90-Day Provision.” See 52 U.S.C. § 20507(c)(2)(A). Under that provision, with some exceptions, “[a] State shall complete, not later than 90 days prior to the date of a primary or general election for Federal office, any program the purpose of which is to systematically remove the names of ineligible voters from the official lists of eligible voters.” Id. § 20507(c)(2)(A). So the district court ruled that Arizona cannot execute H.B. 2243’s provisions requiring the “systematic investigation and removal of registered voters” within 90 days of a federal election. But because the phrase “ineligible voters” in the 90-Day Provision doesn’t include foreign citizens, the provision doesn’t apply to Arizona’s cancellation program. I would thus reverse the district court on this issue. To be sure, the 90-Day Provision uses broad language— applying to “any” program to remove undefined “ineligible voters.” Given these seemingly capacious terms, it’s easy— as the majority does—to just throw up our hands and give the provision its widest implications. See Maj. Op. at 45– 46. But that’s not how we interpret statutes. We don’t read a term “in isolation” or give the statute “the broadest imaginable definitions of its component words.” See Sackett v. EPA, 598 U.S. 651, 674 (2023); Dubin v. United States, 599 U.S. 110, 120 (2023). Instead, our job is to conduct “a careful examination of the ordinary meaning and structure of the law” and keep the “overall statutory scheme” in mind. Food Mktg. Inst. v. Argus Leader Media, 588 U.S. 427, 436 (2019); FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 133 (2000) (simplified). Once we do that, the best MI FAMILIA VOTA V. PETERSEN 123

reading of the statute is that the NVRA’s 90-Day Provision doesn’t apply to the removal of aliens from state voter rolls. Start with the 90-Day Provision’s place within the NVRA’s statutory scheme. It is part of § 20507, also known as Section 8, which addresses “the administration of voter registration.” 52 U.S.C. § 20507. Section 20507 introduces a systematic series of regulations regarding voter rolls. In other words, think of § 20507 as walking the States through each step of the voter-registration process—a process that both enhances participation in elections and ensures the integrity of the vote. It starts with the pre-registration process, then goes to the post-registration process, and ends with voter-removal programs. As in any conversation, what Congress said earlier shapes how we understand what Congress says next. And consistent with the protection of voters’ rights, the NVRA becomes more stringent as we get closer to Election Day. First, the pre-registration process. The first subsection of § 20507 begins with discussion of the “valid voter registration form of the applicant.” Id. § 20507(a)(1)(A)– (D) (emphasis added). Among their responsibilities, States must accept valid voter registration forms from an “applicant” within certain timeframes and provide “notice to each applicant of the disposition of the application.” Id. § 20507(a)(1)–(2) (emphasis added). States must also “inform applicants” of “voter eligibility requirements” and the penalties for providing false voter information. Id. § 20507(a)(5). At this stage, “applicant” must refer to any person who submits a voter registration application, which may include both U.S. citizens and foreign citizens. But before proceeding, this subsection provides an important limitation. 124 MI FAMILIA VOTA V. PETERSEN

Congress instructs the States that they must “ensure that any eligible applicant is registered to vote in an election.” Id. § 20507(a)(1) (emphasis added). In this context, an “eligible applicant” is an “applicant” who is qualified to be registered to vote. See Eligible, Webster’s Third New Int’l Dictionary 736 (1981) (“fitted or qualified to be chosen or used: entitled to something”); Eligible, Oxford English Dictionary 140 (2d ed. 1989) (“Fit or proper to be chosen (for an office or position).”); Eligible, American Heritage Dictionary 280 (4th ed. 2000) (“Qualified to be chosen”). So Congress distinguishes between an “applicant” and an “eligible applicant,” which is a smaller subset of “applicant[s].” States must “ensure” that only “eligible applicant[s]” are “registered to vote.” Id. § 20507(a)(1). Thus, foreign citizens—as ineligible applicants—are weeded out of the statutory process at this stage and may never go further down the regulatory scheme. Second, the post-registration process. After successful “disposition of the application” and an “eligible applicant” is registered to vote, the next subsection calls the person a “registrant.” Id. § 20507(a)(3). As a “registrant,” the person may vote unless the person becomes ineligible because of a criminal conviction, disability, or move. Id. Respecting this, this subsection “provide[s] that . . . a registrant may not be removed from the official list of eligible voters except” by request of the registrant or for a criminal conviction, mental incapacity, death, or change of address. Id. § 20507(a)(3)– (4) (emphasis added). This protection applies only to a “registrant”—again meaning only an “eligible applicant” who was registered to vote. See id. § 20507(a)(3). This definition necessarily excludes foreign citizens, who are never “eligible applicant[s]” having the right to be registered to vote. Thus, § 20507(a)(3) in no way protects foreign MI FAMILIA VOTA V. PETERSEN 125

citizens improperly registered from removal from the voter rolls. See Bell v. Marinko, 367 F.3d 588, 591–92 (6th Cir. 2004) (“In creating a list of justifications for removal, Congress did not intend to bar the removal of names from the official list of persons who were ineligible and improperly registered to vote in the first place.”). Third, removal programs. This phase directs States to conduct programs to purge “ineligible voters” from voter rolls. To begin, States must “conduct a general program that makes a reasonable effort to remove the names of ineligible voters from the official lists of eligible voters by reason of . . . death of the registrant . . . or change in the residence of the registrant.” Id. § 20507(a)(4) (emphasis added). For the first time in § 20507, Congress distinguishes between “eligible voters” and “ineligible voters.” Id. Based on the structure of the preceding subsections and placing the terms within the statutory scheme, these terms must refer to two subcategories of “registrant[s].” The subcategory of “eligible voters” are those “registrants”— “eligible applicants” registered to vote—who remain eligible to vote. The subcategory of “ineligible voters” are those “registrant[s]” who have lost eligibility to vote because of the “death of the registrant,” “change in the residence of the registrant,” or some other intervening event. Id. § 20507(a)(4); see Ineligible, Webster’s New Third Int’l Dictionary 1156 (1981) (“not eligible: not qualified to be chosen for an office : not worthy to be chosen or preferred”); Ineligible, Oxford English Dictionary 904 (2d ed. 1989) (“[i]ncapable of being elected; legally or officially disqualified for election to an office or position”); Ineligible, American Heritage Dictionary 436 (4 ed. 2000) (“[d]isqualified by law or rule”). Thus, Congress itself uses “registrants” to define “ineligible voters.” 126 MI FAMILIA VOTA V. PETERSEN

