League of Women Voters of Florida Inc. v. Florida Secretary of State

66 F.4th 905

• Court: Court of Appeals for the Eleventh Circuit
• Decided: April 27, 2023
• Docket: 22-11143
• Status: Published

Opinion

USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 1 of 79

[PUBLISH]

In the

United States Court of Appeals For the Eleventh Circuit

____________________

No. 22-11143 ____________________

LEAGUE OF WOMEN VOTERS OF FLORIDA INC., et al., Plaintiffs-Appellees, versus FLORIDA SECRETARY OF STATE, et al.,

Defendants-Appellants.

Appeal from the United States District Court for the Northern District of Florida D.C. Docket No. 4:21-cv-00186-MW-MAF ____________________ USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 2 of 79

2 Opinion of the Court 22-11143

Before WILLIAM PRYOR, Chief Judge, and JILL PRYOR and GRANT, Circuit Judges. WILLIAM PRYOR, Chief Judge:

This appeal involves four recently enacted provisions of Florida’s election law, including provisions that regulate ballot drop boxes, the solicitation of voters at the polls, and the delivery of voter-registration forms by third-party voter-registration organ- izations. Several plaintiff organizations sued the Florida Secretary of State, the Florida Attorney General, and several Supervisors of Elections. After a bench trial, the district court enjoined three pro- visions because it found they were adopted with the intent to dis- criminate against black voters in violation of the Fourteenth and Fifteenth Amendments as well as section 2 of the Voting Rights Act. And it imposed a preclearance requirement under section 3(c) of the Act. The district court also ruled that the solicitation provi- sion was unconstitutionally vague and overbroad in violation of the First and Fourteenth Amendments. Finally, it enjoined a provi- sion that required third-party voter-registration organizations to provide a disclaimer to voters who use their services to register to vote, but all parties agree that any appeal of the judgment as to that provision has been rendered moot by the repeal of the provision. Because we hold that the findings of intentional racial discrimina- tion rest on both legal errors and clearly erroneous findings of fact and that only part of the solicitation provision is unconstitutional, we reverse in part, affirm in part, vacate in part, and remand. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 3 of 79

22-11143 Opinion of the Court 3

I. BACKGROUND

In recent decades, the Florida Legislature has amended the election code to make voting more convenient for eligible voters. In 1980, most voters had to cast their ballots in person at their local precincts. Voters could cast an absentee ballot only for one of six reasons: inability to vote in person without assistance, absence from the county on election day, service as an election official in a different precinct, religious observance, change of residency within the state too late to register at the new address, or change of resi- dency outside the state if the voter is unable to vote under the laws of the new state. FLA. STAT. § 101.64 (1980). State law required the voter to attest that he or she qualified to vote absentee, and the attestation had to be notarized or witnessed by two adults. Id. Since then, much has changed. By 2001, Florida no longer required any excuse to vote absentee. Ch. 2001-40, § 53, Laws of Fla. In 2004, the state eliminated the requirement that absentee ballots be wit- nessed. Ch. 2004-232, § 1, Laws of Fla. Also in 2004, Florida began allowing no-excuse early voting. Ch. 2004-252, § 13, Laws of Fla. Most recently, the state required that ballot drop boxes be made available in every county. Ch. 2019-162, § 20, Laws of Fla.

As the Legislature has expanded opportunities for voting, the State of Florida has also become more racially and ethnically diverse. In 1980, the projected voting-age population was about 88 percent white and 12 percent black. BUREAU OF THE CENSUS, U.S. DEP’T OF COM., PROJECTIONS OF THE POPULATION OF VOTING AGE USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 4 of 79

4 Opinion of the Court 22-11143

FOR STATES: NOVEMBER 1980, at 6 (1980). About 6 percent of the voting-age population was of “Spanish origin” based on the previ- ous 1970 census. Id. at 7. According to the organizations’ expert, Mr. Cooper, white voters in 2019 made up about 78 percent of the citizen voting-age population, and black voters made up 15 per- cent. Hispanic or Latino voters constituted about 21 percent of the citizen voting-age population.

Florida’s election code continues to evolve. In the regular session immediately following the 2020 election, the Florida Legis- lature adopted Senate Bill 90. According to the district court, the new law “made a sweeping set of changes to Florida’s election code, with a specific focus on [vote-by-mail]” procedures. The bill incorporated input from a wide array of stakeholders. The county supervisors of elections, through their trade organization, influ- enced the final version. In fact, their lobbyist testified that “proba- bly 80 percent of the provisions . . . have a tweak that [was] [the supervisors’] suggestion[] on how to operate.”

S.B. 90 was a substantively wide-ranging bill. The enrolled version spanned 48 pages and addressed various topics, including procedures for challenging a provision of the election code, testing protocols for the online voter-registration system, live turnout data reports, guidelines for the duplication of damaged vote-by-mail bal- lots, and rules for the inspection of ballot materials. See generally USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 5 of 79

22-11143 Opinion of the Court 5

S.B. 90, 123d Leg. Sess. (Fla. 2021). Only a subset of its provisions was challenged in this action.

The district court enjoined four provisions. First, the district court enjoined enforcement of the drop-box provision. Florida law allows voters who request vote-by-mail ballots to return those bal- lots at secure intake stations, colloquially known as drop boxes. The drop-box provision requires that “secure ballot intake sta- tion[s]” be “monitored in person by an employee of the supervi- sor’s office”; limits the hours of drop-box availability to early voting hours, except for drop boxes located “at an office of the [county] supervisor [of elections]”; and establishes a $25,000 civil penalty against the supervisor “[i]f any secure ballot intake station is left accessible for ballot receipt other than as authorized by this sec- tion.” FLA. STAT. § 101.69(2)–(3). Second, the district court enjoined enforcement of the solicitation provision, which prohibits any “person, political committee, or other group or organization” from “solicit[ing] voters inside the polling place or within 150 feet of a secure ballot intake station or the entrance to any polling place” or other voting location. Id. § 102.031(4)(a). It defines “solicit” and “so- licitation” to include, among other things, “engaging in any activity with the intent to influence or effect of influencing a voter.” Id. § 102.031(4)(b). Third, the district court enjoined enforcement of the registration-delivery provision. Florida law allows third-party voter-registration organizations to collect voter-registration forms and deliver them to election officials. The registration-delivery pro- vision requires that the organization “promptly deliver[]” the USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 6 of 79

6 Opinion of the Court 22-11143

registration forms “to the division or the supervisor of elections in the county in which the applicant resides within 14 days after the application was completed by the applicant, but not after registra- tion closes for the next ensuing election.” Id. § 97.0575(3)(a). Fi- nally, the district court enjoined enforcement of the now-repealed registration-disclaimer provision, which required that a third-party voter-registration organization “notify the applicant at the time the application is collected that the organization might not deliver the application . . . in less than 14 days or before registration closes for the next ensuing election,” “advise the applicant that he or she may deliver the application in person or by mail,” and “inform the ap- plicant how to register online with the [Division of Elections] and how to determine whether the application has been delivered.” Id. § 97.0575(3)(a), repealed by Ch. 2022-73, § 7, Laws of Fla.

Four sets of organizations challenged these and other provi- sions of S.B. 90 in four cases, which were consolidated at trial and on appeal. The district court identified the four actions by their lead plaintiffs: the Florida State Conference of Branches and Youth Units of the NAACP, Florida Rising Together, the League of Women Voters of Florida, and the Harriet Tubman Freedom Fighters. The defendants included the Secretary of State, the Attor- ney General, and several Supervisors of Elections. The Republican National Committee and the National Republican Senatorial Com- mittee intervened as defendants. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 7 of 79

22-11143 Opinion of the Court 7

After a bench trial, the district court determined that four of S.B. 90’s challenged provisions violated the Constitution and the Voting Rights Act. It found that the drop-box provision, the regis- tration-delivery provision, and the solicitation provision all vio- lated the Equal Protection Clause of the Fourteenth Amendment and abridged the right to vote on the basis of race in violation of the Fifteenth Amendment. The district court determined that alt- hough the Legislature did not intend to discriminate against Latino voters, the Legislature did intend to discriminate against black vot- ers. It also determined that the challenged provisions violated sec- tion 2 of the Voting Rights Act. The district court permanently en- joined the enforcement of the three provisions. It determined that the solicitation provision was impermissibly vague in violation of the Fourteenth Amendment and overbroad in violation of the First Amendment. And it permanently enjoined enforcement of the so- licitation provision on that basis, as well. The district court deter- mined that the registration-disclaimer provision violated the First Amendment by compelling speech and permanently enjoined its enforcement. The district court also subjected the State of Florida to limited preclearance for ten years under section 3(c) of the Vot- ing Rights Act. See 52 U.S.C. § 10302(c).

The district court also rejected some of the organizations’ claims and declined to reach others. For example, it determined that two provisions that “reduce[d] the duration of a voter’s [vote- by-mail] ballot request” and required certain identifying infor- mation in a vote-by-mail ballot request were not unconstitutional USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 8 of 79

8 Opinion of the Court 22-11143

and did not violate the Voting Rights Act. It rejected a challenge to the drop-box provision, the solicitation provision, and the vote-by- mail request provision under the Americans with Disabilities Act. The district court declined to reach the question whether the drop- box, solicitation, and registration-delivery provisions unduly bur- den the right to vote, without respect to race, under the First and Fourteenth Amendments. And it rejected the argument that the vote-by-mail provisions did so. It also declined to reach Florida Ris- ing Together’s claim that the registration-delivery provision vio- lated the First Amendment as applied to its voter-registration activ- ities and the Harriet Tubman Freedom Fighters’ claim that certain penalties in the registration-disclaimer provision were unconstitu- tionally vague.

After the filing of this appeal, the Legislature repealed the registration-disclaimer provision. See Ch. 2022-73, § 7, Laws of Fla. The parties agree that any appeal of the judgment respecting the constitutionality of this provision is moot. The state officials and Republican intervenors contend that the judgment as to this provi- sion should be vacated, but the organizations oppose vacatur.

We stayed the judgment pending appeal. League of Women Voters of Fla., Inc. v. Fla. Sec’y of State, 32 F.4th 1363, 1369 (11th Cir. 2022). Our analysis was strongly influenced by the Purcell prin- ciple, which cautions district courts against “enjoin[ing] state elec- tion laws in the period close to an election” and encourages appel- late courts to stay such injunctions. Id. at 1371 (quoting Merrill v. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 9 of 79

22-11143 Opinion of the Court 9

Milligan, 142 S. Ct. 879, 879 (2022) (Kavanaugh, J., concurring)); see also Purcell v. Gonzalez, 549 U.S. 1 (2006). We now address the merits.

II. STANDARD OF REVIEW

In reviewing a judgment following a bench trial, we review de novo both conclusions of law and the application of the law to the facts. U.S. Commodity Futures Trading Comm’n v. S. Tr. Met- als, Inc., 894 F.3d 1313, 1322 (11th Cir. 2018). And we review find- ings of fact for clear error. Id. “We will not find clear error unless our review of the record leaves us with the definite and firm con- viction that a mistake has been committed.” Id. (citation and inter- nal quotation marks omitted).

III. DISCUSSION

We divide our discussion into six parts. First, we explain that the drop-box, solicitation, and registration-delivery provisions do not violate the Fourteenth or Fifteenth Amendment to the Consti- tution. Second, we explain that those provisions also do not violate section 2 of the Voting Rights Act. Third, we hold that the district court erred in subjecting Florida to preclearance under section 3 of the Voting Rights Act. Fourth, we explain that the solicitation pro- vision is, in part, unconstitutionally vague. Fifth, we vacate the judgment against the now-repealed registration-disclaimer provi- sion. Finally, we remand the case for the district court to determine whether the drop-box and registration-delivery provisions unduly USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 10 of 79

10 Opinion of the Court 22-11143

burden the right to vote under the First and Fourteenth Amend- ments.

A. The Drop-Box, Solicitation, and Registration-Delivery Provi- sions Do Not Violate the Fourteenth or Fifteenth Amendment.

We review an alleged violation of equal-protection rights under the Fourteenth Amendment or an alleged denial or abridg- ment on account of race of the right to vote under the Fifteenth Amendment using a two-step burden-shifting test. Greater Bir- mingham Ministries v. Sec’y of State for the State of Ala., 992 F.3d 1299, 1321 (11th Cir. 2021). First, the plaintiffs must prove both that the law will have a discriminatory impact and that it was adopted with discriminatory intent. Id. Second, “the burden shifts to the law’s defenders to demonstrate that the law would have been en- acted without this racial discrimination factor.” Id. (quoting Hunter v. Underwood, 471 U.S. 222, 228 (1985)) (alteration adopted).

In determining whether a “law has both a discriminatory in- tent and effect,” we rely on the guidance in Village of Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252 (1977). See Greater Birmingham, 992 F.3d at 1321. Arlington Heights and later caselaw require considering several factors about the law and its adoption: (1) the impact of the challenged law; (2) the historical background; (3) the specific sequence of events lead- ing up to its passage; (4) procedural and substantive USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 11 of 79

22-11143 Opinion of the Court 11

departures; . . . (5) the contemporary statements and actions of key legislators[;] . . . (6) the foreseeability of the disparate impact; (7) knowledge of that impact[;] and (8) the availability of less discriminatory alterna- tives.

