948 F.3d 989
United States Court of Appeals, Ninth Circuit.
The DEMOCRATIC NATIONAL COMMITTEE; DSCC, aka Democratic Senatorial Campaign Committee; The Arizona Democratic Party, Plaintiffs-Appellants,
v.
Katie HOBBS, in her official capacity as Secretary of State of Arizona; Mark Brnovich, Attorney General, in his official capacity as Arizona Attorney General, Defendants-Appellees,
The Arizona Republican Party; Bill Gates, Councilman; Suzanne Klapp, Councilwoman; Debbie Lesko, Sen.; Tony Rivero, Rep., Intervenor-Defendants-Appellees.
No. 18-15845
|
Argued and Submitted En Banc March 27, 2019 San Francisco, California
|
Filed January 27, 2020
OPINION
W. FLETCHER, Circuit Judge:
The right to vote is the foundation of our democracy. Chief Justice Warren wrote in his autobiography that the precursor to one person, one vote, Baker v. Carr, 369 U.S. 186, 82 S.Ct. 691, 7 L.Ed.2d 663 (1962), was the most important case decided during his tenure as Chief Justice—a tenure that included Brown v. Board of Education, 347 U.S. 483, 74 S.Ct. 686, 98 L.Ed. 873 (1954). Earl Warren, The Memoirs of Earl Warren 306 (1977). Chief Justice Warren wrote in Reynolds v. Sims, 377 U.S. 533, 555, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964): “The right to vote freely for the candidate of one’s choice is of the essence of a democratic society, and any restrictions on that right strike at the heart of representative government.” Justice Black wrote in Wesberry v. Sanders, 376 U.S. 1, 17, 84 S.Ct. 526, 11 L.Ed.2d 481 (1964): “No right is more precious in a free country than that of having a voice in the election of those who make the laws under which, as good citizens, we must live. Other rights, even the most basic, are illusory if the right to vote is undermined.”
For over a century, Arizona has repeatedly targeted its American Indian, Hispanic, and African American citizens, limiting or eliminating their ability to vote and to participate in the political process. In 2016, the Democratic National Committee and other Plaintiffs-Appellants (collectively, “DNC” or “Plaintiffs”) sued Arizona’s Secretary of State and Attorney General in their official capacities (collectively, “Arizona”) in federal district court.
DNC challenged, first, Arizona’s policy of wholly discarding, rather than counting or partially counting, ballots cast in the wrong precinct (“out-of-precinct” or “OOP” policy); and, second, House Bill 2023 (“H.B. 2023”), a 2016 statute criminalizing the collection and delivery of another person’s ballot. DNC contends that the OOP policy and H.B. 2023 violate Section 2 of the Voting Rights Act of 1965 as amended (“VRA”) because they adversely and disparately affect Arizona’s American Indian, Hispanic, and African American citizens. DNC also contends that H.B. 2023 violates Section 2 of the VRA and the Fifteenth Amendment to the United States Constitution because it was enacted with discriminatory intent. Finally, DNC contends that the OOP policy and H.B. 2023 violate the First and Fourteenth Amendments because they unduly burden minorities’ right to vote.
Following a ten-day bench trial, the district court found in favor of Arizona on all claims. Democratic Nat’l Comm. v. Reagan, 329 F. Supp. 3d 824 (D. Ariz. 2018) (Reagan). DNC appealed, and a divided three-judge panel of our court affirmed. Democratic Nat’l Comm. v. Reagan, 904 F.3d 686 (9th Cir. 2018) (DNC). A majority of non-recused active judges voted to rehear this case en banc, and we vacated the decision of the three-judge panel. Democratic Nat’l Comm. v. Reagan, 911 F.3d 942 (9th Cir. 2019).
We review the district court’s conclusions of law de novo and its findings of fact for clear error. Gonzalez v. Arizona, 677 F.3d 383, 406 (9th Cir. 2012) (en banc). We may “correct errors of law, including those that may infect a so-called mixed finding of law and fact, or a finding of fact that is predicated on a misunderstanding of the governing rule of law.” Thornburg v. Gingles, 478 U.S. 30, 79, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986) (internal quotation marks omitted); see Smith v. Salt River Project Agric. Improvement & Power Dist., 109 F.3d 586, 591 (9th Cir. 1997) (Salt River). We review for clear error the district court’s overall finding of vote dilution or vote denial in violation of the VRA. Gingles, 478 U.S. at 78, 106 S.Ct. 2752; Salt River, 109 F.3d at 591.
Reviewing the full record, we conclude that the district court clearly erred. We reverse the decision of the district court. We hold that Arizona’s policy of wholly discarding, rather than counting or partially counting, out-of-precinct ballots, and H.B. 2023’s criminalization of the collection of another person’s ballot, have a discriminatory impact on American Indian, Hispanic, and African American voters in Arizona, in violation of the “results test” of Section 2 of the VRA. We hold, further, that H.B. 2023’s criminalization of the collection of another person’s ballot was enacted with discriminatory intent, in violation of the “intent test” of Section 2 of the VRA and of the Fifteenth Amendment. We do not reach DNC’s First and Fourteenth Amendment claims.
I. Out-of-Precinct Policy and H.B. 2023
DNC challenges (1) Arizona’s policy of wholly discarding, rather than counting or partially counting, ballots cast out-of-precinct (“OOP”), and (2) H.B. 2023, a statute that, subject to certain exceptions, criminalizes the collection of another person’s early ballot. SeeAriz. Rev. Stat. §§ 16-122, -135, -584; H.B. 2023, 52nd Leg., 2d Reg. Sess. (Ariz. 2016), codified asAriz. Rev. Stat. § 16-1005(H), (I).
Arizona offers two methods of voting: (1) in-person voting at a precinct or vote center either on election day or during an early-vote period, or (2) “early voting” whereby the voter receives the ballot via mail and either mails back the voted ballot or delivers the ballot to a designated drop-off location. Arizona’s OOP policy affects in-person voting. H.B. 2023 affects early voting.
We describe in turn Arizona’s OOP policy and H.B. 2023.
A. Out-of-Precinct Policy
1. Policy of Entirely Discarding OOP Ballots
Arizona law permits each county to choose a vote-center or a precinct-based system for in-person voting. Reagan, 329 F. Supp. 3d at 840. In counties using the vote-center system, registered voters may vote at any polling location in the county. Id. In counties using the precinct-based system, registered voters may vote only at the designated polling place in their precinct. Approximately 90 percent of Arizona’s population lives in counties using the precinct-based system.
In precinct-based counties, if a voter arrives at a polling place and does not appear on the voter rolls for that precinct, that voter may cast a provisional ballot. Id.; Ariz. Rev. Stat. §§ 16-122, -135, -584. After election day, county election officials in close elections review all provisional ballots to determine the voter’s identity and address. If, after reviewing a provisional ballot, election officials determine that the voter voted out of precinct, the county discards the OOP ballot in its entirety. In some instances, all of the votes cast by the OOP voter will have been cast for candidates and propositions for which the voter was legally eligible to vote. In other instances, most of the votes cast by the OOP voter will have been cast properly, in the sense that the voter was eligible to vote on those races, but one or more votes for local candidates or propositions will have been cast improperly.
In both instances, the county discards the OOP ballot in its entirety. Reagan, 329 F. Supp. 3d at 840. That is, the county discards not only the votes of an OOP voter for the few local candidates and propositions for which the OOP voter may have been ineligible to vote. The county also discards the votes for races for which the OOP voter was eligible to vote, including U.S. President, U.S. Senator, and (almost always) Member of the U.S. House of Representatives; all statewide officers, including Governor, and statewide propositions; (usually) all countywide officers and propositions; and (often) local candidates and propositions.
2. Comparison with Other States
The district court found that Arizona “consistently is at or near the top of the list of states that collect and reject the largest number of provisional ballots each election.” Id. at 856 (emphasis added). The district court’s finding understates the matter. Arizona is consistently at the very top of the list by a large margin.
Dr. Jonathan Rodden, Professor of Political Science and Senior Fellow at the Hoover Institution at Stanford University, provided expert reports to the district court. The court gave “great weight” to Dr. Rodden’s analysis of the “rates and causes of OOP voting” in Arizona. Id. at 835. Dr. Rodden reported: “Since 2012, Arizona has clearly become the national leader in both provisional ballots cast and especially in provisional ballots rejected among in-person voters.” Jonathan Rodden, Expert Report (Rodden) at 25.
Dr. Rodden reported that, from 2006 to 2010, between 9 to 13 percent of all in-person ballots cast in Arizona were provisional ballots. Id. at 24. In the 2012 general election, more than 22 percent of all in-person ballots cast were provisional ballots. Id. In Maricopa County, Arizona’s most populous county, close to one in three in-person ballots cast in 2012 were provisional ballots. Id. at 27–28. In the 2014 midterm election, over 18 percent of in-person ballots cast in the State were provisional ballots. Id. at 25. These numbers place Arizona at the very top of the list of States in collection of provisional ballots.
Arizona also rejects a higher percentage of provisional ballots than any other State. The district court found:
In 2012 alone “[m]ore than one in every five [Arizona in-person] voters ... was asked to cast a provisional ballot, and over 33,000 of these—more than 5 percent of all in-person ballots cast—were rejected. No other state rejected a larger share of its in-person ballots in 2012.”
Reagan, 329 F. Supp. 3d at 856 (alterations in original) (quoting Rodden at 24–25).
One of the most frequent reasons for rejecting provisional ballots in Arizona is that they are cast out-of-precinct. Id.; see also Rodden at 26–29. From 2008 to 2016, Arizona discarded a total of 38,335 OOP ballots cast by registered voters—29,834 ballots during presidential general elections, and 8,501 ballots during midterm general elections. Reagan, 329 F. Supp. 3d at 856.
Rodden at 26. The percentage of rejected OOP votes in Arizona is eleven times that in Washington, the State with the second-highest percentage.
The percentage of OOP ballots in Arizona, compared to all ballots cast, has declined in recent years. But the percentage of in-person ballots cast, compared to all ballots cast, has declined even more. See Jonathan Rodden, Rebuttal Report (Rodden Rebuttal) at 10. As a result, as a percentage of in-person ballots between 2008 and 2014, the percentage of OOP ballots has increased.
3. Reasons for OOP Ballots
Three key factors leading to OOP ballots are frequent changes in polling locations; confusing placement of polling locations; and high rates of residential mobility. These factors disproportionately affect minority voters. Dr. Rodden summarized:
Voters must invest significant effort in order to negotiate a dizzying array of precinct and polling place schemes that change from one month to the next. Further, Arizona’s population is highly mobile and residential locations are fluid, especially for minorities, young people, and poor voters, which further contributes to confusion around voting locations.
Rodden at 2; see also Reagan, 329 F. Supp. 3d at 857–58 (discussing these reasons).
a. Frequent Changes in Polling Locations
Arizona election officials change voters’ assigned polling places with unusual frequency. Maricopa County, which includes Phoenix, is a striking example. The district court found that between 2006 and 2008, “at least 43 percent of polling locations” changed. Reagan, 329 F. Supp. 3d at 858. Between 2010 and 2012, approximately 40 percent of polling place locations were changed again. Id. These changes continued in 2016, “when Maricopa County experimented with 60 vote centers for the presidential preference election [in March], then reverted to a precinct-based system with 122 polling locations for the May special election, and then implemented over 700 assigned polling places [for] the August primary and November general elections.” Id. The OOP voting rate was 40 percent higher for voters whose polling places were changed. Id. As Chief Judge Thomas put it, “the paths to polling places in the Phoenix area [are] much like the changing stairways at Hogwarts, constantly moving and sending everyone to the wrong place.” DNC, 904 F.3d at 732 (Thomas, C.J., dissenting).
White voters in Maricopa County are more likely than minority voters to have continuity in their polling place location. Rodden at 60–61. Dr. Rodden wrote that between the February and November elections in 2012, “the rates at which African Americans and Hispanics experienced stability in their polling places were each about 30 percent lower than the rate for whites.” Id.
b. Confusing Placement of Polling Locations
Some polling places are located so counterintuitively that voters easily make mistakes. In Maricopa and Pima Counties, many polling places are located at or near the edge of precincts. Id. at 50. An example is the polling place for precinct 222 in Maricopa County during the 2012 election. Dr. Rodden wrote:
[A] group of 44 voters who were officially registered to vote in precinct 222, ... showed up on Election Day at the Desert Star School, the polling location for precinct 173. It is easy to understand how they might have made this mistake. Polling place 173 is the local elementary school, and the only polling place in the vicinity. It is within easy walking distance, and is the polling place for most of the neighbors and other parents at the school, yet due to a bizarre placement of the [polling place at the] Southern border of precinct 222, these voters were required to travel 15 minutes by car (according to [G]oogle maps) to vote in polling location 222, passing four other polling places along the way.
Id. at 47–48.
This map illustrates Dr. Rodden’s point: (see map on Westlaw)
In 2012, approximately 25 percent of OOP voters lived closer to the polling place where they cast their OOP ballot than to their assigned polling place. Id. at 53. Voters who live more than 1.4 miles from their assigned polling place are 30 percent more likely to vote OOP than voters who live within 0.4 miles of their assigned polling place. Id. at 54. American Indian and Hispanic voters live farther from their assigned polling places than white voters. Id. at 60. American Indian voters are particularly disadvantaged. The district court found: “Navajo voters in Northern Apache County lack standard addresses, and their precinct assignments for state and county elections are based upon guesswork, leading to confusion about the voter’s correct polling place.” Reagan, 329 F. Supp. 3d at 873; Rodden Second at 52–53.
c. Renters and Residential Mobility
High percentages of renters and high rates of residential mobility correlate with high rates of OOP voting. Reagan, 329 F. Supp. 3d at 857. The district court found that rates of OOP voting are “higher in neighborhoods where renters make up a larger share of householders.” Id. Between 2000 and 2010, almost 70 percent of Arizonans changed their residential address, the second highest rate of any State. Reagan, 329 F. Supp. 3d at 857; Rodden at 11–12. The district court found that “[t]he vast majority of Arizonans who moved in the last year moved to another address within their current city of residence.” Reagan, 329 F. Supp. 3d at 857.
The need to locate the proper polling place after moving—particularly after moving a short distance in an urban area—leads to a high percentage of OOP ballots. Dr. Rodden wrote:
An individual who faces a rent increase in one apartment complex and moves to another less than a mile away might not be aware that she has moved into an entirely new precinct—indeed, in many cases ... she may still live closest to her old precinct, but may now be required to travel further in order to vote in her new assigned precinct. Among groups for whom residential mobility is common, requirements of in-precinct-voting—as well as the requirement that they update their registration with the state every time that they move even a short distance within a county—can make it substantially more burdensome to participate in elections.
Rodden at 11.
The district court found that minority voters in Arizona have “disproportionately higher rates of residential mobility.” Reagan, 329 F. Supp. 3d at 872. The court found, “OOP voting is concentrated in relatively dense precincts that are disproportionately populated with renters and those who move frequently. These groups, in turn, are disproportionately composed of minorities.” Id.
4. Disparate Impact on Minority Voters
The district court found that Arizona’s policy of wholly discarding OOP ballots disproportionately affects minority voters. Reagan, 329 F. Supp. 3d at 871. During the general election in 2012 in Pima County, compared to white voters, the rate of OOP ballots was 123 percent higher for Hispanic voters, 47 percent higher for American Indian voters, and 37 percent higher for African American voters. Rodden at 43. During the 2014 and 2016 general elections in Apache, Navajo, and Coconino Counties, the vast majority of OOP ballots were in areas that are almost entirely American Indian. Rodden Rebuttal at 53–54, 58; Jonathan Rodden, Second Expert Report (Rodden Second) at 22. In all likelihood, the reported numbers underestimate the degree of disparity. Dr. Rodden wrote, “[A]lthough the racial disparities described ... are substantial, they should be treated as a conservative lower bound on the true differences in rates of out-of-precinct voting across groups.” Rodden Second at 15 (emphasis in original). The district court found, “Dr. Rodden credibly explained that the measurement error for Hispanic probabilities leads only to the under-estimation of racial disparities.” Reagan, 329 F. Supp. 3d at 838.
Racial disparities in OOP ballots in 2016 “remained just as pronounced” as in 2012 and 2014. Rodden Second at 3. For example, the rates of OOP ballots in Maricopa County “were twice as high for Hispanics, 86 percent higher for African Americans, and 73 percent higher for Native Americans than for their non-minority counterparts.” Reagan, 329 F. Supp. 3d at 871–72; Rodden Second at 29. “In Pima County, rates of OOP voting were 150 percent higher for Hispanics, 80 percent higher for African Americans, and 74 percent higher for Native Americans than for non-minorities.” Reagan, 329 F. Supp. 3d at 872. “[I]n Pima County the overall rate of OOP voting was higher, and the racial disparities larger, in 2016 than in 2014.” Id.; Rodden Second at 33.
The district court found:
Among all counties that reported OOP ballots in the 2016 general election, a little over 1 in every 100 Hispanic voters, 1 in every 100 African-American voters, and 1 in every 100 Native American voters cast an OOP ballot. For non-minority voters, the figure was around 1 in every 200 voters.
Reagan, 329 F. Supp. 3d at 872. That is, in the 2016 general election, as in the two previous elections, American Indians, Hispanics, and African Americans voted OOP at twice the rate of whites.
B. H.B. 2023
1. Early Voting and Ballot Collection
Arizona has permitted early voting for over 25 years. Id. at 839. “In 2007, Arizona implemented permanent no-excuse early voting by mail, known as the Permanent Early Voter List (“PEVL”).” Id. Under PEVL, Arizonans may either (a) request an early vote-by-mail ballot on an election-by-election basis, or (b) request that they be placed on the Permanent Early Voter List. See id.; Ariz. Rev. Stat. §§ 16-542, -544. Some counties permit voters to drop their early ballots in special drop boxes. All counties permit the return of early ballots by mail, or in person at a polling place, vote center, or authorized election official’s office. Early voting is by far “the most popular method of voting [in Arizona].” Reagan, 329 F. Supp. 3d at 839. Approximately 80 percent of all ballots cast in the 2016 general election were early ballots. Id. Until the passage of H.B. 2023, Arizona did not restrict collection and drop-off of voted ballots by third parties.
The district court heard extensive testimony about the number of ballots collected and turned in by third parties. Id. at 845. A Maricopa County Democratic Party organizer testified that during the course of her work for the party she personally saw 1,200 to 1,500 early ballots collected and turned in by third-party volunteers. These were only a portion of the total ballots collected by her organization. The organizer testified that during the 2010 election the Maricopa County Democratic Party collected hundreds of ballots from a heavily Hispanic neighborhood in one state legislative district alone. A representative of Citizens for a Better Arizona testified that the organization collected approximately 9,000 early ballots during the 2012 Maricopa County Sheriff’s election. A member of the Arizona Democratic Party testified that the party collected “a couple thousand ballots” in 2014. Id. A community advocate testified before the Arizona Senate Elections Committee that in one election he collected 4,000 early ballots. Id. A Phoenix City Councilmember testified that she and her volunteers collected about 1,000 early ballots in an election in which she received a total of 8,000 votes.
2. Minority Voters’ Reliance on Third-Party Ballot Collection
The district court found “that prior to H.B. 2023’s enactment minorities generically were more likely than non-minorities to return their early ballots with the assistance of third parties.” Id. at 870. The court recounted: “Helen Purcell, who served as the Maricopa County Recorder for 28 years from 1988 to 2016, observed that ballot collection was disproportionately used by Hispanic voters.” Id. Individuals who collected ballots in past elections “observed that minority voters, especially Hispanics, were more interested in utilizing their services.” Id. One ballot collector testified about what she termed a “case study” demonstrating the extent of the disparity. In 2010, she and her fellow organizers collected “somewhere south of 50 ballots” in one area. The area was later redistricted before the next election to add the heavily Hispanic neighborhood of Sunnyslope. In 2012, the organization “pulled in hundreds of ballots, [with the] vast majority from that Sunnyslope area.”
The district court found that, in contrast, the Republican Party has “not significantly engaged in ballot collection as a GOTV [Get Out the Vote] strategy.” Id. The base of the Republican Party in Arizona is white. Id. Individuals who engaged in ballot collection in past elections observed that voters in predominately white areas “were not as interested in ballot collection services.” Id.
Minority voters rely on third-party ballot collection for many reasons. Joseph Larios, a community advocate who has collected ballots in past elections, testified that “returning early mail ballots presents special challenges for communities that lack easy access to outgoing mail services; the elderly, homebound, and disabled voters; socioeconomically disadvantaged voters who lack reliable transportation; voters who have trouble finding time to return mail because they work multiple jobs or lack childcare services; and voters who are unfamiliar with the voting process and therefore do not vote without assistance or tend to miss critical deadlines.” Id. at 847–48 (summarizing Larios’ testimony). These burdens fall disproportionately on Arizona’s minority voters.
Arizona’s American Indian and Hispanic communities frequently encounter mail-related problems that make returning early ballots difficult. In urban areas of heavily Hispanic counties, many apartment buildings lack outgoing mail services. Id. at 869. Only 18 percent of American Indian registered voters have home mail service. Id. White registered voters have home mail service at a rate over 350 percent higher than their American Indian counterparts. Id. Basic mail security is an additional problem. Several witnesses testified that incoming and outgoing mail often go missing. Id. The district court found that especially in low-income communities, frequent mail theft has led to “distrust” in the mail service. Id.
A lack of transportation compounds the issue. “Hispanics, Native Americans, and African Americans ... are significantly less likely than non-minorities to own a vehicle, more likely to rely upon public transportation, [and] more likely to have inflexible work schedules[.]” Id. In San Luis—a city that is 98 percent Hispanic—a major highway separates almost 13,000 residents from their nearest post office. Id. The city has no mass transit, a median income of $22,000, and many households with no cars. Id. On the Navajo Reservation, “most people live in remote communities, many communities have little to no vehicle access, and there is no home incoming or outgoing mail, only post office boxes, sometimes shared by multiple families.” Id. “[R]esidents of sovereign nations often must travel 45 minutes to 2 hours just to get a mailbox.” DNC, 904 F.3d at 751–52 (Thomas, C.J., dissenting). As a result, voting “requires the active assistance of friends and neighbors” for many American Indians. Reagan, 329 F. Supp. 3d at 870 (quoting Rodden Second at 60).
The adverse impact on minority communities is substantial. Without “access to reliable and secure mail services” and without reliable transportation, many minority voters “prefer instead to give their ballots to a volunteer.” Id. at 869. These communities thus end up relying heavily on third-party collection of mail-in ballots. Dr. Berman wrote with respect to Hispanic voters:
[T]he practice of collecting ballots, used principally in Hispanic areas, ha[s] contributed to more votes being cast in those places tha[n] would have been cast without the practice. ... That the practice has increased minority turnout appears to have been agreed upon or assumed by both sides of the issue[.] Democrats and Hispanic leaders have seen reason to favor it, Republicans have not.
Berman, Expert Reply Report at 8–9. Similarly, LeNora Fulton, a member of the Navajo Nation and previous Apache County Recorder, testified that it was “standard practice” in Apache County and the Nation to vote by relying on non-family members with the means to travel. Reagan, 329 F. Supp. 3d at 870.
3. History of H.B. 2023
Before the passage of H.B. 2023, Arizona already criminalized fraud involving possession or collection of another person’s ballot. The district court wrote:
[B]allot tampering, vote buying, or discarding someone else’s ballot all were illegal prior to the passage of H.B. 2023. Arizona law has long provided that any person who knowingly collects voted or unvoted ballots and does not turn those ballots in to an elections official is guilty of a class 5 felony. A.R.S. § 16-1005. Further, Arizona has long made all of the following class 5 felonies: “knowingly mark[ing] a voted or unvoted ballot or ballot envelope with the intent to fix an election;” “receiv[ing] or agree[ing] to receive any consideration in exchange for a voted or unvoted ballot;” possessing another’s voted or unvoted ballot with intent to sell; “knowingly solicit[ing] the collection of voted or unvoted ballots by misrepresenting [one’s self] as an election official or as an official ballot repository or ... serv[ing] as a ballot drop off site, other than those established and staffed by election officials;” and “knowingly collect[ing] voted or unvoted ballots and ... not turn[ing] those ballots in to an election official ... or any ... entity permitted by law to transmit post.” A.R.S. §§ 16-1005(a)–(f). The early voting process also includes a number of other safeguards, such as tamper evident envelopes and a rigorous voter signature verification procedure.
Reagan, 329 F. Supp. 3d at 854 (alterations in original) (internal record citations omitted).
There is no evidence of any fraud in the long history of third-party ballot collection in Arizona. Despite the extensive statutory provisions already criminalizing fraud involving possession or collection of another person’s ballot, and despite the lack of evidence of any fraud in connection with third-party ballot collection, Republican State Senator Don Shooter introduced a bill in February 2011. S.B. 1412, 50th Leg., 1st Reg. Sess. (introduced) (Ariz. 2011), http://www.azleg.gov/legtext/50leg/1r/bills/sb1412p.htm.
Senator Shooter’s bill criminalized non-fraudulent third-party ballot collection. The district court had no illusions about Senator Shooter’s motivation. It found:
Due to the high degree of racial polarization in his district, Shooter was in part motivated by a desire to eliminate what had become an effective Democratic GOTV strategy. Indeed, Shooter’s 2010 election was close: he won with 53 percent of the total vote, receiving 83 percent of the non-minority vote but only 20 percent of the Hispanic vote.
Reagan, 329 F. Supp. 3d at 879–80.
The state legislature amended Senator Shooter’s bill several times, watering it down significantly. As finally enacted, the bill—included as part of a series of election-related changes in Senate Bill 1412 (“S.B. 1412”)—restricted the manner in which unrelated third parties could collect and turn in more than ten voted ballots. S.B. 1412, 50th Leg., 1st Reg. Sess. (engrossed), Sec. 3 at D (Ariz. 2011), https://legiscan.com/AZ/text/SB1412/id/233492/Arizona2011-SB1412-Engrossed.html. If a third-party ballot collector turned in more than ten ballots, the collector was required to provide photo identification. After each election, the Secretary of State was required to compile a statewide public report listing ballot collectors’ information. The bill did not criminalize any violation of its provisions.
When S.B. 1412 became law, Arizona was still subject to preclearance under the Voting Rights Act. S.B. 1412 therefore could not go into effect until it was precleared by the U.S. Department of Justice (“DOJ”) or a three-judge federal district court. On May 18, 2011, the Arizona Attorney General submitted S.B. 1412 to DOJ for preclearance. Arizona Attorney General Thomas Horne, Effect of Shelby County on Withdrawn Preclearance Submissions, (August 29, 2013), https://www.azag.gov/opinions/i13-008-r13-013. On June 27, 2011, DOJ precleared all provisions of S.B. 1412 except the provision regulating third-party ballot collection. Reagan, 329 F. Supp. 3d at 880.
DOJ sent a letter to Arizona concerning the third-party ballot collection provision, stating that the information provided with the preclearance request was “insufficient to enable [DOJ] to determine that the proposed changes have neither the purpose nor will have the effect of denying or abridging the right to vote on account of race, color, or membership in a language minority group.” Id. at 880–81. DOJ requested additional information and stated that it “may object” to the proposed change if no response was received within sixty days. Id. at 881.
Instead of responding with the requested information, the Arizona Attorney General withdrew the preclearance request for the third-party ballot collection provision. Id. The Attorney General did so for good reason. According to DOJ records, Arizona’s Elections Director, who had helped draft the provision, had admitted to DOJ that the provision was “targeted at voting practices in predominantly Hispanic areas.”