In other words, Congress uses these two new terms to subdivide the old group of “registrants” for a new stage of the registration process: post-registration removal programs. But one thing is clear. In all cases, foreign citizens can never be “ineligible voters” or “eligible voters” because they could never have been “registrant[s]”—that is, “eligible applicant[s]” registered to vote. Thus, any limitation Congress places on removal programs doesn’t apply to the removal of non-U.S. citizens. That leads us to the 90-Day Provision—the provision that the district court used to enjoin enforcement of § 16-165 within 90-days of an election. Under that provision, “[a] State shall complete, not later than 90 days prior to the date of a primary or general election for Federal office, any program the purpose of which is to systematically remove the names of ineligible voters from the official lists of eligible voters.” Id. § 20507(c)(2)(A) (emphasis added). The subsection then clarifies that the 90-day quiet period “shall not be construed to preclude . . . the removal of names from official lists of voters on a basis” of (1) a “request of the registrant,” (2) “criminal conviction or mental capacity,” or (3) “the death of the registrant.” Id. § 20507(c)(2)(B). Taken as a whole, this subsection protects only “ineligible voters” from removal within 90 days of election, and “ineligible voters” are simply a subcategory of “registrants.” The 90-Day Provision then doesn’t protect those who were never “registrants”—meaning those who were never “eligible applicants” registered to vote, such as non-U.S. citizens. In other words, § 20507 progresses from (1) “applicant[s]” to (2) “eligible applicant[s]” to (3) “registrant[s]” to (4) “eligible voters” and “ineligible voters.” Each term or set of terms is a subset of its preceding MI FAMILIA VOTA V. PETERSEN 127

term. As explained above, a foreign citizen may be an “applicant” but may not be in the subset of “eligible applicant[s].” Because of this, foreign citizens are excluded from the terms “registrant[s],” “eligible voters” and “ineligible voters.” The following graphic explains this progression of terms: 128 MI FAMILIA VOTA V. PETERSEN

Once placed within the overall statutory scheme, foreign citizens aren’t included in the protection of “ineligible voters” in the 90-Day Provision. Simply, foreign citizens are excluded from the NVRA’s statutory protections during the removal process, and nothing in the NVRA prevents their removal at any point whatsoever. In contrast, reading the 90-Day Provision in a literalist way would lead to absurd results and raise serious constitutional concerns. If foreign citizens are included in the protection of “ineligible voters,” that would mean that States can continue to “systemically remove” those voters convicted of a crime, found mentally incapacitated, or who died—all voters susceptible of being incorrectly removed— within 90 days of the election, but they can’t stop foreign citizens from voting in our elections—a category easier to verify. 52 U.S.C. § 20507(c)(2)(B). And a congressional ban on removing foreign citizens for voting in American elections is absurd. It’s one thing to allow an American citizen who has moved to a new precinct to vote in the wrong district; it’s entirely different to force a State to allow a foreign citizen to vote in its elections. While used only “sparingly,” the absurdity canon means we should “not myopically focus[] on a single” term or phrase and instead we should “evaluate the statute in context.” United States v. Lucero, 989 F.3d 1088, 1098 (9th Cir. 2021). The majority’s acontextual interpretation of § 20507 creates an absurdity that Congress never established in the statutory text. And forcing States to accept foreign citizens in their voting booths would infringe on States’ rights to set voter qualifications and administer elections. Rather than breaking the 90-Day Provision into component parts and reading words in isolation, we should have read the law as a MI FAMILIA VOTA V. PETERSEN 129

whole and understood that it offers no protection for foreign citizens. Lastly, the majority relies on the Eleventh Circuit’s purpose-based analysis in Arcia v. Florida Secretary of State, 772 F.3d 1335 (11th Cir. 2014). According to the Eleventh Circuit, the 90-Day Provision “strikes a careful balance” of the NVRA’s purposes—“[i]t permits systematic removal programs at any time except for the 90 days before an election because that is when the risk of disfranchising eligible voters is the greatest.” Id. at 1346. As explained above, it’s a mistake to overly rely on purpose in interpreting statutes. Even so, this supposed “balanc[ing]” test fails to explain why voters who are convicted of a crime, have a disability, or have died receive no protections at all but foreign citizens are immune from removal. As the Sixth Circuit considered, by finding foreign citizens protected by the NVRA’s removal program regulations, we “effectively grant, and then protect, the franchise of persons not eligible to vote.” Bell, 367 F.3d at 592. It’s hard to see how that’s consistent with the NVRA’s purposes. Because the 90-Day Provision doesn’t apply to foreign citizens, we should have reversed the district court’s injunction of § 16-165(A)(10). VI. Birthplace and Citizen Checkbox Requirements H.B. 2492 requires a state-form voter-registration applicant to provide a “place of birth,” along with the applicant’s name, address, birthdate, and signature “to be properly registered to vote.” Ariz. Rev. Stat. § 16- 121.01(A). It also requires the applicant to place a “checkmark” in a box indicating that the applicant is a U.S. 130 MI FAMILIA VOTA V. PETERSEN

citizen. Id. If any of this information is “incomplete or illegible,” “the registration cannot be completed” and the county recorder must give the applicant notice and opportunity to supply the information. Id. § 16-134(B). The district court held that the birthplace and citizen-checkbox requirements violate the Materiality Provision of the Civil Rights Act, 52 U.S.C. § 10101(a)(2)(B). It permanently enjoined Arizona election officials from enforcing these requirements and from rejecting applicants for the lack of birthplace or citizen-checkbox information if the applicant is otherwise eligible. I would reverse in part and affirm in part. When the Civil Rights Act was enacted, local election officials exploited “hypertechnical[] or entirely invented” errors to reject Black applicants. Justin Levitt, Resolving Election Error: The Dynamic Assessment of Materiality, 54 Wm. & Mary L. Rev. 83, 148 (2012). For example, one applicant was rejected because, when required to provide her age in years, months, and days, she “missed the mark by one day because the day had not yet ended.” Id. Similarly, “[a]nother application was rejected because the applicant’s state was misspelled as ‘Louiseana.’” Id. In another anecdote, a Black schoolteacher in Alabama had her voter- registration form “rejected because she omitted a date in one question—even though she gave the same information elsewhere on the form.” Hearings on S. 1731 and S. 1750 Before the S. Comm. on the Judiciary, 88th Cong. 101–02 (1963) (Statement of Att’y Gen. Robert F. Kennedy). The list goes on. See Levitt, Materiality, at 148 (collecting examples). Congress thus sought to deny the use of irrelevant errors as pretext to hide election officials’ discriminatory intent to deny voters their right to vote. MI FAMILIA VOTA V. PETERSEN 131

The Materiality Provision provides that—

No person acting under color of law shall deny the right of any individual to vote in any election because of an error or omission on any record or paper relating to any application, registration, or other act requisite to voting, if such error or omission is not material in determining whether such individual is qualified under State law to vote in such election[.]