Id. at 1321–22. We review these factors in the same order as the district court addressed them, and our review establishes that the district court committed reversible error.

1. Historical Background

The district court delved deep into Florida’s past. It began with an overview of racist voting laws enacted after the Civil War and discrimination enduring into the twentieth century. It then dis- cussed twenty-first century examples of allegedly racially moti- vated voter-roll “purges” and changes to Florida voting laws. In some of the cases that the district court cited, federal courts ruled that these laws were not racially motivated, and in others the courts never reached the question.

In none of the cases from this century cited by the district court did a court determine that a challenged Florida election law resulted from intentional discrimination. But the district court was persuaded otherwise. “Once is an accident, twice is a coincidence, [and] three times is a pattern,” it wrote. “At some point, when the Florida Legislature passes law after law disproportionately burden- ing Black voters, this Court can no longer accept that the effect is incidental.” The district court found that Florida’s “long history of USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 12 of 79

12 Opinion of the Court 22-11143

racial discrimination against Black . . . Floridians” “informs its pre- sent.”

The district court also discussed “socioeconomic disparities . . . between racial groups.” It framed such disparities as “the stark results of a political system that, for well over a century, has overrepresented White Floridians and underrepresented Black . . . Floridians.” The organizations contend that such statistics serve as “evidence of ‘the lingering effects of past discrimination.’” (Quot- ing Rogers v. Lodge, 458 U.S. 613, 626 (1982)). From the start, the district court erred. As we have ex- plained, a federal court must remain “mindful of the danger of al- lowing the old, outdated intentions of previous generations to taint [Florida]’s legislative action forevermore on certain topics.” Greater Birmingham, 992 F.3d at 1325. We have rejected the argu- ment that “a racist past is evidence of current intent.” Id.; see also City of Mobile v. Bolden, 446 U.S. 55, 74 (1980) (plurality opinion) (“[P]ast discrimination cannot, in the manner of original sin, con- demn governmental action that is not itself unlawful.”), superseded in part by statute, Voting Rights Act Amendments of 1982, Pub. L. No. 97-205, 96 Stat. 131, as recognized in Thornburg v. Gingles, 478 U.S. 30, 43–44 (1986). Instead, we have explained that “the princi- ples of equal sovereignty counsel[] against . . . disparate treatment of” a state based on its history “and guide[] us to look at the precise circumstances surrounding the passing of the” law in question. Greater Birmingham, 992 F.3d at 1325 (citing Shelby Cnty. v. Holder, 570 U.S. 529, 553 (2013)). And we apply “the presumption USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 13 of 79

22-11143 Opinion of the Court 13

of legislative good faith” even in the light of “a finding of past dis- crimination.” Id. (quoting Abbott v. Perez, 138 S. Ct. 2305, 2324 (2018)). The district court did not heed our precedent explaining the proper scope of a historical inquiry.

The organizations maintain in their briefs that a state’s his- tory of discrimination and socioeconomic disparities are relevant to an analysis of discriminatory intent. (Citing Rogers, 458 U.S. at 624–26; United States v. Marengo Cnty. Comm’n, 731 F.2d 1546, 1567–68 (11th Cir. 1984)). But we have read the Arlington Heights factor at issue “not [to] provid[e] an unlimited look-back to past discrimination.” Greater Birmingham, 992 F.3d at 1325 (citing Ar- lington Heights, 429 U.S. at 267). Evidence of historical discrimina- tion imported through socioeconomic data is no exception. And under our precedent, this history cannot support a finding of dis- criminatory intent in this case.

Florida’s more recent history does not support a finding of discriminatory intent. The only pieces of legislation cited by the district court that were adopted since the year 2000 offer no sup- port for its finding of discriminatory intent. For instance, the dis- trict court discussed H.B. 1355, a law adopted in 2011 that reduced early-voting days. The District Court for the District of Columbia declined to preclear the law because the State did not establish that the law would have a “nonretrogressive effect,” but it did not reach the question whether H.B. 1355 was adopted with discriminatory intent. See Florida v. United States, 885 F. Supp. 2d 299, 337, 351 USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 14 of 79

14 Opinion of the Court 22-11143

(D.D.C. 2012). The District Court for the Middle District of Florida determined that plaintiffs who sought to enjoin the same bill “failed to demonstrate a substantial likelihood of success on their claim that” the bill was adopted with discriminatory intent, Brown v. Detzner, 895 F. Supp. 2d 1236, 1239, 1249 (M.D. Fla. 2012), and the District Court for the Northern District of Florida enjoined differ- ent provisions on other grounds, see League of Women Voters of Fla. v. Browning, 863 F. Supp. 2d 1155, 1157–58 (N.D. Fla. 2012). In the end, the Legislature “restored the pre-2011 early voting hours,” which the organizations’ expert Dr. Austin described as “a sign of something positive.” Likewise, a 2019 law required felons to pay any financial obligations arising from their sentences before Florida restored their voting rights, and the District Court for the Northern District of Florida determined that the bill was “not mo- tivated by race.” Jones v. DeSantis, 462 F. Supp. 3d 1196, 1238 (N.D. Fla. 2020), rev’d on other grounds sub nom. Jones v. Governor of Fla., 975 F.3d 1016 (11th Cir. 2020) (en banc). Voter roll “purges,” which the district court also referenced to support its finding of dis- criminatory intent, are not conducted by the Legislature, and a challenged 2001 law that authorized a new procedure for identify- ing felons ineligible to vote was repealed before any court reached judgment.

Finally, the district court discussed the close relationship be- tween racial identification and political affiliation among Floridi- ans. According to the district court, “for White and Black voters in Florida, separating race from politics only works in science fiction.” USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 15 of 79

22-11143 Opinion of the Court 15

But the Supreme Court has warned against conflating discrimina- tion on the basis of party affiliation with discrimination on the basis of race. See Brnovich v. Democratic Nat’l Comm., 141 S. Ct. 2321, 2349 (2021) (“[P]artisan motives are not the same as racial mo- tives.”). To be sure, as the organizations point out, “[i]ntentionally targeting a particular race’s access to the franchise because its members vote for a particular party” is impermissible. (Quoting N.C. State Conf. of the NAACP v. McCrory, 831 F.3d 204, 222 (4th Cir. 2016)). But we must be careful not to infer that racial targeting is, in fact, occurring based solely on evidence of partisanship. Evi- dence of race-based discrimination is necessary to establish a con- stitutional violation.

2. Specific Sequence of Events Leading up to Passage

Next, the district court reviewed the events leading up to the passage of S.B. 90. It considered minority voters’ increased use of vote-by-mail ballots, some of which are deposited in drop boxes, in the 2020 election. It examined, and was unconvinced by, the prof- fered justifications for S.B. 90’s reforms.

The district court summarized recent changes in the de- mographics of vote-by-mail voters. Historically, white and Latino voters were more likely than black voters to use vote by mail. In 2020, amid the COVID-19 pandemic, voters across the demo- graphic spectrum relied more heavily on vote by mail. Compared with the rate of use from 2014 to 2020, white voters’ use of vote by mail in 2020 increased from just over 30 percent to about 45 USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 16 of 79

16 Opinion of the Court 22-11143

percent, and black voters’ use increased from about 20 percent to about 40 percent. Republican voters’ use of vote by mail minimally increased from 40 percent to 42 percent, while Democratic voters’ use increased from 35.5 percent to 61 percent. The district court presented these statistics to suggest that black voters’ increased re- liance on vote by mail prompted the election reforms.

At most, the statistics suggest that Democratic voters’ in- creased use may have been a motivating factor. In 2020, Florida Democrats relied on vote by mail more than Florida Republicans; but by the district court’s own account of the statistics, Florida black voters relied on vote by mail less than Florida white voters. Once again, partisan discrimination must not be conflated with ra- cial discrimination. See Brnovich, 141 S. Ct. at 2349.

The district court then reviewed the sponsors’ and support- ers’ proffered justifications for the enactment of S.B. 90. It stated that “the exact justification for SB 90 as a whole, and for its constit- uent parts, is difficult to pin down, with sponsors and supporters offering conflicting or nonsensical rationales.” But that difficulty is to be expected when examining the subjective intent of a multi- member body. Cf. Wis. Pub. Intervenor v. Mortier, 501 U.S. 597, 620 (1991) (Scalia, J., concurring in judgment) (explaining that com- mittee reports “do[] not necessarily say anything about what Con- gress as a whole thought”).

The district court expressed skepticism of the supporters’ professed motivation: preventing voter fraud. It cited high voter USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 17 of 79

22-11143 Opinion of the Court 17

confidence in Florida’s 2020 election and the fact that the Legisla- ture did not amend the law when voter fraud was alleged in 2018. The district court found a lack of “evidence before the Legislature that fraud is even a marginal issue in Florida elections.” The organ- izations agree, describing S.B. 90 as “the proverbial solution in search of a problem,” which “support[s] the inference that the true purpose of the bill was impermissible discrimination.”

This analysis is flawed for three reasons. First, our precedent does not require evidence of voter fraud to justify adopting legisla- tion that aims to prevent fraud. Second, even if it did, the record establishes that fraud, including vote-by-mail fraud, has plagued Florida elections in the past. Third, the record establishes that sup- porters of S.B. 90 sought to prevent the type of fraud that had been observed in Florida and other jurisdictions through this legislation.

First, the district court’s reasoning—implicitly requiring ev- idence of voter fraud in Florida to justify prophylactic measures— does not follow our precedents. “[T]he Supreme Court has already held that deterring voter fraud is a legitimate policy on which to enact an election law, even in the absence of any record evidence of voter fraud.” Greater Birmingham, 992 F.3d at 1334. For exam- ple, the Supreme Court has explained that even though a “record contain[ed] no evidence of any such fraud actually occurring in [a particular state] at any time,” “flagrant examples of such fraud in other parts of the country” can help “demonstrate that not only is the risk of voter fraud real but that it could affect the outcome of a USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 18 of 79

18 Opinion of the Court 22-11143

close election.” Crawford v. Marion Cnty. Election Bd., 553 U.S. 181, 194–96 (2008) (plurality opinion); see also Brnovich, 141 S. Ct. at 2348 (holding, in a Voting Rights Act analysis, that “it should go without saying that a State may take action to prevent election fraud without waiting for it to occur and be detected within its own borders”). Even if there were no evidence of voter fraud in Florida, our precedents would not require it before a bill like S.B. 90 could be adopted.

Second, the record includes undisputed evidence of fraud— including vote-by-mail fraud in Florida—in any event. Dr. Moreno, an expert for the state officials and Republican Party intervenors, detailed in his report several instances of fraud that had occurred in Florida since the 1990s. For example, in 1993, a judge ordered a new mayoral election in the City of Hialeah because of pervasive absentee-ballot fraud. In a 1997 Miami mayoral race, a court “threw out all the absentee ballots cast in the election,” which reversed the election’s outcome. In 2013, a congressional chief of staff “went to jail after being implicated in a sophisticated scheme to manipulate the previous year’s primary elections by submitting hundreds of fraudulent absentee-ballot requests.” Dr. Moreno also collected examples of smaller-scale viola- tions that occurred in more recent elections: a “ballot broker” who pleaded guilty to absentee-ballot fraud charges after she allegedly exploited hundreds of elderly Hispanic voters; an election worker who pleaded guilty to marking some of the vote-by-mail ballots she opened; and a Miami Commission candidate who allegedly USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 19 of 79

22-11143 Opinion of the Court 19

harassed and pressured elderly voters in a public housing complex, among other incidents. When convictions had been obtained, it was not within the district court’s discretion to question whether these instances of fraud took place. Cf. Emich Motors Corp. v. Gen. Motors Corp., 340 U.S. 558, 569 (1951); In re Raiford, 695 F.2d 521, 523 (11th Cir. 1983). In fact, absentee voter fraud was an issue of such concern in Miami-Dade County that in 2012 the State Attor- ney’s Office convened a grand jury to address the problem. See FINAL REPORT OF THE MIAMI-DADE COUNTY GRAND JURY, IN CIRCUIT COURT OF THE ELEVENTH JUDICIAL CIRCUIT OF FLORIDA, SPRING TERM 2012. Representative Ingoglia, a sponsor, also dis- cussed some of these examples on the House floor. To the extent that the district court purported to find no history of voter fraud in Florida, it committed clear error.

Even the 2020 election gave rise to allegations of voter fraud. For instance, Maria Matthews, Director of the Florida Division of Elections, estimated that her department forwarded “upwards of 75” complaints of election fraud to law enforcement from the 2020 election cycle. The Division would only forward a complaint if it “found some legal sufficiency to the complaint.” This evidence of suspected fraud undermines the finding that voter fraud is not a problem in Florida sufficient to justify legislative action.

Third, the record establishes that the sponsors and support- ers of S.B. 90 repeatedly asserted that they were motivated by con- cerns over voter fraud. Contrary to the organizations’ argument on USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 20 of 79

20 Opinion of the Court 22-11143

appeal, the purpose of “promot[ing] election integrity” is by no means a “post-hoc litigation position.” Even one of the organiza- tions’ experts, Dr. Herron, stated that “the dominant theme artic- ulated by supporters of SB 90 was election integrity.” Another of the organizations’ experts, Dr. Burch, agreed. In her words, “[t]he proponents of this bill were consistent in their messaging about the need to proactively tighten Florida’s elections,” though she “ques- tion[ed] their sincerity.” Whether S.B. 90 was the best way to achieve that policy objective is not for us to decide.