The state legislature formally repealed the provision after receiving the letter from DOJ. Withdrawing a preclearance request was not common practice in Arizona. Out of 773 proposals that Arizona submitted for preclearance over almost forty years, the ballot collection provision of S.B. 1412 was one of only six that Arizona withdrew. Id.
Two years later, on June 25, 2013, the United States Supreme Court decided Shelby County v. Holder, 570 U.S. 529, 133 S.Ct. 2612, 186 L.Ed.2d 651 (2013). The Court declared unconstitutional the formula in Section 4(b) of the VRA for determining “covered jurisdictions,” thereby eliminating preclearance under Section 5 for any previously covered jurisdiction, including Arizona. On June 19, 2013, Arizona’s Governor had signed a new bill, H.B. 2305, which entirely banned partisan ballot collection and required non-partisan ballot collectors to complete an affidavit stating that they had returned the ballot. Reagan, 329 F. Supp. 3d at 881; H.B. 2305, 51st Leg., 1st Reg. Sess. (engrossed), at Secs. 3 and 5 (Ariz. 2013), https://legiscan.com/AZ/text/HB2305/id/864002. Violation of H.B. 2305 was a criminal misdemeanor.
H.B. 2305 “was passed along nearly straight party lines in the waning hours of the legislative session.” Reagan, 329 F. Supp. 3d at 881. “Shortly after its enactment, citizen groups organized a referendum effort[.]” Id. They “collected more than 140,000 signatures”—significantly more than the required amount—“to place H.B. 2305 on the ballot for a straight up-or-down [statewide] vote” in the next election. Id. Arizona law provided that repeal by referendum prevented the legislature from enacting future related legislation without a supermajority vote. Moreover, any such future legislation could only “further[ ]”—not undercut—“the purposes” of the referendum. Ariz. Const. art. IV, pt. 1, § 1(6)(C), (14). “Rather than face a referendum, Republican legislators ... repealed their own legislation along party lines.” Reagan, 329 F. Supp. 3d at 881. The primary sponsor of H.B. 2305, then-State Senator Michele Reagan (a future Secretary of State of Arizona and an original defendant in this action), “admitted that the legislature’s goal [in repealing H.B. 2305] was to break the bill into smaller pieces and reintroduce individual provisions ‘a la carte.’ ” Id.
During the 2015 and 2016 legislative sessions, Republican legislators again sought to criminalize ballot collection by third parties, culminating in 2016 in the passage of H.B. 2023, the measure challenged in this suit. The district court found that Republican legislators had two motivations for passing H.B. 2023. First, Republican legislators were motivated by the “unfounded and often farfetched allegations of ballot collection fraud” made by former State Senator Shooter—who had introduced the bill to limit third-party ballot collection in 2011. Id. at 880 (finding Shooter’s allegations “demonstrably false”). Second, Republican legislators were motivated by a “racially-tinged” video known as the “LaFaro Video.” Id.
The video gave proponents of H.B. 2023 their best and only “evidence” of voter fraud. During legislative hearings on previous bills criminalizing third-party collection, the district court wrote, “Republican sponsors and proponents [had] expressed beliefs that ballot collection fraud regularly was occurring but struggled with the lack of direct evidence substantiating those beliefs.” Id. at 876. In 2014, Republicans’ “perceived ‘evidence’ arrived in the form of a racially charged video created by Maricopa County Republican Chair A.J. LaFaro ... and posted on a blog.” Id. The court summarized:
The LaFaro Video showed surveillance footage of a man of apparent Hispanic heritage appearing to deliver early ballots. It also contained a narration of “Innuendos of illegality ... [and] racially tinged and inaccurate commentary by ... LaFaro.” LaFaro’s commentary included statements that the man was acting to stuff the ballot box; that LaFaro did not know if the person was an illegal alien, a dreamer, or citizen, but knew that he was a thug; and that LaFaro did not follow him out to the parking lot to take down his tag number because he feared for his life.
Id. (alterations in original and internal record citations omitted). A voice-over on the video described “ballot parties” where people supposedly “gather en mass[e] and give their un-voted ballots to operatives of organizations so they can not only collect them, but also vote them illegally.” Id. at 876–77.
The district court found, “The LaFaro Video did not show any obviously illegal activity and there is no evidence that the allegations in the narration were true.” Id. at 877. The video “merely shows a man of apparent Hispanic heritage dropping off ballots and not obviously violating any law.” Id. The video “became quite prominent in the debates over H.B. 2023.” Id. The court wrote:
The LaFaro video also was posted on Facebook and YouTube, shown at Republican district meetings, and was incorporated into a television advertisement—entitled “Do You Need Evidence Terry?”—for Secretary Reagan when she ran for Secretary of State. In the ad, the LaFaro Video plays after a clip of then-Arizona Attorney General Terry Goddard stating he would like to see evidence that there has been ballot collection fraud. While the video is playing, Secretary Reagan’s narration indicates that the LaFaro Video answers Goddard’s request for evidence of fraud.
Id. (internal record citations omitted). The court found, “Although no direct evidence of ballot collection fraud was presented to the legislature or at trial, Shooter’s allegations and the LaFaro Video were successful in convincing H.B. 2023’s proponents that ballot collection presented opportunities for fraud that did not exist for in-person voting[.]” Id. at 880.
The district court found that H.B. 2023 is no harsher than any of the third-party ballot collection bills previously introduced in the Arizona legislature. The court found:
[A]lthough Plaintiffs argue that the legislature made H.B. 2023 harsher than previous ballot collection bills by imposing felony penalties, they ignore that H.B. 2023 in other respects is more lenient than its predecessors given its broad exceptions for family members, household members, and caregivers.
Id. at 881. In so finding, the district court clearly erred. Both S.B. 1412 and H.B. 2305 were more lenient than H.B. 2023.
For example, S.B. 1412, which was presented to DOJ for preclearance, required a third party collecting more than ten voted ballots to provide photo identification. There were no other restrictions on third-party ballot collection. There were no criminal penalties. By contrast, under H.B. 2023 a third party may collect a ballot only if the third party is an official engaged in official duties, or is a family member, household member, or caregiver of the voter. Ariz. Rev. Stat. § 16-1005(H), (I); Reagan, 329 F. Supp. 3d at 839–40. A third party who violates H.B. 2023 commits a class 5 felony.
In 2011, the relatively permissive third-party ballot collection provision of S.B. 1412 was withdrawn from Arizona’s preclearance request when DOJ asked for more information. In 2016, in the wake of Shelby County and without fear of preclearance scrutiny, Arizona enacted H.B. 2023.
II. Section 2 of the VRA
“Congress enacted the Voting Rights Act of 1965 for the broad remedial purpose of ‘rid[ding] the country of racial discrimination in voting.’ ” Chisom v. Roemer, 501 U.S. 380, 403, 111 S.Ct. 2354, 115 L.Ed.2d 348 (1991) (alteration in original) (quoting South Carolina v. Katzenbach, 383 U.S. 301, 315, 86 S.Ct. 803, 15 L.Ed.2d 769 (1966)). “The Act create[d] stringent new remedies for voting discrimination where it persists on a pervasive scale, and ... strengthen[ed] existing remedies for pockets of voting discrimination elsewhere in the country.” Katzenbach, 383 U.S. at 308, 86 S.Ct. 803.
When Section 2 of the Voting Rights Act was originally enacted in 1965, it read:
SEC. 2. No voting qualification or prerequisite to voting, or standard, practice, or procedure shall be imposed or applied by any State or political subdivision to deny or abridge the right of any citizen of the United States to vote on account of race or color.
Chisom, 501 U.S. at 391, 111 S.Ct. 2354 (citing 79 Stat. 437). “At the time of the passage of the Voting Rights Act of 1965, § 2, unlike other provisions of the Act, did not provoke significant debate in Congress because it was viewed largely as a restatement of the Fifteenth Amendment.” Id. at 392, 111 S.Ct. 2354. 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,” and it authorizes Congress to enforce the provision “by appropriate legislation.” U.S. Const. amend. XV. In City of Mobile v. Bolden, 446 U.S. 55, 100 S.Ct. 1519, 64 L.Ed.2d 47 (1980) (plurality), the Supreme Court held that the “coverage provided by § 2 was unquestionably coextensive with the coverage provided by the Fifteenth Amendment; the provision simply elaborated upon the Fifteenth Amendment.” Chisom, 501 U.S. at 392, 111 S.Ct. 2354. That is, the Court held that proof of intentional discrimination was necessary to establish a violation of Section 2. Id. at 393, 111 S.Ct. 2354.
Congress responded to Bolden by amending Section 2, striking out “to deny or abridge” and substituting “in a manner which results in a denial or abridgement of.” Id. (quoting amended Section 2; emphasis added by the Court); see also Gingles, 478 U.S. at 35, 106 S.Ct. 2752. “Under the amended statute, proof of intent [to discriminate] is no longer required to prove a § 2 violation.” Chisom, 501 U.S. at 394, 111 S.Ct. 2354. Rather, plaintiffs can now prevail under Section 2 either by demonstrating proof of intent to discriminate or “by demonstrating that a challenged election practice has resulted in the denial or abridgment of the right to vote based on color or race.” Id. That is, a Section 2 violation can “be established by proof of discriminatory results alone.”Chisom, 501 U.S. at 404, 111 S.Ct. 2354. The Supreme Court summarized: “Congress substantially revised § 2 to make clear that a violation could be proved by showing discriminatory effect alone and to establish as the relevant legal standard the ‘results test.’ ” Gingles, 478 U.S. at 35, 106 S.Ct. 2752 (emphasis added).
A violation of Section 2 may now be shown under either the results test or the intent test. Id. at 35, 44, 106 S.Ct. 2752. In the sections that follow, we analyze Plaintiffs’ challenges under these two tests. First, we analyze Arizona’s OOP policy and H.B. 2023 under the results test. Second, we analyze H.B. 2023 under the intent test.
A. Results Test: OOP Policy and H.B. 2023
1. The Results Test
Section 2 of the VRA “ ‘prohibits all forms of voting discrimination’ that lessen opportunity for minority voters.” League of Women Voters of N.C. v. North Carolina, 769 F.3d 224, 238 (4th Cir. 2014) (quoting Gingles, 478 U.S. at 45 n.10, 106 S.Ct. 2752). As amended in 1982, Section 2 of the VRA provides:
(a) No voting qualification or prerequisite to voting or standard, practice, or procedure shall be imposed or applied by any State or political subdivision in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color, or in contravention of the guarantees set forth in section 10303(f)(2) of this title, as provided in subsection (b).
(b) A violation of subsection (a) is established if, based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a) in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.
52 U.S.C. § 10301 (emphases added).
The results test of Section 2 applies in both vote dilution and vote denial cases. “Vote dilution claims involve challenges to methods of electing representatives—like redistricting or at-large districts—as having the effect of diminishing minorities’ voting strength.” Ohio State Conference of NAACP v. Husted, 768 F.3d 524, 554 (6th Cir. 2014), vacated on other grounds, 2014 WL 10384647 (6th Cir. 2014). A vote denial claim is generally understood to be “any claim that is not a vote dilution claim.” Id. The case now before us involves two vote-denial claims.
The jurisprudence of vote-denial claims is relatively underdeveloped in comparison to vote-dilution claims. As explained by the Fourth Circuit, “[T]he predominance of vote dilution in Section 2 jurisprudence likely stems from the effectiveness of the now-defunct Section 5 preclearance requirements that stopped would-be vote denial from occurring in covered jurisdictions[.]” League of Women Voters, 769 F.3d at 239.
In evaluating a vote-denial challenge to a “standard, practice, or procedure” under the “results test” of Section 2, most courts, including our own, engage in a two-step process. We first did so, in abbreviated fashion, in Smith v. Salt River Project Agricultural Improvement & Power District, 109 F.3d 586 (9th Cir. 1997). We later did so, at somewhat greater length, in Gonzalez v. Arizona, 677 F.3d 383 (9th Cir. 2012) (en banc). Other circuits have subsequently used a version of the two-step analysis. See Veasey v. Abbott, 830 F.3d 216, 244–45 (5th Cir. 2016); League of Women Voters, 769 F.3d at 240 (4th Cir. 2014); Husted, 768 F.3d at 554 (6th Cir. 2014). Compare Frank v. Walker, 768 F.3d 744, 755 (7th Cir. 2014) (“We are skeptical about the second of these steps[.]”).
First, we ask whether the challenged standard, practice or procedure results in a disparate burden on members of the protected class. That is, we ask whether, “as a result of the challenged practice or structure[,] plaintiffs do not have an equal opportunity to participate in the political processes and to elect candidates of their choice.” Gingles, 478 U.S. at 44, 106 S.Ct. 2752. The mere existence—or “bare statistical showing”—of a disparate impact on a racial minority, in and of itself, is not sufficient. See Salt River, 109 F.3d at 595 (“[A] bare statistical showing of disproportionate impact on a racial minority does not satisfy the § 2 ‘results’ inquiry.” (emphasis in original)).
Second, if we find at the first step that the challenged practice imposes a disparate burden, we ask whether, under the “totality of the circumstances,” there is a relationship between the challenged “standard, practice, or procedure,” on the one hand, and “social and historical conditions” on the other. The purpose of the second step is to evaluate a disparate burden in its real-world context rather than in the abstract. As stated by the Supreme Court, “The essence of a § 2 claim is that a certain electoral law, practice, or structure interacts with social and historical conditions to cause an inequality in the opportunities enjoyed by [minority] and white voters to elect their preferred representatives” or to participate in the political process. Gingles, 478 U.S. at 47, 106 S.Ct. 2752; 52 U.S.C. § 10301(b). To determine at the second step whether there is a legally significant relationship between the disparate burden on minority voters and the social and historical conditions affecting them, we consider, as appropriate, factors such as those laid out in the Senate Report accompanying the 1982 amendments to the VRA. Id. at 43, 106 S.Ct. 2752 (“The Senate Report which accompanied the 1982 amendments elaborates on the nature of § 2 violations and on the proof required to establish these violations.”); Veasey, 830 F.3d at 244–45.
The Senate Report provides:
If as a result of the challenged practice or structure plaintiffs do not have an equal opportunity to participate in the political processes and to elect candidates of their choice, there is a violation of this section. To establish a violation, plaintiffs could show a variety of factors, depending on the kind of rule, practice, or procedure called into question.
Typical factors include:
1. the extent of any history of official discrimination in the state or political subdivision that touched the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process;
2. the extent to which voting in the elections of the state or political subdivision is racially polarized;
3. the extent to which the state or political subdivision has used unusually large election districts, majority vote requirements, anti-single shot provisions, or other voting practices or procedures that may enhance the opportunity for discrimination against the minority group;
4. if there is a candidate slating process, whether the members of the minority group have been denied access to that process;
5. the extent to which members of the minority group in the state or political subdivision bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process;
6. whether political campaigns have been characterized by overt or subtle racial appeals;
7. the extent to which members of the minority group have been elected to public office in the jurisdiction.
Additional factors that in some cases have had probative value as part of plaintiffs’ evidence to establish a violation are:
[8.] whether there is a significant lack of responsiveness on the part of elected officials to the particularized needs of the members of the minority group.
[9.] whether the policy underlying the state or political subdivision’s use of such voting qualification, prerequisite to voting, or standard, practice or procedure is tenuous.
S. Rep. No. 97-417 (“S. Rep.”), at 28–29 (1982); see Gingles, 478 U.S. at 36–37, 106 S.Ct. 2752 (quoting the Senate Report).
The Senate Committee’s list of “typical factors” is neither comprehensive nor exclusive. S. Rep. at 29. “[T]here is no requirement that any particular number of factors be proved, or that a majority of them point one way or the other.” Id. “[T]he question whether the political processes are ‘equally open’ depends on a searching practical evaluation of the ‘past and present reality.’ ” Id. at 30. An evaluation of the totality of circumstances in a Section 2 results claim, including an evaluation of appropriate Senate factors, requires “a blend of history and an intensely local appraisal[.]” Gingles, 478 U.S. at 78, 106 S.Ct. 2752 (quoting White v. Regester, 412 U.S. 755, 769–70, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973)). The Senate factors are relevant to both vote-denial and vote-dilution claims. Gingles, 478 U.S. at 45, 106 S.Ct. 2752 (Senate factors will be “pertinent to certain types of § 2 claims,” including vote denial claims, but will be “particularly [pertinent] to vote dilution claims.”).
Our sister circuits have struck down standards, practices, or procedures in several vote-denial cases after considering the Senate factors. In Husted, the Sixth Circuit upheld a district court’s finding that an Ohio law limiting early voting violated the results test of Section 2. The court wrote,
We find Senate factors one, three, five, and nine particularly relevant to a vote denial claim in that they specifically focus on how historical or current patterns of discrimination “hinder [minorities’] ability to participate effectively in the political process.” Gingles, 478 U.S. at 37, 106 S.Ct. 2752 (quoting Senate factor five). All of the factors, however, can still provide helpful background context to minorities’ overall ability to engage effectively on an equal basis with other voters in the political process.
Husted, 768 F.3d at 555. In Veasey, the Fifth Circuit upheld a district court’s finding that Texas’s requirement that a photo ID be presented at the time of voting violated the results test. Veasey, 830 F.3d at 256–64 (considering Senate factors one, two, five, six, seven, eight, and nine). In League of Women Voters, the Fourth Circuit held that the district court had clearly erred in finding that the results test had not been violated by North Carolina’s elimination of same-day registration, and by North Carolina’s practice of wholly discarding out-of-precinct ballots. League of Women Voters, 769 F.3d at 245–46 (considering Senate factors one, three, and nine).
2. OOP Policy and the Results Test
Uncontested evidence in the district court established that minority voters in Arizona cast OOP ballots at twice the rate of white voters. The question is whether the district court clearly erred in holding that Arizona’s policy of entirely discarding OOP ballots does not violate the “results test” of Section 2.
a. Step One: Disparate Burden
The question at step one is whether Arizona’s policy of entirely discarding OOP ballots results in a disparate burden on a protected class. The district court held that Plaintiffs failed at step one. The district court clearly erred in so holding.
Extensive and uncontradicted evidence in the district court established that American Indian, Hispanic, and African American voters are over-represented among OOP voters by a ratio of two to one. See Part II(A), supra. The district court wrote, “Plaintiffs provided quantitative and statistical evidence of disparities in OOP voting through the expert testimony of Dr. Rodden .... Dr. Rodden’s analysis is credible and shows that minorities are over-represented among the small number of voters casting OOP ballots.” Reagan, 329 F. Supp. 3d at 871. Dr. Rodden reported that this pattern was consistent over time and across counties. Based on this evidence, the court found that during the 2016 general election, American Indian, Hispanic, and African American voters were twice as likely as white voters to vote out-of-precinct and not have their votes counted. Id. at 872.
Despite these factual findings, the district court held that Arizona’s policy of entirely discarding OOP ballots does not impose a disparate burden under the results test. The court gave two reasons to support its holding.
First, the district court discounted the disparate burden on the ground that there were relatively few OOP ballots cast in relation to the total number of ballots. Id. at 872. The district court clearly erred in so doing.
The district court pointed out that the absolute number of OOP ballots in Arizona fell between 2012 and 2016. It pointed out, further, that as a percentage of all ballots cast, OOP ballots fell from 0.47 percent to 0.15 percent during that period. Id. The numbers and percentages cited by the district court are accurate. Standing alone, they may be read to suggest that locating the correct precinct for in-person voting has become easier and that OOP ballots, as a percentage of in-person ballots, have decreased accordingly.
However, the opposite is true. Arizona’s OOP policy applies only to in-person ballots. The proper baseline to measure OOP ballots to is thus not all ballots, but all in-person ballots. The district court failed to point out that the absolute number of all in-person ballots fell more than the absolute number of OOP ballots, and that, as a result, as a percentage of in-person ballots, OOP ballots increased rather than decreased.
Even putting aside the potentially misleading numbers and percentages cited by the district court and focusing only on the decline in the absolute number of OOP ballots, the court clearly erred. As indicated above, the vote-denial category encompasses all cases that are not vote-dilution cases. The number of minority voters adversely affected, and the mechanism by which they are affected, may vary considerably. For example, if a polling place denies an individual minority voter her right to vote based on her race or color, Section 2 is violated based on that single denial. However, a different analysis may be appropriate when a facially neutral policy adversely affects a number of minority voters. Arizona’s OOP policy is an example. We are willing to assume in such a case that more than a de minimis number of minority voters must be burdened before a Section 2 violation based on the results test can be found. Even on that assumption, however, we conclude that the number of OOP ballots cast in Arizona’s general election in 2016—3,709 ballots—is hardly de minimis.
We find support for our conclusion in several places. The Department of Justice submitted an amicus brief to our en banc panel in support of Arizona. Despite its support for Arizona, DOJ specifically disavowed the district court’s conclusion that the number of discarded OOP ballots was too small to be cognizable under the results test. DOJ wrote:
[T]he district court’s reasoning was not correct to the extent that it suggested that plaintiffs’ Section 2 claim would fail solely because of the small number of voters affected. ...
That is not a proper reading of the statute. Section 2 prohibits any “standard, practice, or procedure” that “results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color.” 52 U.S.C. 10301(a) (emphasis added); see also Frank v. Walker, 819 F.3d 384, 386 (7th Cir. 2016) (Frank II) (“The right to vote is personal and is not defeated by the fact that 99% of other people can secure the necessary credentials easily.”). Section 2 safeguards a personal right to equal participation opportunities. A poll worker turning away a single voter because of her race plainly results in “less opportunity *** to participate in the political process and to elect representatives of [her] choice.” 52 U.S.C. 10301(b).
DOJ Amicus Brief at 28–29. DOJ’s brief appears to treat as equivalent the case of an individually targeted single minority voter who is denied the right to vote and the case where a facially neutral policy affects a single voter. We do not need to go so far. We need only point out that in the case before us a substantial number of minority voters are disparately affected by Arizona’s OOP policy. As long as an adequate disparate impact is shown, as it has been shown here, and as long as the other prerequisites for finding a Section 2 violate are met, each individual in the affected group is protected under Section 2.
Further, in League of Women Voters, “approximately 3,348 out-of-precinct provisional ballots” cast by African American voters would have been discarded under the challenged North Carolina law. 769 F.3d at 244 (quoting the district court). The district court had held that this was a “minimal” number of votes, and that Section 2 was therefore not violated. The Fourth Circuit reversed, characterizing the district court’s ruling as a “grave error.” Id. at 241.
Finally, in the 2000 presidential election, the official margin of victory for President George W. Bush in Florida was 537 votes. Federal Election Commission, 2000 Official Presidential General Election Results (Dec. 2001), available at https://transition.fec.gov/pubrec/2000presgeresults.htm. If there had been 3,709 additional ballots cast in Florida in 2000, in which minority voters had outnumbered white voters by a ratio of two to one, it is possible that a different President would have been elected.
Second, the district court concluded that Arizona’s policy of rejecting OOP ballots does not impose a disparate burden on minority voters because Arizona’s policy of entirely discarding OOP ballots “is not the cause of the disparities in OOP voting.” Reagan, 329 F. Supp. 3d at 872. The court wrote that Plaintiffs “have not shown that Arizona’s policy to not count OOP ballots causes minorities to show up to vote at the wrong precinct at rates higher than their non-minority counterparts.” Id. at 873. Again, the district court clearly erred.
The district court misunderstood what Plaintiffs must show. Plaintiffs need not show that Arizona caused them to vote out of precinct. Rather, they need only show that the result of entirely discarding OOP ballots has an adverse disparate impact, by demonstrating “a causal connection between the challenged voting practice and a prohibited discriminatory result.” Salt River, 109 F.3d at 595 (emphasis added). Here, “[t]he challenged practice—not counting OOP ballots—results in ‘a prohibited discriminatory result’; a substantially higher percentage of minority votes than white votes are discarded.” DNC, 904 F.3d at 736 (Thomas, C.J., dissenting).
We hold that the district court clearly erred in holding that Arizona’s policy of entirely discarding OOP ballots does not result in a disparate burden on minority voters. We accordingly hold that Plaintiffs have succeeded at step one of the results test.
b. Step Two: Senate Factors
The question at step two is whether, under the “totality of circumstances,” the disparate burden on minority voters is linked to social and historical conditions in Arizona so as “to cause an inequality in the opportunities enjoyed by [minority] and white voters to elect their preferred representatives” or to participate in the political process. Gingles, 478 U.S. at 47, 106 S.Ct. 2752; 52 U.S.C. § 10301(b). The district court wrote that because in its view Plaintiffs failed at step one, discussion of step two was unnecessary. Reagan, 329 F. Supp. 3d at 873. The court nonetheless went on to discuss step two and, after considering various Senate factors, to hold that Plaintiffs failed at this step as well. The district court clearly erred in so holding.
At step two, we consider relevant Senate factors. Some Senate factors are “more important to” vote-denial claims, or to some vote-denial claims, and others, “[i]f present, ... are supportive of, but not essential to” the claim. Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752 (emphasis in original). That is, Senate factors vary in importance depending on whether a court is dealing with a vote-dilution or a vote-denial case. The same factors may also vary in importance from one vote-denial case to another.
We emphasize that the relative importance of the Senate factors varies from case to case. For example, as we will describe in a moment, Arizona has a long and unhappy history of official discrimination connected to voting. Other States may not have such a history, but depending on the existence of other Senate factors they may nonetheless be found to have violated the results test of Section 2.
The district court considered seven of the nine Senate factors: factor one, the history of official discrimination connected to voting; factor two, racially polarized voting patterns; factor five, the effects of discrimination in other areas on minority groups’ access to voting; factor six, racial appeals in political campaigns; factor seven, the number of minorities in public office; factor eight, officials’ responsiveness to the needs of minority groups; and factor nine, the tenuousness of the justification for the challenged voting practice.
We analyze below each of these factors, indicating whether we agree or disagree with the district court’s analysis as to each. Of the various factors, we regard Senate factors five (the effects of discrimination in other areas on minorities access to voting) and nine (the tenuousness of the justification for the challenged voting practices) as particularly important. We also regard factor one (history of official discrimination) as important, as it bears on the existence of discrimination generally and strongly supports our conclusion under factor five. Though “not essential,” Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752, the other factors provide “helpful background context.” Husted, 768 F.3d at 555.
i. Factor One: History of Official Discrimination Connected to Voting
Arizona has a long history of race-based discrimination against its American Indian, Hispanic, and African American citizens. Much of that discrimination is directly relevant to those citizens’ ability “to register, to vote, or otherwise to participate in the democratic process.” Id. We recount the most salient aspects of that history.
Dr. David Berman, a Professor Emeritus of Political Science at Arizona State University, submitted an expert report and testified in the district court. The court found Dr. Berman “credible” and gave “great weight to Dr. Berman’s opinions.” Reagan, 329 F. Supp. 3d at 834. The following narrative is largely drawn from Dr. Berman’s report and the sources on which he relied.