52 U.S.C. § 10101(a)(2)(B). First, the term “material” is used often in the law. We’ve recently reiterated that something “is material if it could have affected or influenced the government’s decision.” United States v. Patnaik, 125 F.4th 1223, 2025 WL 85836, at *3 (9th Cir. 2025) (simplified); see also Material, Oxford English Dictionary Online (defining “material” in legal sense as “significant or influential, esp[ecially] in having affected a person’s decision-making” or “having a logical connection with the facts at issue”). Something need not be essential to be “material” in this context. Vote.Org v. Callanen, 89 F.4th 459, 478 (5th Cir. 2023) (“We reject ‘essential’ as a reasonable meaning” of “material.”) So an “error or omission” is “material” if it could have affected or influenced the decision “whether an individual is qualified under State law to vote.” 52 U.S.C. § 10101(a)(2)(B). The “error or omission” need not be “essential” to the decision to register the person. Second, the Materiality Provision only bars the improper use of an immaterial “error or omission” on voting forms. Id. It doesn’t prevent government officials from requesting 132 MI FAMILIA VOTA V. PETERSEN

the underlying information. States may thus ask for any information they deem necessary in voter-registration forms. The law only applies once an applicant makes an error or omits some information. In other words, whatever preemptive force the Materiality Provision has, it applies only from the use of an “error or omission”—not from the request for the underlying information. Finally, the Materiality Provision is violated only if a voter registration is “reject[ed]” “because of” the immaterial error or omission. 52 U.S.C. § 10101(a)(2)(B). In this context, “because of” means the “‘but-for’ cause.” Univ. of Texas Southwestern Medical Center v. Nassar, 570 U.S. 338, 350 (2013) (simplified). So the law prohibits an immaterial “error or omission” from being the “but-for” cause of rejecting a voter-registration application. The law thus doesn’t prevent government officials from using an immaterial “error or omission” to investigate or further probe the application. Nor does it prevent election officials from requesting corrections. And if investigation uncovers other information revealing that the applicant is ineligible to register to vote under state law, then the “error or omission” certainly becomes material. Given these considerations, I would reverse the district court’s injunction as to the birthplace requirement but affirm the injunction of the citizenship-checkbox requirement. A. Birthplace Requirement After a bench trial, the district court concluded that the birthplace requirement violates the Materiality Provision because it can’t be used to verify citizenship, residence, or identity—all state-law requisites for voting. Mi Familia MI FAMILIA VOTA V. PETERSEN 133

Vota II, 719 F. Supp. 3d at 995. But because some circumstances exist in which an omitted birthplace may affect or influence verification of a person’s registration application, I would reverse the district court’s injunction. As this is a pre-enforcement challenge, opponents of the birthplace requirement bring a facial challenge to the law— a claim that is “hard to win.” Moody v. NetChoice, LLC, 603 U.S. 707, 723 (2024). “Claims of facial invalidity often rest on speculation about the law’s coverage and its future enforcement.” Id. (simplified). And “facial challenges threaten to short circuit the democratic process by preventing duly enacted laws from being implemented[.]” Id. (simplified). Thus, a facial challenge is “the most difficult challenge to mount successfully.” Anderson v. Edwards, 514 U.S. 143, 155 (1995) (simplified). Plaintiffs must “establish[] that no set of circumstances exists under which the Act would be valid, i.e., that the law is [invalid] in all of its applications.” Wash. State Grange v. Wash. State Republican Party, 552 U.S. 442, 449 (2008) (simplified). Indeed, so onerous is the task, a defendant can “defeat [a] facial challenge by conjuring up a single valid application of the law.” City of Chicago v. Morales, 527 U.S. 41, 81 (1999) (Scalia, J., dissenting). To be qualified to vote, an applicant in Arizona must be a U.S. citizen, over 18 years old, and (in most cases) a resident of the State for 29 days before the election. Ariz. Const. art. VII, § 2; Ariz. Rev. Stat. § 16-101(A). And implicitly underlying these qualifications is identity—that the applicant is who he says he is. See Vote.Org, 89 F.4th at 489 (noting that identity is “the most basic qualification to vote”). An omitted birthplace could be material in determining an applicant’s identity in at least two situations. 134 MI FAMILIA VOTA V. PETERSEN

First, an omitted birthplace could be significant when a county recorder comes across what’s called a “soft match.” Once a voter-registration form is received, county recorders must search existing voter records to try to determine if the applicant matches someone already registered to vote. If there’s a match, the new form is treated as a request to update voter information. If there’s no match, the county recorder registers the applicant as a new voter. A “soft match” occurs when an old voting record does not provide enough information to conclusively match a new registration form. One county recorder stated that this situation “happens a lot.” The “birthplace” requirement helps resolve “soft matches.” Before H.B. 2492, applicants only needed to provide their name, address, date of birth, signature, and an affirmation of citizenship. If applicants had a social security or driver’s license number, they were asked to include it too. A “soft match” occurs, for example, when a recorder finds matches between the applications’ listed first name, last name, and birth date or listed first name, birth date, and the last four digits of a social security number. Adding a datapoint—like matching birthplaces—would eliminate some soft matches. Imagine a county recorder receives a voter registration application from “John Doe” born on “April 1, 2000.” There’s a match with an existing voter record—a “John Doe” also born on “April 1, 2000.” This presents a “soft match”—he might or might not be the same person. Now say that the new registration form indicates that “John Doe” was born in “Peoria.” But the registered John Doe was born in “Phoenix.” Now we know they are not the same person. This difference means that county recorders could eliminate the record as a “soft match.” On the other hand, a match between first name, last name, MI FAMILIA VOTA V. PETERSEN 135