The bill’s sponsors, legislative leaders, and the Governor all presented a consistent message about the need for election secu- rity. The sponsors, Representative Blaise Ingoglia and Senator Dennis Baxley, repeatedly stated that the purpose of S.B. 90 was to prevent voter fraud. In committee, Representative Ingoglia insisted that S.B. 90 would “keep our elections safe and secure” and “in- crease election security without suppressing anybody’s vote.” Sen- ator Baxley, although he did not “know of widespread complaints,” asked, “[D]o we have to wait for a debacle? Why can’t we take something that’s working well and put guardrails on it and keep it safe so it doesn’t have a debacle[?]” The Governor’s press release, issued when he signed S.B. 90, quoted the President of the Senate, the Speaker of the House, and Senator Baxley, all of whom echoed the message that the purpose of the bill was to proactively ensure election security and, in Senator Baxley’s words, “make Florida a place where it is easy to vote and very hard to cheat.” At a public event introducing the bill, Governor DeSantis reiterated the USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 21 of 79

22-11143 Opinion of the Court 21

message: “We need to make sure we stay ahead of the curve. We need to make sure our citizens have confidence in the elections.” And Director Matthews testified that the Division of Elections had received questions from legislators during that session about “[i]ssues of voter fraud . . . as it relates to a number of areas.”

The organizations offer selected record citations purporting to prove that S.B. 90’s “sponsors disclaimed that [S.B.] 90 and the Challenged Provisions were intended to address fraud or election integrity.” For example, they appear to reference a statement by Senator Baxley in committee that Florida “had an excellent, excel- lent conducted election” in 2020. But at the same time, Senator Baxley also made clear that keeping elections honest mattered:

And now we’re looking at each juncture and saying, what do we do to reassure for the future that we’ve put the guardrails on the highway so that no one runs off? We had a great journey. What can we do to improve that election security? And that’s what we’re after.

The other quotations that the organizations cite follow roughly the same pattern: the speaker acknowledges that the 2020 election ran smoothly and expresses a desire to ensure security going forward.

Context clarifies the statements that the organizations high- light in their briefs. To be sure, Senator Baxley stated that address- ing “vote-by-mail fraud in the last election” was “not the purpose of our bill.” But his statement does not imply that the legislation USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 22 of 79

22 Opinion of the Court 22-11143

was adopted for a reason other than promoting election integrity. Likewise, although Senator Baxley stated that he was “not trying to build a case on” “examples of fraudulent ballot gathering or changing of ballots” in the 2020 election, he also questioned whether, when third-party organizations gather ballots, “people [are] fully participating, or are . . . being used in a way that’s inap- propriate to participation.” According to the organizations, “Rep- resentative Ingoglia admitted that he saw no issues with election integrity.” In the statement referenced, Representative Ingoglia re- sponded to a question about whether the supervisors of elections had reported issues related to fraud or election integrity. Although Representative Ingoglia was unaware of problems other than ballot harvesting, he reiterated that he sought to “mak[e] sure we’re put- ting [up] the guardrails” and to ensure that “everybody treats, and uses, and deploys their drop boxes in the same way.” During the same floor debate, Representative Ingoglia also pointed out that fraud might be taking place even if it is not detected. The district court acknowledged—but dismissed—S.B. 90’s supporters’ “proffered justifications” for each of the challenged provisions. We now turn to those justifications and the evidence supporting their credibility, and we review the findings of fact for clear error. a. Drop-Box Provision The district court outlined three proffered justifications for the drop-box provision: “(1) without more restrictions, people may tamper with drop boxes”; “(2) the Supervisors were not using drop USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 23 of 79

22-11143 Opinion of the Court 23

boxes properly”; “and (3) ‘the provision was necessary to ensure the chain of custody of the ballot.’” But it rejected the possibility that these concerns reflected the supporters’ sincere motivations. First, the district court was skeptical of concerns about drop- box tampering because it determined that “[n]o evidence was pre- sented to the Legislature that drop box tampering actually occurs.” Senator Baxley stated that he had “never made the case that there’s box tampering.” But at the same time, he elaborated, “I’m not try- ing to present a case that there’s a problem. I’m[] presenting a case that we can prevent ever having a problem.” During the legislative process, Senator Baxley explained the basis for his concerns respecting unattended drop boxes. He stated on the Senate floor that “officials say you’d be surprised what all we find in these drop boxes and what gets dropped in there.” He likewise asserted on the Senate Floor that he “g[o]t a lot of input that says people throw all kind of stuff in these boxes” and that, even though he didn’t have a “name and place[,] . . . [this] is a reg- ular phenomenon that happens.” In a hearing, he referenced prob- lems with drop boxes “across the country.” Similarly, in commit- tee, he stated that “you don’t know what you don’t know because many of these [drop] boxes were actually in places that no one was providing security over them or observing what was going on there.” Senator Baxley’s statements evidence that items other than ballots can be, and have been, deposited in drop boxes, which sug- gests that ballots could be damaged or destroyed as a result. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 24 of 79

24 Opinion of the Court 22-11143

The record also establishes that concerns respecting un- manned drop boxes were valid and were expressed by persons other than S.B. 90’s sponsors. For example, Director Matthews tes- tified at trial that drop boxes had been vandalized in other states. The Lee County Supervisor of Elections testified that he discontin- ued the use of unmanned, overnight drop boxes in his jurisdiction because of security concerns. In particular, he became distressed when he found a drop box so full at the end of a weekend that “somebody could reach in and grab a ballot.” The Okaloosa County Supervisor of Elections stated before a legislative commit- tee that his office did not permit unmanned drop boxes because he was “concerned that [he] could not protect those ballots” and “if someone were to sabotage the box, . . . [he] would not be able to know which voters to contact to correct that.” And the chair of the House Public Integrity and Elections Committee, Representative Grall, later reminded the committee members of the concern— “overwhelmingly from a bipartisan standpoint”—regarding “the proliferation of drop boxes and the security of those drop boxes” expressed at the committee’s “workshop” held to review the 2020 election.

The district court stated that the second proffered justifica- tion for the drop-box provision—that “the Supervisors were not using drop boxes properly”—had “more merit.” After all, the 2020 election marked the first time that drop boxes were used statewide. See Ch. 2019-162, § 20, Laws of Fla. Director Matthews testified that county supervisors were not uniformly following the USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 25 of 79

22-11143 Opinion of the Court 25

Department’s guidance about drop boxes. She stated that she “kept getting questions through emails [and] phone calls” and that “[i]t was quite apparent that the Supervisors had differing interpreta- tion[s] of the drop box law and that they needed guidance.” For example, she explained that some supervisors, contrary to her De- partment’s guidance, had made drop boxes available outside early voting hours at locations other than the supervisors’ offices. A sur- vey of county supervisors also revealed that the times of day at which drop boxes were available to voters, as well as the supervi- sors’ approaches to monitoring and emptying drop boxes, varied considerably. And Representative Ingoglia expressed concern in committee over the lack of uniformity in drop-box policies.

According to the district court, the third justification for the drop-box provision—that “the provision was necessary to ensure the chain of custody of the ballot”—was “nonsensical” because “[m]ost [vote-by-mail] ballots are still deposited through mail- boxes.” But that reasoning is itself unsound. It is more than reason- able to secure one method of delivering a ballot, even if other methods are used more often. As the state officials argue, drop boxes may also be more obvious targets than regular mailboxes for anyone seeking to interfere with an election. Although mailboxes contain many types of mail, drop boxes should contain only ballots. And as Representative Ingoglia recognized, state legislators cannot control federal mailboxes. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 26 of 79

26 Opinion of the Court 22-11143

b. Solicitation Provision Next, the district court rejected the proffered justifications for the solicitation provision. It determined that “the justification appeared to be respect for privacy . . . [o]r perhaps to prohibit po- litical solicitation.” It expressed skepticism of the latter reason, be- cause “Florida law already bans solicitation.” But it ignored evi- dence in the record that existing restrictions were insufficient to maintain order at the polls. Director Matthews testified that the Department of State re- ceived “frequent” “complaints from voters about interference within the 150-foot nonsolicitation zone in past elections.” For ex- ample, the Supervisor of Elections for Miami-Dade County testi- fied that “there’s quite a lot of activity going on” at voting sites, including “performances,” “food trucks,” and “bullhorns”—such that the scene can become “quite chaotic at points.” Not all these disruptive activities would be covered by a narrow definition of so- licitation, such as one that prohibited only partisan electioneering.

The organizations assert that S.B. 90’s supporters had at times suggested that the solicitation provision was designed only to prevent campaigning. For example, Representative Ingoglia ex- plained that the goal was to prevent “campaigning on [the] line,” including activities such as candidates handing out water bottles to voters. And although Representative Ingoglia stated that “we’ve never said that any non-profit organization is trying to influence votes,” he also reiterated that “the intent . . . is to make sure that USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 27 of 79

22-11143 Opinion of the Court 27

nobody is trying to influence the vote while they are on the line.” (Emphasis added.) Handing a voter something of value, even wa- ter, could be a means of influence, he explained. In that context, a categorical bar on soliciting voters—not just solicitation by partisan electioneers—is consistent with the stated justification.

And protecting voter privacy is also a valid state interest. Ac- cording to Senator Baxley, S.B. 90’s supporters sought “to protect this sacred act [of voting], this is a private and individual thing.” The district court disagreed and described voting as an often “com- munitarian act,” especially for black voters. But the Supreme Court and this Court have acknowledged that states have an interest in protecting voters from unwanted interactions with third parties as they enter or exit the polling place. We have explained that if “[t]he State wants peace and order around its polling places, . . . we ac- cord significant value to that desire[,] for it preserves the integrity and dignity of the voting process and encourages people to come and to vote.” Citizens for Police Accountability Pol. Comm. v. Browning, 572 F.3d 1213, 1220 (11th Cir. 2009). And the Supreme Court has recognized that “there must be a substantial regulation of elections if they are to be fair and honest and if some sort of or- der, rather than chaos, is to accompany the democratic processes.” Buckley v. Am. Const. L. Found., Inc., 525 U.S. 182, 187 (1999) (quoting Storer v. Brown, 415 U.S. 724, 730 (1974)). We have even upheld a restriction on the solicitation of voters as they leave the polling place, “envision[ing] polling places awash with exit solici- tors” and voters “refrain[ing] from participating in the election USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 28 of 79

28 Opinion of the Court 22-11143

process merely to avoid the resulting commotion when leaving the polls.” Citizens for Police Accountability, 572 F.3d at 1220. A broad prohibition on soliciting voters serves the State’s interest in pre- serving order at polling places.

c. Registration-Delivery Provision Finally, the district court rejected the proffered justifications for the registration-delivery provision. According to the district court, members of the Legislature “falsely” claimed that the new restrictions on third-party voter-registration organizations were re- quired by a previous court order. (Citing Browning, 863 F. Supp. 2d at 1159.) The district court suggested that the justification for the provision “rested on th[at] false claim.” Even if some supporters misunderstood that court order, such a mistake hardly proves that their concerns were a pretext for discriminatory intent. The record reflects other motivations as well. For example, one supporter described the provision as a “good commonsense regulation which, by the way, is absolutely required by court rul- ing.” (Emphasis added.) The lobbyist for the Florida Supervisors of Elections testified that the registration-delivery provision was a leg- islative “priority” for his client. Third-party voter-registration or- ganizations, he testified, would deposit large numbers of registra- tion forms in populous counties, burdening the supervisors with the task of “separat[ing] out their files for them.” He also expressed these concerns to members of the House Public Integrity and Elec- tions Committee at a hearing. This information may have USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 29 of 79

22-11143 Opinion of the Court 29

influenced members of the Florida Legislature to enact the regis- tration-delivery provision.