(A) Territorial Period
Arizona’s history of discrimination dates back to 1848, when it first became an American political entity as a United States territory. “Early territorial politicians acted on the belief that it was the ‘manifest destiny’ of the Anglos to triumph in Arizona over the earlier Native American and Hispanic civilizations.” David Berman, Expert Report (Berman) at 4. Dr. Berman wrote that from the 1850s through the 1880s there were “blood thirsty efforts by whites to either exterminate” Arizona’s existing American Indian population or “confine them to reservations.” Id. at 5. In 1871, in the Camp Grant Massacre, white settlers “brutal[ly] murder[ed] over 100 Apaches, most of whom were women and children.” Id. Arizona’s white territorial legislature passed a number of discriminatory laws, including anti-miscegenation laws forbidding marriage between whites and Indians. See James Thomas Tucker et al., Voting Rights in Arizona: 1982–2006, 17 S. Cal. Rev. L. & Soc. Just. 283, 283 n.3 (2008) (Tucker et al., Voting Rights). Dr. Berman wrote: “By the late 1880s and the end of th[e] Indian wars, the realities of life for Native Americans in Arizona were confinement to reservations, a continuous loss of resources (water, land, minerals) to settlers, poverty, and pressure to abandon their traditional cultures.” Berman at 5.
White settlers also discriminated against Arizona’s Hispanic population. Dr. Berman wrote:
Although Hispanics in the territory’s early period commonly held prominent roles in public and political life, as migration continued they were overwhelmed by a flood of Anglo-American and European immigrants. While a small group of Hispanics continued to prosper, ... most Hispanics toiled as laborers who made less than Anglos even though they performed the same work.
Id. (footnote omitted). Hispanics in Arizona “found it difficult to receive acceptance or fair treatment in a society that had little tolerance for people of Latin American extraction, and particularly those whose racial make-up included Indian or African blood.” Id. at 5–6 (quoting Oscar J. Martinez, Hispanics in Arizona, in Arizona at Seventy-Five: The Next Twenty-Five Years 88–89 (Ariz. State Univ. Pub. History Program & the Ariz. Historical Soc’y, 1987)).
Pursuant to the Treaty of Guadalupe Hidalgo that ended the Mexican-American War, the United States conferred citizenship on the approximately 100,000 Hispanics living in Arizona. In 1909, the Arizona territorial legislature passed a statute imposing an English language literacy test as a prerequisite to voter registration. Id. at 10. The test was specifically designed to prevent the territory’s Hispanic citizens—who had lower English literacy rates than white citizens—from voting. Id. At the time, Indians were not citizens and were not eligible to vote.
In 1910, Congress passed a statute authorizing Arizona, as a prelude to statehood, to draft a state constitution. Upon approval of its constitution by Congress, the President, and Arizona voters, Arizona would become a State. Id. at 11. Members of Congress viewed Arizona’s literacy test as a deliberate effort to disenfranchise its Hispanic voters. Id. The authorizing statute specifically provided that Arizona could not use its newly adopted literacy test to prevent Arizona citizens from voting on a proposed constitution. Id.
That same year, Arizona convened a constitutional convention. Id. at 7. Although Congress had ensured that Arizona would not use its literacy test to prevent Hispanic citizens from voting on the constitution, Hispanics were largely excluded from the drafting process. With the exception of one Hispanic delegate, all of the delegates to the convention were white. Id. By comparison, approximately one-third of the delegates to the 1910 New Mexico constitutional convention were Hispanic, and one-sixth of the 48 delegates to the 1849 California constitutional convention were Hispanic. Id.
The influence of Hispanic delegates is evident in those States’ constitutions. For example, New Mexico’s constitution provides that the “right of any citizen of the state to vote, hold office or sit upon juries, shall never be restricted, abridged or impaired on account of ... race, language or color, or inability to speak, read or write the English or Spanish languages.” N.M. Const. art. VII, § 3 (1910). It also requires the legislature to provide funds to train teachers in Spanish instruction. N.M. Const. art. XII, § 8 (1910). California’s constitution required all state laws to be published in Spanish as well as English. Cal. Const. art. XI, § 21 (1849).
By contrast, Arizona’s constitution did not include such provisions. Indeed, two provisions required precisely the opposite. The Arizona constitution provided that public schools “shall always be conducted in English” and that “[t]he ability to read, write, speak, and understand the English language sufficiently well to conduct the duties of the office without the aid of an interpreter, shall be a necessary qualification for all State officers and members of the State Legislature.” Ariz. Const. art. XX, §§ 7, 8 (1910).
(B) Early Statehood
(1) Literacy Test
Arizona became a State in 1912. That same year, the Arizona legislature passed a statute reimposing an English literacy test—the test that had been imposed by the territorial legislature in 1909 and that Congress had forbidden the State to use for voting on the state constitution. Berman at 11; see also James Thomas Tucker, The Battle Over Bilingual Ballots: Language Minorities and Political Access Under the Voting Rights Act 20 (Routledge, 2016) (Tucker, Bilingual Ballots). According to Dr. Berman, the statute was enacted “to limit ‘the ignorant Mexican vote.’ ” David R. Berman, Arizona Politics and Government: The Quest for Autonomy, Democracy, and Development 75 (Univ. of Neb. Press, 1998) (Berman, Arizona Politics) (quoting letter between prominent political leaders); Berman at 12.
County registrars in Arizona had considerable discretion in administering literacy tests. Registrars used that discretion to excuse white citizens from the literacy requirement altogether, to give white citizens easier versions of the test, and to help white citizens pass the test. See also Katzenbach, 383 U.S. at 312, 86 S.Ct. 803 (describing the same practice with respect to African American citizens in southern States). In contrast, Hispanic citizens were often required to pass more difficult versions of the test, without assistance and without error. Berman, Arizona Politics at 75; see also Berman at 12.
The literacy test was used for the next sixty years. The year it was introduced, Hispanic registration declined so dramatically that some counties lacked enough voters to justify primaries. Berman at 12. One county had recall campaigns because enough Hispanic voters had been purged from voting rolls to potentially change the electoral result. Id. Arizona would use its literacy test not only against Hispanics, but also against African Americans and, once they became eligible to vote in 1948, against American Indians. The test was finally repealed in 1972, two years after an amendment to the Voting Rights Act banned literacy tests nationwide. Id.
(2) Disenfranchisement of American Indians
In 1912, when Arizona became a State, Indians were not citizens of Arizona or of the United States. In 1924, Congress passed the Indian Citizenship Act, declaring all Indians citizens of the United States and, by extension, of their States of residence. Indian Citizenship Act of 1924, Pub. L. No. 68-175, 43 Stat. 253 (codified at 8 U.S.C. § 1401(b)).
Indian voting had the potential to change the existing white political power structure of Arizona. See Patty Ferguson-Bohnee, The History of Indian Voting Rights in Arizona: Overcoming Decades of Voter Suppression, 47 Ariz. St. L.J. 1099, 1103–04 (2015) (Ferguson-Bohnee). Indians comprised over 14 percent of the population in Arizona, the second-highest percentage of Indians in any State. Id. at 1102 n.19, 1104. Potential power shifts were even greater at the county level. According to the 1910 Census, Indians comprised over 66 percent of the population of Apache County, over 50 percent of Navajo County, over 34 percent of Pinal County, and over 34 percent of Coconino County. Id. at 1104.
Enacted under the Fourteenth and Fifteenth Amendments, the Indian Citizenship Act should have given Indians the right to vote in Arizona elections. The Attorney General of Arizona initially agreed that the Act conferred the right to vote, and he suggested in 1924 that precinct boundaries should be expanded to include reservations. Id. at 1105. However, in the years leading up to the 1928 election, Arizona’s Governor, county officials, and other politicians sought to prevent Indians from voting. Id. at 1106–08. The Governor, in particular, was concerned that Indian voter registration—specifically, registration of approximately 1,500 Navajo voters—would hurt his reelection chances. Id. at 1107–08. The Governor sought legal opinions on ways to exclude Indian voters, id., and was advised to “adopt a systematic course of challenging Indians at the time of election.” Id. at 1108 (quoting Letter from Samuel L. Pattee to George W.P. Hunt, Ariz. Governor (Sept. 22, 1928)). County officials challenged individual Indian voter registrations. Id. at 1107–08.
Prior to the 1928 election, two Indian residents of Pima County brought suit challenging the county’s rejection of their voter registration forms. Id. at 1108. The Arizona Supreme Court sided with the county. The Arizona constitution forbade anyone who was “under guardianship, non compos mentis, or insane” from voting. Ariz. Const. art. VII, § 2 (1910). The Court held that Indians were “wards of the nation,” and were therefore “under guardianship” and not eligible to vote. Porter v. Hall, 34 Ariz. 308, 271 P. 411, 417, 419 (1928).
Arizona barred Indians from voting for the next twenty years. According to the 1940 census, Indians comprised over 11 percent of Arizona’s population. Ferguson-Bohnee at 1111. They were the largest minority group in Arizona. “One-sixth of all Indians in the country lived in Arizona.” Id.
After World War II, Arizona’s Indian citizens returned from fighting the Axis powers abroad to fight for the right to vote at home. Frank Harrison, a World War II veteran and member of the Fort McDowell Yavapai Nation, and Harry Austin, another member of the Fort McDowell Yavapai Nation, filed suit against the State. In 1948, the Arizona Supreme Court overturned its prior decision in Porter v. Hall. Harrison v. Laveen, 67 Ariz. 337, 196 P.2d 456, 463 (1948). Almost a quarter century after enactment of the Indian Citizenship Act of 1924, Indian citizens in Arizona had the legal right to vote.
(C) The 1950s and 1960s
For decades thereafter, however, Arizona’s Indian citizens often could not exercise that right. The Arizona Supreme Court’s decision in Harrison v. Laveen did not result in “a large influx” of new voters because Arizona continued to deny Indian citizens—as well as Hispanic and African American citizens—access to the ballot through other means. Berman at 15.
The biggest obstacle to voter registration was Arizona’s English literacy test. In 1948, approximately 80 to 90 percent of Indian citizens in Arizona did not speak or read English. Tucker et al., Voting Rights at 285; see also Berman at 15. In the 1960s, about half the voting-age population of the Navajo Nation could not pass the English literacy test. Ferguson-Bohnee at 1112 n.88. For Arizona’s Indian—and Hispanic and African American—citizens who did speak and read English, discriminatory administration of the literacy test by county registrars often prevented them from registering. See, e.g., Berman, Arizona Politics at 75 (“As recently as the 1960s, registrars applied the test to reduce the ability of blacks, Indians and Hispanics to register to vote.”).
Voter intimidation during the 1950s and 60s often prevented from voting those American Indian, Hispanic, and African American citizens who had managed to register. According to Dr. Berman:
During the 1960s, it was ... clear that more than the elimination of the literacy test in some areas was going to be needed to protect minorities. Intimidation of minority-group members—Hispanics, African Americans, as well as Native Americans—who wished to vote was ... a fact of life in Arizona. Anglos sometimes challenged minorities at the polls and asked them to read and explain “literacy” cards containing quotations from the U.S. Constitution. These intimidators hoped to frighten or embarrass minorities and discourage them from standing in line to vote. Vote challenges of this nature were undertaken by Republican workers in 1962 in South Phoenix, a largely minority Hispanic and African-American area. ... [In addition,] [p]eople in the non-Native American community, hoping to keep Native Americans away from the polls, told them that involvement could lead to something detrimental, such as increased taxation, a loss of reservation lands, and an end to their special relationship with the federal government.
Berman at 14–15.
Intimidation of minority voters continued throughout the 1960s. For example, in 1964, Arizona Republicans undertook voter intimidation efforts throughout Arizona “as part of a national effort by the Republican Party called ‘Operation Eagle Eye.’ ” Id. at 14. According to one account:
The approach was simple: to challenge voters, especially voters of color, at the polls throughout the country on a variety of specious pretexts. If the challenge did not work outright—that is, if the voter was not prevented from casting a ballot (provisional ballots were not in widespread use at this time)—the challenge would still slow down the voting process, create long lines at the polls, and likely discourage some voters who could not wait or did not want to go through the hassle they were seeing other voters endure.
Id. (quoting Tova Andrea Wang, The Politics of Voter Suppression: Defending and Expanding Americans’ Right to Vote 44–45 (Cornell Univ. Press, 2012)).
Compounding the effects of the literacy test and voter intimidation, Arizona “cleansed” its voting rolls. In 1970, Democrat Raul Castro narrowly lost the election for Governor. (He would win the governorship four years later to become Arizona’s first and only Hispanic Governor.) Castro received 90 percent of the Hispanic vote, but he lost the election because of low Hispanic voter turnout. Dr. Berman explained:
[C]ontributing to that low turnout was “a decision by the Republican-dominated legislature to cleanse the voting rolls and have all citizens reregister. This cleansing of the rolls erased years of registration drives in barrios across the state. It seems certain that many Chicanos did not understand that they had to reregister, were confused by this development, and simply stayed away from the polls.”
Id. at 17 (quoting F. Chris Garcia & Rudolph O. de la Garza, The Chicano Political Experience 105 (Duxbury Press, 1977)).
(D) Voting Rights Act and Preclearance under Section 5
Congress passed the Voting Rights Act in 1965. See Voting Rights Act of 1965, Pub. L. No. 89-110, 79 Stat. 437–446 (codified as amended at 52 U.S.C. §§ 10301–10314, 10501–10508, 10701, 10702). Under Section 4(b) of the Act, a State or political subdivision qualified as a “covered jurisdiction” if it satisfied two criteria. Id. § 4(b). The first was that on November 1, 1964—the date of the presidential election—the State or political subdivision had maintained a “test or device,” such as a literacy test, restricting the opportunity to register or vote. The second was either that (a) on November 1, 1964, less than 50 percent of the voting-age population in the jurisdiction had been registered to vote, or (b) less than 50 percent of the voting-age population had actually voted in the presidential election of 1964. Seven States qualified as covered jurisdictions under this formula: Alabama, Alaska, Georgia, Louisiana, Mississippi, South Carolina, and Virginia. Determination of the Director of the Census Pursuant to Section 4(b)(2) of the Voting Rights Act of 1965, 30 Fed. Reg. 9897-02 (Aug. 7, 1965). Political subdivisions in four additional States—Arizona, Hawai‘i, Idaho, and North Carolina—also qualified as covered jurisdictions. See id.; Determination of the Director of the Census Pursuant to Section 4(b)(2) of the Voting Rights Act of 1965, 30 Fed. Reg. 14,505-02 (Nov. 19, 1965).
Under Section 4(a) of the VRA, covered jurisdictions were forbidden for a period of five years from using a “test or device,” such as a literacy test, as a prerequisite to register to vote, unless a three-judge district court of the District of Columbia found that no such test had been used by the jurisdiction during the preceding five years for the purpose of denying the right to vote on account of race or color. Voting Rights Act of 1965, Pub. L. No. 89-110, § 4(a). Under Section 5, covered jurisdictions were forbidden from changing “any voting qualification or prerequisite to voting, or standard, practice, or procedure with respect to voting” unless the jurisdiction “precleared” that change, by either obtaining approval (a) from a three-judge district court of the District of Columbia acknowledging that the proposed change “neither has the purpose nor will have the effect of denying or abridging the right to vote on account of race or color,” or (b) from the Attorney General if a proposed change has been submitted to DOJ and the Attorney General has not “interposed an objection” within sixty days of the submission. Id. § 5.
Three counties in Arizona qualified as “covered jurisdictions” under the 1965 Act: Apache, Coconino, and Navajo Counties. SeeDetermination of the Director of the Census Pursuant to Section 4(b)(2) of the Voting Rights Act of 1965, 30 Fed. Reg. 9897-02, 14,505-02. Those counties were therefore initially prohibited from using the literacy test as a prerequisite to voter registration. All three counties were majority American Indian, and there was a history of high use of the literacy test and correspondingly low voter turnout. Berman at 12. However, in 1966, in a suit brought by the counties against the United States, a three-judge district court held that there was insufficient proof that a literacy test had been used by the counties in a discriminatory fashion during the immediately preceding five years. See Apache Cty. v. United States, 256 F. Supp. 903 (D.D.C. 1966). The Navajo Nation had sought to intervene and present evidence of discrimination in the district court, but its motion to intervene had been denied. Id. at 906–13.
Congress renewed and amended the VRA in 1970, extending it for another five years. Voting Rights Act of 1970, Pub. L. No. 91-285, 84 Stat. 314 (1970). Under the VRA of 1970, the formula for determining covered jurisdictions under Section 4(b) was changed to add the presidential election of 1968 to the percentage-of-voters criterion. Id. § 4(b). As a result, eight out of fourteen Arizona counties—including Apache, Navajo, and Coconino Counties—qualified as covered jurisdictions. Tucker et al., Voting Rights at 286. Under the 1970 Act, non-covered jurisdictions were forbidden from using a “test or device,” such as a literacy test, to the same degree as covered jurisdictions. The 1970 Act thus effectively imposed a nationwide ban on literacy tests. Voting Rights Act of 1970, Pub. L. No. 91-285, § 201.
Arizona immediately challenged the ban. In Oregon v. Mitchell, 400 U.S. 112, 132, 91 S.Ct. 260, 27 L.Ed.2d 272 (1970), the Court unanimously upheld the ban on literacy tests. Justice Black wrote,
In enacting the literacy test ban ... [,] Congress had before it a long history of the discriminatory use of literacy tests to disfranchise voters on account of their race. ... Congress ... had evidence to show that voter registration in areas with large Spanish-American populations was consistently below the state and national averages. In Arizona, for example, only two counties out of eight with Spanish surname populations in excess of 15% showed a voter registration equal to the state-wide average. Arizona also has a serious problem of deficient voter registration among Indians.
Two years after the Court’s decision, Arizona finally repealed its literacy test. Tucker, Bilingual Ballots, at 21.
In 1975, Congress again renewed and amended the VRA. Voting Rights Act of 1975, Pub. L. No. 94-73, 89 Stat. 400 (1975). Under the VRA of 1975, the formula for determining covered jurisdictions under Section 4(b) was updated to add the presidential election of 1972. Id. § 202. In addition, Congress expanded the definition of “test or device” to address discrimination against language minority groups. Id. § 203 (Section 4(f)). Pursuant to this amended formula and definition, any jurisdiction where a single language minority group (e.g., Spanish speakers who spoke no other language) constituted more than 5 percent of eligible voters was subject to preclearance under Section 5 if (a) the jurisdiction did not offer bilingual election materials during the 1972 presidential election, and (b) less than 50 percent of the voting-age population was registered to vote, or less than 50 percent of the voting-age population actually voted in the 1972 presidential election. Id. §§ 201–203.
Every jurisdiction in Arizona failed the new test. As a result, the entire State of Arizona became a covered jurisdiction. Berman at 20–21.
(E) Continued Obstacles to Voting: The Example of Apache County
The VRA’s elimination of literacy tests increased political participation by Arizona’s American Indian, Hispanic, and African American citizens. However, state and county officials in Arizona continued to discriminate against minority voters. Apache County, which includes a significant part of the Navajo Reservation, provides numerous examples of which we recount only one.
In 1976, a school district in Apache County sought to avoid integration by holding a special bond election to build a new high school in a non-Indian area of the county. See Apache Cty. High Sch. Dist. No. 90 v. United States, No. 77-1815 (D.D.C. June 12, 1980); see also Tucker et al., Voting Rights at 324–26 (discussing the same). Less than a month before the election, the school district, a “covered jurisdiction” under the VRA, sought preclearance under Section 5 for proposed changes in election procedures, including closure of nearly half the polling stations on the Navajo Reservation. Letter from J. Stanley Pottinger, Assistant Attorney Gen., Civil Rights Div., Dep’t of Justice, to Joe Purcell, Gust, Rosenfeld, Divelbess & Henderson (Oct. 4, 1976). DOJ did not complete its review before the election. The school district nonetheless held the bond election using the proposed changes. After the election, DOJ refused to preclear the proposed changes, finding that they had a discriminatory purpose or effect. Id. (and subsequent letters from Assistant Attorney Gen. Drew S. Days III on May 3, 1977, and June 10, 1977). The school district brought suit in a three-judge district court, seeking a declaratory judgment that the election did not violate the VRA.
The district court found that “[t]he history of Apache County reveals pervasive and systemic violations of Indian voting rights.” Apache Cty. High Sch. Dist. No. 90, No. 77-1815, at 6. The court found that the school district’s behavior was neither “random[ ]” nor “unconscious[ ].” Id. at 14–15. “Rather, its campaign behavior served to effectuate the unwritten but manifest policy of minimizing the effect of the Navajos’ franchise, while maximizing the Anglo vote.” Id. at 15.
(F) United States v. Arizona and Preclearance during the 1980s and 1990s
During the following two decades, DOJ refused to preclear numerous proposed voting changes in Arizona. See, e.g., Goddard v. Babbitt, 536 F. Supp. 538, 541, 543 (D. Ariz. 1982) (finding that a state legislative redistricting plan passed by the Arizona state legislature “dilut[ed] the San Carlos Apache Tribal voting strength and divid[ed] the Apache community of interest”); see also Tucker et al., Voting Rights at 326–28 (discussing additional examples). In 1988, the United States sued Arizona, alleging that the State, as well as Apache and Navajo Counties, violated the VRA by employing election standards, practices, and procedures that denied or abridged the voting rights of Navajo citizens. See United States v. Arizona, No. 88-1989 (D. Ariz. May 22, 1989) (later amended Sept. 27, 1993); see also Tucker et al., Voting Rights at 328–30 (discussing the same). A three-judge district court summarized the complaint:
The challenged practices include alleged discriminatory voter registration, absentee ballot, and voter registration cancellation procedures, and the alleged failure of the defendants to implement, as required by Section 4(f)(4), effective bilingual election procedures, including the effective dissemination of election information in Navajo and providing for a sufficient number of adequately trained bilingual persons to serve as translators for Navajo voters needing assistance at the polls on election day.
United States v. Arizona, No. 88-1989, at 1–2.
Arizona and the counties settled the suit under a Consent Decree. Id. at 1–26. The Decree required the defendants to make extensive changes to their voting practices, including the creation of a Navajo Language Election Information Program. See id. at 4–23. More than a decade later, those changes had not been fully implemented. See U.S. Gov’t Accountability Office, Department of Justice’s Activities to Address Past Election-Related Voting Irregularities 91–92 (2004), available at http://www.gao.gov/new.items/d041041r.pdf (identifying significant deficiencies and finding that implementation of the Navajo Language Election Information Program by Apache and Navajo Counties was “inadequate”).
During the 1980s and 1990s, DOJ issued seventeen Section 5 preclearance objections to proposed changes in Arizona election procedures, concluding that they had the purpose or effect of discriminating against Arizona’s American Indian and/or Hispanic voters. See U.S. Dep’t of Justice, Voting Determination Letters for Arizona, https://www.justice.gov/crt/voting-determination-letters-arizona (last updated Aug. 7, 2015). Three of these objections were for statewide redistricting plans, one in the 1980s and two in the 1990s. Id. Other objections concerned plans for seven of Arizona’s fifteen counties. Id. (objections to plans for Apache, Cochise, Coconino, Graham, La Paz, Navajo, and Yuma Counties).
(G) Continuation to the Present Day
Arizona’s pattern of discrimination against minority voters has continued to the present day.
(1) Practices and Policies
We highlight two examples of continued discriminatory practices and policies. First, as the district court found, the manner in which Maricopa County—home to over 60 percent of Arizona’s population—administers elections has “been of considerable concern to minorities in recent years.” Reagan, 329 F. Supp. 3d at 871; Berman at 20. During the 2016 presidential primary election, Maricopa County reduced the number of polling places by 70 percent, from 200 polling places in 2012 to just 60 polling places in 2016. Berman at 20. The reduction in number, as well as the locations, of the polling places had a disparate impact on minority voters. Rodden at 61–68. Hispanic voters were “under-served by polling places relative to the rest of the metro area,” id. at 62, and Hispanic and African American voters were forced to travel greater distances to reach polling places than white, non-Hispanic voters. Id. at 64–68. The reduction in the number of polling places “resulted in extremely long lines of people waiting to vote—some for five hours—and many people leaving the polls, discouraged from voting by the long wait.” Berman at 20.
Second, the district court found that Maricopa County has repeatedly misrepresented or mistranslated key information in Spanish-language voter materials. Reagan, 329 F. Supp. 3d at 875 (“Along with the State’s hostility to bilingual education, Maricopa County has sometimes failed to send properly translated education[al] materials to its Spanish speaking residents, resulting in confusion and distrust from Hispanic voters.”); Berman at 20. In 2012, the official Spanish-language pamphlet in Maricopa County told Spanish-speaking voters that the November 6 election would be held on November 8. Berman at 20. The county did not make the same mistake in its English-language pamphlet. Four years later, Spanish-language ballots in Maricopa County provided an incorrect translation of a ballot proposition. Id.
(2) Voter Registration and Turnout
Voter registration of Arizona’s minority citizens lags behind that of white citizens. In November 2016, close to 75 percent of white citizens were registered to vote in Arizona, compared to 57 percent of Hispanic citizens. See U.S. Census Bureau, Reported Voting and Registration by Sex, Race, and Hispanic Origin for November 2016, tbl. 4b.
Arizona has one of the lowest voter turnout rates in the United States. A 2005 study ranked Arizona forty-seventh out of the fifty States. See Ariz. State Univ., Morrison Inst. for Pub. Policy, How Arizona Compares: Real Numbers and Hot Topics 47 (2005) (relying on Census data); see also Tucker et al., Voting Rights at 359. In 2012, Arizona ranked forty-fourth in turnout for that year’s presidential election. Rodden at 19.
The turnout rate for minority voters is substantially less than that for white voters. In 2002, 59.8 percent of registered Hispanic voters turned out for the election, compared to 72.4 percent of total registered voters. Tucker et al., Voting Rights at 359–60 (relying on Census data). In the 2012 presidential election, 39 percent of Arizona’s Hispanic voting-age population and 46 percent of Arizona’s African American voting-age population turned out for the election, compared to 62 percent of Arizona’s white population. Rodden at 20–21. The national turnout rate for African Americans in that election was 66 percent. Id. In the 2000 and 2004 presidential elections, turnout of Arizona’s American Indian voters was approximately 23 percentage points below the statewide average. Tucker et al., Voting Rights at 360.
(H) District Court’s Assessment of Factor One
The district court recognized Arizona’s history of discrimination, but minimized its significance. Quoting Dr. Berman, the court wrote:
In sum, “[d]iscriminatory action has been more pronounced in some periods of state history than others ... [and] each party (not just one party) has led the charge in discriminating against minorities over the years.” Sometimes, however, partisan objectives are the motivating factor in decisions to take actions detrimental to the voting rights of minorities. “[M]uch of the discrimination that has been evidenced may well have in fact been the unintended consequence of a political culture that simply ignores the needs of minorities.” Arizona’s recent history is a mixed bag of advancements and discriminatory actions.
Id. at 875–76 (alterations in original).
The fact that each party in Arizona “has led the charge in discriminating against minorities” does not diminish the legal significance of that discrimination. Quite the contrary. That fact indicates that racial discrimination has long been deeply embedded in Arizona’s political institutions and that both parties have discriminated when it has served their purposes. Further, the “mixed bag of advancements and discriminatory actions” in “Arizona’s recent history” does not weigh in Arizona’s favor. As Chief Judge Thomas wrote: “Rather, despite some advancements, most of which were mandated by courts or Congress [through Section 5 preclearance], Arizona’s history is marred by discrimination.” DNC, 904 F.3d at 738 (Thomas, C.J., dissenting). The “history of official discrimination” in Arizona and its political subdivisions “touch[ing] the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process” is long, substantial, and unambiguous. Gingles, 478 U.S. at 36–37, 106 S.Ct. 2752 (quoting S. Rep. at 28–29).