birthdate, and birthplace would give further evidence of a “match” and might prompt the county recorder to follow up with the applicant—as one county recorder testified. Opponents of the requirement claim that a birthplace resolving a “soft match” would be rare—as the Civil Rights Division’s expert witness testified. The expert identified only 12 pairs of voter records where incompatible birthplaces would eliminate the “soft match”—out of 4.7 million voter records. Opponents also point out that the databases used by Arizona county recorders do not currently use birthdate to find matches. This is not enough to succeed on a facial challenge. To begin, even a single valid circumstance showing the omission of a birthplace is enough to defeat a facial challenge. See Morales, 527 U.S. at 81 (Scalia, J., dissenting). Further, Arizona hasn’t been allowed to implement H.B. 2492. If birthplace information became mandatory, Arizona could alter how it collects and analyzes that information—advancing its use in the verification process. Instead, the district court relied on the state of affairs in Arizona as it existed before the law changed. On a facial challenge, we are not so backwards looking. Cf. Wash. State Grange, 552 U.S. at 450 (“The State has had no opportunity to implement [the challenged law], and its courts have had no occasion to construe the law in the context of actual disputes arising from the electoral context, or to accord the law a limiting construction to avoid constitutional questions.”). We thus resist facial challenges relying on “premature interpretations of statutes.” Id. (simplified). Second, an omitted birthplace could be material when an applicant submits a birth certificate as proof of citizenship that includes a last name different from the applicant’s 136 MI FAMILIA VOTA V. PETERSEN

current last name. In that case, Arizona’s 2023 Election Procedures Manual instructs county recorders to accept the birth certificate if the applicant’s first and middle names, birthplace, date of birth, and parents’ names match. Once again, omission of birthplace could be dispositive. In conclusion, an omitted birthplace can sometimes pose an obstacle to verifying an applicant’s identity. Opponents of the requirement thus fail to show that “no set of circumstances exists under which the [law] would be valid.” Moody, 603 U.S. at 723 (simplified). We should have lifted the injunction on this part of § 16-121.01(A). B. Citizenship-Checkbox Requirement The district court’s injunction of the citizenship- checkbox requirement is a different matter. First, all parties agree that the citizenship checkbox can help determine an applicant’s citizenship in some cases. The district court acknowledged that the checkbox could be material when an applicant submits no documentary proof of citizenship. It thus permitted Arizona to reject voter- registration applications for failure to check the citizenship box when no documentary proof of citizenship exists. Second, the district court only enjoined the checkbox requirement when two conditions are met: (1) the applicant has provided satisfactory proof of citizenship and (2) county recorders have otherwise established eligibility, including citizenship. Thus, the injunction applies only when there is no doubt about the applicant’s citizenship or eligibility. Third, nothing prevents Arizona from still using the checkbox and investigating applicants who skip it. The MI FAMILIA VOTA V. PETERSEN 137

injunction doesn’t prevent Arizona election officials from contacting applicants who neglected to check the box or asking them to correct the omission. And if investigation leads to other information indicating that the applicant is not the rightful bearer of the citizenship documents or that the person is otherwise ineligible, Arizona may still reject that applicant on those grounds. Once there’s a determination of ineligibility then the injunction simply doesn’t apply by its own terms. So in all cases, election officials may reject ineligible applicants. The injunction would, however, prevent officials from rejecting applicants for failing to check the citizenship box when those officials have already verified the applicant’s citizenship. Proponents of the requirement argue that enjoining the checkbox amounts to an anti-repetition rule—that States can’t enforce requests for duplicative information. They are correct that a sweeping rule against seeking duplicative information would be troubling. Sometimes a belt-and- suspenders approach is appropriate. But focus on this case— it’s hard to see how the failure to check the citizenship box could affect or influence the determination of the applicant’s citizenship when the applicant’s citizenship has already been verified. Rejecting a voter application for omitting a citizenship checkbox at the same time the applicant provides hard proof of citizenship seems more like dinging a voter for misspelling “Louisiana,” which falls into the heart of the Civil Rights Act. We thus properly affirm the injunction of this provision. 138 MI FAMILIA VOTA V. PETERSEN

VII “Reason To Believe” Provision H.B. 2243 requires Arizona county recorders to periodically search a registrant’s citizenship within the USCIS SAVE database if the county recorder has “reason to believe” the registrant is not a U.S. citizen. Ariz. Rev. Stat. § 16-165(I). The district court enjoined this provision under the “Different Standards, Practices, and Procedures” Provision of the Civil Rights Act. I agree with affirming the injunction. Under that provision,

No person acting under color of law shall[,] in determining whether any individual is qualified under State law or laws to vote in any election, apply any standard, practice, or procedure different from the standards, practices, or procedures applied under such law or laws to other individuals within the same county, parish, or similar political subdivision who have been found by State officials to be qualified to vote.

52 U.S.C. § 10101(a)(2)(A). The provision has a relatively straightforward command—election officials can’t use a “different . . . practice[] or procedure” for determining voter eligibility for different groups of “individuals” within the same political unit. Id. While the duty to verify citizenship through the SAVE database is reasonable enough, there’s a problem with it in practice—the SAVE database is only searchable for individuals with an immigration or A-File number. See Mi MI FAMILIA VOTA V. PETERSEN 139

Familia Vota II, 719 F. Supp. 3d at 995. That means that it only contains information about naturalized citizens. Thus, county recorders can only use the SAVE database for naturalized citizens—and never for natural-born citizens. So while the state law may be facially neutral, in “practice” or “procedure” it can be applied only in unequal ways. Say the county recorder has “reason to believe” two registrants are not U.S. citizens. One is a native-born registrant, who the recorder thinks is no longer a U.S. citizen (maybe, the registrant renounced his citizenship). The other is a naturalized citizen born out of the country. Under § 16- 165(I), only the latter can be subject to a SAVE check, meaning the naturalized citizen is subject to a “different . . . practice[] or procedure” than the natural-born citizen. Thus, I would affirm this portion of the district court’s injunction. VIII. Discriminatory-Purpose Challenge to Voter- Verification Laws Opponents of the Voting Laws also challenge the voter- verification laws under the Equal Protection Clause of the Fourteenth Amendment. They claim that the laws were enacted with discriminatory intent. The district court ruled against this challenge. Reviewing what’s known as the Arlington Heights factors, the district court found that these opponents hadn’t overcome the “strong presumption of good faith” we must afford to state legislatures. See United States v. Carrillo-Lopez, 68 F.4th 1133, 1140 (9th Cir. 2023) (simplified). It determined that the laws’ legislative history shows no “motive to discriminate against voters based on race or national origin” and that the laws have no discriminatory impact based on “naturalization status, race, 140 MI FAMILIA VOTA V. PETERSEN