There is also evidence of third-party voter-registration or- ganizations turning in voter-registration forms after the registra- tion deadline. Director Matthews testified that the Department of State had received complaints on “a fairly regular basis” of organi- zations “providing registrations late.” Supervisor White of Miami- Dade County testified that she was aware of instances in which reg- istration forms were delivered late and even times when “third- party voter registration organizations are alleged to have changed a voter’s registration information without their consent.” Supervi- sor Doyle of Lee County testified that “one third party . . . seems to constantly turn in late registration forms.” Three county super- visors—from Pasco, Pinellas, and St. Johns Counties—were even aware of voters who did not vote in an election because a third- party voter-registration organization either turned in the would-be voters’ registration forms late or failed to turn them in at all. For example, Supervisor Oakes of St. Johns County received twelve late applications for voters who, as a result, were not able to vote in the March 2020 election. The findings of fact related to the enactment-history Arling- ton Heights factor do not withstand clear-error review. The district court never stated that it found direct evidence of racial animus, so its finding of discriminatory intent “rel[ied] on circumstantial evi- dence.” It relied on its determination that the justifications put for- ward by S.B. 90’s proponents were not credible as evidence that USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 30 of 79

30 Opinion of the Court 22-11143

they were pretext for nefarious motives. But the record makes clear that the supporters’ justifications were credible. Contrary to the or- ganizations’ narrative, the Florida Legislature did not “pass[] a bill that addresses imaginary concerns or lacks a connection to its an- nounced purposes,” so as to “support[] an inference of discrimina- tion.” Instead, it passed a bill that supporters argued would safe- guard the integrity of elections against non-imaginary threats. The wisdom of the Legislature’s policy choices is not ours to judge. 3. Procedural Departures Any procedural departures in the legislative history, the dis- trict court determined, “show only that SB 90 was highly partisan.” Once again, “partisan motives are not the same as racial motives.” See Brnovich, 141 S. Ct. at 2349. It is unnecessary to address this factor because it did not contribute to the district court’s finding of intentional racial discrimination or disparate impact. 4. Contemporary Statements and Actions of Key Legislators The district court also considered the “contemporary state- ments and actions of key legislators.” It detailed multiple state- ments that it recognized yielded little support to a finding of inten- tional racial discrimination. It relied heavily on one statement by Senator Baxley, taken out of context, to find that this Arlington Heights factor favored a finding of discriminatory intent. The district court described multiple exchanges among leg- islators before deciding that they shed little light on the Legisla- ture’s intent. For example, it acknowledged that a text message USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 31 of 79

22-11143 Opinion of the Court 31

thread between two legislators—reproduced in the opinion— “do[es] not show a racially discriminatory intent” but instead “sug- gest[s] that the Legislature passed SB 90 with partisan purpose.” A connection between race and partisan voting patterns is not enough to transform evidence of partisan purpose into evidence of racially discriminatory intent. The district court also mentioned a discussion on the Senate floor, in which two senators purportedly invoked a “racial trope[]” by insinuating that people who do not vote fail to do so because they are “lazy.” But the district court de- termined that the exchange “does not tell this Court much about the Legislature’s motivations as a whole.” The district court focused heavily on Senator Baxley, the bill’s sponsor. It described certain positions he had taken before as “deeply troubling,” but it determined that the sponsor’s personal history was of “marginal relevance” to the Arlington Heights in- quiry. Instead, the district court highlighted a statement that Sena- tor Baxley made on the Senate floor. When asked whether he knew that restrictions in the bill would “have a disparate impact on black voters,” Senator Baxley responded that he had “a hard time hearing somebody even say that” and denied that S.B. 90 would “disenfran- chise[] anyone.” He continued, “Now to look at patterns of use and say, well, you may have to go about it a little different way. There’s a learning curve.” The district court relied heavily on this state- ment to infer that the contemporary statements and actions of key legislators weighed in favor of a finding of discriminatory intent. But, in context, the statement gives rise to no inference of USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 32 of 79

32 Opinion of the Court 22-11143

discriminatory intent. Senator Baxley specifically rejected any sug- gestion that the bill would have a disparate impact on black voters. Senator Baxley made other statements that counsel against inferring a nefarious purpose. For example, in a Rules Committee meeting, he responded to a question about whether S.B. 90 would be “helpful to reduce . . . Black voter turnout” with a resolute de- nial: I certainly hope not. That’s not the intention on my part; I can assure you. I’m very proud that we’ve opened up, during my tenure on working in public policy in this arena, that we have made more and more ways to access, if you will[,] participate. . . . I don’t buy the whole Jim Crow story. I’m sorry that that’s out there.

So, again, the district court relied heavily on a single statement by the sponsor that, in context, offers no evidence of discriminatory intent. And in any event, the explanatory value of an isolated state- ment would be limited. See United States v. O’Brien, 391 U.S. 367, 384 (1968) (“What motivates one legislator to make a speech about a statute is not necessarily what motivates scores of others to enact it . . . .”).

That the statement was made by the sponsor adds little to its significance. As we have explained, “a sponsor is only one vote” out of many. Greater Birmingham, 992 F.3d at 1324. And many votes favored S.B. 90. The bill passed by a margin of 77 to 40 votes in the House and 23 to 17 votes in the Senate. See SB 90, FLA. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 33 of 79

22-11143 Opinion of the Court 33

SENATE, https://www.flsenate.gov/Session/Bill/2021/90. One senator does not speak for all the supporters of S.B. 90.

5. Impact of the Challenged Law

Next, the district court considered the likely impact of the challenged provisions. It determined that all three provisions would have a disparate impact on black voters. But again, this find- ing does not withstand even our deferential review.

a. Drop-Box Provision

The district court determined that the drop-box provision would have a disparate impact on black voters. It “conclude[d] that SB 90 will burden voters who use drop boxes” and that “these vot- ers are disproportionately likely to be Black.” Although it acknowl- edged that “race’s effect on drop-box use appears less pronounced than the effect of party on drop-box usage,” the district court found that race still “appears to have a meaningful and remarkably con- sistent connection with drop-box usage.” And it determined that black voters “use [drop boxes] in precisely the ways SB 90 prohib- its”—that is, outside of early voting hours.

The district court clearly erred. The record does not support a finding of disparate impact. The district court based its finding that black voters are more likely to use drop boxes on a statistically insignificant correlation between the number of black vote-by-mail voters and drop-box usage across counties; inconsistent drop-box USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 34 of 79

34 Opinion of the Court 22-11143

usage records from five counties; a survey showing a 1.3 percent- age point difference in the rate at which black and white voters use drop boxes; and the lower drop-box-to-population ratio in “large, racially heterogenous counties.”

First, the district court relied on a positive but statistically insignificant correlation between the number of black vote-by-mail voters and drop-box usage across counties. Even the organizations’ expert, Dr. Herron, admitted that this correlation was only “sug- gestive of a relationship between race and drop box use but not definitive on this point.” A statistically insignificant relationship is not statistically distinguishable from chance, and it is impossible to say whether any relationship is causal based on a mere correlation. And the analysis included data from only 46 of 67 counties because not all counties produced such information. Perhaps this evidence could, as the organizations suggest, bolster other consistent evi- dence, but the other evidence in the record is as weak.

Second, Dr. Herron reviewed drop-box usage by race in the five out of sixty-seven counties that provided such information. He separately analyzed one set of three counties and another set of two smaller counties. The district court discussed the results of only the former analysis.

Dr. Herron performed logistic regression to determine the relationship between race and drop-box usage in only three coun- ties. Based on his analysis, the district court concluded that black vote-by-mail voters in those counties “had, on average, 14 [percent] USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 35 of 79

22-11143 Opinion of the Court 35

greater odds” than white vote-by-mail voters “of voting by drop box in the general election” and “48 [percent] and 25 [percent] greater odds . . . of voting via drop box” in “the presidential prefer- ence primary and the primary elections,” respectively. But the data from the three counties on which he relied—Sarasota, Santa Rosa, and Columbia—cannot support a meaningful, representative anal- ysis.

The data on which Dr. Herron relied came from a small and unrepresentative sample, as even the district court acknowledged. About 14 percent of voters in Columbia County are black. But black voters make up only 3.24 and 4.45 percent of registered vot- ers in Sarasota and Santa Rosa Counties, respectively. Based on the district court’s statistics, black voters make up roughly 13 percent of the registered voters in Florida. Dr. Herron’s analysis included data from eight elections: the 2020 primary and general elections in three counties and the 2020 presidential preference primaries in two counties. In three of those eight elections, there were fewer than one hundred black drop-box users, including one election in which there were only three black drop-box users. Dr. Herron con- tended that he was even “more confident” in his conclusion be- cause “he detected a correlation between being Black and using a drop box despite the unrepresentatively small Black populations in the counties for which he had data.” Although it is true that a sta- tistically significant relationship can sometimes be detected even when analyzing a small dataset, any relationship found in a small, unrepresentative sample is, by definition, not reliably descriptive of USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 36 of 79

36 Opinion of the Court 22-11143

the population. Barbara Illowsky & Susan Dean, Introductory Sta- tistics 20 (2018) (“[S]amples that are not representative of the pop- ulation give results that are inaccurate and not valid.”).

Worse still, the statistical relationship that Dr. Herron found is predominantly driven by data from a single county. The statisti- cal analysis pools the data from the three counties across general, primary, and presidential preference elections. As a result, data from Sarasota County—which has nearly twice as many registered voters as Santa Rosa and Columbia Counties combined—largely drove the findings. White voters used drop boxes at higher rates than black voters in all but one of the elections analyzed in Santa Rosa and Columbia Counties. Among the three counties, only in Sarasota County did black voters consistently use drop boxes at higher rates. We have no way to know which county, if any of the three, is representative of voter behavior across the State. But we do know that, among the three counties, black voters constitute the smallest percentage of registered voters in Sarasota County. No statistically valid conclusions about a state in which black voters make up approximately thirteen percent of registered voters can be drawn from an analysis driven predominantly by data from a single county in which black voters make up less than four percent of reg- istered voters.

Our reservations respecting the reliance on Dr. Herron’s analysis extend beyond the quality of the underlying data. Dr. Her- ron described the importance of positive coefficients—that is, a USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 37 of 79

22-11143 Opinion of the Court 37

positive relationship between a voter’s race and likelihood to use a drop box—over the “precise magnitudes of the coefficient esti- mates” in his logistic regression, “because these estimates are not easily interpreted.” Nonetheless, the district court itself attempted the very calculation against which Dr. Herron had advised. The district court, not Dr. Herron, estimated that a black vote-by-mail voter in those three counties had, “on average, 14 [percent] greater odds than a White [vote-by-mail] voter of voting by drop box” in a general election and even higher odds in primaries. For reference, the raw difference in the rates at which black and white voters used drop boxes in Sarasota County averaged roughly 11 percentage points, but the difference in the one Columbia County election in which black voters used drop boxes at a higher rate than white vot- ers was less than one percentage point. We question the district court’s decision to rely on its own calculations as meaningful evi- dence, despite Dr. Herron’s cautionary statement.

Dr. Herron separately reviewed data from Madison and Franklin Counties, which he acknowledged “are relatively small in terms of [vote-by-mail] counts.” Data was available in those coun- ties from only the 2020 general election. And there were only sev- enty-eight and fifty black drop-box voters in Franklin and Madison Counties, respectively. In Franklin County, 37.5 percent of black vote-by-mail voters used drop boxes compared with 33.61 percent of white voters—a difference of 3.89 percentage points. In Madison County, 6.24 percent of black vote-by-mail voters used drop boxes compared with 4.03 percent of white voters—a difference of 2.21 USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 38 of 79

38 Opinion of the Court 22-11143

percentage points. The difference in race-based drop-box usage in these two counties was neither of large magnitude nor statistically significant.

Third, the evidence of statewide drop-box use clarifies little. Dr. Burch, an expert for the organizations, cited a survey by the 2020 Cooperative Election Study. It found that “an estimated 29.6 percent [of black respondents in Florida who voted by mail] used [drop] boxes compared with 28.3 percent of [white respondents in Florida].” A difference of only 1.3 percentage points is not substan- tial.

Fourth, the lower drop-box-to-population ratio in “large, ra- cially heterogenous counties” proves little. Dr. Herron explained that “the more populous counties in Florida have more locations at which drop boxes may be situated per SB 90,” but “the rate at which these counties have more locations does not keep up with the sizes of the counties’ registered voter pools.” Even assuming the accuracy of this statement, we question its significance to the present inquiry. When there is a low drop-box-to-population ratio in populous counties, Dr. Herron suggested, voters may face “greater congestion” around drop boxes. For this reason, a de- crease in the number of drop boxes available might have a dispar- ate impact on black voters, who tend to reside in more populous areas. But voters in large population centers are not the only voters who would be affected by a decrease in the number of drop-box locations. For example, where there are few drop boxes per square USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 39 of 79

22-11143 Opinion of the Court 39

mile in rural areas, voters may have to drive considerable distances to deposit their ballots. So, the drop-box-to-population ratio in pop- ulous counties does not necessarily establish that the restrictions on drop-box placement imposed by S.B. 90 will have a disparate impact on black voters.

The finding that black voters are more likely to use drop boxes outside of early voting hours rests on equally flimsy evi- dence. Dr. Smith, an expert for the organizations, relied on data from the two out of sixty-seven counties that tracked when voters deposited ballots as well as voters’ identifying information. In Co- lumbia County, he found that 52.4 percent of ballots deposited by black voters and 50.2 percent of ballots deposited by white voters were deposited outside of early voting days. In Manatee County, he found that 13.5 percent of ballots deposited by black voters and 11.4 percent of ballots deposited by white voters were deposited outside of business hours. Even at face value, the disparities are so small—just over two percentage points in both counties—that they cannot support a finding of disparate impact. But the analyses are also flawed in other ways.

In Columbia County, Dr. Smith only measured rates of drop-box use outside of early voting days, not necessarily outside of business hours. Dr. Smith’s analysis establishes that black voters in Columbia County were slightly more likely than white voters to deposit their ballots on days that fell outside the early-voting pe- riod. But it does not explain why black voters would USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 40 of 79

40 Opinion of the Court 22-11143

disproportionately struggle to use drop boxes during the early-vot- ing period, if drop boxes were only available during that time. The early-voting period includes all seven days of the week, see FLA. STAT. § 101.657(d), so any racial disparities in the flexibility of work- ing hours and access to transportation do not impact black voters’ ability to deposit ballots in drop boxes during the early-voting pe- riod. A finding that black voters in one county are slightly more likely to deposit ballots in drop boxes outside of early-voting days is not evidence that the drop-box provision will have a disparate impact on black voters in Florida.