The district court clearly erred in minimizing the strength of this factor in Plaintiffs’ favor.
ii. Factor Two: Racially Polarized Voting Patterns
Voting in Arizona is racially polarized. The district court found, “Arizona has a history of racially polarized voting, which continues today.” Reagan, 329 F. Supp. 3d at 876. In recent years, the base of the Republican party in Arizona has been white. Putting to one side “landslide” elections, in statewide general elections from 2004 to 2014, 59 percent of white Arizonans voted for Republican candidates, compared with 35 percent of Hispanic voters. The district court found that in the 2016 general election, exit polls “demonstrate that voting between non-minorities and Hispanics continues to be polarized along racial lines.” Id. In the most recent redistricting cycle, the Arizona Independent Redistricting Commission “found that at least one congressional district and five legislative districts clearly exhibited racially polarized voting.” Id.
Voting is particularly polarized when Hispanic and white candidates compete for the same office. In twelve non-landslide district-level elections in 2008 and 2010 between a Hispanic Democratic candidate and a white Republican candidate, an average of 84 percent of Hispanics, 77 percent of American Indians, and 52 percent of African Americans voted for the Hispanic candidate compared to an average of only 30 percent of white voters.
The district court did not clearly err in assessing the strength of this factor in Plaintiffs’ favor.
iii. Factor Five: Effects of Discrimination
It is undisputed that “members of the minority group[s]” in Arizona “bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process.” Gingles, 478 U.S. at 37, 106 S.Ct. 2752 (quoting S. Rep. at 28–29). The district court found, “Racial disparities between minorities and non-minorities in socioeconomic standing, income, employment, education, health, housing, transportation, criminal justice, and electoral representation have persisted in Arizona.” Reagan, 329 F. Supp. 3d at 876.
The district court made factual findings in four key areas—education, poverty and employment, home ownership, and health. The district court concluded in each area that the effects of discrimination “hinder” minorities’ ability to participate effectively in the political process.
First, the district court wrote:
From 1912 until the Supreme Court’s decision in Brown v. Board of Education, segregated education was widespread throughout Arizona and sanctioned by both the courts and the state legislature. In fact, the Tucson Public Schools only recently reached a consent decree with the DOJ over its desegregation plan in 2013. The practice of segregation also extended beyond schools; it was common place to have segregated public spaces such as restaurants, swimming pools, and theaters. Even where schools were not segregated, Arizona enacted restrictions on bilingual education. As recently as 2000, Arizona banned bilingual education with the passage of Proposition 203.
Arizona has a record of failing to provide adequate funding to teach its non-English speaking students. This underfunding has taken place despite multiple court orders instructing Arizona to develop an adequate funding formula for its programs, including a 2005 order in which Arizona was held in contempt of court for refusing to provide adequate funding for its educational programs. “According to the Education Law Center’s latest National Report Card that provided data for 2013, Arizona ranked 47th among the states in per-student funding for elementary and secondary education.”
Id. at 874–75 (internal citations omitted).
White Arizonans “remain more likely than Hispanics, Native Americans, and African Americans to graduate from high school, and are nearly three times more likely to have a bachelor’s degree than Hispanics and Native Americans.” Id. at 868. “[I]n a recent survey, over 22.4 percent of Hispanics and 11.2 percent of Native Americans rated themselves as speaking English less than ‘very well,’ as compared to only 1.2 percent of non-minorities.” Id. The district court found that, due to “lower levels of [English] literacy and education, minority voters are more likely to be unaware of certain technical [voting] rules, such as the requirement that early ballots be received by the county recorder, rather than merely postmarked, by 7:00 p.m. on Election Day.” Id.
Second, Hispanics and African Americans in Arizona live in poverty at nearly two times the rate of whites. American Indians live in poverty at three times the rate of whites. Id. “Wages and unemployment rates for Hispanics, African Americans, and Native Americans consistently have exceeded non-minority unemployment rates for the period of 2010 to 2015.” Id. The district court found that minority voters are more likely to work multiple jobs, less likely to own a car, and more likely to lack reliable access to transportation, id. at 869, all of which make it more difficult to travel to a polling place—or between an incorrect polling place and a correct polling place.
Third, the district court found that “[i]n Arizona, 68.9 percent of non-minorities own a home, whereas only 32.3 percent of African Americans, 49 percent of Hispanics, and 56.1 percent of Native Americans do so.” Id. at 868. Lower rates of homeownership and correspondingly higher rates of renting and residential mobility contribute to higher rates of OOP voting.
Fourth, the district court found that “[a]s of 2015, Hispanics, Native Americans, and African Americans fared worse than non-minorities on a number of key health indicators.” Id. at 868–69. “Native Americans in particular have much higher rates of disability than non-minorities, and Arizona counties with large Native American populations have much higher rates of residents with ambulatory disabilities.”Id. at 869. “For example, ‘17 percent of Native Americans are disabled in Apache County, 22 percent in Navajo County, and 30 percent in Coconino County.’ ” Id. “Further, ‘11 percent [of individuals] have ambulatory difficulties in Apache County, 13 percent in Navajo County, and 12 percent in Coconino County, all of which contain significant Native American populations and reservations.’ ” Id. (alteration in original). Witnesses credibly testified that ambulatory disabilities—both alone and combined with Arizona’s transportation disparities—make traveling to and between polling locations difficult.
The district court did not clearly err in assessing the strength of this factor in Plaintiffs’ favor.
iv. Factor Six: Racial Appeals in Political Campaigns
Arizona’s “political campaigns have been characterized by overt [and] subtle racial appeals” throughout its history. Gingles, 478 U.S. at 37, 106 S.Ct. 2752 (quoting S. Rep. at 28–29). The district court found that “Arizona’s racially polarized voting has resulted in racial appeals in campaigns.” Reagan, 329 F. Supp. 3d at 876.
For example, when Raul Castro, a Hispanic man, successfully ran for governor in the 1970s, Castro’s opponent, a white man, urged voters to support him instead because “he looked like a governor.” Id. “In that same election, a newspaper published a picture of Fidel Castro with a headline that read ‘Running for governor of Arizona.’ ” Id. In his successful 2010 campaign for State Superintendent of Public Education, John Huppenthal, a white man running against a Hispanic candidate, ran an advertisement in which the announcer said that Huppenthal was “one of us,” was opposed to bilingual education, and would “stop La Raza,” an influential Hispanic civil rights organization. Id. When Maricopa County Attorney Andrew Thomas, a white man, ran for governor in 2014, he ran an advertisement describing himself as “the only candidate who has stopped illegal immigration.” Id. The advertisement “simultaneously show[ed] a Mexican flag with a red strikeout line through it superimposed over the outline of Arizona.” Id. Further, “racial appeals have been made in the specific context of legislative efforts to limit ballot collection.” Id. The district court specifically referred to the “racially charged” LaFaro Video, falsely depicting a Hispanic man, characterized as a “thug,” “acting to stuff the ballot box.” Id.
The district court did not clearly err in assessing the strength of this factor in Plaintiffs’ favor.
v. Factor Seven: Number of Minorities in Public Office
The district court recognized that there has been a racial disparity in elected officials but minimized its importance. The court wrote, “Notwithstanding racially polarized voting and racial appeals, the disparity in the number of minority elected officials in Arizona has declined.” Id. at 877. Citing an expert report by Dr. Donald Critchlow—an expert whose opinion the court otherwise afforded “little weight,” id. at 836—the court wrote, “Arizona has been recognized for improvements in the number of Hispanics and Native Americans registering and voting, as well as in the overall representation of minority elected officials,” id. at 877.
As recounted above, it is undisputed that American Indian, Hispanic, and African American citizens are underrepresented in public office in Arizona. Minorities make up 44 percent of Arizona’s total population, but they hold 25 percent of Arizona’s elected offices. Id. Minorities hold 22 percent of state congressional seats and 9 percent of judgeships. No American Indian or African American has ever been elected to represent Arizona in the United States House of Representatives. Only two minorities have been elected to statewide office in Arizona since the passage of the VRA. Arizona has never elected an American Indian candidate to statewide office. No American Indian, Hispanic, or African American candidate has ever been elected to serve as a United States Senator representing Arizona.
Arizona’s practice of entirely discarding OOP ballots is especially important in statewide and United States Senate elections. Some votes for local offices may be improperly cast in an OOP ballot, given that the voter has cast the ballot in the wrong precinct. But no vote for statewide office or for the United States Senate is ever improperly cast in an OOP ballot. Arizona’s practice of wholly discarding OOP ballots thus has the effect of disproportionately undercounting minority votes, by a factor of two to one, precisely where the problem of under-representation in Arizona is most acute.
The district court clearly erred in minimizing the strength of this factor in Plaintiffs’ favor.
vi. Factor Eight: Officials’ Responsiveness to the Needs of Minority Groups
The district court found that “Plaintiffs’ evidence ... is insufficient to establish a lack of responsiveness on the part of elected officials to particularized needs of minority groups.” Id. In support of its finding, the court cited the activity of one organization, the Arizona Citizens Clean Elections Commission, which “engages in outreach to various communities, including the Hispanic and Native American communities, to increase voter participation” and “develops an annual voter education plan in consultation with elections officials and stakeholders,” and whose current Chairman is an enrolled member of the San Carlos Apache Tribe. Id.
The district court’s finding ignores extensive undisputed evidence showing that Arizona has significantly underserved its minority population. “Arizona was the last state in the nation to join the Children’s Health Insurance Program, which may explain, in part, why forty-six states have better health insurance coverage for children.” DNC, 904 F.3d at 740 (Thomas, C.J., dissenting). Further, “Arizona’s public schools are drastically underfunded; in fact, in 2016 Arizona ranked 50th among the states and the District of Columbia in per pupil spending on public elementary and secondary education.” Id. “Given the well-documented evidence that minorities are likelier to depend on public services[,] ... Arizona’s refusal to provide adequate state services demonstrates its nonresponsiveness to minority needs.” Id.; cf. Myers v. United States, 652 F.3d 1021, 1036 (9th Cir. 2011) (holding that the district court clearly erred when it ignored evidence contradicting its findings).
Further, the district court’s finding is contradicted elsewhere in its own opinion. Earlier in its opinion, the court had written that Arizona has a “political culture that simply ignores the needs of minorities.” Id. at 876 (citation omitted). Later in its opinion, the court referred to “Arizona’s history of advancing partisan objectives with the unintended consequence of ignoring minority interests.” Id. at 882.
The district court clearly erred in finding that this factor does not weigh in Plaintiffs’s favor.
vii. Factor Nine: Tenuousness of Justification of the Policy Underlying the Challenged Restriction
The ninth Senate factor is “whether the policy underlying the state or political subdivision’s use of such voting qualification, prerequisite to voting, or standard, practice or procedure is tenuous.”Gingles, 478 U.S. at 37, 106 S.Ct. 2752 (quoting S. Rep. at 28). The district court found that Arizona’s policy of entirely discarding OOP ballots is justified by the importance of Arizona’s precinct-based system of elections. The court held:
Precinct-based voting helps Arizona counties estimate the number of voters who may be expected at any particular precinct, allows for better allocation of resources and personnel, improves orderly administration of elections, and reduces wait times. The precinct-based system also ensures that each voter receives a ballot reflecting only the races for which that person is entitled to vote, thereby promoting voting for local candidates and issues and making ballots less confusing. Arizona’s policy to not count OOP ballots is one mechanism by which it strictly enforces this system to ensure that precinct-based counties maximize the system’s benefits. This justification is not tenuous.
Reagan, 329 F. Supp. 3d at 878.
The court misunderstood the nature of Plaintiffs’ challenge. Plaintiffs do not challenge Arizona’s precinct-based system of voting. Indeed, their challenge assumes both its importance and its continued existence. Rather, their challenge is to Arizona’s policy, within that system, of entirely discarding OOP ballots. The question before the district court was not the justification for Arizona’s precinct-based system. The question, rather, was the justification for Arizona’s policy of entirely discarding OOP ballots.
There is no finding by the district court that would justify, on any ground, Arizona’s policy of entirely discarding OOP ballots. There is no finding that counting or partially counting OOP ballots would threaten the integrity of Arizona’s precinct-based system. Nor is there a finding that Arizona has ever sought to minimize the number of OOP ballots. The lack of such findings is not surprising given the extreme disparity between OOP voting in Arizona and such voting in other states, as well as Arizona’s role in causing voters to vote OOP by, for example, frequently changing the location of polling places.
The only plausible justification for Arizona’s OOP policy would be the delay and expense entailed in counting OOP ballots, but in its discussion of the Senate factors, the district court never mentioned this justification. Indeed, the district court specifically found that “[c]ounting OOP ballots is administratively feasible.” Id. at 860.
Twenty States, including Arizona’s neighboring States of California, Utah, and New Mexico, count OOP ballots. Id.; Cal. Elec. Code §§ 14310(a)(3), 14310(c)(3), 15350; Utah Code Ann. § 20A-4-107(1)(b)(iii), 2(a)(ii), 2(c); N.M. Stat. Ann § 1-12-25.4(F); N.M. Admin. Code 1.10.22.9(N). The district court wrote: “Elections administrators in these and other states have established processes for counting only the offices for which the OOP voter is eligible to vote.” Reagan, 329 F. Supp. 3d at 861. “Some states, such as New Mexico, use a hand tally procedure, whereby a team of elections workers reviews each OOP ballot, determines the precinct in which the voter was qualified to vote, and marks on a tally sheet for that precinct the votes cast for each eligible office.” Id.; seeN.M. Admin Code 1.10.22.9(H)–(N). “Other states, such as California, use a duplication method, whereby a team of elections workers reviews each OOP ballot, determines the precinct in which the voter was qualified to vote, obtains a new paper ballot for the correct precinct, and duplicates the votes cast on the OOP ballot onto the ballot for the correct precinct.”Reagan, 329 F. Supp. 3d at 861. “Only the offices that appear on both the OOP ballot and the ballot for the correct precinct are copied. The duplicated ballot then is scanned through the optical scan voting machine and electronically tallied.” Id.
Arizona already uses a duplication system, similar to that used in California, for provisional ballots cast by voters eligible to vote in federal but not state elections, as well as for damaged or mismarked ballots that cannot be read by an optical scanner. Id. The district court briefly discussed the time that might be required to count or partially count OOP ballots, but it did not connect its discussion to its consideration of the Senate factors. The court cited testimony of a Pima County election official that the county’s duplication procedure “takes about twenty minutes per ballot.” Id. The court did not mention that this same official had stated in his declaration that the procedure instead takes fifteen minutes per ballot. The court also did not mention that a California election official had testified that it takes a very short time to count or partially count the valid votes on an OOP ballot. That official testified that it takes “several minutes” in California to confirm the voter’s registration—which is done for all provisional ballots, in Arizona as well as in California. Once that is done, the official testified, it takes one to three minutes to duplicate the ballot.
The district court clearly erred in finding that this factor does not weigh in Plaintiffs’ favor.
viii. Assessment of Senate Factors
The district court’s “overall assessment” of the Senate factors was: “In sum, of the germane Senate Factors, the Court finds that some are present in Arizona and others are not.” Id. at 878. Based on this assessment, the court held that Plaintiffs had not carried their burden at step two. The district court clearly erred in so holding. The district court clearly erred in minimizing the strength in favor of Plaintiffs of Senate factors one (official history of discrimination) and seven (number of minorities in public office). Further, the district court clearly erred in finding that Senate factors eight (officials’ responsiveness to the needs of minority groups) and nine (tenuousness of the justification of the policy underlying the challenged provision) do not favor Plaintiffs. Plaintiffs have successfully shown that all of the considered Senate factors weigh in their favor. Most important, plaintiffs have shown that the most pertinent factors, five and nine, weigh very strongly in their favor.
c. Summary
We hold that the district court clearly erred in holding that Plaintiffs’ challenge to Arizona’s OOP policy failed under the results test. We hold that Plaintiffs have carried their burden at both steps one and two. First, they have shown that Arizona’s OOP policy imposes a significant disparate burden on its American Indian, Hispanic, and African American citizens, resulting in the “denial or abridgement of the right” of its citizens to vote “on account of race or color.” 52 U.S.C. § 10301(a). Second, they have shown that, under the “totality of circumstances,” the discriminatory burden imposed by the OOP policy is in part caused by or linked to “social and historical conditions” that have or currently produce “an inequality in the opportunities enjoyed by [minority] and white voters to elect their preferred representatives” and to participate in the political process. Gingles, 478 U.S. at 47, 106 S.Ct. 2752; 52 U.S.C. § 10301(b).
We therefore hold that Arizona’s OOP policy violates the results test of Section 2 of the Voting Rights Act.
3. H.B. 2023 and the Results Test
Uncontested evidence in the district court established that, prior to the enactment of H.B. 2023, a large and disproportionate number of minority voters relied on third parties to collect and deliver their early ballots. Uncontested evidence also established that, beginning in 2011, Arizona Republicans made sustained efforts to limit or eliminate third-party ballot collection. The question is whether the district court clearly erred in holding that H.B. 2023 does not violate the “results test” of Section 2.
a. Step One: Disparate Burden
The question at step one is whether H.B. 2023 results in a disparate burden on a protected class. The district court held that Plaintiffs failed at step one. The district court clearly erred in so holding.
Extensive and uncontradicted evidence established that prior to the enactment of H.B. 2023, third parties collected a large and disproportionate number of early ballots from minority voters. Neither the quantity nor the disproportion was disputed. Numerous witnesses testified without contradiction to having personally collected, or to having personally witnessed the collection of, thousands of early ballots from minority voters. There is no evidence that white voters relied to any significant extent on ballot collection by third parties.
The district court recognized the disparity in third-party ballot collection between minority and white citizens. It wrote that “[t]he Democratic Party and community advocacy organizations ... focused their ballot collection efforts on low-efficacy voters, who trend disproportionately minority.” Reagan, 329 F. Supp. 3d at 870. “In contrast,” the court wrote, “the Republican Party has not significantly engaged in ballot collection as a GOTV strategy.” Id.
The district court nonetheless held that this evidence was insufficient to establish a violation at step one. To justify its holding, the court wrote, “[T]he Court finds that Plaintiffs’ circumstantial and anecdotal evidence is insufficient to establish a cognizable disparity under § 2.” Id. at 868. The court wrote further:
Considering the vast majority of Arizonans, minority and non-minority alike, vote without the assistance of third-parties who would not fall within H.B. 2023’s exceptions, it is unlikely that H.B. 2023’s limitations on who may collect an early ballot cause a meaningful inequality in the electoral opportunities of minorities as compared to non-minorities.
Id. at 871.
First, the court clearly erred in discounting the evidence of third-party ballot collection as merely “circumstantial and anecdotal.” The evidence of third-party ballot collection was not “circumstantial.” Rather, as recounted above, it was direct evidence from witnesses who had themselves acted as third-party ballot collectors, had personally supervised third-party ballot collection, or had personally witnessed third-party ballot collection by others. Nor was the evidence merely “anecdotal.” As recounted above, numerous witnesses provided consistent and uncontradicted testimony about third-party ballot collection they had done, supervised, or witnessed. This evidence established that many thousands of early ballots were collected from minority voters by third parties. The court itself found that white voters did not significantly rely on third-party ballot collection. No better evidence was required to establish that large and disproportionate numbers of early ballots were collected from minority voters.
Second, the court clearly erred by comparing the number of early ballots collected from minority voters to the much greater number of all ballots cast “without the assistance of third parties,” and then holding that the relatively smaller number of collected early ballots did not cause a “meaningful inequality.” Id. at 871. In so holding, the court repeated the clear error it made in comparing the number of OOP ballots to the total number of all ballots cast. Just as for OOP ballots, the number of ballots collected by third parties from minority voters surpasses any de minimis number.
We hold that H.B. 2023 results in a disparate burden on minority voters, and that the district court clearly erred in holding otherwise. We accordingly hold that Plaintiffs have succeeded at step one of the results test.
b. Step Two: Senate Factors
The district court did not differentiate between Arizona’s OOP policy and H.B. 2023 in its discussion of step two. Much of our analysis of the Senate factors for Arizona’s OOP policy applies with equal force to the factors for H.B. 2023. Again, we regard Senate factors five (the effects of discrimination in other areas on minorities access to voting) and nine (the tenuousness of the justification for the challenged voting practices) as particularly important, given the nature of Plaintiffs’ challenge to H.B. 2023. We also regard factor one (history of official discrimination) as important, as it strongly supports our conclusion under factor five. Though “not essential,” Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752, the other less important factors provide “helpful background context.” Husted, 768 F.3d at 555.
We do not repeat here the entirety of our analysis of Arizona’s OOP policy. Rather, we incorporate that analysis by reference and discuss only the manner in which the analysis is different for H.B. 2023.
i. Factor One: History of Official Discrimination Connected to Voting
We recounted above Arizona’s long history of race-based discrimination in voting. H.B. 2023 grows directly out of that history. During the Republicans’ 2011 attempt to limit ballot collection by third parties, Arizona was still subject to preclearance under Section 5. When DOJ asked for more information about whether the relatively innocuous ballot-collection provision of S.B. 1412 had the purpose or would have the effect of denying minorities the right to vote and requested more information, Arizona withdrew the preclearance request. It did so because there was evidence in the record that the provision intentionally targeted Hispanic voters. In 2013, public opposition threatened to repeal H.B. 2305 by referendum. If passed, the referendum would have required that any future bill on the same topic pass the legislature by a supermajority. Republicans repealed H.B. 2305 rather than face a referendum. Finally, after the Supreme Court’s decision in Shelby County eliminated preclearance, Arizona enacted H.B. 2023, making third-party ballot collection a felony. The campaign was marked by race-based appeals, most prominently in the LaFaro Video described above.
As it did with respect to OOP voting, the district court clearly erred in minimizing the strength of this factor in Plaintiffs’ favor.
ii. Factor Two: Racially Polarized Voting Patterns
H.B. 2023 connects directly to racially polarized voting patterns in Arizona. The district court found that “H.B. 2023 emerged in the context of racially polarized voting.” Reagan, 329 F. Supp. 3d at 879. Senator Shooter, who introduced the bill that became S.B. 1412—the predecessor to H.B. 2023—was motivated by the “high degree of racial polarization in his district” and introduced the bill following a close, racially polarized election. Id.
The district court did not clearly err in assessing the strength of this factor in Plaintiffs’ favor.
iii. Factor Five: Effects of Discrimination
H.B. 2023 is closely linked to the effects of discrimination that “hinder” the ability of American Indian, Hispanic, and African American voters “to participate effectively in the political process.” Gingles, 478 U.S. at 37, 106 S.Ct. 2752. The district court found that American Indian, Hispanic, and African American Arizonans “are significantly less likely than non-minorities to own a vehicle, more likely to rely upon public transportation, more likely to have inflexible work schedules, and more likely to rely on income from hourly wage jobs.” Reagan, 329 F. Supp. 3d at 869. In addition, “[r]eady access to reliable and secure mail service is nonexistent in some minority communities.” Id. Minority voters in rural communities disproportionately lack access to outgoing mail, while minority voters in urban communities frequently encounter unsecure mailboxes and mail theft. Id. These effects of discrimination hinder American Indian, Hispanic, and African American voters’ ability to return early ballots without the assistance of third-party ballot collection.
The district court did not clearly err in assessing the strength of this factor in Plaintiffs’ favor.
iv. Factor Six: Racial Appeals in Political Campaigns
The enactment of H.B. 2023 was the direct result of racial appeals in a political campaign. The district court found that “racial appeals [were] made in the specific context of legislative efforts to limit ballot collection.” Id. at 876. Proponents of H.B. 2023 relied on “overt or subtle racial appeals,” Gingles, 478 U.S. at 37, 106 S.Ct. 2752, in advocating for H.B. 2023, including the “racially tinged” LaFaro Video, Reagan, 329 F. Supp. 3d at 876–77 (characterizing the LaFaro Video as one of the primary motivators for H.B. 2023). The district court concluded, “[Senator] Shooter’s allegations and the LaFaro video were successful in convincing H.B. 2023’s proponents that ballot collection presented opportunities for fraud that did not exist for in-person voting.” Reagan, 329 F. Supp. 3d at 880.
The district court did not clearly err in assessing the strength of this factor in Plaintiff’s favor.
v. Factor Seven: Number of Minorities in Public Office
Because Arizona’s OOP policy had a particular connection to the election of minorities to statewide office and to the United States Senate, we concluded that the factor of minorities in public office favored Plaintiffs. That particular connection to statewide office does not exist between H.B. 2023 and election of minorities. However, H.B. 2023 is likely to have a pronounced effect in rural counties with significant American Indian and Hispanic populations who disproportionately lack reliable mail and transportation services, and where a smaller number of votes can have a significant impact on election outcomes. In those counties, there is likely to be a particular connection to election of American Indian and Hispanic candidates to public office.
As it did with respect to OOP voting, the district court clearly erred in minimizing the strength of this factor in Plaintiffs’ favor.
vi. Factor Eight: Officials’ Responsiveness to the Needs of Minority Groups
The district court found that “Plaintiffs’ evidence ... is insufficient to establish a lack of responsiveness on the part of elected officials to particularized needs of minority groups.”Id. at 877. As discussed above, this finding ignores extensive evidence to the contrary and is contradicted by the court’s statements elsewhere in its opinion.
The district court clearly erred in finding that this factor does not weigh in Plaintiffs’ favor.
vii. Factor Nine: Tenuousness of Justification of the Policy Underlying the Challenged Restriction
The district court relied on two justifications for H.B. 2023: That H.B. 2023 is aimed at preventing ballot fraud “by creating a chain of custody for early ballots and minimizing the opportunities for ballot tampering, loss, and destruction”; and that H.B. 2023 is aimed at improving and maintaining “public confidence in election integrity.” Id. at 852. We address these justifications in turn.
First, third-party ballot collection was permitted for many years in Arizona before the passage of H.B. 2023. No one has ever found a case of voter fraud connected to third-party ballot collection in Arizona. This has not been for want of trying. The district court described the Republicans’ unsuccessful attempts to find instances of fraud:
The Republican National Lawyers Association (“RNLA”) performed a study dedicated to uncovering cases of voter fraud between 2000 and 2011. The study found no evidence of ballot collection or delivery fraud, nor did a follow-up study through May 2015. Although the RNLA reported instances of absentee ballot fraud, none were tied to ballot collection and delivery. Likewise, the Arizona Republic conducted a study of voter fraud in Maricopa County and determined that, out of millions of ballots cast in Maricopa County from 2005 to 2013, a total of 34 cases of fraud were prosecuted. Of these, 18 involved a felon voting without her rights first being restored. Fourteen involved non-Arizona citizens voting. The study uncovered no cases of fraud perpetrated through ballot collection.
Id. at 853 (internal citations omitted).
The district court wrote, “[T]here has never been a case of voter fraud associated with ballot collection charged in Arizona.” Id. at 852. “No specific, concrete example of voter fraud perpetrated through ballot collection was presented by or to the Arizona legislature during the debates on H.B. 2023 or its predecessor bills.” Id. at 852–53. “No Arizona county produced evidence of confirmed ballot collection fraud in response to subpoenas issued in this case, nor has the Attorney General’s Office produced such information.” Id. at 853.
Ballot-collection-related fraud was already criminalized under Arizona law when H.B. 2023 was enacted. Collecting and failing to turn in someone else’s ballot was already a class 5 felony. Ariz. Rev. Stat. § 16-1005(F). Marking someone else’s ballot was already a class 5 felony. Id. § 16-1005(A). Selling one’s own ballot, possessing someone else’s ballot with the intent to sell it, knowingly soliciting the collection of ballots by misrepresenting one’s self as an election official, and knowingly misrepresenting the location of a ballot drop-off site were already class 5 felonies. Id. § 16-1005(B)–(E). These criminal prohibitions are still in effect. Arizona also takes measures to ensure the security of early ballots, such as using “tamper evident envelopes and a rigorous voter signature verification procedure.” Reagan, 329 F. Supp. 3d at 854.
