League of Women Voters of Florida Inc. v. Florida Secretary of StateLeague of Women Voters of Florida Inc. v. Florida Secretary of State
Opinion of the Court
BY THE COURT:
A petition for rehearing having been filed and a member of this Court in active service having requested a poll on whether this appeal should be reheard by the Court sitting en banc, and a majority of the judges in active service on this Court having voted against granting rehearing en banc, IT IS ORDERED that this appeal will not be reheard en banc.
* Judges Rosenbaum and Abudu recused themselves and did not participate in the en banc poll.
WILLIAM PRYOR, Chief Judge, and GRANT and BRASHER, Circuit Judges, respecting the denial of rehearing en banc:
In 2021, in the aftermath of a presidential election conducted during a pandemic and marked by partisan rancor, Florida enacted several election reforms—especially for voting by mail—with broad legislative support. That legislation, S.B. 90, passed by large majorities of 77 to 40 votes in the Florida House of Representatives and 23 to 17 votes in the Florida Senate. See S.B. 90, FLA. SENATE, https://perma.cc/U2YB-3J4M. Each House member represented approximately 156,000 individuals, see 2010 House District Summary Statistics, FLA. SENATE, https://perma.cc/TY82-F5ST, and each Senator 470,000, see 2010 Senate District Summary Statistics, FLA. SENATE, https://perma.cc/3V6M-ZGJ4. By
A year later, a single district judge refused to afford those elected officials a presumption of legislative good faith and instead found that three provisions were enacted with an intent to discriminate against black voters in violation of the Fourteenth and Fifteenth Amendments and section 2 of the Voting Rights Act. A panel of this Court stayed that injunction, and another panel, after full briefing and oral argument, reversed most of it as based on clearly erroneous findings of fact and misapplications of settled law. See League of Women Voters of Fla. Inc. v. Flа. Sec‘y of State, 66 F.4th 905 (11th Cir. 2023). The Court has since voted to deny rehearing en banc.
Three of our colleagues register a histrionic dissent from the denial of rehearing en banc. They accuse the panel of having disregarded historical evidence of discrimination beginning in the Civil War era and argue that we insufficiently weighed the lingering effects of a “political and economic culture disadvantaging one race over the other.” Dissent at 7. They suggest that we ignored
evidence of pretext in the legislators’ justifications. Dissent at 8-11. And they complain that we failed to defer to the district judge‘s factual findings, especially his analysis of statistical evidence. Dissent at 11-12. These criticisms are meritless.
A. Distant vs. Recent Historical Background
The dissent accuses the panel of effectively prohibiting consideration of historical evidencе when evaluating a legislature‘s discriminatory intent. Not true. We did not hold that historical evidence is “irrelevant.” See Dissent at 7. We instead applied settled precedent that courts must not allow “the old, outdated intentions of previous generations to taint Florida‘s legislative action forevermore,” and that we must “look at the precise circumstances surrounding the passing of the law in question.” League of Women Voters, 66 F.4th at 923 (alterations adopted) (citations and internal quotation marks omitted). Past discrimination is relevant, but historical background is but “one evidentiary source” and not to be overweighed. Abbott v. Perez, 138 S. Ct. 2305, 2325 (2018) (citation and internal quotation marks omitted). We acknowledge thаt laws affect people differently across political and socioeconomic lines. See Dissent at 7. But that fact does not make every historical event that contributed to disparities among racial groups relevant to an analysis of discriminatory intent on the part of a particular legislature.
We faithfully applied controlling precedent when evaluating the effect of historical discrimination on present-day legislative intent. Cf. Village of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252 (1977); Greater Birmingham Ministries v. Sec‘y of State for Ala.,
992 F.3d 1299 (11th Cir. 2021). We gave little weight to distant history—from the Civil War era into the last century—because the Supreme Court
The record reveals a stark lack of evidence of discriminatory intent of the present Florida Legislature. To the contrary, recent history shows that, as Florida has become increasingly racially diverse, its elections have become increasingly open and accessible. See League of Women Voters, 66 F.4th at 919. And in each of three modern judicial decisions that the district court cited as pertaining to allegedly racially motivated laws, federal courts either ruled that the laws were not proved to be enacted with discriminatory intent or never reached the question. Id. at 922. The district court did not even attempt to independently apply the relevant legal tests to those laws: instead, it cursorily declared that “[o]nce is an accident, twice is a coincidence, [and] three times is a pattern.” Id.