In Manatee County, Dr. Smith measured drop-box use out- side of business hours, but he employed a flawed methodology. The Manatee Supervisor of Elections “recorded (with a timestamp) each [vote-by-mail] ballot it received.” Dr. Smith stated that the su- pervisor’s “staff presumably collected and processed after-hour [vote-by-mail] ballots deposited in drop boxes late at night or the ensuing morning.” So, he classified a ballot as having been depos- ited after business hours if it was “processed before 10[:00] [a.m.]” But this methodology was far from precise. The Manatee County Supervisor of Election’s Office was open from “7:00 [a.m.] to 7:00 [p.m.] during the early voting period.” A ballot processed by elec- tion officials at 10:00 a.m. may have been deposited outside of those hours or it may have been deposited by a voter on his way to work between 7:00 and 10:00 a.m. As a result, the analysis likely overestimates the number of ballots deposited outside of business hours. It also reduces our confidence in Dr. Smith’s conclusion that USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 41 of 79

22-11143 Opinion of the Court 41

black voters are more likely than white voters to deposit ballots outside of business hours. It is impossible to tell whether any er- ror—ballots mislabeled as deposited outside of business hours—is evenly distributed among black and white voters or whether the results have been skewed. As a result, there is simply no reliable evidence from which to infer that black voters are more likely than white voters to deposit ballots in drop boxes outside of business hours.

Other evidence related to drop-box use also cannot support a finding of disparate impact. Evidence related to drop-box use out- side the early-voting period or outside of business hours—absent information about the race of the voters who deposited their bal- lots at those times—does not prove that the drop-box provision will have a disparate impact on black voters. Likewise, testimony that some counties will offer fewer drop boxes in the future cannot establish disparate impact without reliable evidence of voting pat- terns as to race. Because there is no reliable evidence in the record that black voters are meaningfully more likely to use drop boxes or more likely to use drop boxes outside of business hours, the finding that the drop-box provision would have a disparate impact on black voters was clear error.

b. Solicitation Provision

Next, the district court found that the solicitation provision would have a disparate impact on black voters. “[I]n practical terms, the solicitation definition discourages groups who give food, USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 42 of 79

42 Opinion of the Court 22-11143

water, and other forms of encouragement to voters waiting in long lines,” it determined. So, the district court measured disparate im- pact by assessing racial disparities in the time voters wait in line at the polls. It found that “minority voters in Florida are, on average, more likely to wait in long lines to vote.”

Again, the evidence on which the district court relied was flawed. One of the organizations’ experts, Dr. Smith, reported that, during the 2020 early-voting period in Miami-Dade County, “24.8 [percent] of Black voters had wait times of 30 minutes or longer” and “15.2 [percent] of White voters had wait times of 30 minutes or longer.” Not only was this report based on an analysis of a single county, but it also derives from an incomplete dataset, even within that county. According to his expert report, Dr. Smith obtained seven screenshots taken across five days of a county website show- ing the wait times at all thirty-three early voting locations. Dr. Smith examined the relationship between whether the polling place had a long “wait time at some point during the day” accord- ing to the snapshots and the demographics of the voters who cast their ballots at each location on those days. The district court relied on this data—specifically, an amended table admitted at trial—as evidence that black voters were more likely to wait in long lines at the polls.

But this evidence is fatally imprecise, as wait times at polling places can vary dramatically throughout the day. The length of a line at a given polling location at 3:00 p.m., for example, tells us USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 43 of 79

22-11143 Opinion of the Court 43

nothing about the line voters face at 6:00 p.m., when many people stop to vote on the way home from work. We do not know whether black voters are more likely to vote at those polling places when the lines are long or short. A slightly more precise version of Dr. Smith’s analysis identified voters who cast their ballots within an hour of each snapshot, but that data is even more limited—and it appears to evidence that Hispanic voters suffered the brunt of the long wait times. Other evidence suggested that wait times in Mi- ami-Dade County were “modest” after the first day of early voting and that “no voter wait[ed] more than one hour to vote” on Elec- tion Day.

In two other counties, the district court acknowledged that Dr. Smith’s data was too limited to serve as the basis for meaning- ful conclusions. In Orange County, data derived from a single snap- shot evidenced that white voters faced the longest wait times, though the finding was to some degree driven by a single polling place where the line was unusually long. An incomplete set of screenshots in Lee County showed that black voters were more likely than white voters to wait for less than 15 or more than 30 minutes, whereas white voters were more likely to wait between 15 and 30 minutes. But the Lee County analysis was based on “daily screenshots,” most of which were “taken around mid-day” and some of which were missing certain polling locations or specific timestamps. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 44 of 79

44 Opinion of the Court 22-11143

Finally, in his supplemental report, Dr. Smith relied on news media reports to assess wait times in Lee, Hillsborough, and Mi- ami-Dade Counties. These media reports are not reliable evidence of disparate impact. Dr. Smith appears to have offered no addi- tional empirical evidence that the long lines in these counties had a disparate impact on black voters.

In sum, Dr. Smith’s analysis was based on an extremely lim- ited—and not necessarily representative—dataset. And even if his analysis were methodologically reliable, it yields mixed evidence about whether black voters faced longer wait times than white vot- ers at the polls.

Other evidence also cannot sustain the district court’s find- ing. Two of the organizations’ other experts, Dr. Burch and Dr. Herron, also analyzed voter wait times based on statewide survey data. Dr. Burch relied on data from the 2020 Cooperative Election Study. She testified that 5.3 percent of black voters as compared to 4 percent of white voters reported waiting in “long lines”—more than an hour long—in 2020. But a difference of 1.3 percentage points is hardly meaningful. Dr. Herron relied on data from the Survey of the Performance of American Elections, which evi- denced that white voters in the 2020 and 2016 elections—but not the 2012 election—were more likely to report waiting in line for more than 30 minutes compared with black voters. Again, the evi- dence was mixed, and the identified racial disparities were small. The district court also cited Dr. Smith’s testimony respecting the USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 45 of 79

22-11143 Opinion of the Court 45

academic literature on voting lines, particularly a study of the 2012 election, for the proposition that minority voters face “dispropor- tionately long lines” in Florida. That study measured the race only of voters who cast ballots after early voting lines were cut off at 7:00 p.m., not racial disparities in wait times overall, and it found that the racial disparity observed in 2012 effectively “did not appear in 2016.” Also, the 2012 election was anomalous because the early- voting period had been reduced, and it has since been restored.

And even if the evidence established that black voters were more likely to wait in lines at the polls, that finding would not alone support a conclusion that the solicitation provision has a disparate impact on black voters. The district court determined that “the so- licitation definition will have a disparate impact on minority voters because minority voters are disproportionately likely to wait in line to vote, and because the provision discourages third parties from helping those waiting to vote.” The district court assumed, without making any factual findings supporting the assumption, that by re- stricting the ability of third parties to hand out water bottles and snacks, the solicitation provision makes it harder for voters waiting in line to cast their ballots. The organizations make a similar as- sumption, supported at best by evidence that “excessive wait times cause voters to leave without voting”—not evidence that a bottle of water will convince them to stay in line. In sum, the finding that the solicitation provision will have a disparate impact on black vot- ers was clear error. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 46 of 79

46 Opinion of the Court 22-11143

c. Registration-Delivery Provision

Finally, the district court found that the registration-delivery provision would have a disparate impact on black voters. The pro- vision, it reasoned, “impos[ed] additional costs on [third-party voter-registration organizations], thus limiting the number of vot- ers each [organization] can reach.” The district court cited post- 2012 voter registration data that evidenced that “15.37 [percent] of Black voters” and 2.79 percent of white voters registered using third-party voter-registration organizations. A similar, but smaller, disparity exists across party lines: “10.48 [percent] of Democrats . . . and only 3.9 [percent] of Republicans registered using [third-party voter-registration organizations].” And evidence in the record sug- gested that the registration-delivery provision imposed compliance costs on the third-party organizations. The finding that the regis- tration-delivery provision will have a disparate impact on black voters is not clearly erroneous. But absent “a clear pattern, unex- plainable on grounds other than race,” a finding of “discriminatory impact alone is not determinative” of whether a provision violates the Fourteenth or Fifteenth Amendment. Greater Birmingham, 992 F.3d at 1322 (quoting Arlington Heights, 429 U.S. at 266) (alter- ation adopted).

6 & 7. The Foreseeability of the Disparate Impact & Legislators’ Knowledge of that Impact

The district court determined not only that S.B. 90 would have a disparate impact but also that this impact was foreseeable USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 47 of 79

22-11143 Opinion of the Court 47

to—and foreseen by—the Legislature. Because we hold that the finding that the drop-box and solicitation provisions will have a dis- parate impact on black voters is clear error, we are skeptical that the Legislature could have foreseen a disparate impact. In any event, the evidence of foreseeability on which the district court re- lied was deficient.

The district court found that the Legislature enjoyed access to voter statistics supplied by the Division of Elections. It implied that those statistics would reveal that the challenged provisions would have a discriminatory impact on black voters. But Director Matthews’s testimony confirmed that the Legislature received raw data, not easy-to-read summaries. Even the district court recog- nized that “[p]erhaps the raw data itself is not enough.”

Even so, the district court found that the Florida legislators could—and did—foresee that the drop-box provision would have a disparate impact on black voters. “Director Matthews testified that the Legislature wanted to know ‘who uses drop boxes,’” it ex- plained. Although Director Matthews also testified that she be- lieved that the Legislature “just wanted to know how many people were using the drop boxes,” the district court found—in a trial con- ducted over the video-conferencing software Zoom—that her “face and body language” suggested that she “recognized her slip in testifying truthfully” that the Legislature was really interested in demographics. The district court inferred not only that the Legis- lature sought information on the demographics of drop-box users, USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 48 of 79

48 Opinion of the Court 22-11143

but also that S.B. 90’s regulation of drop-boxes was infused with discriminatory intent.

The district court made multiple inferential leaps, some of which were entirely unfounded. First, it inferred from Director Matthews’s “face and body language,” over Zoom, that the witness was lying. We ordinarily defer to the factfinder’s credibility assess- ment of witnesses, so we accept that determination. Crystal Ent. & Filmworks, Inc. v. Jurado, 643 F.3d 1313, 1320 (11th Cir. 2011). Next, it inferred that, in fact, the Legislature sought information about the demographics of drop-box users. Finally, it inferred that the legislators either sought to craft legislation that would have a disparate impact on black voters or were made aware that S.B. 90 would have such an impact. But this interpretation of—and extrap- olation from—Director Matthews’s testimony contradicts her pre- vious, unambiguous testimony that she did not “recall . . . being asked” “to break down the data by racial demographics of voters and the modality chosen to cast a ballot.” So, even if we accept the assessment of Director Matthews’s credibility, the record does not support the chain of inferences and finding that the district court drew from her testimony.

The finding that “the Legislature” sought to uncover “who uses drop boxes” begs an important question: how can a multi- member body ask a question? Based on Director Matthews’s testi- mony, an unspecified number of individual legislators requested information by phone, by email, at hearings, and the like. The USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 49 of 79

22-11143 Opinion of the Court 49

district court made no finding of fact respecting how many legisla- tors might have made such an inquiry. Although the contemporary statements of key legislators are relevant to an Arlington Heights analysis, a statement or inquiry by a single legislator would consti- tute little evidence of discriminatory intent on the part of the legis- lature. Cf. Greater Birmingham, 992 F.3d at 1324 (“It is also ques- tionable whether the sponsor speaks for all legislators.”). Thus— even if we were to accept the tenuous inferential chain outlined above—the evidence in the record does not appear to substantially support the major inference that “the Legislature” sought infor- mation about the demographics of drop-box users.

It is also unclear whether the legislators could have obtained this demographic information even if they wanted to do so. The district court explained that “the Division does not keep infor- mation on drop boxes, and thus presumably did not provide any information to the Legislature.” But individual supervisors of elec- tions have access to such information, and the district court found that “[t]he Legislature also asked individual supervisors of elections for information.” Based on these facts—but no ascertainable evi- dence that any specific legislators, except perhaps Senator Baxley, requested this information of the supervisors, much less that most did—the district court determined that “the Legislature likely had the same drop box data before it that is now before this Court.” The state officials correctly contend that there is also no evidence that the legislators were “aware of, let alone motivated by, the so- phisticated math [respecting drop-box use] the district court found USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 50 of 79

50 Opinion of the Court 22-11143

persuasive.” Finally, even if the legislators had access to all the data in the record respecting drop-box usage, it would not prove the Legislature knew that the drop-box provision would have disparate impact on black voters. As we have already explained, the data do not support a finding that the drop-box provision will have a dis- parate impact.