The history of H.B. 2023 shows that its proponents had other aims in mind than combating fraud. H.B. 2023 does not forbid fraudulent third-party ballot collection. It forbids non-fraudulent third-party ballot collection. To borrow an understated phrase, the anti-fraud rationale advanced in support of H.B. 2023 “seems to have been contrived.” Dep’t of Commerce v. New York, ––– U.S. ––––, 139 S. Ct. 2551, 2575, 204 L.Ed.2d 978 (2019).
Second, we recognize the importance of public confidence in election integrity. We are aware that the federal bipartisan Commission on Federal Election Reform, charged with building public confidence, recommended inter alia that States “reduce the risks of fraud and abuse in absentee voting by prohibiting ‘third-party’ organizations, candidates, and political party activists from handling absentee ballots.” Building Confidence in U.S. Elections § 5.2 (Sept. 2005). We are aware of the recent case of voter fraud in North Carolina involving collection and forgery of absentee ballots by a political operative hired by a Republican candidate. And we are aware that supporters of H.B. 2023 and its predecessor bills sought to convince Arizona voters, using false allegations and racial innuendo, that third-party ballot collectors in Arizona have engaged in fraud.
Without in the least discounting either the common sense of the bipartisan commission’s recommendation or the importance of public confidence in the integrity of elections, we emphasize, first, that the Supreme Court has instructed us in Section 2 cases to make an “intensely local appraisal.” Gingles, 478 U.S. at 78, 106 S.Ct. 2752. The third-party ballot collection fraud case in North Carolina has little bearing on the case before us. We are concerned with Arizona, where third-party ballot collection has had a long and honorable history, and where the acts alleged in the criminal indictment in North Carolina were illegal under Arizona law before the passage of H.B. 2023, and would still be illegal if H.B. 2023 were no longer the law.
We emphasize, further, that if some Arizonans today distrust third-party ballot collection, it is because of the fraudulent campaign mounted by proponents of H.B. 2023. Those proponents made strenuous efforts to persuade Arizonans that third-party ballot collectors have engaged in election fraud. To the degree that there has been any fraud, it has been the false and race-based claims of the proponents of H.B. 2023. It would be perverse if those proponents, who used false statements and race-based innuendo to create distrust, could now use that very distrust to further their aims in this litigation.
The district court clearly erred in finding that this factor does not weigh in Plaintiffs’ favor. This factor either weighs in Plaintiffs’ favor or is, at best, neutral.
viii. Assessment
The district court made the same overall assessment of the Senate factors in addressing H.B. 2023 as in addressing Arizona’s policy of discarding OOP ballots. As it did with respect to OOP ballots, the court concluded that Plaintiffs had not carried their burden at step two. Here, too, the district court’s conclusion was clearly erroneous. Contrary to the court’s conclusion, Plaintiffs have successfully shown that six of the Senate factors weigh in their favor and that the remaining factor weighs in their favor or is neutral.
c. Summary
We hold that the district court clearly erred in holding that Plaintiffs’ challenge to H.B. 2023 failed under the results test. We hold that Plaintiffs have carried their burden at both steps one and two. First, they have shown that H.B. 2023 imposes a disparate burden on American Indian, Hispanic, and African American citizens, resulting in the “denial or abridgement of the right” of its citizens to vote “on account of race or color.” 52 U.S.C. § 10301(a). Second, they have shown that, under the “totality of circumstances,” the discriminatory burden imposed by H.B. 2023 is in part caused by or linked to “social and historical conditions” that have or currently produce “an inequality in the opportunities enjoyed by [minority] and white voters to elect their preferred representatives” and to participate in the political process. Gingles, 478 U.S. at 47, 106 S.Ct. 2752; 52 U.S.C. § 10301(b).
We therefore conclude that H.B. 2023 violates the results test of Section 2 of the Voting Rights Act.
B. Intent Test: H.B. 2023
As indicated above, uncontested evidence in the district court established that before enactment of H.B. 2023, a large and disproportionate number of minority voters relied on third parties to collect and deliver their early ballots. Uncontested evidence also established that, beginning in 2011, Arizona Republicans made sustained efforts to outlaw third-party ballot collection. After a racially charged campaign, they finally succeeded in passing H.B. 2023. The question is whether the district court clearly erred in holding that H.B. 2023 does not violate the “intent test” of Section 2.
1. The Intent Test
Village of Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252, 97 S.Ct. 555, 50 L.Ed.2d 450 (1977), provides the framework for analyzing a claim of intentional discrimination under Section 2. See, e.g., N.C. State Conference of NAACP v. McCrory, 831 F.3d 204, 220–21 (4th Cir. 2016). Under Arlington Heights, Plaintiffs have an initial burden of providing “[p]roof of racially discriminatory intent or purpose.” Arlington Heights, 429 U.S. at 265, 97 S.Ct. 555. Plaintiffs need not show that discriminatory purpose was the “sole[ ]” or even a “primary” motive for the legislation. Id. Rather, Plaintiffs need only show that discriminatory purpose was “a motivating factor.” Id. at 265–66, 97 S.Ct. 555 (emphasis added).
“Determining whether invidious discriminatory purpose was a motivating factor demands a sensitive inquiry into such circumstantial and direct evidence of intent as may be available.” Id. at 266, 97 S.Ct. 555. “[D]iscriminatory purpose may often be inferred from the totality of the relevant facts, including the fact, if it is true, that the law bears more heavily on one race than another.” Washington v. Davis, 426 U.S. 229, 242, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976). Because “[o]utright admissions of impermissible racial motivation are infrequent[,] ... plaintiffs often must rely upon other evidence,” including the broader context surrounding passage of the legislation. Hunt v. Cromartie, 526 U.S. 541, 553, 119 S.Ct. 1545, 143 L.Ed.2d 731 (1999). “In a vote denial case such as the one here, where the plaintiffs allege that the legislature imposed barriers to minority voting, this holistic approach is particularly important, for ‘[d]iscrimination today is more subtle than the visible methods used in 1965.’ ” N.C. State Conference of NAACP, 831 F.3d at 221 (quoting H.R. Rep. No. 109–478, at 6 (2006)).
Arlington Heights provided a non-exhaustive list of factors that a court should consider. Arlington Heights, 429 U.S. at 266, 97 S.Ct. 555. The factors include (1) the historical background; (2) the sequence of events leading to enactment, including any substantive or procedural departures from the normal legislative process; (3) the relevant legislative history; and (4) whether the law has a disparate impact on a particular racial group. Id. at 266–68, 97 S.Ct. 555.
“Once racial discrimination is shown to have been a ‘substantial’ or ‘motivating’ factor behind enactment of the law, the burden shifts to the law’s defenders to demonstrate that the law would have been enacted without this factor.” Hunter v. Underwood, 471 U.S. 222, 228, 105 S.Ct. 1916, 85 L.Ed.2d 222 (1985). In determining whether a defendant’s burden has been carried, “courts must scrutinize the legislature’s actual non-racial motivations to determine whether they alone can justify the legislature’s choices.” N.C. State Conference of NAACP, 831 F.3d at 221 (emphases in original) (citing Mt. Healthy City Sch. Dist. Bd. of Educ. v. Doyle, 429 U.S. 274, 287, 97 S.Ct. 568, 50 L.Ed.2d 471 (1977); Miss. Univ. for Women v. Hogan, 458 U.S. 718, 728, 102 S.Ct. 3331, 73 L.Ed.2d 1090 (1982)). “In the context of a § 2 discriminatory intent analysis, one of the critical background facts of which a court must take notice is whether voting is racially polarized.” Id. “[I]ntentionally targeting a particular race’s access to the franchise because its members vote for a particular party, in a predictable manner, constitutes discriminatory purpose.” Id. at 222.
2. H.B. 2023 and the Intent Test
a. Arlington Heights Factors and Initial Burden of Proof
The district court wrote, “Having considered [the Arlington Heights] factors, the Court finds that H.B. 2023 was not enacted with a racially discriminatory purpose.” Reagan, 329 F. Supp. 3d at 879. The court then went on to discuss each of the four factors, but did not attach any particular weight to any of them. In holding that the Plaintiffs had not shown that discriminatory purpose was “a motivating factor,” the district court clearly erred.
We address the Arlington Heights factors in turn.
i. Historical Background
“A historical pattern of laws producing discriminatory results provides important context for determining whether the same decisionmaking body has also enacted a law with discriminatory purpose.” N.C. State Conference of NAACP, 831 F.3d at 223–24; see Arlington Heights, 429 U.S. at 267, 97 S.Ct. 555 (“The historical background of the decision is one evidentiary source, particularly if it reveals a series of official actions taken for invidious purposes.”). As recounted above, the Arizona legislature has a long history of race-based discrimination, disenfranchisement, and voter suppression, dating back to Arizona’s territorial days. Further, the history of H.B. 2023 itself reveals invidious purposes.
In addressing the “historical background” factor, the district court mentioned briefly the various legislative efforts to restrict third-party ballot collection that had been “spearheaded” by Senator Shooter, described briefly Senator Shooter’s allegations of third-party ballot fraud, and alluded to the “racially-tinged” LaFaro Video. Reagan, 329 F. Supp. 3d at 879–80. But the district court discounted their importance. We discuss the court’s analysis below, under the third Arlington Heights factor.
ii. Sequence of Events Leading to Enactment
“The specific sequence of events leading up to the challenged decision ... may shed some light on the decisionmaker’s purposes.” Arlington Heights, 429 U.S. at 267, 97 S.Ct. 555. We recounted above the sequence of events leading to the enactment of H.B. 2023. The district court acknowledged this history but again discounted its importance. We discuss the court’s analysis below, under the third Arlington Heights factor.
iii. Relevant Legislative History
“The legislative ... history may be highly relevant, especially where there are contemporary statements by members of the decisionmaking body[.]” Id. at 268, 97 S.Ct. 555. The district court found that legislators voted for H.B. 2023 in response to the “unfounded and often farfetched allegations of ballot collection fraud” made by former Senator Shooter, and the “racially-tinged LaFaro Video.” Reagan, 329 F. Supp. 3d at 880. As Chief Judge Thomas wrote: “Because there was ‘no direct evidence of ballot collection fraud ... presented to the legislature or at trial,’ the district court understood that Shooter’s allegations and the LaFaro Video were the reasons the bill passed.” DNC, 904 F.3d at 748 (Thomas, C.J., dissenting) (quoting Reagan, 329 F. Supp. 3d at 880) (emphasis in original).
Senator Shooter was one of the major proponents of the efforts to limit third-party ballot collection and was influential in the passage of H.B. 2023. Reagan, 329 F. Supp. 3d at 879. According to the district court, Senator Shooter made “demonstrably false” allegations of ballot collection fraud. Id. at 880. Senator Shooter’s efforts to limit ballot collection were motivated in substantial part by the “high degree of racial polarization in his district.” Id. at 879. He was “motivated by a desire to eliminate” the increasingly effective efforts to ensure that Hispanic votes in his district were collected, delivered, and counted. Id.
The LaFaro Video provides even stronger evidence of racial motivation. Maricopa County Republican Chair LaFaro produced a video showing “a man of apparent Hispanic heritage”—a volunteer with a get-out-the-vote organization—apparently dropping off ballots at a polling place. Id. at 876. LaFaro’s voice-over narration included unfounded statements, id. at 877, “that the man was acting to stuff the ballot box” and that LaFaro “knew that he was a thug,” id. at 876. The video was widely distributed. It was “shown at Republican district meetings,” “posted on Facebook and YouTube,” and “incorporated into a television advertisement.” Id. at 877.
The district court used the same rationale to discount the importance of all of the first three Arlington Heights factors. It pointed to the “sincere belief,” held by some legislators, that fraud in third-party ballot collection was a problem that needed to be addressed. The district court did so even though it recognized that the belief was based on the false and race-based allegations of fraud by Senator Shooter and other proponents of H.B. 2023. The court wrote: “Shooter’s allegations and the LaFaro Video were successful in convincing H.B. 2023’s proponents that ballot collection presented opportunities for fraud that did not exist for in-person voting[.]” Id. at 880.
We accept the district court’s conclusion that some members of the legislature who voted for H.B. 2023 had a sincere, though mistaken, non-race-based belief that there had been fraud in third-party ballot collection, and that the problem needed to be addressed. However, as the district court found, that sincere belief had been fraudulently created by Senator Shooter’s false allegations and the “racially-tinged” LaFaro video. Even though some legislators did not themselves have a discriminatory purpose, that purpose may be attributable to their action under the familiar “cat’s paw” doctrine. The doctrine is based on the fable, often attributed to Aesop, in which a clever monkey induces a cat to use its paws to take chestnuts off of hot coals for the benefit of the monkey.
For example, we wrote in Mayes v. WinCo Holdings, Inc., 846 F.3d 1274 (9th Cir. 2017):
[T]he animus of a supervisor can affect an employment decision if the supervisor “influenced or participated in the decisionmaking process.” Dominguez-Curry [v. Nev. Transp. Dep’t], 424 F.3d [1027,] 1039–40 [ (9th Cir. 2017)]. Even if the supervisor does not participate in the ultimate termination decision, a “supervisor’s biased report may remain a causal factor if the independent investigation takes it into account without determining that the adverse action was, apart from the supervisor’s recommendation, entirely justified.” Staub v. Proctor Hosp., 562 U.S. 411, 421, 131 S.Ct. 1186, 179 L.Ed.2d 144 (2011).
Id. at 1281; see also Poland v. Chertoff, 494 F.3d 1174, 1182 (9th Cir. 2007) (“[I]f a subordinate ... sets in motion a proceeding by an independent decisionmaker that leads to an adverse employment action, the subordinate’s bias is imputed to the employer if the plaintiff can prove that the allegedly independent adverse employment decision was not actually independent because the biased subordinate influenced or was involved in the decision or decisionmaking process.”).
The good-faith belief of these sincere legislators does not show a lack of discriminatory intent behind H.B. 2023. Rather, it shows that well meaning legislators were used as “cat’s paws.” Convinced by the false and race-based allegations of fraud, they were used to serve the discriminatory purposes of Senator Shooter, Republican Chair LaFaro, and their allies.
We hold that the district court clearly erred in discounting the importance of the first three Arlington Heights factors. We hold that all three factors weigh in favor of showing that discriminatory intent was a motivating factor in enacting H.B. 2023.
iv. Disparate Impact on a Particular Racial Group
“The impact of the official action[,] whether it ‘bears more heavily on one race than another,’ may provide an important starting point. Sometimes a clear pattern, unexplainable on grounds other than race, emerges from the effect of the state action even when the governing legislation appears neutral on its face.” Arlington Heights, 429 U.S. at 266, 97 S.Ct. 555 (internal citation omitted). As described above, uncontested evidence shows that H.B. 2023 has an adverse and disparate impact on American Indian, Hispanic, and African American voters. The district court found that the legislature “was aware” of the impact of H.B. 2023 on what the court called “low-efficacy minority communities.” Reagan, 329 F. Supp. 3d at 881.
It appears that the district court weighed this factor in favor of showing discriminatory intent as a motivating factor in enacting H.B. 2023. The court did not clearly err in so doing.
v. Assessment
We hold that the district court clearly erred in holding that Plaintiffs failed to carry their initial burden of proof of showing that racial discrimination was a motivating factor leading to the enactment of H.B. 2023. We hold that all four of the Arlington Heights factors weigh in favor of Plaintiffs. Our holding does not mean that the majority of the Arizona state legislature “harbored racial hatred or animosity toward any minority group.” N.C. State Conference of NAACP, 831 F.3d at 233. “But the totality of the circumstances”—Arizona’s long history of race-based voting discrimination; the Arizona legislature’s unsuccessful efforts to enact less restrictive versions of the same law when preclearance was a threat; the false, race-based claims of ballot collection fraud used to convince Arizona legislators to pass H.B. 2023; the substantial increase in American Indian and Hispanic voting attributable to ballot collection that was targeted by H.B. 2023; and the degree of racially polarized voting in Arizona—“cumulatively and unmistakably reveal” that racial discrimination was a motivating factor in enacting H.B. 2023. Id.
b. Would H.B. 2023 Otherwise Have Been Enacted
At the second step of the Arlington Heights analysis, Arizona has the burden of showing that H.B. 2023 would have been enacted without racial discrimination as a motivating factor. Because the district court held that Plaintiffs had not carried their initial burden, it did not reach the second step of the Arlington Heights analysis.
Although there is no holding of the district court directed to Arlington Heights’ second step, the court made a factual finding that H.B. 2023 would not have been enacted without racial discrimination as a motivating factor. The court specifically found that H.B. 2023 would not have been enacted without Senator Shooter’s and LaFaro’s false and race-based allegations of voter fraud. The court wrote, “The legislature was motivated by a misinformed belief that ballot collection fraud was occurring, but a sincere belief that mail-in ballots lacked adequate prophylactic safeguards as compared to in-person voting.” Reagan, 329 F. Supp. 3d at 882. That is, members of the legislature, based on the “misinformed belief” created by Shooter, LaFaro, and their allies and serving as their “cat’s paws,” voted to enact H.B. 2023. See Poland, 494 F.3d at 1182. Based on the court’s finding, we hold that Arizona has not carried its burden of showing that H.B. 2023 would have been enacted without the motivating factor of racial discrimination.
c. Summary
We hold that the district court clearly erred in holding that Plaintiffs failed to establish that H.B. 2023 violates the intent test of Section 2 of the VRA. A holding that H.B. 2023 violates the intent test of Section 2 necessarily entails a holding that it also violates the Fifteenth Amendment.
III. Response to Dissents
We respectfully disagree with our dissenting colleagues. For the most part, our response to their contentions is contained in the body of our opinion and needs no elaboration. Several contentions, however, merit a specific response.
A. Response to the First Dissent
Our first dissenting colleague, Judge O’Scannlain, makes several mistakes.
First, our colleague contends that H.B. 2023 does not significantly change Arizona law. Our colleague writes:
For years, Arizona has restricted who may handle early ballots. Since 1992, Arizona has prohibited anyone but the elector himself from possessing “that elector’s unvoted absentee ballot.” 1991 Ariz. Legis. Serv. Ch. 310, § 22 (S.B. 1390) (West). In 2016, Arizona enacted a parallel regulation, H.B. 2023 (the “ballot-collection” policy), concerning the collection of early ballots.
Diss. Op. at –––– – –––– (emphases added).
Our colleague appends a footnote to the first sentence in the passage just quoted:
The majority’s effort to deny history can easily be dismissed. Maj. Op. 104–105. As Judge Bybee’s dissent ably recounts, not only Arizona but 21 other states have restricted early balloting for years. Bybee, J. Diss. Op. –––– – ––––.
Our colleague fails to recognize the distinction between “unvoted” and “voted” ballots. Contrary to our colleague’s contention, H.B. 2023 is not “a parallel regulation” to already existing Arizona law. Under prior Arizona law, possession of an “unvoted absentee ballot” was forbidden. Arizona law in no way restricted non-fraudulent possession of voted absentee ballots (absentee ballots on which the vote had already been indicated). Unlike our colleague, the district court recognized the distinction. It wrote:
Since 1997, it has been the law in Arizona that “[o]nly the elector may be in possession of that elector’s unvoted early ballot.” A.R.S. § 16-542(D). In 2016, Arizona amended A.R.S. § 16-1005 by enacting H.B. 2023, which limits who may collect a voter’s voted or unvoted early ballot.
Reagan, 329 F. Supp. 3d at 839 (emphases added). H.B. 2023 for the first time forbade non-fraudulent collection of voted ballots. It was not a “parallel regulation.” It was a fundamental change in Arizona law.
Second, our colleague repeats the potentially misleading numbers and percentages of OOP voting recounted by the district court. Our colleague writes:
Only 0.47 percent of all ballots cast in the 2012 general election (10,979 out of 2,323,579) were not counted because they were cast out of the voter’s assigned precinct. [Reagan, 329 F. Supp. 3d] at 872. In 2016, this fell to 0.15 percent (3,970 out of 2,661,497). Id.
Diss. Op. at –––– – ––––. Our colleague, like the district court, see Reagan, 329 F. Supp. 3d at 872, fails to mention that, as a percentage of all in-person ballots, OOP ballots increased between 2012 and 2016.
Third, our colleague quotes from a sentence in a footnote in the Supreme Court’s opinion in Gingles. Based on that sentence, he insists that “substantial difficulty electing representatives of their choice” is the governing standard for the Section 2 results test in the case before us. Our colleague writes:
[In Gingles], the Court observed that “[i]t is obvious that unless minority group members experience substantial difficulty electing representatives of their choice, they cannot prove that a challenged electoral mechanism impairs their ability ‘to elect.’ ” Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752 (quoting 52 U.S.C. § 10301(b)) (emphasis added).
Diss. Op. at –––– (emphasis in original). He later writes:
Given the lack of any testimony in the record indicating that the ballot-collection policy would result in minority voters ‘experienc[ing] substantial difficulty electing representatives of their choice,’ Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752, the district court did not clearly err[.]
Id. at –––– (emphasis added).
Our colleague fails to distinguish between a vote dilution case and a vote denial case. As we noted above, a vote dilution case is one in which multimember electoral districts have been formed, or in which district lines have been drawn, so as to dilute and thereby diminish the effectiveness of minority votes. Vote denial cases are all other cases, including cases in which voters are prevented from voting or in which votes are not counted. Gingles was a vote dilution case, and the case before us is a vote denial case. Our colleague fails to quote the immediately preceding sentence in the Gingles footnote, which makes clear that the Court was addressing vote dilution cases. The Court wrote, “In recognizing that some Senate Report factors are more important to multimember district vote dilution claims than others, the Court effectuates the intent of Congress.” Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752 (emphasis added).
The standard in a vote denial case is different, as recognized by DOJ in its amicus brief in this case, and in League of Women Voters where the Fourth Circuit struck down a state statute that would have prevented the counting of OOP ballots in North Carolina without inquiring into whether the number of affected ballots was likely to affect election outcomes. See769 F.3d at 248–49. As we noted above, there may be a de minimis number in vote denial cases challenging facially neutral policies or law, but the 3,709 OOP ballots in our case is above any such de minimis number.
Citing our en banc decision in Gonzalez, our colleague contends that our case law does not differentiate between vote denial and vote dilution cases. But the very language from Gonzalez that he quotes belies his contention. We wrote in text:
[A] § 2 challenge “based purely on a showing of some relevant statistical disparity between minorities and whites,” without any evidence that the challenged voting qualification causes that disparity, will be rejected.
Gonzalez, 677 F.3d at 405. We then appended a footnote, upon which our colleague relies:
This approach applies both to claims of vote denial and of vote dilution. [Smith v. Salt River Project Agric. Improvement & PowerDist., 109 F.3d 586,] 596 n.8 [ (9th Cir. 1997)].
Id. at 405 n.32. The quoted language makes the obvious point that in both vote denial and vote dilution cases, we require evidence of a causal relation between a challenged voting qualification and any claimed statistical disparity between minority and white voters. However, this language does not tell us that the predicate disparity, and its effect, are the same in vote denial and vote dilution cases.
B. Response to the Second Dissent
Our second dissenting colleague, Judge Bybee, writes “to make a simple point: The Arizona rules challenged here are part of an ‘electoral process that is necessarily structured to maintain the integrity of the democratic system.’ ” Diss. Op. at –––– (quoting Burdick v. Takushi, 504 U.S. 428, 441, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992)). We respectfully disagree. There is nothing in Arizona’s policy of discarding OOP votes or about H.B. 2023 that is necessary “to maintain the integrity” of Arizona’s democratic system.
Our colleague writes, further, “Parties of all stripes should have an equal interest in rules that are both fair on their face and fairly administered.” Id. at ––––. Our colleague misunderstands the purpose of the VRA’s results test of Section 2. The results test looks past the facial fairness of a law to its actual results.
We take these two points in turn.
1. Integrity of Arizona’s Democratic System
First, our colleague uses his “simple point” to justify Arizona’s OOP policy and H.B. 2023 on the ground that they are necessary to protect the integrity of Arizona’s system.
Our colleague argues that eliminating Arizona’s OOP policy will “lower[ ] the cost to voters of determining where they are supposed to vote, but only as to presidential, U.S. Senate, and statewide races,” and will have “its own consequences.” Id. at ––––, ––––. To illustrate those consequences, our colleague imagines a voter from Tuscon who votes in Phoenix. Based on his imagined voter, he posits “two predictable ways” in which future elections in Arizona will be “skew[ed]” if OOP votes are counted for the elections in which the voter is entitled to vote. Id. at ––––. Because his imagined voter cares only about national elections, that voter “may vote with impunity in the wrong precinct.” Id. at ––––. This will result, first, in “overvalu[ing]” national elections, and, second, in “undervalu[ing]” local elections. Id.
Our colleague speculates that Arizona’s OOP policy will result in voters either finding the right precinct, or voting by mail. He writes:
Under Arizona’s current OOP rule, a voter, having gone to the trouble of going to a precinct to vote in person and suffering the indignity of having to fill out a provisional ballot, is less likely to make the same mistake next year. A voter who has had a ballot disqualified is more likely to figure out the correct precinct next time—or, better yet, sign up for the convenience of early voting, a measure that avoids the conundrum of OOP altogether.
Id. at ––––.
Our colleague’s speculation leads him to predict that Arizona’s OOP policy will lead to increased in-precinct voting. There is nothing in the record that remotely supports our colleague’s predicted consequences. Instead, the record clearly shows the opposite. Arizona’s OOP policy has been in place since at least 1970. Reagan, 329 F. Supp. 3d at 840. The record shows that, despite its long-standing policy, Arizona has consistently had by far the highest rate of OOP voting of any State—in 2012, eleven times greater than the second-place State. See Figure 6, supra at ––––; see also Rodden at 26 (describing OOP voting as a “persistent problem” in Arizona).
Contrary to our colleague’s speculation, OOP voters are unlikely ever to discover the “indignity” of having their provisional ballots discarded. Our colleague quotes from an Arizona statute requiring county recorders to establish a “method” by which a voter casting a provisional ballot be notified that his or ballot was not counted, and giving a reason why it was not counted. Diss. Op. at –––– n.9. However, there is nothing in the record showing that county recorders have in fact established, or followed, such a “method.” Instead, there was uncontradicted testimony in the district court by OOP voters that they were not directed to their proper polling place and were never told that their vote would not be counted if cast out of precinct. See Reagan, 329 F. Supp. 3d at 858 (finding that poll workers neither directed OOP voters to the correct precinct nor told voters that OOP ballots would be discarded).
The persistence of OOP voting is unsurprising given the actions of Arizona. Arizona changes polling places with extraordinary frequency, and often locates them in inconvenient and misleading places. This produces a high rate of OOP voting, particularly in urban areas and particularly for voters with high rates of residential mobility. The uncontested result is that minority voters cast OOP votes twice as often as white voters.
Our colleague further argues that H.B. 2023 is an appropriate measure to protect against voter fraud. He begins by pointing out that many States forbid third-party ballot collection. Diss. Op. at –––– – ––––. But a simple numerical comparison with other states fails to take into account, as the VRA says we must, the particular geography, ethnic patterns, and long history of third-party ballot collection in Arizona. See Gingles, 478 U.S. at 78, 106 S.Ct. 2752 (a Section 2 analysis requires “a blend of history and an intensely local appraisal”). Evidence in the record shows that third-party ballot collection has long had a unique role in Arizona, given the large numbers of Hispanic and American Indian voters who have unreliable or non-existent in-home mail service, who have unreliable means of transportation, who live long distances from polling places, and who have long-standing cultural traditions of ballot collection. Evidence in the record shows that Arizona has never, in its long history of third-party ballot collection, found a single case of fraud.