The dissent cites the recent decision in Allen v. Milligan as approving the use of historical evidence to prove discriminatory intent. 143 S. Ct. 1487 (2023). But that decision involved section 2 of the Voting Rights Act, which “turns on the presence of
discriminatory effects, not discriminatory intent.” Id. at 1507 (emphasis added). Allen cited the “extensive history of repugnant racial and voting-related discrimination” in Alabama as relevant to whether the political process today is “equally open” to minority voters. Id. at 1506 (citation and internal quotation marks omitted). But relying on distant history to prove that black voters are still affected by unequal access to the political process is different from relying on that history to establish that present-day legislators acted with discriminatory intent. To be sure, when a law has a disparate racial impact, that impact is relevant to an analysis of discriminatory intent, but disparate impact is a standalone factor distinct from historical background under Arlington Heights. See Greater Birmingham, 992 F.3d at 1322.
B. The Absence of Legislative Pretext
The dissent also accuses the panel of improperly dismissing the district judge‘s finding that the Florida legislators’ justifications for S.B. 90 were a pretext for racist intent. But the legislators’ justifications—election security and preventing voter fraud—were presumptively lawful and credible, and the district court clearly erred by finding otherwise. The sponsors and supporters of S.B. 90 repeatedly asserted that they were motivated by concerns of electoral integrity. The proponents were “consistent in their messaging” about the need to assure voters that the process remained “safe and secure” in the face of then-prominent allegations of voter fraud. League of Women Voters, 66 F.4th at 926.
Under settled precedent, evidencе of existing voter fraud is unnecessary for legislation that aims to prevent future fraud. See Greater Birmingham, 992 F.3d at 1334 (“[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.” (citing Crawford v. Marion Cnty. Election Bd., 553 U.S. 181, 192-97 (2008))). So even if voter fraud had never occurred in Florida, the legislators’
C. The District Judge‘s Clearly Erroneous Factual Findings
Finally, the dissent criticizes the panel for improperly reweighing statistical evidence cited by the district court. But if anything, findings based on statistical evidence lend themselves best to appellate review. When a factfinder reviews eyewitness testimony, the judge who personally observes the testimony is in a better position than an appellate judge to assess credibility. By contrast, mathematical facts are empirically provable and the best statistical аnalyses are replicable. If a district judge found that two plus two equals ten, we would lose no sleep over reversing for clear error,
even if the judge relied on the testimony of a college mathematics professor. See, e.g., Miles v. M.N.C. Corp., 750 F.2d 867, 872-73 (11th Cir. 1985) (rejecting district court findings on statistical evidence); E.E.O.C. v. Joe‘s Stone Crab, Inc., 220 F.3d 1263, 1276-78 (11th Cir. 2000) (same). And reversing in such an instance would certainly not be “usurp[ing] the authority” of the district court to “weigh the merits of the expert testimony.” Dissent at 12.
The district judge‘s factual findings were clearly erroneous because they relied on fatally flawed statistical analysis; both the underlying data and the inferences drawn therefrom were irredeemably defective. Not even an expert can draw reliable population-level conclusions by analyzing a sample that is materially unrepresentative. Yet the district judge relied on expert analyses of small, unrepresentative samples of voting behavior to draw broad inferences about the relationship between race and the effect of the challenged S.B. 90 provisions. See League of Women Voters, 66 F.4th at 933. For example, to “prove” the supposed relationship between race and drop-box usage, one expert relied on data drawn predominantly from Sarasota County, where black voters made up 3.24 percent of the population compared to the statewide average of 13 percent. But places where black residents make up an especially high or low percentage of the population tend, as a descriptive matter, to differ according to other metrics, such as wealth and population density. Those disparities may well give rise to differences in voter behavior. And the unrepresentativeness of the sample is only one of several problems we identified with that expert‘s analysis and the district judge‘s assessment of it. The combined
flaws established that the district court clearly erred by relying on that evidence.
To be sure, our precedents rightly require deference to a district judge‘s factual findings. But review for clear error is not a rubber stamp. We must ensure that the district judge did not clearly err and fail to accord the elected representatives of the people of Florida a presumption of legislative good faith. That review requires a close look at the underlying evidence without blinders. We could not perform our duty as an appellate court otherwise.