The district court likewise determined that Florida legisla- tors could—and did—foresee that the registration-delivery provi- sion would have a disparate impact on black voters. It found that “the Legislature asked for and received demographic information about [third-party voter-registration organization] use[, and] . . . that the Legislature knew that the registration[-delivery] provision would have a disparate impact on minority voters.” This finding appears to be based on Senator Farmer’s testimony that “[w]e were in possession of statistical evidence that showed that voter registra- tion groups registered about 10 percent of Black voters . . . but only 1 percent of White voters.” Senator Farmer testified to this fact without explaining where the statistics came from or how widely they were distributed among the legislators. The district court in- ferred that Senator Farmer must have obtained the information from the Division of Elections because it was based on nonpublic data, and it further inferred that “the Legislature asked for” this in- formation. The district court also highlighted Director Matthews’s testimony that members of the Legislature “wanted to know about third-party voter registration organizations.” But that statement is even more ambiguous than the statement that members of the USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 51 of 79

22-11143 Opinion of the Court 51

Legislature wanted to know “who uses the drop boxes.” It could mean that members of the Legislature sought any type of infor- mation about these organizations. Because it is unclear how many legislators even had access to or considered the information, it can- not support a finding of discriminatory intent.

The district court even determined that legislators could— and did—foresee that the solicitation provision would have a dis- parate impact on black voters. But the record does not make clear whether any information respecting the solicitation provision’s im- pact was before the Legislature, other than the statements of the bill’s opponents. The organizations maintain that these statements were enough to bridge what the state officials characterize as a “logical leap[]” between voting-line length and the alleged dispar- ate impact of the solicitation provision. We disagree.

The district court and the organizations emphasize that the statements of S.B. 90’s opponents should have put legislators on notice that all the challenged provisions would have a disparate im- pact on black voters. But the concerns expressed by political oppo- nents during the legislative process are not reliable evidence of leg- islative intent. Cf. Ernst & Ernst v. Hochfelder, 425 U.S. 185, 203 n.24 (1976) (explaining that warnings of the potentially vast impact of a bill by “legislative opponents[—]who [i]n their zeal to defeat a bill . . . understandably tend to overstate its reach”—should be “en- titled to little weight” (internal quotation marks omitted)). Several lobbying organizations also distributed studies explaining S.B. 90’s USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 52 of 79

52 Opinion of the Court 22-11143

potential impact “to the Legislature,” but the district court deter- mined that “[i]t is unclear . . . whether these letters were ever con- sidered.”

In sum, there is no substantial evidence that the Legislature foresaw that the challenged provisions would have a disparate im- pact. The finding to the contrary was clear error.

8. The Availability of Less Discriminatory Alternatives

Finally, the district court found that “less discriminatory al- ternatives” to the challenged provisions “not only were available but were presented to and rejected by the Legislature.” To be sure, as the organizations point out, various amendments to S.B. 90 were offered. The organizations contend that some of these amend- ments “would have reduced the Challenged Provisions’ racially dis- parate impact.” But the fact that “the [Florida] [L]egislature did not include the alternative option[s] that Plaintiffs would have pre- ferred” is not evidence of discriminatory intent. Greater Birming- ham, 992 F.3d at 1327. The legislative branch is not hamstrung by judicial review to adopt any amendment that a bill’s opponents claim would improve it.

In fact, the record establishes that S.B. 90’s proponents were receptive to input during the legislative process. The lobbyist for the Supervisors of Elections—who did not formally support or op- pose the bill as a whole—“testified that 75 [percent] to 80 [percent] of the alterations to SB 90 during its trip through the Florida USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 53 of 79

22-11143 Opinion of the Court 53

Legislature were based on his recommendations.” Representative Tracie Davis, a Democrat, acknowledged in an Appropriations Committee meeting that “the sponsor of this bill has been working with all the stakeholders,” that the “bill has changed[,] . . . and [that] it’s going into a decent direction”—though she still ques- tioned its content. Likewise, Democratic Representative Ben Dia- mond expressed his appreciation for some of the changes made, although he did not support the legislation overall. And multiple changes were made to S.B. 90 that loosened the restrictions that it would have otherwise imposed. For example, earlier versions “pro- posed a complete elimination of drop boxes” or required “a voter to provide [his or her] driver’s license number [or an attestation] when dropping off a ballot at a drop box.” Plainly, the Legislature adopted some alternatives that were more palatable to the bill’s op- ponents. It did not accept all of them, nor was it required to do so. See Greater Birmingham, 992 F.3d at 1327.

Finally, the district court failed to identify viable alternatives to the challenged provisions that would have achieved the same objectives, which might have served as evidence of discriminatory intent under Arlington Heights. See Greater Birmingham, 992 F.3d at 1327. As for the drop-box provision, the district court pointed to a failed amendment that would have permitted video surveillance of 24-hour drop boxes. But unless the live video feed is monitored and security personnel are nearby, a video record of vandalism would not serve the purpose of preventing the destruction of the ballots inside the box or provide the other benefits of a live person USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 54 of 79

54 Opinion of the Court 22-11143

to assist voters at the drop box. With respect to the solicitation pro- vision, the district court referenced proposed amendments “that would have allowed civic organizations to hand out food and wa- ter.” But if, as Senator Baxley suggested, the Legislature sought to prevent a captive audience of voters in line from being approached by third parties in the interest of privacy, this alternative would not achieve the goal. As for the registration-delivery provision, the dis- trict court suggested that the Legislature could simply have “do[ne] nothing.” It reasoned that there was no legitimate justification for the provision, so no alternative was necessary. But, as described above, legitimate motivations did exist. As a result, the finding that the Legislature rejected less discriminatory alternatives to the three provisions was clear error.

***

Weighing this evidence, the district court found that the Legislature intended to target black voters. The motivation, it de- termined, was to “secure an electoral advantage for the Republican Party.” In particular, the district court found that the drop-box pro- vision targeted black voters because it “effectively bans drop-box use at the specific times and the specific days that Black voters . . . are most likely to use them.” Likewise, it found that the solicitation and registration-delivery provisions targeted black voters because “White Democrats do not wait in long lines, nor do they use [third- party voter-registration organizations] to register.” USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 55 of 79

22-11143 Opinion of the Court 55

As we have explained, the district court’s finding based on this evidence does not withstand examination. The district court relied on fatally flawed statistical analyses, out-of-context state- ments by individual legislators, and legal premises that do not fol- low our precedents. The organizations contend that “divorc[ing] individual threads of evidence from the larger ‘calculus-of-voting’ framework” unfairly deconstructs their argument. On the con- trary, examining the record reveals that the finding of intentional discrimination rests on hardly any evidence.

The organizations bore the burden of proving both discrim- inatory impact and discriminatory intent. Greater Birmingham, 992 F.3d at 1321. Because the record does not contain evidence suf- ficient to sustain a finding of either disparate impact or discrimina- tory intent for the solicitation provision and drop-box provision, neither provision violates the Constitution.

The registration-delivery provision presents a closer ques- tion. Sufficient evidence exists in the record to uphold the finding, on clear-error review, that the provision will have a disparate im- pact on black voters. But a finding of disparate impact alone cannot support a finding that the registration-delivery provision violates the Constitution. Id. at 1322. Other evidence, at most, establishes that some legislators knew that black voters are more likely than white voters to register to vote using third-party voter-registration organizations. That evidence does not establish that the Legislature acted with discriminatory intent. So, the organizations’ USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 56 of 79

56 Opinion of the Court 22-11143

constitutional claims against the registration-delivery provision must also fail.

And even if the organizations established a prima facie case against the registration-delivery provision, the burden would shift to the state officials to “demonstrate that the law would have been enacted without this [racial discrimination] factor.” Id. at 1321 (quoting Hunter, 471 U.S. at 228). Undisputed evidence in the rec- ord establishes that a valid justification for the registration-delivery provision existed. After all, the lobbyist for the Supervisors of Elec- tions testified that the provision was a “priority” because organiza- tions would deposit large numbers of registration forms in popu- lous counties, and Director Matthews testified that election offi- cials received complaints on “a fairly regular basis” about organiza- tions “providing registrations late.” The only less discriminatory al- ternative mentioned by the district court was “doing nothing.” As a result, the registration-delivery provision does not violate the Constitution.

In sum, based on this record—and even in the light of the deferential standard of review we must apply to the findings of fact—the district court clearly erred in finding that the challenged provisions were enacted with discriminatory intent in violation of the Fourteenth and Fifteenth Amendments. As a result, we reverse the decision of the district court in part. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 57 of 79

22-11143 Opinion of the Court 57

B. The Drop-Box, Solicitation, and Registration-Delivery Provi- sions Do Not Violate the Voting Rights Act.

The district court determined that the challenged provisions violated section 2 of the Voting Rights Act. It concluded that a find- ing of discriminatory impact was unnecessary to establish a section 2 violation. Assuming instead that a finding of discriminatory intent would suffice, the district court found that the challenged provi- sions violated the Voting Rights Act because the Legislature en- acted them with discriminatory intent. It declined to reach the question whether the challenged provisions, considering the total- ity of the circumstances, failed a discriminatory-results test.

The district court erred. A finding of discriminatory impact is necessary and sufficient to establish a section 2 violation. Section 2’s “results test requires an inquiry into the totality of the circum- stances.” Greater Birmingham, 992 F.3d at 1329 (quoting Chisom v. Roemer, 501 U.S. 380, 394 (1991)). A violation exists if “members of a protected class ‘have less opportunity than other members of the electorate to participate in the political process and to elect rep- resentatives of their choice.’” Chisom, 501 U.S. at 388 (quoting lan- guage now codified at 52 U.S.C. § 10301(b)). A finding of discrimi- natory intent alone will not suffice.

Our precedents respecting the proper standard are admit- tedly inconsistent. In Johnson v. DeSoto County Board of Commis- sioners, we held that the “statutory language expressly requires a showing of discriminatory results, and it admits of no exception for USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 58 of 79

58 Opinion of the Court 22-11143

situations in which there is discriminatory intent but no discrimi- natory results.” 72 F.3d 1556, 1563 (11th Cir. 1996). The next year, without purporting to overrule Johnson, we held that a “statutory claim under [s]ection 2 may be established by proof that the chal- lenged methods of election either have a discriminatory purpose or effect.” Askew v. City of Rome, 127 F.3d 1355, 1373 (11th Cir. 1997). Later, citing Johnson, we again held that “discriminatory in- tent alone, in the absence of a showing of discriminatory effect, is insufficient to establish a violation of [section] 2.” Brooks v. Miller, 158 F.3d 1230, 1237 (11th Cir. 1998). But in 2004, we held that a finding of either discriminatory intent or discriminatory impact would suffice. Osburn v. Cox, 369 F.3d 1283, 1289 (11th Cir. 2004). We cited both Brooks—in which we had actually reached the op- posite conclusion—and a portion of a 1994 en banc decision that the Johnson Court had rejected as “dictum,” “joined only by two . . . members of this Court,” and “inconsistent with [an] express contrary holding by the Supreme Court.” Johnson, 72 F.3d at 1564 n.8; see Nipper v. Smith, 39 F.3d 1494, 1520 (11th Cir. 1994) (en banc). We never overruled Johnson, our earliest binding prece- dent, so we are obliged by stare decisis to follow it. “Under our prior panel precedent rule, we are bound to follow a prior panel’s holding unless and until it is overruled or undermined to the point of abrogation by an opinion of the Supreme Court or of this Court sitting en banc.” United States v. Gillis, 938 F.3d 1181, 1198 (11th Cir. 2019). Neither court has overruled or abrogated Johnson. And USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 59 of 79

22-11143 Opinion of the Court 59

“when we have conflicting precedents, we follow our oldest prec- edent.” CSX Transp., Inc. v. Gen. Mills, Inc., 846 F.3d 1333, 1338 (11th Cir. 2017) (citation omitted) (alterations adopted).

Johnson also satisfies Supreme Court precedent. The Su- preme Court has stated that under section 2 as amended, “[t]he ‘right’ question . . . is whether ‘as a result of the challenged practice or structure plaintiffs do not have an equal opportunity to partici- pate in the political processes and to elect candidates of their choice.’” Gingles, 478 U.S. at 44 (citations omitted); see also Voino- vich v. Quilter, 507 U.S. 146, 157 (1993) (“We hold only that, under [section] 2 of the Voting Rights Act . . . plaintiffs can prevail on a dilution claim only if they show that . . . the State’s apportionment scheme has the effect of diminishing or abridging the voting strength of the protected class.” (emphasis added)). So, a finding of discriminatory impact is necessary to establish a violation of section 2 of the Voting Rights Act. Our decisions in Askew and Osburn are not—and never were—good law to the extent that they purported to hold otherwise.

None of the challenged provisions violates section 2 of the Voting Rights Act. For the reasons already explained, the record does not support a finding that the drop-box provision or solicita- tion provision will have a disparate impact on black voters. And although there is some evidence that the registration-delivery pro- vision will have a disparate impact, it is not enough to meet section 2’s high standard. The test would require a finding that, because of USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 60 of 79

60 Opinion of the Court 22-11143

the registration-delivery provision, “political processes leading to . . . election in [Florida] . . . are not equally open to” black voters “in that [they] have less opportunity than other members of the electorate to participate in the political process and to elect repre- sentatives of their choice.” Wis. Legislature v. Wis. Elections Comm’n, 142 S. Ct. 1245, 1248 (2022) (quoting 52 U.S.C. § 10301(b)). The record does not come close to meeting that stand- ard. So, we reverse the decision of the district court in relevant part.