Our colleague also argues that Arizona should not ignore the recommendation of the report of the bipartisan commission, Building Confidence in U.S. Elections (2005). Diss. Op. at –––– – ––––. This is a reasonable argument, but it has limited force when applied to Arizona. Forbidding third-party ballot collection protects against potential voter fraud. But such protection is not necessary, or even appropriate, when there is a long history of third-party ballot collection with no evidence, ever, of any fraud and such fraud is already illegal under existing Arizona law. Such protection is undesirable, even illegal, when a statute forbidding third-party ballot collection produces a discriminatory result or is enacted with discriminatory intent. The commission was concerned with maintaining “confidence” in our election system, as indicated by the title of its report. If there is a lack of confidence in third-party ballot collection in Arizona, it is due to the fraudulent, race-based campaign mounted by the proponents of H.B. 2023.
Finally, our colleague points to third-party ballot collection fraud perpetrated by a Republican political operative in North Carolina. Id. at –––– – ––––. Our colleague’s argument ignores the different histories and political cultures in Arizona and North Carolina, and puts to one side as irrelevant the long and honorable history of third-party ballot collection in Arizona. The argument also ignores the fact that Arizona had long had statutes prohibiting fraudulent handling of both unvoted and voted ballots by third parties, even before the enactment of H.B. 2023. The actions of the North Carolina Republican operative, if performed in Arizona, would have been illegal under those statutes. H.B. 2023 does not forbid fraudulent third-party ballot collection. Such fraud is forbidden by other provisions of Arizona law. H.B. 2023 forbids non-fraudulent third-party ballot collection.
2. Rules that Are Fair on Their Face
Second, our colleague defends Arizona’s OOP policy and H.B. 2023 as “rules that are ... fair on their face.” Id. at ––––. The results test of Section 2 of the VRA is based on the understanding that laws that are “fair on their face” can, as in this case, produce discriminatory results. Indeed, Congress added the results test to the VRA precisely to address laws that were fair on their face but whose result was unfair discrimination.
Arizona’s OOP policy and H.B. 2023 both fail the results test. The result of Arizona’s OOP policy is that twice as many minority ballots as white ballots are thrown away. Prior to the enactment of H.B. 2023, third-party ballot collectors, acting in good faith, collected many thousands of valid ballots cast by minority voters. White voters rarely relied on third-party ballot collection. The result of H.B. 2023 is that many thousands of minority ballots will now not be collected and counted, while white ballots will be largely unaffected.
IV. Conclusion
We hold that Arizona’s OOP policy violates the results test of Section 2. We hold that H.B. 2023 violates both the results test and the intent test of Section 2. We hold that H.B. 2023 also violates the Fifteenth Amendment. We do not reach Plaintiffs’ other constitutional challenges.
We reverse the judgment of the district court and remand for further proceedings consistent with this opinion.
REVERSED and REMANDED.
WATFORD, Circuit Judge, concurring:
I join the court’s opinion to the extent it invalidates Arizona’s out-of-precinct policy and H.B. 2023 under the results test. I do not join the opinion’s discussion of the intent test.
O’SCANNLAIN, Circuit Judge, with whom CLIFTON, BYBEE, and CALLAHAN, Circuit Judges, join, dissenting:
We have been asked to decide whether two current Arizona election practices violate the Voting Rights Act or the First, Fourteenth, or Fifteenth Amendments to the United States Constitution.1 Based on the record before us and relevant Supreme Court and Ninth Circuit precedent, the answer to such question is clear: they do not. The majority, however, draws factual inferences that the evidence cannot support and misreads precedent along the way. In so doing, it impermissibly strikes down Arizona’s duly enacted policies designed to enforce its precinct-based election system and to regulate third-party collection of early ballots.
I respectfully dissent.
I
Given the abundant discussion by the district court and the en banc majority, I offer only a brief summary of the policies at issue here and discuss the district court’s factual findings as pertinent to the analysis below.
A
Arizona offers voters several options: early mail ballot, early in-person voting, and in-person Election Day voting. Democratic Nat’l Comm. v. Reagan (“DNC”), 329 F. Supp. 3d 824, 838 (D. Ariz. 2018).
1
Since at least 1970, Arizona has required that in-person voters “cast their ballots in their assigned precinct and has enforced this system by counting only those ballots cast in the correct precinct.” Id. at 840. A voter who arrives at a precinct in which he or she is not listed on the register may cast a provisional ballot, but Arizona will not count such ballot if it determines that the voter does not live in the precinct in which he or she voted. Id. For shorthand, I refer to this rule as Arizona’s “out-of-precinct” or “OOP” policy.
Most Arizona voters, however, do not vote in person on Election Day. Id. at 845. Arizona law permits all registered voters to vote early by mail or in person at an early voting location in the 27 days before an election. Ariz. Rev. Stat. §§ 16-121(A), 16-541(A), 16-542(D). All Arizona counties operate at least one location for early in person voting. DNC, 329 F. Supp. 3d at 839. Rather than voting early in person, any voter may instead request an early ballot to be delivered to his or her mailbox on an election-by-election or permanent basis. Id. In 2002, Arizona became the first state to make available an online voter registration option, which also permits voters to enroll in permanent early voting by mail. Id. Voters who so enroll will be sent an early ballot no later than the first day of the 27-day early voting period. Id. Voters may return early ballots in person at any polling place, vote center, or authorized office without waiting in line or may return their early ballots by mail at no cost. Id. To be counted, however, an early ballot must be received by 7:00 p.m. on Election Day. Id.
2
For years, Arizona has restricted who may handle early ballots.2 Since 1992, Arizona has prohibited anyone but the elector himself from possessing “that elector’s unvoted absentee ballot.” 1991 Ariz. Legis. Serv. Ch. 310, § 22 (S.B. 1390) (West). In 2016, Arizona enacted a parallel regulation, H.B. 2023 (the “ballot-collection” policy), concerning the collection of early ballots.3DNC, 329 F. Supp. 3d at 839. Under the ballot-collection policy, only a “family member,” “household member,” “caregiver,” “United States postal service worker” or other person authorized to transmit mail, or “election official” may return another voter’s completed early ballot. Id. at 839–40 (citing Ariz. Rev. Stat. § 16-1005(H)–(I)).
B
In April 2016, the Democratic National Committee, the Democratic Senatorial Campaign Committee, and the Arizona Democratic Party (together, “DNC”) sued the State of Arizona to challenge the OOP policy and the ballot-collection policy. The district court denied DNC’s motions to enjoin preliminarily enforcement of both polices, and DNC asked our court to issue injunctions pending appeal of such denials. After expedited proceedings before three-judge and en banc panels, our court denied the motion for an injunction against the OOP policy but granted the parallel motion against the ballot-collection policy. Feldman v. Ariz. Sec’y of State’s Office, 840 F.3d 1165 (9th Cir. 2016) (en banc) (mem.) (per curiam); Feldman v. Ariz. Sec’y of State’s Office (Feldman III), 843 F.3d 366 (9th Cir. 2016) (en banc). The Supreme Court, however, stayed our injunction against the ballot-collection policy and the OOP and ballot-collection policies functioned in usual fashion. Ariz. Sec’y of State’s Office v. Feldman, ––– U.S. ––––, 137 S. Ct. 446, 196 L.Ed.2d 326 (2016) (mem.).
In 2017, the district court proceeded to the merits of DNC’s suit. In May 2018, after a ten-day bench trial, the district court issued a decision supported by thorough findings of fact and conclusions of law. DNC, 329 F. Supp. 3d at 832. The district court found that DNC failed to prove any violation of the Voting Rights Act or the United States Constitution and issued judgment in the state’s favor. Id. at 882–83.
DNC timely appealed, and a three-judge panel of our court affirmed the decision of the district court in its entirety. Democratic Nat’l Comm. v. Reagan (“DNC”), 904 F.3d 686 (9th Cir. 2018), vacated by order granting rehearing en banc, 911 F.3d 942 (9th Cir. 2019) (mem.). But today, the en banc panel majority reverses the decision of the district court and holds that the OOP and ballot-collection policies violate § 2 of the Voting Rights Act and that the ballot-collection policy was enacted with discriminatory intent in violation of the Fifteenth Amendment.
II
The first mistake of the en banc majority is disregarding the critical standard of review. Although the majority recites the appropriate standard, it does not actually engage with it.4 Maj. Op. –––– – ––––. The standard is not complex. We review de novo the district court’s conclusions of law, but may review its findings of fact only for clear error. Navajo Nation v. U.S. Forest Serv., 535 F.3d 1058, 1067 (9th Cir. 2008) (en banc).
The majority’s disregard of such standard and, thus, our appellate role, infects its analysis of each of DNC’s claims. The demanding clear error standard “plainly does not entitle a reviewing court to reverse the finding of the trier of fact simply because it is convinced that it would have decided the case differently.” Anderson v. City of Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985). Rather, we may reverse a finding only if, “although there is evidence to support it, [we are] left with the definite and firm conviction that a mistake has been committed.” Id. (quoting United States v. U. S. Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 92 L.Ed. 746 (1948)). To do otherwise “oversteps the bounds of [our] duty under [Federal Rule of Civil Procedure] 52(a)” by “duplicat[ing] the role of the lower court.” Id. at 573, 105 S.Ct. 1504. As explained in Parts III and IV, I fail to see how on the record before us one could be “left with a definite and firm conviction” that the district court erred.
III
DNC first contends that Arizona’s policies violate § 2 of the Voting Rights Act. A district court’s determination of whether a challenged practice violates § 2 of the Voting Rights Act is “intensely fact-based”: the court assesses the “totality of the circumstances” and conducts “a ‘searching practical evaluation of the past and present reality.’ ” Smith v. Salt River Project Agric. Improvements & Power Dist. (“Salt River”), 109 F.3d 586, 591 (9th Cir. 1997) (quoting Thornburg v. Gingles, 478 U.S. 30, 79, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986)). Thus, “[d]eferring to the district court’s superior fact-finding capabilities, we review only for clear error its ultimate finding of no § 2 violation.” Id. at 591 (emphasis added).
In relevant part, § 2 provides:
(a) No voting qualification or prerequisite to voting or standard, practice, or procedure shall be imposed or applied by any State ... in a manner which results in a denial or abridgment of the right of any citizen of the United States to vote on account of race or color ....
(b) A violation of subsection (a) is established if, based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State ... are not equally open to participation by members of a class of citizens protected by subsection (a) in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.
52 U.S.C. § 10301 (emphasis added). “The essence of a § 2 claim is that a certain electoral law, practice, or structure interacts with social and historical conditions to cause an inequality in the opportunities enjoyed by black and white voters to elect their preferred representatives.” Gingles, 478 U.S. at 47, 106 S.Ct. 2752. To determine whether a practice violates § 2, courts employ a two-step analysis. See Ohio Democratic Party v. Husted, 834 F.3d 620, 637 (6th Cir. 2016); Veasey v. Abbott, 830 F.3d 216, 244 (5th Cir. 2016); Frank v. Walker, 768 F.3d 744, 754–55 (7th Cir. 2014); League of WomenVoters of N.C. v. North Carolina, 769 F.3d 224, 240 (4th Cir. 2014).
The first step is asking whether the practice provides members of a protected class “less ‘opportunity’ than others ‘to participate in the political process and to elect representatives of their choice.’ ” Chisom v. Roemer, 501 U.S. 380, 397, 111 S.Ct. 2354, 115 L.Ed.2d 348 (1991) (alteration in original) (quoting 52 U.S.C. § 10301). In other words, the challenged practice “must impose a discriminatory burden on members of a protected class.” League of Women Voters, 769 F.3d at 240 (emphasis added). To prevail at step one, the plaintiff therefore “must show a causal connection between the challenged voting practice and [a] prohibited discriminatory result.” Salt River, 109 F.3d at 595 (alteration in original) (quoting Ortiz v. City of Phila. Office of City Comm’rs Voter Registration Div., 28 F.3d 306, 312 (3d Cir. 1994)); see also Ohio Democratic Party, 834 F.3d at 638. If a discriminatory burden is established, then—and only then—do we consider whether the burden is “caused by or linked to ‘social and historical conditions’ that have or currently produce discrimination against members of the protected class.” League of Women Voters, 769 F.3d at 240 (quoting Gingles, 478 U.S. at 47, 106 S.Ct. 2752).
The majority agrees that this two-step analysis controls but mistakenly applies it. According to the majority, DNC has shown that the OOP policy and the ballot-collection policy fail at both steps—and, presumably, that the district court clearly erred in finding otherwise. Under an appropriately deferential analysis, however, DNC cannot prevail even at step one: it has simply failed to show that either policy erects a discriminatory burden.
A
As to the facially neutral OOP policy, DNC argues, erroneously, that wholly discarding, rather than partially counting, ballots that are cast out-of-precinct violates § 2 of the Voting Rights Act because such policy imposes a discriminatory burden on minority voters related to Arizona’s history of discrimination. The district court, quite properly, found that DNC failed to carry its burden at step one—that the practice imposes a discriminatory burden on minority voters—for two reasons. DNC, 329 F. Supp. 3d at 873.
1
First, the district court determined that DNC failed to show “that the racial disparities in OOP voting are practically significant enough to work a meaningful inequality in the opportunities of minority voters as compared to non-minority voters.” Id. Thus, it ruled that DNC failed to show that the precinct-based system has a “disparate impact on the opportunities of minority voters to elect their preferred representatives.” Id. at 872. To the contrary, the district court made the factual finding that out-of-precinct “ballots represent ... a small and ever-decreasing fraction of the overall votes cast in any given election.” Id.
Furthermore, the district court determined that “the burdens imposed by precinct-based voting ... are not severe. Precinct-based voting merely requires voters to locate and travel to their assigned precincts, which are ordinary burdens traditionally associated with voting.” Id. at 858. Indeed, the numbers found by the district court support such conclusion. Only 0.47 percent of all ballots cast in the 2012 general election (10,979 out of 2,323,579) were not counted because they were cast out of the voter’s assigned precinct. Id. at 872. In 2016, this fell to 0.15 percent (3,970 out of 2,661,497). Id. And of those casting ballots in-person on Election Day, approximately 99 percent of minority voters and 99.5 percent of non-minority voters cast their ballots in their assigned precincts. Id. Given that the overwhelming majority of all voters complied with the precinct-based voting system during the 2016 election, it is difficult to see how the district court’s finding could be considered clearly erroneous. See also Crawford v. Marion Cty. Election Bd., 553 U.S. 181, 198, 128 S.Ct. 1610, 170 L.Ed.2d 574 (2008) (plurality opinion) (discussing “the usual burdens of voting”). And it further ruled that DNC “offered no evidence of a systemic or pervasive history of minority voters being given misinformation regarding the locations of their assigned precincts, while non-minority voters were given correct information” to suggest that the burden of voting in one’s assigned precinct is more significant for minority voters than for non-minority voters. DNC, 329 F. Supp. 3d at 873.
As Judge Ikuta explained in her now-vacated majority opinion for the three-judge panel:
If a challenged election practice is not burdensome or the state offers easily accessible alternative means of voting, a court can reasonably conclude that the law does not impair any particular group’s opportunity to “influence the outcome of an election,” even if the practice has a disproportionate impact on minority voters.
DNC, 904 F.3d at 714 (citation omitted) (quoting Chisom, 501 U.S. at 397 n.24, 111 S.Ct. 2354). The “bare statistic[s]” presented may indeed show a disproportionate impact on minority voters, but we have held previously that such showing is not enough. Salt River, 109 F.3d at 595 (“[A] bare statistical showing of disproportionate impact on a racial minority does not satisfy the § 2 ‘results’ inquiry.” (emphasis in original)). A court must evaluate the burden imposed by the challenged voting practice—not merely any statistical disparity that may be shown. The Supreme Court’s interpretation of § 2 in Gingles suggests the same. There, the Court observed that “[i]t is obvious that unless minority group members experience substantial difficulty electing representatives of their choice, they cannot prove that a challenged electoral mechanism impairs their ability ‘to elect.’ ” Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752 (emphasis added) (quoting 52 U.S.C. § 10301(b)). Furthermore, because “[n]o state has exactly equal registration rates, exactly equal turnout rates, and so on, at every stage of its voting system,” it cannot be the case that pointing to a mere statistical disparity related to a challenged voting practice is sufficient to “dismantle” that practice. Frank, 768 F.3d at 754; see also Salt River, 109 F.3d at 595.
The majority, however, contends that “the district court discounted the disparate burden on the ground that there were relatively few OOP ballots cast in relation to the total number of ballots.” Maj. Op. ––––. In the majority’s view, the district court should have emphasized that the percentage of in-person ballots that were cast out-of-precinct increased, thus isolating the specific impact of the OOP policy amongst in-person voters bound by the precinct-system requirements.
Contrary to the majority’s assertion, however, the legal review at hand does not require that we isolate the specific challenged practice in the manner it suggests. Rather, at step one of the § 2 inquiry, we only consider whether minority voters “experience substantial difficulty electing representatives of their choice,” Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752, “based on the totality of circumstances,” 52 U.S.C. § 10301(b).5 Although the majority would like us to believe that the increasing percentage of in-person ballots cast out-of-precinct demonstrates that minorities are disparately burdened by the challenged policy, the small number of voters who chose to vote in-person and the even smaller number of such voters who fail to do so in the correct precinct demonstrate that any minimal burden imposed by the policy does not deprive minority voters of equal opportunities to elect representatives of their choice. A conclusion otherwise could not be squared with our determination that a mere statistical showing of disproportionate impact on racial minorities does not satisfy the challenger’s burden. See Salt River, 109 F.3d at 595. If such statistical impact is not sufficient, it must perforce be the case that the crucial test is the extent to which the practice burdens minority voters as opposed to non-minority voters. But the en banc majority offers no explanation for how or why the burden of voting in one’s assigned precinct is severe or beyond that of the burdens traditionally associated with voting.
The majority argues that there may be a “de minimis number” below which no § 2 violation has occurred.6 Maj. Op. ––––. But we know from our own precedent that “a bare statistical showing of disproportionate impact on a racial minority does not satisfy the § 2 ... inquiry.” Salt River, 109 F.3d at 595 (emphasis in original). And Chisom makes clear that § 2 “claims must allege an abridgment of the opportunity to participate in the political process and to elect representatives of one’s choice.” 501 U.S. at 398, 111 S.Ct. 2354 (emphasis in original). As such, the inquiry must require consideration of both the scope of the burden imposed by the particular policy—not merely how many voters are impacted by it—and the difficulty of accessing the political process in its entirety.
Thus, it cannot be true, as the majority suggests, that simply showing that some number of minority voters’ ballots were not counted as a result of an individual policy satisfies step one of the § 2 analysis for a facially neutral policy.
2
Second, the district court made the factual finding that “Arizona’s policy to not count OOP ballots is not the cause of [any identified] disparities in OOP voting.” DNC, 329 F. Supp. 3d at 872. According to the OOP policy that is challenged by DNC, a ballot is not counted if it is cast outside of the voter’s assigned precinct. And the district court pointed to several factors that result in higher rates of out-of-precinct voting among minorities. For example, the district court found that “high rates of residential mobility are associated with higher rates of OOP voting,” and minorities are more likely to move more frequently. Id. at 857, 872. Similarly, “rates of OOP voting are higher in neighborhoods where renters make up a larger share of householders.” Id. at 857. The precinct-system may also pose special challenges for Native American voters, because they may “lack standard addresses” and there may be additional “confusion about the voter’s correct polling place” where precinct assignments may differ from assignments for tribal elections. Id. at 873. “Additionally”, the district court found, Arizona’s “changes in polling locations from election to election, inconsistent election regimes used by and within counties, and placement of polling locations all tend to increase OOP voting rates.” Id. at 858.
But the burden of complying with the precinct-based system in the face of any such factors is plainly distinguishable from the consequence imposed should a voter fail to comply. Indeed, as the district court found, “there is no evidence that it will be easier for voters to identify their correct precincts if Arizona eliminated its prohibition on counting OOP ballots.” Id. Although “the consequence of voting OOP might make it more imperative for voters to correctly identify their precincts,” id., such consequence does not cause voters to cast their ballots out-of-precinct or make it more burdensome for voters to cast their ballots in their assigned precincts.
The majority goes astray by failing to recognize the distinction between the burden of complying and the consequence of failing to do so. In fact, the majority undercuts its own claim by citing the same host of reasons identified by the district court as the reasons why a minority voter is more likely to vote out-of-precinct. Maj Op. –––– – ––––. All the factors the majority seizes upon, however, stem from the general requirement that a voter cast his or her ballot in the assigned precinct—not the policy that enforces such requirement. The importance of such distinction is made clear by the relief that DNC seeks: DNC does not request that Arizona be made to end its precinct-based system or to assign its precincts differently, but instead requests that Arizona be made to count those ballots that are not cast in compliance with the OOP policy.7 Removing the enforcement policy, however, would do nothing to minimize or to extinguish the disparity that exists in out-of-precinct voting.
Consider another basic voting requirement: in order to cast a ballot, a voter must register. If a person fails to register, his or her vote will not count. Any discriminatory result from such a policy would need to be addressed in a challenge to that policy itself. For example, if minorities are underrepresented as a segment of registered voters, perhaps they could challenge some discriminatory aspect of the registration system. But they surely could not prevail by challenging simply the state’s enforcement of the registration policy by refusing to count unregistered voters’ ballots. Minorities in a jurisdiction may very well be underrepresented as members of the registered electorate, but the discrepancy between the protected class as a segment of the general population and as a segment of the registered voting population would not require that a state permit unregistered voters to cast valid ballots on Election Day.
Similarly, the fact that a ballot cast by a voter outside of his or her assigned precinct is discarded does not cause minorities to vote out-of-precinct disproportionately. But DNC does not challenge the general requirement that one vote in his or her precinct or take issue with the assignment of precinct locations—the very requirements that could lead to a disproportionate impact. It may indeed be the case in a precinct-based voting system that a state’s poor assignment of districts, distribution of inadequate information about voting requirements, or other factors have some material effect on election practices such that minorities have less opportunity to elect representatives of their choice as a result of the system. But, in the words of the majority, DNC’s challenge “assumes both [the] importance and [the] continued existence” of “Arizona’s precinct-based system of voting.” Maj. Op. ––––. Instead, DNC challenges only Arizona’s enforcement of such system. Thus, even if there were a recognizable disparity in the opportunities of minority voters voting out-of-precinct, it would nonetheless not be the result of the policy at issue before us.
3
I reject the suggestion implicit in the majority opinion that any facially neutral policy which may result in some statistical disparity is necessarily discriminatory under step one of the § 2 inquiry. We have already held otherwise. Salt River, 109 F.3d at 595. And the majority itself concedes that “more than a de minimis number of minority voters must be burdened before a Section 2 violation based on the results test can be found.” Maj. Op. ––––. Furthermore, I fail to see how DNC—and the majority—can concede the importance and continued existence of a precinct-based system, yet argue that the enforcement mechanism designed to maintain such system is impermissible.
Because DNC has failed to meet its burden under step one of the Voting Rights Act § 2 inquiry—that the district court’s findings were clearly erroneous—our analysis of its OOP claim should end here.
B
As to the facially neutral ballot-collection policy, DNC argues, erroneously, that it violates § 2 because there is “extensive evidence” demonstrating that minority voters are more likely to have used ballot-collection services and that they would therefore be disproportionately burdened by limitations on such services. Specifically, DNC relies on anecdotal evidence that ballot collection has disproportionately occurred in minority communities, that minority voters were more likely to be without home mail delivery or access to transportation, and that ballot-harvesting efforts were disproportionately undertaken by the Democratic Party in minority communities. And, DNC claims, such burden is caused by or linked to Arizona’s history of discrimination.
The district court, quite properly, rejected such argument, making the factual finding that DNC failed to establish at step one that the ballot-collection policy imposed a discriminatory burden on minority voters. DNC, 329 F. Supp. 3d at 866, 871. Once again, the question is whether such finding was clearly erroneous. Salt River, 109 F.3d at 591.
1
The district court found broadly that the non-quantitative evidence offered by DNC failed to show that the ballot-collection policy denied minority voters of “meaningful access to the political process.” DNC, 329 F. Supp. 3d at 871. As Judge Ikuta observed, to determine whether the challenged policy provides minority voters “less opportunity to elect representatives of their choice, [we] must necessarily consider the severity and breadth of the law’s impacts on the protected class.” DNC, 904 F.3d at 717.
But no evidence of that impact has been offered. “In fact, no individual voter testified that [the ballot-collection policy’s] limitations on who may collect an early ballot would make it significantly more difficult to vote.” DNC, 329 F. Supp. 3d at 871 (emphasis added). Anecdotal evidence of how voters have chosen to vote in the past does not establish that voters are unable to vote in other ways or would be burdened by having to do so. The district court simply found that “prior to the [ballot-collection policy’s] enactment minorities generically were more likely than non-minorities to return their early ballots with the assistance of third parties,” id. at 870, but, once again, the disparate impact of a challenged policy on minority voters is insufficient to establish a § 2 violation, see Salt River, 109 F.3d at 594–95.
The majority simply does not address the lack of evidence as to whether minority voters have less opportunity than non-minority voters now that ballot collection is more limited. Instead, the majority answers the wrong question by pointing to minority voters’ use of ballot collection in the past. The majority offers no record-factual support for its conclusion that the anecdotal evidence presented demonstrates that compliance with the ballot-collection policy imposes a disparate burden on minority voters—a conclusion that must be reached in order to satisfy step one of the § 2 inquiry—let alone evidence that the district court’s contrary finding was “clearly erroneous.”
Given the lack of any testimony in the record indicating that the ballot-collection policy would result in minority voters “experienc[ing] substantial difficulty electing representatives of their choice,” Gingles, 478 U.S. at 48 n.15, 106 S.Ct. 2752, the district court did not clearly err in finding that, “for some voters, ballot collection is a preferred and more convenient method of voting,” but a limitation on such practice “does not deny minority voters meaningful access to the political process.” DNC, 329 F. Supp. 3d at 871.
2
The district court further found that the ballot-collection policy was unlikely to “cause a meaningful inequality in the electoral opportunities of minorities” because only “a relatively small number of voters have used ballot collection services” in the past at all. DNC, 329 F. Supp. 3d at 870–71. And, the district court noted, DNC “provided no quantitative or statistical evidence comparing the proportion that is minority versus non-minority.” Id. at 866. “Without this information,” the district court explained, “it becomes difficult to compare the law’s impact on different demographic populations and to determine whether the disparities, if any, are meaningful.” Id. at 867. Thus, from the record, we do not know either the extent to which voters may be burdened by the ballot-collection policy or how many minority voters may be so burdened.