or ethnicity.” See Mi Familia Vota II, 719 F. Supp. 3d at 1016. Not enough—the majority reverses the district court and all but finds that Arizona legislators enacted H.R. 2243 for a discriminatory purpose. In reversing the district court’s finding, the majority commits two errors. First, it neglects Article III standing doctrine. Only two non-profit organizations, Promise Arizona and Southwest Voter Registration Education Project, appeal the district court’s ruling. But neither organization has standing to bring this challenge. Second, the majority substitutes the district court’s factfinding for its own and lowers the evidentiary burden to the floor—flipping the strong presumption of good faith we give to legislative action and essentially requiring the State to disprove any discriminatory motive. A. Article III Standing Before reaching the merits, we must first decide whether the non-profits have Article III standing. See Mendoza v. Strickler, 51 F.4th 346, 354 n.5 (9th Cir. 2022). Organizations, like Promise Arizona and Southwest Voter Registration Education Project, can claim two paths to standing. See Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. (“SFFA”), 600 U.S. 181, 199 (2023) (simplified). The first path—known as “organizational standing”—is for the organization to show that it directly satisfies the Article III standing requirements. Id. The second path—known as “associational” or “representational standing”—is for it to assert “standing solely as the representative of its members.” Id. (simplified). Promise Arizona and Southwest Voter Registration MI FAMILIA VOTA V. PETERSEN 141

Education Project claim both paths. Neither leads them to standing. 1. Organizational Standing The Ninth Circuit long viewed organizational standing as “an ever-expanding universe.” See E. Bay Sanctuary Covenant v. Biden, 993 F.3d 640, 693 (9th Cir. 2021) (Bumatay, J., dissenting for the denial of rehearing en banc). Ignoring the traditional need for an injury in fact, we have continuously “loosen[ed] organizational standing requirements” to “increase our own authority to adjudicate policy disputes.” Id. Under our precedent, all an organization had to do was declare some voluntary “diversion of its resources” in response to a policy objection and it got a ticket into federal court. See, e.g., Fair Hous. Council of San Fernando Valley v. Roommate.com, LLC, 666 F.3d 1216, 1219 (9th Cir. 2012); id. at 1224 (Ikuta, J., dissenting in part). But a self- inflicted injury cannot establish standing. See Clapper v. Amnesty Int’l USA, 568 U.S. 398, 416 (2013) (“[R]espondents cannot manufacture standing merely by inflicting harm on themselves[.]”); Nat’l Fam. Plan. and Reprod. Health Ass’n, Inc. v. Gonzales, 468 F.3d 826, 831 (D.C. Cir. 2006) (“We have consistently held that self- inflicted harm doesn’t satisfy the basic requirements for standing.”); Pennsylvania v. New Jersey, 426 U.S. 660, 664 (1976) (per curiam) (“The injuries to the plaintiffs’ fiscs were self-inflicted . . . . No state can be heard to complain about damage inflicted by its own hand.”). The Supreme Court has finally declared enough is enough. In FDA v. Alliance for Hippocratic Medicine, 602 142 MI FAMILIA VOTA V. PETERSEN

U.S. 367 (2024), the Court reined in this expansive view of organizational standing. No longer will an organization’s “sincere legal, moral, ideological, and policy objections” to a law be sufficient to grant it Article III standing. See id. at 386. Now, an organization “cannot spend its way into standing simply by expending money to gather information and advocate against the defendant’s action.” Id. at 394. Nor can it “manufacture its own standing in that way.” Id. Instead, an organization may only assert standing when a challenged policy “directly affect[s] and interfere[s] with [its existing] core business activities.” Id. at 395; see also Ariz. All. for Retired Americans v. Mayes, 117 F.4th 1165, 1177 (9th Cir. 2024) (To confer organizational standing, “the organization must show that the new policy directly harms its already-existing core activities.”). Promise Arizona and Southwest Voter Registration Education Project assert that they are non-profit organizations seeking to empower Latino communities through their vote and increase their participation in the electoral process. To do this, they assist with voter registration, voter education, and turn-out-the-vote operations. In other words, their mission is to help Latinos navigate voting laws. To establish organizational standing, the organizations claim H.B. 2243 may cause them to reallocate resources to train staff and voters on the new voting laws, will require them to assist voters whose registration is erroneously cancelled, and might deter Latinos from registering to vote. In particular, they worry that H.B. 2243’s requirement for periodic verification of citizens might lead to inaccurate removal of eligible voters too close to an election to be corrected. They believe they may need to spend money to remedy this and to educate voters. MI FAMILIA VOTA V. PETERSEN 143

This is hardly an injury in fact to the organizations. It is nothing more than the diversion-of-resources theory of standing rejected in Alliance for Hippocratic Medicine. Simply, organizations can’t assert standing “based on their incurring costs to oppose” the voter-verifications laws. See All. for Hippocratic Med., 602 U.S. at 394 (holding that no organizational standing exists when organizations engage in “public advocacy” and “public education” on the effects of governmental action). At most, the new voter-verification laws may mean that the organizations will need to update their voter-registration operations—a completely voluntary move consistent with their mission. Such voluntary actions in no way interfere with their “core business activit[y]” of registering new voters. Id. at 395. Unlike “a retailer who sues a manufacturer for selling defective goods to the retailer,” these organizations are merely diverting resources to oppose a law they dislike. Id. “With or without” H.R. 2243, the non-profits “can still register and educate voters— in other words, continue their core activities that they have always engaged in.” Ariz. All. for Retired Americans, 117 F.4th at 1178. They can’t “attempt to spend their way into Article III standing by taking new actions in response to what they view as a disfavored policy.” Id. And they can’t manufacture standing based on their speculation that county recorders may erroneously reject voter applications. Standing isn’t based on a “highly attenuated chain of possibilities” premised on the presumption of erroneous actions by government officials. See Clapper, 568 U.S. at 410. Granting the organizations standing to challenge H.B. 2243 just because county recorders might make mistakes “would be an unprecedented and limitless approach and would allow [non-profits] to sue in federal court to challenge almost any policy affecting” 144 MI FAMILIA VOTA V. PETERSEN