C. The District Court Erred by Subjecting Florida to Preclearance Under Section 3(c) of the Voting Rights Act.

The organizations requested—and the district court or- dered—that Florida be subjected to a preclearance requirement un- der section 3(c) of the Voting Rights Act. Section 3(c) provides that if “the court finds that violations of the [F]ourteenth or [F]ifteenth [A]mendment justifying equitable relief have occurred . . . , the court . . . shall retain jurisdiction for such period as it may deem appropriate” and exercise preclearance power over new laws re- lated to voting. 52 U.S.C. § 10302(c). After finding violations, the district court determined that that “without preclearance, Florida [could] pass unconstitutional restrictions . . . with impunity.” So, it subjected Florida to preclearance for ten years with respect to any “law or regulation governing [third-party voter-registration organ- izations], drop boxes, or ‘line warming’ activities.”

Preclearance may not be imposed under these circum- stances. Neither this Court nor the Supreme Court has ever USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 61 of 79

22-11143 Opinion of the Court 61

meaningfully interpreted section 3 of the Voting Rights Act of 1965. But the Supreme Court has described the remedy of preclearance as “a drastic departure from basic principles of federalism,” justified by the “exceptional conditions” Congress confronted when the law was enacted. Shelby Cnty., 570 U.S. at 535. And the text of the stat- ute is unequivocal: section 3 only applies “to enforce the voting guarantees of the [F]ourteenth or [F]ifteenth [A]mendment.” 52 U.S.C. § 10302(c) (emphasis added). To succeed on a Fourteenth or Fifteenth Amendment claim, the organizations needed to prove discriminatory intent as well as discriminatory effect. See Greater Birmingham, 992 F.3d at 1321. Because the Florida Legislature did not adopt the challenged provisions with an intent to discriminate, the decision to impose preclearance pursuant to section 3(c) of the Voting Rights Act was incorrect as a matter of law. So, we reverse this decision of the district court.

D. The District Court Correctly Concluded that the Solicitation Provision Is Unconstitutionally Vague.

S.B. 90 expanded the scope of the prohibition against solicit- ing voters who are waiting in line to cast their votes. The new stat- ute provides that “[n]o person, political committee, or other group or organization may solicit voters inside the polling place or within 150 feet of a secure ballot intake station or the entrance to any USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 62 of 79

62 Opinion of the Court 22-11143

polling place.” FLA. STAT. § 102.031(4)(a). It defines “solicitation” to include the following several activities:

seeking or attempting to seek any vote, fact, opinion, or contribution; distributing or attempting to distrib- ute any political or campaign material, leaflet, or handout; conducting a poll except as specified in this paragraph; seeking or attempting to seek a signature on any petition; selling or attempting to sell any item; and engaging in any activity with the intent to influ- ence or effect of influencing a voter.

Id. § 102.031(4)(b). At issue is the constitutionality of the final clause, which prohibits “engaging in any activity with the intent to influence or effect of influencing a voter.” Id. The district court de- termined that this clause was unconstitutionally vague and over- broad.

Before reaching the merits, we address the organizations’ contention that the state officials and Republican Party intervenors lack standing to appeal the district court’s invalidation of the solic- itation provision. This Court has an obligation to ensure that its jurisdiction is proper “at each stage of the proceedings.” Cuban Am. Bar Ass’n, Inc. v. Christopher, 43 F.3d 1412, 1422 (11th Cir. 1995) (citation omitted). The organizations state that “[t]he District Court’s judgment . . . enjoined only one defendant from enforcing the Solicitation Definition: Bay County Supervisor of Elections Mark Andersen.” Supervisor Andersen did not appeal, and the Su- pervisors who did appeal are not enjoined and are thus free to USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 63 of 79

22-11143 Opinion of the Court 63

enforce the provision, according to the organizations. And the Sec- retary of State, they assert, “has no role in enforcing the Solicitation Definition.”

The Republican Party intervenors respond that “[b]eing bound by an injunction is sufficient for appellant standing, but it’s not necessary.” They cite West Virginia v. Environmental Protec- tion Agency for the proposition that standing on appeal can be based on the effect of the judgment below: In considering a litigant’s standing to appeal, the ques- tion is whether it has experienced an injury “fairly traceable to the judgment below.” If so, and a “favor- able ruling” from the appellate court “would redress that injury,” then the appellant has a cognizable Arti- cle III stake.

142 S. Ct. 2587, 2606 (2022) (internal citations omitted) (quoting Food Mktg. Inst. v. Argus Leader Media, 139 S. Ct. 2356, 2362 (2019)) (alterations adopted).

The Secretary has standing to appeal the judgment with re- spect to the solicitation provision. He need not be bound by an in- junction nor even bear the primary responsibility for enforcing the solicitation provision to enjoy the requisite interest. The Secretary is not merely a “concerned bystander” without a “personal stake in defending [the law’s] enforcement.” Hollingsworth v. Perry, 570 U.S. 693, 707 (2013) (citations and internal quotation marks omit- ted). He has a statutory obligation to uniformly administer USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 64 of 79

64 Opinion of the Court 22-11143

elections according to the election code adopted by the Legislature. See FLA. STAT. § 97.012(1). What is more, the Attorney General is a defendant in this suit. Although she sought to have the claims against her dismissed, the district court denied her motion. So, she remained a defendant. The Attorney General appealed the judgment, as the organizations acknowledge. And the Attorney General has the authority to “ap- pear in and attend to, in behalf of the state, all suits or prosecutions, civil or criminal or in equity, in which the state may be a party, or in anywise interested” in federal court. FLA. STAT. § 16.01(4)–(5). She is empowered to represent Florida in this action.

Federal courts must respect states’ strong interests in de- fending the constitutionality of their laws. Here, if the district court’s decision “is left undisturbed, [Florida] will be bound by the conclusive adjudication that [the solicitation provision] is unconsti- tutional.” Maine v. Taylor, 477 U.S. 131, 137 (1986). “[A] State clearly has a legitimate interest in the continued enforceability of its own statutes.” Id. In fact, the Supreme Court recently cautioned that “federal courts should rarely question that a State’s interests will be practically impaired or impeded if its duly authorized repre- sentatives are excluded from participating in federal litigation chal- lenging state law.” Berger v. N.C. State Conf. of the NAACP, 142 S. Ct. 2191, 2201 (2022) (explaining why state officials should be permitted to intervene in a case). The State’s participation facili- tates “a full and fair adversarial testing of [its] interests and argu- ments.” Id. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 65 of 79

22-11143 Opinion of the Court 65

We now turn to the merits. The district court found that the language of the solicitation provision, “on its face, does not provide anyone fair notice of what’s prohibited, nor does it provide precise guidance to the law’s enforcers to prevent arbitrary or discrimina- tory enforcement.” As a result, it declared the provision’s “ban on ‘engaging in any activity with the intent to influence or effect of influencing a voter’ [to be] unconstitutionally vague under the Due Process Clause of the Fourteenth Amendment.” (Quoting FLA. STAT. § 102.031(4)(b)).

The Supreme Court has held that a statute may be “imper- missibly vague” for two reasons. Hill v. Colorado, 530 U.S. 703, 732 (2000). “First, if it fails to provide people of ordinary intelligence a reasonable opportunity to understand what conduct it prohibits. Second, if it authorizes or even encourages arbitrary and discrimi- natory enforcement.” Id. Courts should not lightly declare laws to be void for vagueness. “Facial vagueness occurs when a statute is utterly devoid of a standard of conduct so that it simply has no core and cannot be validly applied to any conduct.” High Ol’ Times, Inc. v. Busbee, 673 F.2d 1225, 1228 (11th Cir. 1982) (internal quotation marks omitted). If a law “implicates no constitutionally protected conduct,” a court “should uphold [a facial vagueness] challenge . . . if the enactment is impermissibly vague in all of its applications.” Vill. of Hoffman Ests. v. Flipside, Hoffman Ests., Inc., 455 U.S. 489, 494–95 (1982). USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 66 of 79

66 Opinion of the Court 22-11143

The challenged clause contains two operative phrases. One prohibits “engaging in any activity with the intent to influence . . . a voter.” FLA. STAT. § 102.031(4)(b). The other prohibits “engaging in any activity with the . . . effect of influencing a voter.” Id. The former is constitutionally permissible, but the latter is unconstitu- tionally vague.

The first half of the challenged clause prohibits acting with intent to “influence a voter,” a phrase that is not devoid of content. Definitions do not exist in a vacuum. The Republican Party inter- venors persuasively point out that “[n]o one contends that a flat ban on ‘soliciting voters’ would be vague.” And the meaning of “solicitation” clarifies the meaning of “influence.” Black’s Law Dic- tionary defines “solicitation,” in relevant part, as “[t]he act or an instance of requesting or seeking to obtain something; a request or petition.” Solicitation, BLACK’S LAW DICTIONARY (11th ed. 2019). The statute does not prohibit “any activity [conducted] with the intent to influence . . . a voter”; instead, it prohibits any activity that qualifies as solicitation that is conducted “with the intent to influ- ence . . . a voter.” See FLA. STAT. § 102.031(4)(b). The plain mean- ing of “solicitation” is definite enough to give notice to ordinary citizens and to prevent arbitrary enforcement.

And the phrase does not prohibit all activities that influence voters; it instead prohibits engaging in activity with the intent to influence a voter. We have explained that “the inclusion of a spe- cific mens rea element may alleviate a law’s vagueness with respect USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 67 of 79

22-11143 Opinion of the Court 67

to providing fair notice to the accused that certain conduct is pro- hibited.” High Ol’ Times, 673 F.2d at 1229; see also Colautti v. Franklin, 439 U.S. 379, 395 (1979) (collecting cases), abrogated on other grounds by Dobbs v. Jackson Women’s Health Org., 142 S. Ct. 2228 (2022). The mens rea element also undermines any asser- tion that this phrase is unconstitutionally vague.

The second half of the challenged clause—which prohibits “engaging in any activity with the . . . effect of influencing a voter”—presents a different question. See FLA. STAT. § 102.031(4)(b) (emphasis added). How is an individual seeking to comply with the law to anticipate whether his or her actions will have the subjective effect of influencing a voter? Knowing what it means to influence a voter does not bestow the ability to predict which actions will influence a voter. As a result, the district court correctly determined that this phrase in the solicitation provision “both fails to put Floridians of ordinary intelligence on notice of what acts it criminalizes and encourages arbitrary and discrimina- tory enforcement, making this provision vague to the point of un- constitutionality.” We need not address whether the organizations are engaging in constitutionally protected conduct because the phrase “is impermissibly vague in all of its applications.” Vill. of Hoffman Ests., 455 U.S. at 494–95.

The Supreme Court has held other laws unconstitutionally vague for similar reasons. For example, the organizations aptly compare the phrase at issue to an ordinance that prohibited USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 68 of 79

68 Opinion of the Court 22-11143

conduct that was “annoying to persons passing by.” Coates v. City of Cincinnati, 402 U.S. 611, 612 (1971). According to the Supreme Court, because “[c]onduct that annoys some people does not an- noy others,” the ordinance specified “no standard of conduct . . . at all.” Id. at 614. Likewise, a person of reasonable intelligence might struggle to identify in advance what conduct would have the “ef- fect of influencing” a voter. Some supervisors of elections also stated that they and their staff would struggle to make the requisite judgment call, which could lead to arbitrary enforcement.

The Republican Party intervenors’ arguments to the con- trary are unavailing. They contend that “[w]hether someone’s con- duct had th[e] effect” of influencing “a voter is a true-or-false deter- mination.” “While it might be difficult to prove that effect in many cases (unless a specific voter comes forward and testifies),” they ar- gue, “that difficulty does not implicate the vagueness doctrine.” They invoke the admonition that “[w]hat renders a statute vague is not the possibility that it will sometimes be difficult to determine whether the incriminating fact it establishes has been proved; but rather the indeterminacy of precisely what that fact is.” United States v. Williams, 553 U.S. 285, 306 (2008). But this argument all but concedes the key point. If the best—or perhaps only—way to determine what activity has the “effect of influencing” a voter is to ask the voter, then the question of what activity has that effect is a “wholly subjective judgment[] without statutory definition[], nar- rowing context, or settled legal meaning[].” Id. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 69 of 79

22-11143 Opinion of the Court 69

The intervenors’ remaining arguments are also unpersua- sive. We will not rely on the assumption that a state court enforc- ing the law would impose a mens rea requirement, apply the law with lenity, and “require that the defendant’s conduct . . . had th[e] natural and probable effect” of influencing the voter. The purpose of the vagueness doctrine is to prevent a person of “ordinary intel- ligence” from being subject to a law that is so vague he cannot de- termine “what conduct it prohibits” or that authorizes “arbitrary and discriminatory enforcement.” Hill, 530 U.S. at 732. Although close cases should be “addressed, not by the doctrine of vagueness, but by the requirement of proof beyond a reasonable doubt,” Wil- liams, 553 U.S. at 306, the promise of due process later on does not obliterate the vagueness doctrine altogether.