The Court made the right decision by denying rehearing en banc. Based on this record and controlling precedents, the district
WILSON, Circuit Judge, dissenting from the denial of rehearing en banc, joined by JILL PRYOR, Circuit Judge, and joined as to Part II by JORDAN, Circuit Judge:
One-hundred and fifty-three years ago, in 1870, President Grant hopefully remarked that ratification of the Fifteenth Amendment to the Constitution of the United States would “complete[] the greatest civil change and constitute[] the most important event that has occurred since the nation came into life.”1 Nearly one hundred years later, a similar sentiment would be expressed by Chairman of the United States Commission on Civil Rights, and President of the University of Notre Dame, The Reverend Theodore M. Hesburgh. Reflecting upon the passage of thе Voting Rights Act of 1965, he hailed it as the “most successful civil rights law in the nation‘s history” and “one of the most important legislative enactments of all time.”2
This case demonstrates that nearly sixty years later, despite the promise of the Reconstruction Amendments and the successes
of the Voting Rights Act, the struggle to purge our democracy of discrimination on the basis of race continues. In this case, plaintiffs challenge several provisions of Florida Senate Bill 90 (S.B. 90) because they allege S.B. 90 was enacted with the intent to discriminate against voters of color. The district court conducted a twoweek bench trial, heard from forty-two witnesses, rеceived thousands of pages of documentary evidence, and ultimately issued a 288-page opinion finding that multiple provisions of S.B. 90 violated the Constitution and laws of the United States. Of relevance here, the district court enjoined three provisions because it found they were passed with the intent to discriminate against Black voters.
On review of that order, a divided panel of this court reversed. League of Women Voters of Fla. v. Fla. Sec‘y of State, 66 F.4th 905, 919 (11th Cir. 2023). Incredibly, the panel opinion rejected nearly all of the district court‘s factual findings. This, despite our deferential clear error standard of review for factual findings, which asks whеther the district court‘s findings are simply “plausible in light of the entire record.” Brnovich v. Democratic Nat‘l Comm., 141 S. Ct. 2321, 2349 (2021).
Concluding that so much of a thorough, careful district court opinion fails clear error review should raise eyebrows. Unfortunately, it is not all that surprising. In recent years, this court has picked up a troubling habit of too easily overriding district court factual findings. See Adams ex rel. Kasper v. Sch. Bd. of St. Johns Cnty., 57 F.4th 791, 828 (11th Cir. 2022) (en banc) (Jordan, J.,
dissenting) (collecting cases). More concerning, however, is the panel opinion‘s narrowing of the holistic, multi-factor Arlington Heights3 inquiry for evaluating claims of intentional discrimination.
The panel opinion thus puts this court‘s imprimatur on a law that a federal district judge already concluded intentionally targeted millions of Black Floridians’ rights to vote. This imposition on the right of Floridians to vote their conscience in free elections is itself an issue of “exceptional importance” warranting rehearing under оur rules.
* * *
Before addressing my concerns in greater detail, some background information on the challenged provisions is necessary. The
challenged law, S.B. 90, is a wide-ranging update to Florida‘s voting laws. The three provisions still relevant in this appeal are the dropbox provision, the solicitation provision, and the registration-delivery provision. First, the drop-box provision generally heightens the requirements for county Supervisors of Elections to operate ballot drop-boxes by restricting the hours that the boxes may be available and by requiring that they be monitored in-person when available.
After evaluating all the evidence, the district court concluded that each of these provisions increased the burdens on voting or those assisting others with voting, and that this was done with the intent to target Black Florida voters. With this background in mind, I turn to my concerns about the panel opinion‘s reasoning.
I.
Of the panel‘s errors, the misapplication of the Arlington Heights factors has the widest-reaching impact and is the strongest justification for en banc review, and so I begin there.
To prove a case of vote denial under the Fifteenth Amendment, or a violation of the Equal Protection Clause of the Fourteenth Amendment, our precedents require the plaintiff to show both discriminatory intent and effect. Greater Birmingham Ministries v. Sec‘y of State for Ala., 992 F.3d 1299, 1321 (11th Cir. 2021). To evaluate whether a law was passed with discriminatory intent, we consider the multi-factor test established in the Supreme
(1) the impact of the challenged law; (2) the historical background; (3) the specific sequence of events leading up to its passage; (4) procedural and substantive 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 alternatives.