voter registration. See All. for Hippocratic Med., 602 U.S. at 391–92. After all, “that is not what the law requires or what any county recorder would reasonably be expected to do.” Ariz. All. for Retired Americans, 117 F.4th at 1179. And it is even more speculative to claim that H.B. 2243 might injure the organizations’ “abstract social interest[]” in encouraging Latino-voter registration. See id. at 1177 (simplified). Thus, all these arguments amount to “a diversion-of-resources theory by another name.” Id. at 1180. 2. Associational Standing Associational standing doesn’t help the non-profit organizations either. To pursue associational standing, an organization must show that “(a) its members would otherwise have standing to sue in their own right; (b) the interests it seeks to protect are germane to the organization’s purpose; and (c) neither the claim asserted nor the relief requested requires the participation of individual members in the lawsuit.” All. for Hippocratic Medicine, 602 U.S. at 398 (Thomas, J., concurring) (simplified). We must be just as careful in granting organizations associational standing as well. Justice Thomas raises some valid concerns. First, “associational standing conflicts with Article III by permitting an association to assert its members’ injuries instead of its own.” Id. at 399. It does seem odd that we allow an association to “seek relief for its entire membership” when a single member suffers an injury—“even if the association has tens of millions of other, non-injured members.” Id. Likewise, “associational- standing doctrine does not appear to comport with the requirement that the plaintiff present an injury that the court can redress.” Id. at 400. If a single member has suffered an MI FAMILIA VOTA V. PETERSEN 145

injury, why then do we provide redress to the organization, which hasn’t sustained an injury itself? Anomalously, the actual injured party may not receive any relief himself. Thus, we mustn’t relax any standing requirements just because an organization presses a claim on behalf of an injured member. Here, Promise Arizona and Southwest Voter Registration Education Project fail to show that their “members would otherwise have standing to sue in their own right.” See SFFA, 600 U.S. at 199 (simplified). Promise Arizona claims 1,043 members, including an unspecified number of voters who are naturalized citizens. Promise Arizona hooks onto the H.B. 2243 provision that requires county recorders to conduct monthly SAVE checks on registered voters whom the county recorder has “reason to believe” are not U.S. citizens. Promise Arizona argues that its naturalized members will suffer an injury in fact if a SAVE check is run against them and if they are improperly removed from the voter rolls. The majority buys this argument—claiming that Promise Arizona’s members are in danger of losing the right to vote. This isn’t sufficient for associational standing. Promise Arizona has not plausibly alleged a “real and immediate threat of” future injury to its members. City of Los Angeles v. Lyons, 461 U.S. 95, 103 (1983). Rather, it only posits conjectural allegations of potential injuries that require a “long chain of hypothetical contingencies.” Lake v. Fontes, 83 F.4th 1199, 1204 (9th Cir. 2023) (per curiam) (simplified). First, Promise Arizona doesn’t specify how many naturalized members it has. All we know is that the number is between 2 and 1,043. So we are left to wonder what the 146 MI FAMILIA VOTA V. PETERSEN

chances are that one of its members will be subject to a SAVE check. Second, we must guess the possibility that a county recorder will somehow have “reason to believe” one of Promise Arizona’s naturalized members is not a U.S. citizen. Third, we must calculate the unlikely probability that the SAVE database will erroneously show that the naturalized member is not a U.S. citizen. Keep in mind that the district court found that the SAVE database is not “unreliable” and it doesn’t “contain[] severely inaccurate or outdated citizenship information.” Mi Familia Vota II, 719 F. Supp. 3d at 955. While the SAVE database can take one or two days to update, the district court found that Arizona has procedures to ensure that county recorders seek the latest information on citizenship. Id. 5 Fourth, we must predict the chances that the county recorder will not catch the error in citizenship for that naturalized member.

5 Promise Arizona doesn’t seem to assert an injury from the simple fact of a member’s name being run through the SAVE database. In any case, it’s hard to imagine what the injury would be if the SAVE database then confirms the member’s U.S. citizenship and nothing happens to the member’s voting status. Further, Promise Arizona doesn’t say how its member would find out about any database check and so its implausible that the check itself would lead to injury. To the extent that the member could assert some sort of “stigmatic injury” based on the database check, Promise Arizona will have to show much more. See, e.g., Allen v. Wright, 468 U.S. 737, 757 n.22 (1984) (A “stigmatic injury” demands “identification of some concrete interest with respect to which respondents are personally subject to discriminatory treatment.”). MI FAMILIA VOTA V. PETERSEN 147

Fifth, because Arizona law lets registrants correct any error, we must then presume that the naturalized member will not persuade the county recorder to fix the problem. And finally, we must then assess the likelihood that the naturalized member will be denied the vote because of all these hypothetical screw-ups. This is the kind of speculation that stretches the concept of imminence of harm beyond recognition. We can’t manufacture injury based on “conjecture about the behavior of other parties”—here, county recorders. Ecological Rts. Found. v. Pac. Lumber Co., 230 F.3d 1141, 1152 (9th Cir. 2000). Simply, Promise Arizona’s “conjectural allegations of potential injuries . . . are insufficient to plead a plausible real and immediate threat of” voter suppression. Lake, 83 F.4th at 1204 (simplified). As we recently said, Promise Arizona fails to “support[] a plausible inference that [its members’] individual votes in future elections will be adversely affected by” H.B. 2243, “particularly given the robust safeguards in Arizona law.” Id. Thus, Promise Arizona and Southwest Voter Registration Education Project can’t establish standing to appeal the equal protection claim against H.B. 2243. We should have ended the appeal here. B. Discriminatory Purpose Analysis While Promise Arizona and Southwest Voter Registration Education Project lack standing to raise this appeal, the majority disagrees and reaches the merits of the equal protection challenge. Unfortunately, they all but find discriminatory intent based on the weakest of evidence. Simply, the majority views any voter-verification 148 MI FAMILIA VOTA V. PETERSEN

requirements as discriminatory voter suppression. Because the majority decides the merits, I am compelled to address the serious flaws in its analysis. In seeking to overturn a duly enacted law based on a legislature’s discriminatory purpose, the plaintiff bears the burden to prove that purpose “by an evidentiary preponderance.” Carrillo-Lopez, 68 F.4that 1139 (simplified). In line with our respect for the separation of powers and federalism, we must accord a “strong presumption of good faith” to state legislative enactments. Id. at 1140 (simplified). Several non-exhaustive factors guide the inquiry:

(1) the impact of the official action and whether it bears more heavily on one race than another; (2) the historical background of the decision; (3) the specific sequence of events leading to the challenged action; (4) the defendant’s departures from normal procedures or substantive conclusions; and (5) the relevant legislative or administrative history.