Nonetheless, the district court considered circumstantial and anecdotal evidence offered by DNC and determined that “the vast majority of Arizonans, minority and non-minority alike, vote without the assistance of third-parties who would not fall within [the ballot-collection policy’s] exceptions.”Id. at 871. DNC—and the majority—argue that such finding is not supported by the record, but, given the lack of quantitative or statistical evidence before us, it is difficult to conclude that such finding is clearly erroneous. The district court itself noted that it could not “speak in more specific or precise terms” given the sparsity of the record. Id. at 870. Drawing from anecdotal testimony, the district court estimated that fewer than 10,000 voters used ballot-collection services in any election. Id. at 845. Drawing even “the unjustified inference that 100,000 early mail ballots were collected” during the 2012 general election, the district court found that such higher total would nonetheless be “relatively few early voters” as compared to the 1.4 million early mail ballots returned or 2.3 million total votes cast. Id. at 845. The majority further argues that the district court erred in “discounting the evidence of third-party ballot collection as merely ‘circumstantial and anecdotal’ ” Maj. Op. ––––. But the district court did nothing of the sort. To the contrary, the district court considered whether the ballot-collection policy violated § 2 by making these estimates—and even generous estimates—from the anecdotal evidence offered. And the district court’s subsequent conclusion that the limitation of third-party ballot collection would impact only a “relatively small number of voters,” id. at 870, is clearly plausible on this record, see Bessemer City, 470 U.S. at 573, 105 S.Ct. 1504.
The majority also argues that the total number of votes affected is not the relevant inquiry; the proper test is whether the number of ballots collected by third parties surpasses any de minimis number. Maj. Op. ––––. But we already know “that a bare statistical showing” that an election practice has a “disproportionate impact on a racial minority does not satisfy” step one of the § 2 inquiry. Salt River, 109 F.3d at 595 (emphasis in original). And, even if such impact were sufficient, the record offers no evidence from which the district court could determine the extent of the discrepancy between minority voters as a proportion of the entire electorate versus minority voters as a proportion of those who have voted using ballot-collection services in the past. DNC, 329 F. Supp. 3d at 866–67.
3
As Judge Bybee keenly observed in a previous iteration of this case (and indeed in his dissent in this case), “[t]here is no constitutional or federal statutory right to vote by absentee ballot.” Feldman III, 843 F.3d at 414 (Bybee, J., dissenting) (citing McDonald v. Bd. of Election Comm’rs of Chi., 394 U.S. 802, 807–08, 89 S.Ct. 1404, 22 L.Ed.2d 739 (1969)); accord Bybee, J. Diss. Op. ––––. Both today and in the past, Arizona has chosen to provide a wide range of options to voters. But Arizona’s previous decision to permit a particular mechanism of voting does not preclude Arizona from modifying its election system to limit such mechanism in the future so long as such modification is made in a constitutional manner. And, in fact, Arizona’s modification here was made in compliance with “the recommendation of the bipartisan Commission on Federal Election Reform.” DNC, 329 F. Supp. 3d at 855. Without any evidence in the record of the severity and breadth of the burden imposed by this change to the ballot-collection policy, we cannot be “left with the definite and firm conviction” that the district court erred in finding that DNC failed to show that the policy violated § 2. See Bessemer City, 470 U.S. at 573, 105 S.Ct. 1504; see also Salt River, 109 F.3d at 591.
C
Because I disagree with the majority’s conclusion that DNC has satisfied its burden at step one of the § 2 Voting Rights Act inquiry, I would not reach step two. I therefore do not address the majority’s consideration of the so-called “Senate Factors” in determining whether the burden is “in part caused by or linked to ‘social and historical conditions’ that have or currently produce discrimination against members of the protected class.” League of Women Voters, 769 F.3d at 240 (quoting Gingles, 478 U.S. at 47, 106 S.Ct. 2752). These factors—and the majority’s lengthy history lesson on past election abuses in Arizona—simply have no bearing on this case. Indeed, pages 47 to 81 of the majority’s opinion may properly be ignored as irrelevant.
IV
DNC also contends that the ballot-collection policy violates the Fifteenth Amendment to the United States Constitution.8 To succeed on a claim of discriminatory intent under the Fifteenth Amendment, the challenger must demonstrate that the state legislature “selected or reaffirmed a particular course of action at least in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an identifiable group.” Pers. Adm’r of Mass. v. Feeney, 442 U.S. 256, 279, 99 S.Ct. 2282, 60 L.Ed.2d 870 (1979). Because discriminatory intent “is a pure question of fact,” we again review only for clear error. Pullman-Standard v. Swint, 456 U.S. 273, 287–88, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982). “Determining whether invidious discriminatory purpose was a motivating factor demands a sensitive inquiry into such circumstantial and direct evidence of intent as may be available.” Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 266, 97 S.Ct. 555, 50 L.Ed.2d 450 (1977).
The district court concluded that the ballot-collection policy did not violate the Fifteenth Amendment because it made the factual finding that the legislature “was not motivated by a desire to suppress minority voters,” although “some individual legislators and proponents of limitations on ballot collection harbored partisan motives” that “did not permeate the entire legislative process.” DNC, 329 F. Supp. 3d at 879, 882 (emphasis added). Instead, “[t]he legislature was motivated by ... a sincere belief that mail-in ballots lacked adequate prophylactic safeguards as compared to in-person voting.” Id. at 882. In analyzing DNC’s appeal from such finding, the majority, once again, completely ignores our demanding standard of review and instead conducts its own de novo review. Maj. Op. ––––. Our duty is only to consider whether the district court clearly erred in its finding that the ballot-collection policy was not enacted with discriminatory intent. See Bessemer City, 470 U.S. at 573, 105 S.Ct. 1504. And “to be clearly erroneous, a decision must ... strike [a court] as wrong with the force of a five-week old, unrefrigerated dead fish.” Ocean Garden, Inc. v. Marktrade Co., Inc., 953 F.2d 500, 502 (9th Cir. 1991) (quoting Parts & Elec. Motors, Inc. v. Sterling Elec., Inc., 866 F.2d 228, 233 (7th Cir. 1988)).
The majority therefore fails to offer any basis—let alone a convincing one—for the conclusion that it must reach in order to reverse the decision of the district court: that the district court committed clear error in its factual findings. Given the failure of the majority to conduct its review in the proper manner, I see no reason to engage in a line-by-line debate with its flawed analysis. Rather, it is enough to note two critical errors made by the majority in ignoring the district court’s determinations that while some legislators were motivated by partisan concerns, the legislature as a body was motivated by a desire to enact prophylactic measures to prevent voter fraud.
A
First, the majority fails to distinguish between racial motives and partisan motives. Even when “racial identification is highly correlated with political affiliation,” a party challenging a legislative action nonetheless must show that racial motives were a motivating factor behind the challenged policy. Cooper v. Harris, ––– U.S. ––––, 137 S. Ct. 1455, 1473, 197 L.Ed.2d 837 (2017) (quoting Easley v. Cromartie, 532 U.S. 234, 243, 121 S.Ct. 1452, 149 L.Ed.2d 430 (2001)). Nonetheless, the majority suggests that a legislator motivated by partisan interest to enact a law that disproportionately impacts minorities must necessarily have acted with racially discriminatory intent as well. For example, the district court noted that Arizona State Senator Don Shooter was, “in part motivated by a desire to eliminate what had become an effective Democratic [Get Out The Vote] strategy.” DNC, 329 F. Supp. 3d at 879. The majority simply concludes that such finding shows racially discriminatory intent as a motivating factor. But the majority’s unsupported inference does not satisfy the required showing. And the majority fails to cite any evidence demonstrating that the district court’s finding to the contrary was not “plausible in light of the record viewed in its entirety.” Bessemer City, 470 U.S. at 574, 105 S.Ct. 1504.
B
Second, in defiance of Supreme Court precedent to the contrary, the majority assumes that a legislature’s stated desire to prevent voter fraud must be pretextual when there is no direct evidence of voter fraud in the legislative record. In Crawford, the Court rejected the argument that actual evidence of voter fraud was needed to justify the State’s decision to enact prophylactic measures to prevent such fraud. Crawford, 553 U.S. at 195–96, 128 S.Ct. 1610. There, the Court upheld an Indiana statute requiring in-person voters to present government-issued photo identification in the face of a constitutional challenge. Id. at 185, 128 S.Ct. 1610. Although “[t]he record contain[ed] no evidence of [voter] fraud actually occurring in Indiana at any time in its history,” the Supreme Court nonetheless determined that the State had a legitimate and important interest “in counting only the votes of eligible voters.” Id. at 194, 196, 128 S.Ct. 1610; see also id. at 195 nn.11–13, 128 S.Ct. 1610 (citing “fragrant examples of” voter fraud throughout history and in recent years). Given its interest in addressing its valid concerns of voter fraud, Arizona was free to enact prophylactic measures even though no evidence of actual voter fraud was before the legislature. Yet the majority does not even mention Crawford, let alone grapple with its consequences on this case.
And because no evidence of actual voter fraud is required to justify an anti-fraud prophylactic measure, the majority’s reasoning quickly collapses. The majority cites Senator Shooter’s “false and race-based allegations” and the “LaFaro video,” which the district court explained “showed surveillance footage of a man of apparent Hispanic heritage appearing to deliver early ballots” and “contained a narration of [i]nnuendos of illegality ... [and] racially tinged and inaccurate commentary by ... LaFaro.” DNC, 329 F. Supp. 3d at 876 (second, third, and fourth alterations in original). The majority contends that although “some members of the legislature who voted for H.B. 2023 had a sincere, though mistaken, non-race-based belief that there had been fraud in third-party ballot collection, and that the problem needed to be addressed,” a discriminatory purpose may be attributable to all of them as a matter of law because any sincere belief was “created by Senator Shooter’s false allegations and the ‘racially tinged’ LaFaro video.” Maj. Op. ––––. The majority claims that these legislators were used as “cat’s paws” to “serve the discriminatory purposes of Senator Shooter, Republican Chair LaFaro, and their allies.” Maj. Op. ––––. Yet, the majority’s reliance on such employment discrimination doctrine is misplaced because, unlike employers whose decision may be tainted by the discriminatory motives of a supervisor, each legislator is an independent actor, and bias of some cannot be attributed to all members. The very fact that some members had a sincere belief that voter fraud needed to be addressed is enough to rebut the majority’s conclusion. To the contrary, the underlying allegations of voter fraud did not need to be true in order to justify the “legitimacy or importance of the State’s interest in counting only the votes of eligible voters.” Crawford, 553 U.S. at 196, 128 S.Ct. 1610. And the majority provides no support for its inference of pretext where there is a sincere and legitimate interest in addressing a valid concern. Maj. Op. at –––– – ––––. Instead, the majority accepts the district court’s finding that some legislators “had a sincere, non-race-based belief that there was fraud” that needed to be addressed. Nevertheless, unable to locate any discriminatory purpose, it simply attributes one to them using the inapplicable “cat’s paw doctrine.” Maj. Op. ––––. Such argument demonstrates the extraordinary leap in logic the majority must make in order to justify its conclusion.
Let me restate the obvious: we may reverse the district court’s intensely factual determination as to discriminatory intent only if we determine that such finding was clearly erroneous. Thus, even if the majority disagrees with the district court’s finding, it must demonstrate that the evidence was not “plausible in light of the record viewed in its entirety.” Bessemer City, 470 U.S. at 574, 105 S.Ct. 1504. Perhaps if the majority had reminded itself of our appellate standard, it would not have simply re-weighed the same evidence considered by the district court to arrive at its own findings on appeal.
V
The district court properly determined that neither Arizona’s out-of-precinct policy nor its ballot-collection policy violates § 2 of the Voting Rights Act and the Fifteenth Amendment to the Constitution.9 In concluding otherwise, the majority misperceives the inquiry before us and fails to narrow the scope of its review, instead insisting on acting as a de novo trial court. That, of course, is not our role.
I would therefore affirm the judgment of the district court and must respectfully dissent from the majority opinion.
BYBEE, Circuit Judge, with whom O’SCANNLAIN, CLIFTON, and CALLAHAN, Circuit Judges, join, dissenting:
The right to vote is the most fundamental of our political rights and the basis for our representative democracy. “No right is more precious” because it is a meta-right: it is the means by which we select “those who make the laws under which, as good citizens, we must live.”Wesberry v. Sanders, 376 U.S. 1, 17, 84 S.Ct. 526, 11 L.Ed.2d 481 (1964). “Other rights, even the most basic, are illusory if the right to vote is undermined.” Id. Almost as fundamental as the right to vote is the need for the electorate to have confidence in the rules by which elections are conducted.
I write separately to make a simple point: The Arizona rules challenged here are part of an “electoral process that is necessarily structured to maintain the integrity of the democratic system.” Burdick v. Takushi, 504 U.S. 428, 441, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992).1 The Constitution entrusts the “Times, Places and Manner of holding Elections” to state legislatures, subject to laws enacted by Congress to “make or alter such Regulations.” U.S. Const. art. I, § 4, cl. 1. “ ‘Times, Places, and Manner,’ ... are ‘comprehensive words,’ which ‘embrace authority to provide a complete code for ... elections.’ ” Arizona v. Inter Tribal Council of Ariz., Inc., 570 U.S. 1, 8–9, 133 S.Ct. 2247, 186 L.Ed.2d 239 (2013) (quoting Smiley v. Holm, 285 U.S. 355, 366, 52 S.Ct. 397, 76 L.Ed. 795 (1932)); see Rucho v. Common Cause, ––– U.S. ––––, 139 S. Ct. 2484, 2495, 204 L.Ed.2d 931 (2019).
“[A]s a practical matter, there must be a substantial regulation of elections if they are to be fair and honest and if some sort of order, rather than chaos, is to accompany the democratic processes.” To achieve these necessary objectives, States have enacted comprehensive and sometimes complex election codes. Each provision of these schemes, whether it governs the registration and qualifications of voters, the selection and eligibility of candidates, or the voting process itself, inevitably affects—at least in some degree—the individual’s right to vote and his right to associate with others for political ends. Nevertheless, the State’s important regulatory interests are generally sufficient to justify reasonable, nondiscriminatory restrictions.
Anderson v. Celebrezze, 460 U.S. 780, 788, 103 S.Ct. 1564, 75 L.Ed.2d 547 (1983) (citation omitted) (quoting Storer v. Brown, 415 U.S. 724, 730, 94 S.Ct. 1274, 39 L.Ed.2d 714 (1974)).
Time, place, and manner restrictions are fundamentally differently from provisions that affect the “Qualifications requisite for Electors,” U.S. Const. art. I, § 2, cl. 1, and state apportionments “according to their respective Numbers,” id. art. I, § 2, cl. 3. The Constitution restricts with exactness the qualifications states may require of their voters. See id. amend. XV, § 1 (“race, color, or previous condition of servitude”); amend. XIX (sex); amend. XXIV (“failure to pay any poll tax or other tax”); amend. XXVI (those “eighteen years of age or older, ... on account of age”); Kramer v. Union Free Sch. Dist. No. 15, 395 U.S. 621, 89 S.Ct. 1886, 23 L.Ed.2d 583 (1969) (property ownership). Similarly, the constitutional imperative for one person, one vote demands that apportionment be subject to precision approaching “absolute population equality,” Karcher v. Daggett, 462 U.S. 725, 732, 103 S.Ct. 2653, 77 L.Ed.2d 133 (1983), “as nearly as practicable,” Kirkpatrick v. Preisler, 394 U.S. 526, 531, 89 S.Ct. 1225, 22 L.Ed.2d 519 (1969).
Time, place, and manner restrictions stand on different footing from status-based restraints on vote qualifications and legislative malapportionment. State requirements respecting when and where we vote and how ballots will be counted are “generally-applicable and evenhanded restrictions that protect the integrity and reliability of the electoral process itself.”Anderson, 460 U.S. at 788 n.9, 103 S.Ct. 1564. By contrast, for example, “redistricting differs from other kinds of state decisionmaking in that the legislature always is aware of race when it draws district lines, just as it is aware of age, economic status, religions and political persuasion, and a variety of other demographic factors.” Shaw v. Reno, 509 U.S. 630, 646, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993). Time, place, and manner restrictions are the rules of the game, announced in advance, so that all voters will know what they must do. Parties of all stripes should have an equal interest in rules that are both fair on their face and fairly administered.
Two such rules are challenged here: the rule about how Arizona will count out-of-precinct votes (OOP) and the rule about who may file another person’s absentee ballot (H.B. 2023). As rules of general applicability, they apply to all voters, without “account of race or color.” 52 U.S.C. § 10301(a).2 Rather than simply recognizing that Arizona has enacted neutral, color-blind rules, the majority has embraced the premise that § 2 of the VRA is violated when any minority voter appears to be adversely affected by Arizona’s election laws. Although the majority abjures this premise for now, claiming that it does “not need to go so far” as equating “the case of an individually targeted single minority voter who is denied the right to vote and the case where a facially neutral policy affects a single voter,” Maj. Op. at ––––, its analysis necessarily rests on that premise. The majority has no limiting principle for identifying a de minimis effect in a facially neutral time, place, or manner rule. The premise finds its clearest expression in the Fourth Circuit’s opinion in League of Women Voters of N.C. v. North Carolina, 769 F.3d 224, 244 (4th Cir. 2014) (emphasis added): “[W]hat matters for purposes of Section 2 is not how many minority voters are being denied equal electoral opportunities but simply that ‘any’ minority voter is being denied equal electoral opportunities.” See Maj. Op. at –––– – ––––, –––– – ––––, –––– (relying on League of Women Voters). Such a premise insists on a precision that we have never demanded before.
By contrast, the Supreme Court explained that following City of Mobile v. Bolden, 446 U.S. 55, 100 S.Ct. 1519, 64 L.Ed.2d 47 (1980), “Congress substantially revised § 2 to make clear that a violation could be proved by showing discriminatory effect alone and to establish as the relevant legal standard the ‘results test,’ applied ... in White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973).” Thornburg v. Gingles, 478 U.S. 30, 35, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986). Yet in White, the Court made clear that it “did not hold ... that any deviations from absolute equality, however small, must be justified to the satisfaction of the judiciary to avoid invalidation under the Equal Protection Clause.” 412 U.S. at 763–64, 93 S.Ct. 2332. Rather, the Court recognized that any rule in an election scheme might suffer “relatively minor population deviations .... ‘based on legitimate considerations incident to the effectuation of a rational state policy.’ ” Id. at 764, 93 S.Ct. 2332 (quoting Reynolds v. Sims, 377 U.S. 533, 579, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964)).
A “rational state policy” surely includes the need for a consistent, neutral set of time, place, and manner rules. The majority’s reading of the Voting Rights Act turns § 2 into a “one-minority-vote-veto rule” that may undo any number of time, place, and manner rules. It is entirely results-bound, so much so that under the majority’s reading of the Voting Rights Act, the same rules the majority strikes down in Arizona may be perfectly valid in every other state, even states within our circuit. It all depends on the numbers. Indeed, so diaphonous is the majority’s holding, that it may be a temporary rule for Arizona. If Arizona were to reenact these provisions again in, say, 2024, the numbers might come out differently and the OOP and ballot collection rules would be lawful once again.
The two Arizona rules at issue here—OOP and H.B. 2023—are rules of general applicability, just like the rules governing voting on the day of the election, registering with the Secretary of State, and bringing identification with you. Such “ ‘evenhanded restrictions that protect the integrity and reliability of the electoral process itself’ are not invidious.” Crawford v. Marion Cty. Election Bd., 553 U.S. 181, 189–90, 128 S.Ct. 1610, 170 L.Ed.2d 574 (2008) (plurality opinion) (quoting Anderson, 460 U.S. at 788 n.9, 103 S.Ct. 1564). Both rules the majority strikes down today have widely-held, well-recognized—even distinguished—pedigrees. As I show in Part I, the OOP is a long-standing rule that remains in place in a majority of American jurisdictions. The rule the majority prefers is a minority rule in the United States and, more importantly, disregards Arizona’s interest in encouraging voting in local elections and, in application, may actually disadvantage minority voters. In Part II, I demonstrate that, although H.B. 2023 is of more recent vintage, similar rules are in place in other American jurisdictions, and H.R. 2023 follows carefully the recommendation of a bi-partisan commission on the integrity of American elections.
I
It has long been a feature of American democracy that, on election day, voters must vote in person at an assigned polling venue—an election precinct.
[I]t is the well established practice in nearly every state to divide the county or city into a number of geographical districts for the purpose of holding elections. Each elector is required to vote at the polling place of his own precinct, which by custom is ordinarily located within the precinct, and, in cities, within a few blocks of his residence.
Joseph P. Harris, Election Administration in the United States 206–07 (1934). Like most American jurisdictions, Arizona’s election rules require a non-absentee voter’s personal presence at the polling place. Ariz. Rev. Stat. § 16-411(A) (“The broad of supervisors of each county ... shall establish a convenient number of election precincts in the county and define the boundaries of the precincts.”). The reasons for such a venue rule are
significant and numerous: it caps the number of voters attempting to vote in the same place on election day; it allows each precinct ballot to list all of the votes a citizen may cast for all pertinent federal, state, and local elections, referenda, initiatives, and levies; it allows each precinct ballot to list only those votes a citizen may cast, making ballots less confusing; it makes it easier for election officials to monitor votes and prevent election fraud; and generally puts polling places in closer proximity to voter residences.
Sandusky Cty. Democratic Party v. Blackwell, 387 F.3d 565, 569 (6th Cir. 2004).3 Precincts help to secure the orderly administration of elections, which then assures all voters of the integrity of the election.
A
Arizona’s out of precinct rule (OOP) is a standard feature of American democracy. Under Arizona’s election code, “[n]o person shall be permitted to vote unless such person’s name appears as a qualified elector in both the general county register and in the precinct register.” Ariz. Rev. Stat. § 16-122. The election code provides extensive instructions for electors who have changed their residence or whose name does not appear on the precinct register; if there is any question of the elector’s eligibility to vote in that precinct, Arizona authorizes the filing of a provisional ballot. See, e.g., Ariz. Rev. Stat. §§ 16-135, 16-583, 16-584, 16-592.
There is nothing unusual about Arizona’s OOP rule.4 Although there are variations in the way the rule is formulated, by my count, twenty-six states, the District of Columbia, and three U.S. territories disqualify ballots cast in the wrong precinct.5 These states represent every region of the country: The Northeast (Connecticut, Vermont), the mid-Atlantic (Delaware, District of Columbia, West Virginia), the South (Alabama, Florida, Kentucky, Mississippi, South Carolina, Tennessee, Virginia, Virgin Islands), the mid-West (Illinois, Indiana, Iowa, Michigan, Missouri, Nebraska, South Dakota, Wisconsin), the Southwest (Arizona, Oklahoma, Texas), the Mountain States (Montana, Wyoming), and the West (American Samoa, Hawaii, Nevada, Northern Mariana Islands). Twenty states and two territories will count out of precinct ballots, although the states are not uniform in what they will count.6 They also represent a broad spectrum of the country: The Northeast (Maine, Massachusetts, New York, Rhode Island), the mid-Atlantic (Maryland, New Jersey, Pennsylvania), the South (Arkansas, Louisiana, North Carolina, Georgia, Puerto Rico), the mid-West (Ohio, Kansas), the Southwest (New Mexico), the Mountain States (Colorado, Utah), and the West (Alaska, California, Guam, Oregon, Washington).7
Nowhere in its discussion of the “totality of the circumstances” has the majority considered that Arizona’s OOP provision is a widely held time, place, or manner rule. It is not a redistricting plan, see Cooper v. Harris, ––– U.S. ––––, 137 S. Ct. 1455, 197 L.Ed.2d 837 (2017); League of United Latin Am. Citizens v. Perry, 548 U.S. 399, 126 S.Ct. 2594, 165 L.Ed.2d 609 (2006); Shaw v. Reno, 509 U.S. 630, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993); a multimember district, see Chisom v. Roemer, 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d 348 (1991); Gingles, 478 U.S. 30, 106 S.Ct. 2752; or an at-large system, seeRogers v. Lodge, 458 U.S. 613, 102 S.Ct. 3272, 73 L.Ed.2d 1012 (1982). Those “circumstances” are as unique as a fingerprint, subject to manipulation, and require “an intensely local appraisal” of the state’s plan. Gingles, 478 U.S. at 78, 106 S.Ct. 2752 (internal quotation marks and citation omitted). Arizona’s OOP applies statewide; it is not a unique rule, but a traditional rule, common to the majority of American states. The OOP rule, as a rule of general applicability, is part of a “political process[ ] ... equally open to participation” by all Arizona voters. 52 U.S.C. § 10301(b).
B
The majority asserts that “counting or partially counting OOP ballots would [not] threaten the integrity of Arizona’s precinct-based system.” Maj. Op. at ––––. Effectively, the majority holds that Arizona must abandon its traditional polling venue rules and accept the ballots of voters who cast their ballot in the wrong precinct, at least for national and state-wide offices. Id. at –––– – –––– (citing the rules of California, Utah, and New Mexico as an example of states partially counting OOP ballots). Under the majority’s preferred scheme, Arizona must count all votes for offices that are not precinct dependent. As to the remainder of the ballot, Arizona may—in accordance with its traditional rule—disqualify the ballot for all offices for which the political geography of the precinct matters. The majority has failed to take into account that the rule it prefers has its own consequences, including adverse consequences for minority voters.
Let’s review an example to consider the unintended consequences of the majority’s haste. Under Arizona’s traditional rules, the state would disqualify the ballot of a voter from Tucson who votes in any precinct other than his assigned precinct. Under the majority’s new rule, a voter from Tucson may cross precinct lines and vote in any precinct in Arizona—for instance, in Phoenix. His cross-precinct ballot will be counted for those offices which are common to ballots in his precinct-in-law in Tucson and his new precinct-in-fact in Phoenix—such offices would include the presidency, the U.S. Senate, and any statewide offices. His ballot will be disqualified, however, for all state and local offices defined by geographic boundaries that are not common to the two precincts—for example, the U.S. House of Representatives, the state legislature, and municipal offices such as mayor, city council, and school board.
The majority’s rule will skew future elections in Arizona in two predictable ways. First, it overvalues national elections. Ballots for the presidency, the U.S. Senate, and any state offices that would otherwise be disqualified must be counted. Voters—whether intentionally or carelessly—may vote with impunity in the wrong precinct, knowing that their vote will count for the national and statewide offices.
Second, it undervalues local elections. Those same ballots will not be counted toward those federal, state, and local offices that are defined by geographic boundaries and for which the voters from the outside precinct are not eligible. Non-conscientious voters—voters who care more about a national or a statewide race than the local races—are permitted to vote wherever they please, while conscientious voters—those concerned with all the offices on the ballot—are burdened by the requirement that they find their way to their proper precinct. And if the conscientious voter can’t get to the polling place on time, he will have cast no ballot for any office, national, state, or local.
The net result is that the majority has lowered the cost to voters of determining where they are supposed to vote, but only as to presidential, U.S. Senate, and statewide races. As the majority no doubt intends, persons who didn’t know or were confused about their polling place will have their vote counted, but only in select races. But as the majority may not have thought through, anyone in Arizona, including people who know where they are supposed to vote in an election (but for one reason or another would not have otherwise voted because it was inconvenient or impossible to vote at their home precinct), will also be able to vote—but again, only in select races. Arizona can thus expect more votes in the presidential, senatorial, and state races than would be cast under its traditional rules. I suppose that in theory that’s a good thing. What the majority has not counted on is the effect its order will have on the races that depend on geographic boundaries within Arizona: congressional, state-legislative, and local offices. When voters do not go to their local precincts to vote, they cannot vote in those races. Voters who do not take the time to determine their appropriate precinct—for whatever reason—and vote out of precinct have disenfranchised themselves with respect to the local races. That’s a bad thing.