League of Women Voters, 66 F.4th at 922 (alterations in original) (quoting Greater Birmingham, 992 F.3d at 1322). These factors reflect a “sensitive inquiry” considering all “circumstantial and direct evidence of intent as may be available.” Arlington Heights, 429 U.S. at 266. Thus, properly construed, the Arlington Heights inquiry
requires a holistic evaluation, without overreliance or under-reliance on any single factor. See Abbott v. Perez, 138 S. Ct. 2305, 2325 (2018) (noting “historical background” is but “one evidentiary source” and not to be overweighed). Indeed, that is exactly what the district court did in this case as it recognized our admonition that “[t]he Arlington Heights factors require a fact intensive examination of the record.” League of Women Voters of Fla. v. Lee, 595 F. Supp. 3d 1042, 1077 (N.D. Fla. 2022) (alteration in original) (quoting Greater Birmingham, 992 F.3d at 1322 n.33).
Of particular concern is the panel opinion‘s treatment of the historical background evidence under the second factor and its treatment of the legislative-statements evidence under both the sequence-of-events and contemporary-statements factors. I address each in turn.
A.
With regard to historical evidence, this court has cautioned that there is some “danger” in “allowing the old, outdated intentions” of the past to “taint” a government‘s present efforts. Greater Birmingham, 992 F.3d at 1325. Nonetheless, the Supreme Court has explained that the “historical background” of a government action is probative of discriminatory intent, “particularly if it reveals a series of official actions taken for invidious purpоses.” Arlington Heights, 429 U.S. at 267. This commonsense notion reflects that past actions taken with discriminatory intent are probative of whether the government in the present day is acting pursuant to a pattern of racial discrimination.
In this case, the panel opinion faulted the district court for considering Florida‘s storied post-Reconstruction history of discriminating against voters of color. League of Women Voters, 66 F.4th at 923. It rejected the district court‘s careful explanation of how this history of discriminatory laws working hand-in-hand with mob violence to suppress Black Floridians’ rights to vote is evinced in present-day disparities in socioeconomic and political data. Id.
By the panel opinion‘s telling, this history was irrelevant to the issue of whether S.B. 90 was today enacted with discriminatory intent. This cannot be the case. While it may be that “outdated” discriminatory intentions are of little probative weight in the present, in my view, those intentions must actually be outdated and abandoned before we can cast them out as evidence. Where, as here, the district court has carefully noted how a long history of intentional racial discrimination has produced a political and economic culture disadvantaging one race over the other and finds that this culture both persists into the present day and motivates present-day government actors,
Further, outright rejection of this evidence is inconsistent with the holistic and “sensitive” inquiry required under Arlington Heights. It defies the Supreme Court‘s instruction to consider all “circumstantial and direct evidence of intent as may be available.” Arlington Heights, 429 U.S. at 266. In a case decided just this term, the Supreme Court reaffirmed the viability of assessing a government‘s history of discriminatory actions. See Allen v. Milligan, 143
S. Ct. 1487, 1506 (2023) (approving the district court‘s consideration of “Alabama‘s extensive history of repugnant racial and voting-related discrimination“).5 It is easy to see how a history of discrimination, when evinced in presеnt-day data, is relevant not just to the “historical background” factor, but also to the consideration of both the foreseeability of a disparate impact and the government‘s knowledge of that impact (factors 6 and 7). The Fifteenth Amendment attacks not only the “simple-minded modes of discrimination,” but also the more subtle “sophisticated” ones as well. See Lane v. Wilson, 307 U.S. 268, 275 (1939). When a present-day state government enacts a law that is neutral on its face, it is very much relevant whether or not the “neutral” criteria it purports to utilize are in fact built upon a history of past racial discrimination. Simply put, our Constitution does not require us to overlook the truth thаt this nation‘s history of discrimination is still reflected in the present.
B.
My other primary area of concern is the panel opinion‘s evaluation of the evidence reflecting the legislature‘s non-discriminatory justifications for S.B. 90. Under our construction of the Arlington Heights inquiry, both the sequence of events leading up to the passage of the challenged law and the contemporary statements of key legislators are relevant factors for assessing the legislature‘s intent. Greater Birmingham, 992 F.3d at 1322. Though we have cautioned that courts should be careful when extrapolating a collective
body‘s intent from the intent of its individual members, we have nonetheless recognized thаt these factors encompass some of the most direct evidence of the legislature‘s intent. See id. at 1322, 1324.