Arce v. Douglas, 793 F.3d 968, 977 (9th Cir. 2015) (citing Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 266–68 (1977)). The discriminatory-purpose analysis demands a “sensitive inquiry into . . . circumstantial and direct evidence” of intent. Reno v. Bossier Par. Sch. Bd., 520 U.S. 471, 488 (1997) (quoting Arlington Heights, 429 U.S. at 266). We review the district court’s discriminatory-purpose finding for clear error. Brnovich, 594 U.S. at 687. If the district court’s finding was “plausible,” we “may not reverse MI FAMILIA VOTA V. PETERSEN 149

even if . . . [we] would have weighed the evidence differently in the first instance.” Id. “Where there are two permissible views of the evidence, the factfinder’s choice between them cannot be clearly erroneous.” Id. (simplified). The district court’s finding on discriminatory intent had ample support in the record. In the district court’s view, opponents of the law didn’t prove Arizona had a discriminatory purpose in enacting the voter-verification laws and H.B. 2243 based on several factual findings:

• While Arizona has a long-ago history of discriminating against people of color, opponents identified no “persuasive nexus between Arizona’s history of animosity toward marginalized communities and the Legislature’s enactment of the Voting Laws.” • Analysis of the legislative hearings “evince[s] [no] motive to discriminate against voters based on race or national origin.” • Any concern for non-citizens voting in elections doesn’t amount to “community animus” to “impute a discriminatory motive” to the Legislature. • Although the Free Enterprise Club, a major supporter of the voter-verification laws, used the term “illegals” in lobbying materials, no evidence showed that the Legislature relied on “coded appeals” or sought to “prevent anyone other than non-citizens from voting.” 150 MI FAMILIA VOTA V. PETERSEN

• One legislator’s allegedly discriminatory comments are not enough to impute intent to the “Arizona Legislature as a whole.” • Opponents “have not shown that the Voting Laws will have any significant discriminatory impact based on naturalization status, race, or ethnicity.” • At most, database checks will require only 0.001% of voters to produce documentary proof of citizenship. • Although H.B. 2243 was passed “abrupt[ly]” after the Arizona governor’s veto, it wasn’t “so abrupt” to show improper motive because related legislation was passed “through the ordinary legislative process.” • Arizona has had proof-of-citizenship requirements since 2005 and the provisions of H.R. 2243 “supplement” and “expand[]” on Arizona’s “existing practice[s].”

Mi Familia Vota II, 719 F. Supp. 3d at 1014–18 (simplified). Under the totality of the circumstances, the record is more than enough to support the district court’s finding of a lack of discriminatory purpose. Given our strong presumption of good faith, we have no basis to overturn the district court’s factual determination. Despite this thorough analysis, the majority grasps at straws to find some error. It settles on some odd notion that MI FAMILIA VOTA V. PETERSEN 151

the district court tried to “directly link” the evidence presented by the opponents of the law to “the motive of the Legislature.” Maj. Op. at 64. Although unclear, it seems the majority believes that the district court should have been more pliable to “circumstantial” evidence. See id. But the district court examined circumstantial evidence—it just found it unconvincing. While circumstantial evidence “may . . . be more certain, satisfying and persuasive than direct evidence” of discriminatory intent, Desert Palace, Inc. v. Costa, 539 U.S. 90, 100 (2003) (emphasis added) (simplified), circumstantial evidence must still convince us of animus—and it can fall short. See, e.g., Abbott, 585 U.S. at 610–11 (finding circumstantial evidence of quick passage of redistricting legislation unconvincing). And to be clear: at no point did the district court conclude that only direct evidence could suffice. It even stated explicitly that community animus, a form of circumstantial evidence, “can support a finding of discriminatory motives by government officials . . . .” Mi Familia Vota II, 719 F. Supp. 3d at 1016 (emphasis added) (quoting Ave. 6E Invs., LLC v. City of Yuma, Ariz., 818 F.3d 493, 504 (9th Cir. 2016)). So the majority’s differences with the district court, in the end, are factual. While the majority clearly would have found discriminatory intent here, our job is not to substitute our will for the factfinder’s. Start with the majority’s critique of the district court’s treatment of the “historical background” prong. The district court acknowledged that “Arizona does have a long history of discriminating against people of color,” but decided that this history was of “little probative value” because it was long ago—mostly up to the 1970s. Id. at 1014 (simplified). The district court thus found no “persuasive nexus” between this history and the enactment of H.B. 2243. Id. The 152 MI FAMILIA VOTA V. PETERSEN

majority attacks the district court for not considering how this history may be “circumstantial evidence” of discriminatory intent and calls the district court’s attempt to find any “nexus” an overly “onerous” inquiry. Maj. Op. 66. But the majority misunderstands the historical inquiry. By its nature, distant “history” is circumstantial evidence. After all, looking to past events—when current legislators weren’t alive, were infants, or not in office—must be circumstantial. Thus, distant incidents, dissimilar to current circumstances, offer only weak circumstantial evidence. As the Court has said, “unless historical evidence is reasonably contemporaneous with the challenged decision, it has little probative value.” McCleskey v. Kemp, 481 U.S. 279, 298 n.20 (1987). We can’t simply “accept official actions taken long ago as evidence of current intent.” Id. And so the district court’s weighing of the weak historical evidence was no clear error. The majority opinion gets even more baffling when it comes to legislative history. Again, the majority faults the district court for not analyzing the totality of the evidence. Maj. Op. at 66. But it’s the majority that cherry-picks events. The majority focuses on the fact that the Legislature conducted an audit that found no voter fraud as evidence that the voter-verification laws must have been a product of discriminatory intent. Id. It also relies on the Free Enterprise Club’s use of the word “illegals” to conclude the passage of the laws was racially motived. Id. at 69. The district court fully accounted for both facts. But reviewing the totality of the evidence, including the legislative hearings, public comments made about non-citizen voting, the Free Enterprise Club lobbying materials, and statements made by legislators, the district court found insufficient evidence to attribute animus to the Arizona Legislature as a whole. Mi MI FAMILIA VOTA V. PETERSEN 153