Next, we consider whether the solicitation provision is over- broad. We have already held that the second phrase in the chal- lenged clause is unconstitutionally vague, so we need not reach the question of overbreadth as to the second phrase.

The first phrase—which prohibits engaging in activity with the “intent to influence” a voter—is not overbroad. A law is uncon- stitutionally overbroad if its “application to protected speech [is] substantial, not only in an absolute sense, but also relative to the scope of the law’s plainly legitimate applications.” Virginia v. Hicks, 539 U.S. 113, 119–20 (2003) (internal quotation marks omit- ted). Because the first phrase is not ambiguous, it does not “con- sume[] vast swaths of core First Amendment speech,” contrary to USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 70 of 79

70 Opinion of the Court 22-11143

the district court’s ruling. And both this Court and the Supreme Court have upheld various restrictions, even of First Amendment activity, around polling places. See generally Minn. Voters All. v. Mansky, 138 S. Ct. 1876, 1883 (2018) (“Today, all 50 States and the District of Columbia have laws curbing various forms of speech in and around polling places on Election Day.”); Citizens for Police Accountability, 572 F.3d at 1221 (recognizing “our country’s long history of election regulation, the consensus emerging from that history, and the practical need to keep voters and voting undis- turbed”). The phrase’s “application to protected speech” is not “substantial,” particularly “relative to the scope of the law’s plainly legitimate applications.” Hicks, 539 U.S. at 119–20 (citation omit- ted).

Finally, we hold that the second phrase can be severed from the rest of the statute. The district court determined that the entire challenged clause—“engaging in any activity with the intent to in- fluence or effect of influencing a voter”—was severable from the rest of the statute. Although we hold that only the phrase “or effect of influencing” is unconstitutionally vague, we conclude that this phrase is severable from the whole of the statute. Under Florida law, an unconstitutional provision is severable “so long as four re- quirements are met”:

(1) the unconstitutional provisions can be separated from the remaining valid provisions, (2) the legislative purpose expressed in the valid provisions can be ac- complished independently of those which are void, (3) USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 71 of 79

22-11143 Opinion of the Court 71

the good and the bad features are not so inseparable in substance that it can be said that the Legislature would have passed the one without the other and, (4) an act complete in itself remains after the invalid provisions are stricken.

Jones v. Governor of Fla., 950 F.3d 795, 831 (11th Cir. 2020) (quot- ing Smith v. Dep’t of Ins., 507 So. 2d 1080, 1089–90 (Fla. 1987)). The second phrase can be separated from the rest of the statute without undermining the legislative purpose or the coherence of the act. And the prohibition of engaging in activity with the “intent to in- fluence” a voter stands alone; it in no way depends on the reference to the “effect of influencing” a voter for its meaning or operation. So, we affirm in part and reverse in part the judgment of the district court.

E. We Vacate the Judgment Concerning the Registration-Dis- claimer Provision.

The district court ruled that the registration-disclaimer pro- vision violated the First Amendment by compelling speech and granted a permanent injunction against its enforcement. Soon after an appeal was filed, the Florida Legislature repealed the registra- tion-disclaimer provision. See S.B. 524, 124th Leg. Sess. § 7 (Fla. 2022). The parties agree that any appeal of the district court’s judg- ment as to the constitutionality of the registration-disclaimer USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 72 of 79

72 Opinion of the Court 22-11143

provision is now moot. But they disagree regarding whether the judgment should be vacated.

The “ordinary practice in disposing of a case that has be- come moot on appeal is to vacate the judgment with directions to dismiss.” Lewis v. Cont’l Bank Corp., 494 U.S. 472, 482 (1990); see United States v. Munsingwear, Inc., 340 U.S. 36, 39 (1950). But the organizations correctly point out that an exception to this doctrine exists. “Where mootness results from settlement, . . . the losing party has voluntarily forfeited his legal remedy by the ordinary pro- cesses of appeal or certiorari, thereby surrendering his claim to the equitable remedy of vacatur.” U.S. Bancorp Mortg. Co. v. Bonner Mall P’ship, 513 U.S. 18, 25 (1994). In those cases, whether vacatur should be granted is an equitable determination, “and exceptional circumstances may conceivably counsel in favor of such a course.” Id. at 29. But the Supreme Court’s decision in Bancorp did not alter the rule that vacatur is generally proper when the mootness re- sulted from “circumstances unattributable to any of the parties.” Id. at 23 (citation omitted).

The Florida Legislature repealed the registration-disclaimer provision through a bill that the Governor signed. Neither the Leg- islature nor the Governor is party to this appeal. The organizations nonetheless argue that vacatur is not warranted because “the cir- cumstances surrounding [S.B.] 524’s repeal of the Registration Dis- claimer [provision] suggest that Appellants counseled the Legisla- ture to repeal the provision when it became obvious that USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 73 of 79

22-11143 Opinion of the Court 73

Defendants would lose on that provision by the end of the presen- tation of evidence at trial.”

We must decide whether to treat the actions of the Legisla- ture and Governor as solely their own or whether to inquire into what influence parties to this appeal might have had. This set of facts—a challenged law altered or repealed by a legislative body while an appeal is pending—is not unusual. Supreme Court prece- dent, the weight of authority in our sister circuits, and separation- of-powers principles all support attributing the passage of legisla- tion only to the officials with the constitutional authority to partic- ipate in the legislative process in Florida: members of the Legisla- ture and the Governor.

Supreme Court precedent suggests that we should not at- tribute the actions of unrelated, nonparty government officials to other officials who are party to litigation. For instance, after the Court granted certiorari in a recent case, the State and City of New York amended their firearms laws to, in effect, afford “the precise relief that petitioners requested.” N.Y. State Rifle & Pistol Ass’n, Inc. v. City of N.Y., 140 S. Ct. 1525, 1526 (2020). The Court vacated the decision of the Court of Appeals without even mentioning Ban- corp. Id. at 1527. Bancorp itself “express[ed] no view on Mun- singwear’s implicit conclusion that repeal of administrative regula- tions cannot fairly be attributed to the Executive Branch when it litigates in the name of the United States.” Bancorp, 513 U.S. at 25 n.3. If the Supreme Court reads Munsingwear to caution against USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 74 of 79

74 Opinion of the Court 22-11143

attributing the actions of officials within the same branch of gov- ernment to one another, it is hard to imagine how it could be proper to attribute the actions of officials within different branches of government to one another.

And our sister circuits have, in similar contexts, distin- guished between the executive officials who are party to a suit chal- lenging a law and the legislators whose official acts mooted the ap- peal. Cf. Chem. Producers & Distribs. Ass’n v. Helliker, 463 F.3d 871, 879 (9th Cir. 2006) (collecting relevant cases from the Third, Fourth, and D.C. Circuits), overruled on other grounds by Bd. of Trs. of Glazing Health & Welfare Tr. v. Chambers, 941 F.3d 1195, 1199 (9th Cir. 2019); see also, e.g., Valero Terrestrial Corp. v. Paige, 211 F.3d 112, 121 (4th Cir. 2000) (“In this case, the mootness was, as noted, caused by the state legislature’s amendment of statutory provisions that it had earlier enacted, and not by the actions of any of the defendants before this court, all of whom are state executive officials, none of whom is the Governor.”).

The two Fifth Circuit precedents that the organizations cite do not compel a contrary conclusion. In one case, an official of the defendant county placed a controversial statue in storage, mooting the action. Staley v. Harris Cnty., 485 F.3d 305, 307, 313 (5th Cir. 2007) (en banc). But in that case, the distinction between branches of government was not present. In the other case, the governor was a party to the action mooted by new legislation. Hall v. Loui- siana, 884 F.3d 546, 553 (5th Cir. 2018). Even though he signed the USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 75 of 79

22-11143 Opinion of the Court 75

new law, the court declined to attribute “fault” to the governor be- cause “there [was] no evidence that he was the moving force be- hind the legislation.” Id. Because the governor was part of the leg- islative process, under Bancorp, it was reasonable for the Fifth Cir- cuit at least to examine his overall role in the adoption of the new law. Similarly, the Eighth Circuit—which avoided reaching the question “whether a governor who signs the mooting legislation into law is at ‘fault’ for vacatur purposes”—stated that “the South Dakota attorney general and secretary of state apparently had nothing to do with [the relevant law’s] enactment.” SD Voice v. Noem, 987 F.3d 1186, 1190 (8th Cir. 2021).

The decision that most plausibly favors the organizations’ position comes from the D.C. Circuit. It suggested the possibility that legislative action could be attributed to a member of another branch of government if “additional evidence of an illegitimate mo- tive” exists. Nat’l Black Police Ass’n v. District of Columbia, 108 F.3d 346, 354 (D.C. Cir. 1997). But the D.C. Circuit still recognized that, in general, the “presumption of integrity that attaches to leg- islative action and the difficulties that separation of powers creates for attributing one branch’s actions to another support not apply- ing the Bancorp rule to situations where the party seeking vacatur is the government and mootness results on appeal because of leg- islative action.” Id.

Finally, separation-of-powers principles favor not attrib- uting the acts of the legislature and governor to other state officials. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 76 of 79

76 Opinion of the Court 22-11143

As the Ninth Circuit explained, “[t]he principle that legislation is attributed to the legislature alone is inherent in our separation of powers.” Chem. Producers & Distribs., 463 F.3d at 879. It does not matter what the state officials may have “counseled the Legisla- ture” to do, as the organizations suggest. Cf. id. (“Lobbying Con- gress or a state legislature cannot be viewed as ‘causing’ subsequent legislation for purposes of the vacatur inquiry. Attributing the ac- tions of a legislature to third parties rather than to the legislature itself is of dubious legitimacy . . . .”); Rio Grande Silvery Minnow v. Bureau of Reclamation, 601 F.3d 1096, 1131 (10th Cir. 2010) (“Even assuming that the appellant[] [private organizations] ac- tively lobbied the New Mexico congressional delegation, . . . the case became moot as a consequence of the actions of a third party, Congress.”). We must presume that state legislators carry out their duties according to their own judgment.

The organizations assert that vacatur is improper because the state officials and Republican Party intervenors have “shown no compelling reason in favor of vacatur” that would cause an eq- uitable analysis to tip in their favor. They cite Bancorp for the prop- osition that “[j]udicial precedents are . . . valuable to the legal com- munity . . . and should stand unless a court concludes that the pub- lic interest would be served by a vacatur.” Bancorp, 513 U.S. at 26 (citation omitted). But they misread the decision.

In Bancorp, “[t]he parties . . . agree[d] that vacatur must be decreed for those judgments whose review is, in the words of USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 77 of 79

22-11143 Opinion of the Court 77

Munsingwear, ‘prevented through happenstance’—that is to say, where a controversy presented for review has ‘become moot due to circumstances unattributable to any of the parties.’” Id. at 23 (quoting Karcher v. May, 484 U.S. 72, 82–83 (1987)) (emphasis added). The only “contested question [was] whether courts should vacate where mootness results from a settlement.” Id. In that con- text and analogous ones, allowing the judgment to stand should be the default approach based on equitable considerations. But if a general interest in providing guidance to the legal community al- ways sufficed, the doctrine of Munsingwear vacatur would be all but a nullity. When the parties did not cause the mootness, equita- ble considerations tip toward vacatur because those parties “ought not in fairness be forced to acquiesce in the judgment.” Id. at 25. Because the repeal of the registration-disclaimer provision is “un- attributable to any of the parties” to this case, see id. at 23 (citation omitted), we vacate the judgment of the district court in relevant part. IV. CONCLUSION We REVERSE the judgment that the drop-box, solicitation, and registration-delivery provisions violate the Fourteenth and Fif- teenth Amendments and section 2 of the Voting Rights Act. We also REVERSE the imposition of a preclearance requirement under section 3(c) of the Voting Rights Act. We AFFIRM the judgment declaring unconstitutionally vague the second phrase in the solici- tation provision’s challenged clause, but we REVERSE the judg- ment invalidating the first phrase in the clause. We VACATE the USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 78 of 79

78 Opinion of the Court 22-11143

judgment finding the registration-disclaimer provision unconstitu- tional. And we REMAND to the district court for further proceed- ings consistent with this opinion. USCA11 Case: 22-11143 Document: 193-1 Date Filed: 04/27/2023 Page: 79 of 79

22-11143 JILL PRYOR, J., dissenting 1

JILL PRYOR, Circuit Judge, dissenting: I would affirm the district court’s injunction prohibiting the enforcement of S.B. 90’s drop-box, solicitation, and registration-de- livery provisions. In my view, the district court, in its thorough and well-reasoned order, committed no reversible error when it con- cluded that these provisions violated the Fourteenth and Fifteenth Amendments to the Constitution, as well as section 2 of the Voting Rights Act. Given these violations, the district court did not abuse its discretion when it ordered the State of Florida to submit to pre- clearance under section 3 of the Voting Rights Act. I respectfully dissent. 1

1 Because I conclude that the solicitation provision violated the Fourteenth and Fifteenth Amendments and section 2 of the Voting Rights Act, I would not reach the vagueness or overbreadth challenges. I agree with the majority that the challenge to the registration-disclaimer pro- vision is moot and that the portion of the district court’s judgment pertaining to this challenge should be vacated.