Arizona’s longstanding, neutral rule gives voters an incentive to figure out where their polling place is, which, in turn, encourages voters to cast ballots in national, state, and local elections. In effect, Arizona has stapled national and statewide elections to other state and local elections. The opportunity to vote in any one race is the opportunity to vote in all races. It’s strong medicine, but Arizona’s rule is a self-protective rule; it helps encourage voting and, presumably, interest in local elections. The majority’s preferred rule gives voters an incentive to vote wherever it is convenient for them which increases the likelihood they will vote in certain national and statewide races, but decreases the likelihood they will vote in other state and local races. It places a burden on voters who wish to exercise their right to vote on all matters to which they are entitled, a burden that simply would not exist for the less-engaged voter. The majority’s rule contradicts our most basic principles of federalism by deeming elections for national and statewide offices more important than those for lesser offices.
The majority’s concern is based on the fact that voters who vote in the wrong precinct are more likely to be minorities. Maj. Op. at –––– – ––––. If that fact holds true in the future—and it may not because, as I have explained, any voter in Arizona (including those who know where to vote) may take advantage of the majority’s new rule—then minority ballots will be underrepresented in the local races. Under the majority’s preferred scheme, it is thus likely that more minorities will fail to vote in local elections—elections that most directly affect the daily lives of ordinary citizens, and often provide the first platform by which citizen-candidates, not endowed with personal wealth or name recognition, seek on the path to obtaining higher office. In any event, the court has just put a big thumb on the scale of the Arizona elections—national, state, and local—with unclear results.
These concerns are magnified when we consider the relatively small number of OOP ballots. See Democratic Nat’l Comm. v. Reagan, 329 F. Supp. 3d 824, 873 (D. Ariz. 2018). It is more likely that these ballots would make a difference in a local election than in a national or statewide election. Arizona’s rule encourages its OOP voters—white, African-American, Hispanic, or other—to vote in the correct precinct. Under Arizona’s current OOP rule, a voter, having gone to the trouble of going to a precinct to vote in person and suffering the indignity of having to fill out a provisional ballot, is less likely to make the same mistake the next year.8 A voter who has had a ballot disqualified is more likely to figure out the correct precinct next time—or, better yet, sign up for the convenience of early voting, a measure that avoids the conundrum of OOP altogether.9 The voter who only votes where it is convenient has disenfranchised himself from local elections.
States such as California, Utah, and New Mexico have made the same choice the majority forces on Arizona. Those states may or may not have made the calculus I have set out here and they may or may not have measured the costs and benefits of their new rule; it’s theirs to experiment with. They may conclude that the new rule is the right one; they may not. And if any of those states decides that the count-the-ballots-partially rule is not the best rule, those states will be free to adopt a different rule, including the OOP rule the majority strikes down today. After today’s decision, Arizona has no such recourse.
II
H.B. 2023 presents a different set of considerations. There is no constitutional or federal statutory right to vote by absentee ballot. See McDonald v. Bd. of Election Comm’rs of Chi., 394 U.S. 802, 807–08, 89 S.Ct. 1404, 22 L.Ed.2d 739 (1969) (“It is thus not the right to vote that is at stake here but a claimed right to receive absentee ballots. ... [T]he absentee statutes, which are designed to make voting more available to some groups who cannot easily get to the polls, do not themselves deny ... the exercise of the franchise ....”); see also Crawford, 553 U.S. at 209, 128 S.Ct. 1610 (Scalia, J., concurring in the judgment) (“That the State accommodates some voters by permitting (not requiring) the casting of absentee or provisional ballots, is an indulgence—not a constitutional imperative that falls short of what is required.”); Griffin v. Roupas, 385 F.3d 1128, 1130 (7th Cir. 2004) (rejecting the claim that there is “a blanket right of registered voters to vote by absentee ballot” because “it is obvious that a federal court is not going to decree weekend voting, multi-day voting, all-mail voting, or Internet voting”).10 Nevertheless, if a state is going to offer absentee ballots, it must do so on an equal basis. Arizona’s absentee ballot rule, like its OOP rule, is a neutral time, place, or manner provision to help ensure the integrity of the absentee voting process. In fact, what is at issue here is not the right of Arizona voters to obtain and return an absentee ballot, but the question of who can physically return the ballot.
A
H.B. 2023 provides that “[a] person who knowingly collects voted or unvoted early ballots from another person is guilty of a class 6 felony.” Ariz. Rev. Stat. Ann. § 16-1005(H) (codifying H.B. 2023). The law does not apply to three classes of persons: (1) “[a]n election official,” (2) “a United States postal service worker or any other person who is allowed by law to transmit United States mail,” and (3) “[a] family member, household member or caregiver of the voter.” Id. § 16-1005(H)–(I)(2).
The Arizona provision is substantially similar to the laws in effect in many other states. In Indiana, for example, it is a felony for anyone to collect a voter’s absentee ballot, with exceptions for members of the voter’s household, the voter’s designated attorney in fact, certain election officials, and mail carriers. Ind. Code § 3-14-2-16(4). Connecticut also restricts ballot collection, permitting only the voter, a designee of an ill or disabled voter, or the voter’s immediate family members to mail or return an absentee ballot. Conn. Gen. Stat. § 9-140b(a). New Mexico likewise permits only the voter, a member of the voter’s immediate family, or the voter’s caregiver to mail or return an absentee ballot. N.M. Stat. Ann. § 1-6-10.1. At least seven other states (Georgia, Missouri, Nevada, North Carolina, Oklahoma, Ohio, and Texas) similarly restrict who can personally deliver an absentee ballot to a voting location. Ga. Code Ann. § 21-2-385(a) (limiting who may personally deliver an absentee ballot to designees of ill or disabled voters or family members); Mo. Rev. Stat. § 115.291(2) (restricting who can personally deliver an absentee ballot); Nev. Rev. Stat. Ann. § 293.330(4) (making it a felony for anyone other than the voter or the voter’s family member to return an absentee ballot); Okla. Stat. tit. 26, § 14-108(C) (voter delivering a ballot must provide proof of identity); Ohio Rev. Code Ann. § 3509.05(A) (limiting who may personally deliver an absent voter’s ballot); Tex. Elec. Code Ann. § 86.006(a) (permitting only the voter to personally deliver the ballot).11
Other states are somewhat less restrictive than Arizona because they permit a broader range of people to collect early ballots from voters but restrict how many ballots any one person can collect and return. Colorado forbids anyone from collecting more than ten ballots. Colo. Rev. Stat. § 1-7.5-107(4)(b). North Dakota prohibits anyone from collecting more than four ballots, N.D. Cent. Code § 16.1-07-08(1); New Jersey, N.J. Stat. Ann. § 19:63-4(a), and Minnesota, Minn. Stat. Ann. § 203B.08 sbd. 1, three; Arkansas, Ark. Code Ann. § 7-5-403(a)(1), Nebraska, Neb. Rev. Stat. § 32-943(2), and West Virginia, W. Va. Code § 3-3-5(k), two. South Dakota prohibits anyone from collecting more than one ballot without notifying “the person in charge of the election of all voters for whom he is a messenger.” S.D. Codified Laws § 12-19-2.2.
Still other states have adopted slightly different restrictions on who may collect early ballots. California, Maine, and North Dakota, for example, make it illegal to collect an absentee ballot for compensation. Cal. Elec. Code § 3017(e)(1); Me. Rev. Stat. Ann. tit. 21-A, § 791(2)(A) (making it a crime to receive compensation for collecting absentee ballots); N.D. Cent. Code § 16.1-07-08(1) (prohibiting a person from receiving compensation for acting as an agent for an elector). Florida and Texas make it a crime to receive compensation for collecting certain numbers of ballots. Fla. Stat. Ann. § 104.0616(2) (making it a misdemeanor to receive compensation for collecting more than two vote-by-mail ballots); Tex. Elec. Code Ann. § 86.0052(a)(1) (criminalizing compensation schemes based on the number of ballots collected for mailing).
Some of these laws are stated as a restriction on how the early voter may return a ballot. In those states, the voter risks having his vote disqualified. See, e.g., Wrinn v. Dunleavy, 186 Conn. 125, 440 A.2d 261, 272 (1982) (disqualifying ballots and ordering a new primary election when an unauthorized individual mailed absentee ballots). In other states, as in Arizona, the statute penalizes the person collecting the ballot. SeeInd. Code Ann. § 3-14-2-16 (making it a felony knowingly to receive a ballot from a voter); Nev. Rev. Stat. Ann. § 293.330(4) (making it a felony for unauthorized persons to return an absentee ballot); Tex. Elec. Code Ann. § 86.006(f)–(g) (making it a crime for an unauthorized person to possess an official ballot); see also Murphy v. State, 837 N.E.2d 591, 594–96 (Ind. Ct. App. 2005) (affirming a denial of a motion to dismiss a charge for unauthorized receipt of a ballot from an absentee voter); People v. Deganutti, 348 Ill.App.3d 512, 284 Ill.Dec. 538, 810 N.E.2d 191, 198 (2004) (affirming conviction for absentee ballot violation). In those states, the ballot, even if collected improperly, may be valid. See In re Election of Member of Rock Hill Bd. of Educ., 76 Ohio St.3d 601, 669 N.E.2d 1116, 1122–23 (1996) (holding that a ballot will not be disqualified for a technical error).
In sum, although states have adopted a variety of rules, Arizona’s ballot collection rule is fully consonant with the broad range of rules throughout the United States.12
B
Even more striking than the number of other states with similar provision is that H.B. 2023 follows precisely the recommendation of the bi-partisan Carter-Baker Commission on Federal Election Reform.13 The Carter-Baker Commission found:
Absentee ballots remain the largest source of potential voter fraud. ... Absentee balloting is vulnerable to abuse in several ways: ... Citizens who vote at home, at nursing homes, at the workplace, or in church are more susceptible to pressure, overt and subtle, or to intimidation. Vote buying schemes are far more difficult to detect when citizens vote by mail. States therefore should reduce the risks of fraud and abuse in absentee voting by prohibiting “third-party” organizations, candidates, and political party activists from handling absentee ballots.
Comm’n on Fed. Elections Reform, Building Confidence in U.S. Elections 46 (2005) (“Building Confidence”) (footnote omitted). The Carter-Baker Commission recommended that “States ... should reduce the risks of fraud and abuse in absentee voting by prohibiting ‘third-party’ organizations, candidates, and political party activists from handling absentee ballots.” Id. It made a formal recommendation:
State and local jurisdictions should prohibit a person from handling absentee ballots other than the voter, an acknowledged family member, the U.S. Postal Service or other legitimate shipper, or election officials. The practice in some states of allowing candidates or party workers to pick up and deliver absentee ballots should be eliminated.
Id. at 47 (Recommendation 5.2.1).
The Carter-Baker Commission recommended that states limit the persons, other than the voter, who handle or collect absentee ballots to three classes of persons: (1) family members, (2) employees of the U.S. Postal Service or another recognized shipper, and (3) election officials. H.B. 2013 allows two classes of persons to collect absentee ballots: (1) election officials and (2) employees of the U.S. Postal Service “or any other person who is allowed by law to transmit United States mail.” Ariz. Rev. Stat. § 16-1005(H). H.B. 2023 also provides that the prior restriction on collection of ballots does not apply to “[a] family member, household member or caregiver of the voter.” Id. § 16-1005(I)(2). With respect to election officials and mail delivery workers, Arizona tracks exactly the recommendation from the Commission. With respect to family, however, Arizona’s provision is more generous than the Carter-Baker Commission’s recommendation. Whereas the Commission recommended that only family members be permitted to handled a voter’s absentee ballot, Arizona expanded the class of absentee ballot handlers to “household member[s]” and “caregiver[s].”
I don’t see how Arizona can be said to have violated the VRA when it followed bipartisan recommendations for election reform in an area the Carter-Baker Commission found to be fraught with the risk of voter fraud. Nothing could be more damaging to confidence in our elections than fraud at the ballot box. And there is evidence that there is voter fraud in the collecting of absentee ballots. As the Seventh Circuit described it: “Voting fraud is a serious problem in U.S. elections generally ... and it is facilitated by absentee voting. ... [A]bsentee voting is to voting in person as a take-home exam is to a proctored one.” Griffin, 385 F.3d at 1130–31; see also Wrinn, 440 A.2d at 270 (“[T]here is considerable room for fraud in absentee voting and ... a failure to comply with the regulatory provision governing absentee voting increases the opportunity for fraud.” (citation omitted)); Qualkinbush v. Skubisz, 357 Ill.App.3d 594, 292 Ill.Dec. 745, 826 N.E.2d 1181, 1197 (2004) (“[T]he integrity of a vote is even more susceptible to influence and manipulation when done by absentee ballot.”); Adam Liptak, Error and Fraud at Issue as Absentee Voting Rises, N.Y. Times (Oct. 6, 2012), http://nyti.ms/QUbcrg (discussing a variety of problems in states).14
Organized absentee ballot fraud of sufficient scope to corrupt an election is no doomsday hypothetical: it happened as recently as 2018 in North Carolina. In the state’s Ninth Congressional District, over 282,000 voters cast ballots, either in person or absentee. See Brief of Dan McCready at 7, In re Investigation of Election Irregularities Affecting Ctys. Within the 9th Cong. Dist. (N.C. State Bd. of Elections Feb. 12, 2019) [hereinafter McCready Br.]. North Carolina permits “[a]ny qualified voter” in the state to vote by absentee ballot. N.C. Gen. Stat. § 163A-1295. However, like Arizona, the state adheres to the Commission’s recommendations and restricts the categories of persons who may collect a voter’s absentee ballot. It is a Class I felony in North Carolina for “any person except the voter’s near relative or the voter’s verifiable legal guardian to assist the voter to vote an absentee ballot.” Id. § 163A-1298.
In last year’s election in the Ninth Congressional District, evidence suggested that a political activist hired by the Republican nominee paid employees to collect absentee ballots—possibly more than 1,000—from voters in violation of § 163A-1298. See Indictment, State v. Dowless, No. 19CRS001934 (N.C. Super. Ct. July 30, 2019); McCready Br. at app. 2–3. An employee of the suspected activist testified that she personally collected about three dozen ballots. See Transcript of Evidentiary Hearing at 150, In re Investigation of Election Irregularities Affecting Ctys. Within the 9th Cong. Dist. (N.C. State Bd. of Elections Feb. 18, 2019). She also helped fill in about five or ten incomplete, unsealed ballots in favor of Republican candidates. Id. at 67, 99, 152–53. The ballots were kept at the activist’s home and office for days or longer before they were turned in. Id. at 69. A voter testified that she turned over her blank ballot to the activist’s employees in an unsealed envelope, trusting that the activist would make a good decision for her. Id. at 207–08, 214–15.
This coordinated ballot fraud led the state Board of Elections to invalidate the results of the election, which had been decided by only 905 votes—fewer than the amount of suspected fraudulent ballots. Order at 10, 44–45, In re Investigation of Election Irregularities Affecting Ctys. Within the 9th Cong. Dist. (N.C. State Bd. of Elections Mar. 13, 2019). The residents of the district—some 778,447 Americans—were thus unrepresented in the House of Representatives for the better part of a year. Perhaps the more devastating injury will be the damage this episode does to North Carolinians’ confidence in their election system.
The majority acknowledges that the Democratic Party disproportionately benefits from get-out-the-vote efforts by collecting mail-in ballots. See, e.g., Maj. Op. at –––– (quoting Reagan, 329 F. Supp. 3d at 870). Further, the majority acknowledges that Democratic activists have often led such collection efforts. Id. Yet the experience of North Carolina with Republican activists shows starkly the inherent danger to allowing political operatives to conduct collections of mail-in ballots. Arizona is well within its right to look at the perils endured by its sister states and enact prophylactic measures to curtail any similar schemes. By prohibiting overtly political operatives and activists from playing a role in the ballot-collection process, Arizona mitigates this risk. And the State’s well-acknowledged past sins should not prevent it from using every available avenue to keep safe the public’s trust in the integrity of electoral outcomes.
Indeed, Arizona does not have to wait until it has proof positive that its elections have been tainted by absentee ballot fraud before it may enact neutral rules. “Legislatures ... should be permitted to respond to potential deficiencies in the electoral process with foresight rather than reactively.” Munro v. Socialist Workers Party, 479 U.S. 189, 195, 107 S.Ct. 533, 93 L.Ed.2d 499 (1986). In Crawford, the Supreme Court quoted with approval the Carter-Baker Commission:
There is no evidence of extensive fraud in U.S. elections or of multiple voting, but both occur, and it could affect the outcome of a close election. The electoral system cannot inspire public confidence if no safeguards exist to deter or detect fraud or to confirm the identity of voters.
Crawford, 553 U.S. at 194, 128 S.Ct. 1610 (quoting Building Confidence at 18) (footnote omitted).
The majority today holds that, as a matter of federal law, Arizona may not enforce a neutrally drawn statute recommended by a bi-partisan commission criminalizing the very conduct that produced a fraudulent outcome in a race for Congress less than a year ago. When the Voting Rights Act requires courts to consider the “totality of the circumstances,” it is a poor understanding of the Act that would strike common time, place, and manner restrictions designed to build confidence in the very voting system that it now leaves vulnerable.
III
As citizens of a democratic republic, we understand intuitively that we have a legal right and a moral duty to cast a ballot in free elections. The states have long had the power to fashion the rules by which its citizens vote for their national, state, and local officials. Once we consider that “totality of the circumstances” must take account of long-held, widely adopted measures, we must conclude that Arizona’s time, place, and manner rules are well within our American democratic-republican tradition. Nothing in the Voting Rights Act makes “ ‘evenhanded restrictions that protect the integrity and reliability of the electoral process’ ... invidious.” Crawford, 553 U.S. at 189–90, 128 S.Ct. 1610 (quoting Anderson, 460 U.S. at 788 n.9, 103 S.Ct. 1564).
I would affirm the judgment of the district court, and I respectfully dissent.
Appendix A
State and Territory Laws Regarding Treatment of Out-of-Precinct Provisional Ballots
(See Westlaw or contact the library for assistance)
Appendix B
State and Territory Treatment of Out-of-Precinct Provisional Ballots15
(See Westlaw or contact the library for assistance)
Appendix C
State and Territory Laws Regarding the Collection of Absentee Ballots
(See Westlaw or contact the library for assistance)
All Citations
948 F.3d 989, 2020 WL 414448, 20 Cal. Daily Op. Serv. 630
Footnotes |
|
Judge Berzon was drawn to replace Judge Graber. Judge Berzon has read the briefs, reviewed the record, and watched the recording of oral argument held on March 27, 2019. |
|
Section 2 of the Voting Rights Act prohibits a State from adopting an election practice that “results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color.” 52 U.S.C. § 10301(a). |
|
The majority’s effort to deny history can easily be dismissed. Maj. Op. –––– – ––––. As Judge Bybee’s dissent ably recounts, not only Arizona but 21 other states have restricted early balloting for years. Bybee, J. Diss. Op. –––– – ––––. |
|
While the majority refers to the legislation as “H.B. 2023,” I prefer to call it the ballot-collection policy by which it is commonly known and will do so throughout the dissent. |
|
As the majority admits, we review the district court’s “overall finding of vote dilution” under § 2 of the Voting Rights Act only for clear error. Thornburg v. Gingles, 478 U.S. 30, 79, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986) (emphasis added); Maj. Op. –––– – ––––. The majority quotes an elaboration of this standard by the Supreme Court in Gingles. Maj. Op. –––– – ––––. But the Court in Gingles actually held that the district court’s ultimate finding was not clearly erroneous. Gingles, 478 U.S. at 80, 106 S.Ct. 2752. |
|
The majority correctly asserts that Gingles was a vote dilution not vote denial case. However, it incorrectly claims the standard in a vote denial case is different and, without stating such standard, it simply concludes that the 3,709 ballots cast out of precinct in the 2016 general election in Arizona is more than any “de minimis number” below which there is no Section 2 violation, without ever revealing what such minimum threshold might be. Maj. Op. ––––. The majority cites League of Women Voters, a vote denial case, to reach this conclusion. See769 F.3d at 248–49. Yet, in that case, the Fourth Circuit relies on Gingles throughout to determine that the same analysis applies to vote denial and vote dilution cases. Id. at 238–40. Earlier in its opinion, the majority itself uses Gingles as the standard for analyzing a § 2 violation in a vote denial case. Maj. Op. ––––. The distinction the majority attempts to draw fails because, contrary to what the majority implies, “a § 2 challenge based purely on a showing of some relevant statistical disparity between minorities and whites, without any evidence that the challenged voting qualification causes that disparity, will be rejected,” Gonzalez v. Arizona, 677 F.3d 383, 495 (9th Cir. 2012) (internal quotation marks omitted), and “[t]his approach applies both to claims of vote denial and vote dilution.” Id. at 495 n. 32, 74 S.Ct. 686. |
|
As Judge Ikuta explained, “an election rule requiring voters to identify their correct precinct in order to have their ballots counted does not constitute a ‘disenfranchisement’ of voters.” DNC, 904 F.3d at 730 n.33; see also id. at 724 n.27. |
|
The majority suggests that DNC challenges only “Arizona’s policy, within that system, of entirely discarding OOP ballots” as opposed to counting or partially counting them. Maj. Op. ––––. But this is not a compromise position: there is no difference between counting and partially counting a ballot cast out-of-precinct. Counting an OOP ballot would entail evaluating the ballot to determine on which issues the person would have been qualified to vote in his or her assigned precinct and discarding the person’s votes as to issues on which he or she would not have been qualified to vote. Certainly, the majority isn’t suggesting that a person would ever be allowed to vote on issues which he or she would not have been eligible to vote even in the assigned precinct. It is difficult to discern any other possible meaning for what the majority refers to as entirely “counting” out-of-precinct ballots. |
|
The Fifteenth Amendment authorizes Congress to enforce its guarantee that the right “to vote shall not be denied or abridged ... by appropriate legislation.” U.S. Const. amend. XV. Section 2 of the Voting Rights Act is such legislation. Shelby Cty. v. Holder, 570 U.S. 529, 536, 133 S.Ct. 2612, 186 L.Ed.2d 651 (2013). |
|
Because the majority concludes that the OOP policy and the ballot-collection policy violate § 2 of the Voting Rights Act and the Fifteenth Amendment to the United States Constitution, it does not reach DNC’s claim that such policies also violate the First and Fourteenth Amendments to the United States Constitution. I will not belabor such claims here; for these purposes, it is sufficient to say that—for many of the reasons and based on much of the evidence cited above—I would also conclude that neither practice violates the First and Fourteenth Amendments. |
|
I join in full Judge O’Scannlain’s dissent. I write separately to place the majority’s decision today in context of the American democratic tradition. |
|
In relevant part, § 2 of the Voting Rights Act provides that “[n]o voting qualification or prerequisite to voting or standard, practice, or procedure shall be imposed or applied by any State ... in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color.” 52 U.S.C. § 10301(a). A violation of § 2(a) may be shown “based on the totality of the circumstances ... [if] the political processes leading to nomination or election in the State ... are not equally open to participation by members of a class of citizens [on account of race or color].” Id. § 10301(b). |
|
“One of the major voting innovations in certain states was the increase in the number of polling places.” Robert J. Dinkin, Voting in Revolutionary America: A Study of Elections in the Original Thirteen States, 1776–1789, at 96 (1982). Among the states, New York led the way, “enacting a law in 1778 which stated that all future elections should be held ‘not by counties but by boroughs, towns, manors, districts, and precincts.’ ” Id. at 97 (quoting Laws of New York, sess. 1, chap. 16 (1778)). In early America, polling places were located where the people were: |
|
For many years, a voter was not even permitted to cast a provisional ballot in a precinct other than her own. See Harris, Election Administration in the United States, at 287–88. The Help America Vote Act (HAVA) now requires states to permit voters to cast a provisional ballot. 52 U.S.C. § 21082(a). HAVA, however, does not affect a state’s rules about how to process a provisional ballot. It does provide that states must create a toll-free number that “any individual who casts a provisional ballot may access to discover whether the vote of that individual was counted, and, if the vote was not counted, the reasons that the vote was not counted.” 52 U.S.C. § 21082(a)(5)(B); see Blackwell, 387 F.3d at 576 (“HAVA is quintessentially about being able to cast a provisional ballot. ... [B]ut the ultimate legality of the vote cast provisionally is generally a matter of state law.”). |
|
I have listed all fifty states, the District of Columbia, and U.S. territories, with relevant citations to their treatment of out of precinct votes, in Appendix A. In Appendix B, I have categorized the jurisdictions by rule. |
|
For example, five states will count an out-of-precinct vote, but only if the ballot is filed in the voter’s county (Kansas, New Mexico, Pennsylvania, Utah) or town (Massachusetts). Louisiana and Rhode Island will only count votes for federal office. Puerto Rico will count only votes for Governor and Resident Commissioner. |
|
Four states (Idaho, Minnesota, New Hampshire, North Dakota) are not accounted for in either list because they allow same-day registration and do not use provisional ballots. |
|
The Majority dismisses this point by highlighting how Arizona has frequently changed polling places in some localities. Maj. Op. at –––– (referring to Arizona’s high rate of OOP voting). But there is no evidence in the record that the same voters’s ballots are excluded as OOP year after year. My point is that a voter who has had her ballot excluded as OOP is more likely to exercise greater care in finding the right polling location next time. |
|
The Majority worries that OOP voters may never come to know that their votes were in fact rejected and, hence, will never learn from the situation. Maj. Op. at ––––. Whatever the cause for the Majority’s concern, Arizona’s statutory law is not to blame. Arizona law specifically requires county recorders to establish “a method of notifying the provisional ballot voter at no cost to the voter whether the voter’s ballot was verified and counted and, if not counted, the reason for not counting the ballot.” Ariz. Rev. Stat. Ann. § 16-584(F) (2019). Thus, voters should have the opportunity to find out whether their vote was counted. |
|
“The exercise of a public franchise by proxy was illegal at common law.” Cortlandt F. Bishop, History of Elections in the American Colonies 129 (1893). The Colonies experimented with proxy votes, with varying degrees of success. Proxy voting was not a success in at least one colony. A 1683 letter to the Governor of South Carolina warned: |
|
Until recently, two other states had similar provisions on the books. California formerly limited who could return mail ballots to the voter’s family or those living in the same household. CompareCal. Elec. Code § 3017(a)(2) (West 2019), withCal. Elec. Code § 3017(a) (West 2015). It only amended its law in 2016. 2016 Cal. Legis. Serv. ch. 820 (West). Illinois also used to make it a felony for anyone but the voter, his or her family, or certain licensed delivery companies to mail or deliver an absentee ballot. 10 Ill. Comp. Stat. Ann. 5/19-6 (1996); 10 Ill. Comp. Stat. 5/29-20(4). Illinois amended that provision in 2015 to let voters authorize others to mail or deliver their ballots. 10 Ill. Comp. Stat. Ann. 5/19-6 (2015). |
|
For context, Appendix C provides the relevant provisions of the laws from all fifty states, the District of Columbia, and the U.S. territories regarding the collection and mailing of absentee ballots. |
|
The Commission on Federal Election Reform was organized by American University’s Center for Democracy and Election Management and supported by the Carnegie Corporation of New York, The Ford Foundation, the John S. and James L. Knight Foundation, and the Omidyar Network. It was co-chaired by former President Jimmy Carter and former Secretary of State James Baker. |
|
Pressure on absentee voters has long been noted. See Harris, Election Administration in the United States, at 302 (“The amount of intimidation now exercised by the precinct captain in many sections of large cities is very great; with mail voting it would be enormously increased. The overbearing and dominant precinct captain would insist upon seeing how each voter under obligation to him had marked his ballot, and the voter would have no protection against such tactics.”). |
|
Idaho, Minnesota, New Hampshire, and North Dakota are not included because they do not use provisional ballots. See supra Appendix A. |
|