Familia Vota II, 719 F. Supp. 3d at 1014–16. Rather than conclude that the Arizona Legislature attempted to suppress voters after the 2020 election (as the majority does), the district court considered how legislators have long required proof of citizenship and how legislators wanted to revive the requirement after the Supreme Court seemed to open the door to it in ITCA. Id. at 1015 (citing ITCA, 570 U.S. at 12, 16). The district court also concluded that the other circumstantial evidence here—public concern over “illegals” voting, potentially “offensive” language in Free Enterprise Fund materials, and allegedly derogatory comments by a single state senator—failed to support an inference of discriminatory intent for the dozens of legislators in Arizona’s Legislature. Id. at 1015–16. Thus, the majority failed to look at the totality of the evidence when seeking to reverse the district court’s factual findings. Next, the majority relies on the accelerated passage of H.B. 2243 after the Governor’s veto to suggest improper motive. Maj. Op. at 71–72. But the majority discounted the fact that a related bill, H.B. 2617, had gone through the normal legislative process, because, in the majority’s view, the “amended bill contained many substantive changes.” Maj. Op. at 72. The district court explicitly considered the substance of H.B. 2243 and found it to be more of a “supplement” to valid existing laws than a stark departure indicative of discriminatory purpose. Mi Familia Vota II, 719 F. Supp. 3d at 1018. And the majority ignores that speed alone is poor evidence of animus. See Abbott, 585 U.S. at 610–11 (“[W]e do not see how the brevity of the legislative process can give rise to an inference of bad faith—and certainly not an inference that is strong enough to overcome the presumption of legislative good faith[.]”). 154 MI FAMILIA VOTA V. PETERSEN

Finally, the majority’s criticism of the district court’s “impact on a minority group” analysis is even more off base. The majority attacks the district court’s analysis as “troubling” for suggesting that “[e]vidence of a law’s disparate impact is generally insufficient alone to evidence a legislature’s discriminatory motive.” Maj. Op. 73 (quoting Mi Familia Vota II, 719 F. Supp. 3d at 1016). But there’s a problem with that. The district court was essentially paraphrasing our precedent. “[W]hile [d]isproportionate impact is not irrelevant,” we have said that “it is generally not dispositive, and there must be other evidence of a discriminatory purpose.” Carrillo-Lopez, 68 F.4th at 1141 (emphasis added) (simplified). And the majority ignores that the district court did consider impact of the laws on minorities. Perhaps because it doesn’t fit its narrative, the majority ignores that the district court found that “Plaintiffs have not shown that the Voting Laws will have any significant discriminatory impact.” Mi Familia Vota II, 719 F. Supp. 3d at 1016 (emphasis added). The district court continued on to find the other evidence of intent similarly unconvincing. See id. at 1016–17. So although the majority claims the district court wrongly “view[ed] evidence of the Voting Laws’ disparate impact alone” or “dispositve[ly],” it’s not clear what more the district court could have done. Maj. Op. at 72. In sum, the district court properly considered all relevant evidence, piece by piece, but ultimately concluded that the record only presented a weak array of circumstantial evidence. Because these findings are plausible, the majority is left to accuse the district court of “viewing each piece of evidence in isolation” and failing to consider the “totality of the circumstances.” Id. at 73. But this criticism is just sleight of hand. The district court did view the evidence in MI FAMILIA VOTA V. PETERSEN 155

context—and concluded that it was unpersuasive. Simply, the majority wants to equate any legislative action to prevent foreign citizens from voting in Arizona’s elections with evidence of discriminatory intent. In doing so, the majority essentially flips the strong presumption of good faith we grant to legislative action and requires the State to disprove any discriminatory motive. This is inconsistent with the law and the facts. IX. Waiver of Legislative Privilege The district court ordered Warren Petersen, President of the Arizona Senate, and Ben Toma, Speaker of the Arizona House of Representatives, to sit for depositions and produce privileged documents. On appeal, the Arizona legislators challenge these orders as violations of legislative privilege. See Lee v. City of Los Angeles, 908 F.3d 1175, 1187–88 (9th Cir. 2018) (explaining that “plaintiffs are generally barred from deposing local legislators, even in extraordinary circumstances”) (simplified). But their challenge is moot. Why? Because they have already complied with the discovery orders. So even if the district court were wrong to compel the legislators to provide evidence for trial, that trial already happened and a favorable ruling wouldn’t help the legislators. I understand that we denied the legislators the opportunity to appeal the order immediately and they faced sanctions if they didn’t comply with the district court’s order. But in law as in life, sometimes there are no “take backs.” It’s straightforward that “[c]ompliance with a discovery order renders moot an appeal of that order.” Richmark Corp. v. Timber Falling Consultants, 959 F.2d 1468, 1479 (9th Cir. 1992); see also Fraunhofer- Gesellschaft zur Förderung der angewandten Forschung 156 MI FAMILIA VOTA V. PETERSEN

E.V. v. Sirius XM Radio Inc., 59 F.4th 1319, 1322 (D.C. Cir. 2023) (noting this rule is the consensus rule of the circuits). Because we cannot redress the legislators’ injury, Article III bars us from hearing their claim. See Friends of the Earth, Inc. v. Laidlaw Env’t Servs. (TOC), Inc., 528 U.S. 167, 180 (2000). X. Conclusion I join the judgment on three issues. First, I agree with enjoining the “reason to believe” provision of Arizona Revised Statute § 16-165(I) under the Different Standards, Practices, and Procedures Provision of the Civil Rights Act, 52 U.S.C. § 10101(a)(2)(A). Second, I agree that the citizenship-checkbox requirement under Arizona Revised Statute § 16-121.01(A) violates the Materiality Provision of the Civil Rights Act, 52 U.S.C. § 10101(a)(2)(B), when accompanied by satisfactory proof of citizenship. And third, I agree that the appeal of the district court’s discovery order on Arizona’s legislative leaders is moot. I strongly disagree with the judgment on all other issues. Except as noted above, we should have vacated this sweeping injunction. I respectfully dissent.