Here, the district court, after considering all the evidence, concluded that the legislators’ justifications for S.B. 90 were often “conflicting.” 595 F. Supp. 3d at 1089. It determined that the primary proffered justification, fraud, was unsupported by the evidence. Indeed, it noted that one of the key proponents of S.B. 90 at times disavowed fraud as a justification for S.B. 90. Id. at 1089-90. The panel opinion holds that the district court erred by “implicitly requiring evidence of voter fraud in Florida” to justify S.B. 90. League of Women Voters, 66 F.4th at 925.
I recognize the Supreme Court‘s instruction that, even in the absence of documented cases of voter fraud, the states retain a legitimate interest in guarding against future cases of voter fraud prophylactically. See Crawford v. Marion Cnty. Election Bd., 553 U.S. 181, 196 (2008) (plurality); Brnovich, 141 S. Ct. at 2348. In Greater Birmingham, applying Crawford‘s teachings, we held that the government defendant was not required to produce evidence of voter
unremarkable proposition that the burden of proving intentional discrimination rests on the plaintiff and that the state is not required to produce proof in order to invoke its legitimate interests.
It is one thing to hold, as those cases do, that the state government has a legitimate interest in preventing voter fraud, regardless of actual evidence of fraud. But the panel‘s rule goes further than that. It blocks consideration of the lack of evidence of voter fraud when assessing the credibility and motivations of the legislature as part of the holistic and “sensitive” Arlington Heights inquiry. Further, this narrower rule is unsupported by our case law and is, in my view, wrong. A statе may have many legitimate interests, but if the state was not actually, as a factual matter, motivated by those interests, the Arlington Heights inquiry may take that into consideration when determining if the legislature‘s true motivation was racial discrimination. Here, it was this type of analysis that the district court engaged in. The shifting, contradictory, and conflicting explanations offered by the key legislators in the time leading up to the passage of S.B. 90 convinced the district court that the legislature was actually motivated by something else. Namely, it concluded, a desire to aid the Republican Party by targeting the voting rights of Black Floridians. 595 F. Supp. 3d at 1097-98, 1117.
Our case law should not require district courts tо accept, without scrutiny, talismanic invocations of voter fraud. It is common sense that when a party says it did something for one reason, but that reason is unsupported by the facts and the party is constantly providing shifting, contradictory alternative explanations,
the district court may draw the inference that the party did not really act for that reason. The Supreme Court‘s guidance on this topic, permissive towards the state as it may be, does not require us to naively accept every invocation of voter fraud that is proffered regardless of its factual support.
II.
I turn now to a more basic error of the panel‘s oрinion: its willingness to override the detailed factual findings of the district court. When reviewing factual findings, we should reverse them only if we conclude they are clearly erroneous. Brnovich, 141 S. Ct. at 2348-49. “If the district court‘s view of the evidence is plausible in light of the entire record, an appellate court may not reverse even if it is convinced that it would have weighed the evidence differently in the first instance.” Id. at 2349 (emphasis added). Although not the only place where this occurs, the panel opinion‘s error is most acutely seen in its evaluation of the statistical analysis under the disparate impact factor of Arlington Heights. League of Women Voters, 66 F.4th at 933-38. The plaintiffs put on two experts in statistical analysis to address the impact each оf the challenged provisions of S.B. 90 has on voters of color. The defendants failed to put on any rebuttal expert evidence for these issues and in some places barely conducted cross-examination of the plaintiffs’ experts.
The panel opinion concludes that the district court clearly erred because the statistics that it relied on were “fatally
that some of the mathematical premises it relies upon necessarily lead to the conclusions that the panel opinion draws. But it is immaterial what I, or the panel opinion, think about mathematical principles. In this situation, the experts testified and submitted reports into evidence before the district court. The experts were available both for cross-examination on the substance of their analyses and on the methods and procedures that led to those analyses. The district court was best situated to consider any defects in this evidence and to consider what weight, if any, to give it. In this case, it elected to largely credit the plaintiffs’ experts on the impact of the challenged laws and to find that the laws disparately impacted voters of color.
In concluding оtherwise, the panel opinion necessarily usurped the authority to weigh the merits of the expert testimony. The authority to weigh the evidence and make factual findings rightfully belongs to the district court as the court of original jurisdiction. See Brnovich, 141 S. Ct. at 2349. There will of course be cases where the district court‘s factual findings or conclusions are objectively wrong, and it abuses the authority entrusted to it. In such cases, even our deferential standard of review, of course, does not require us to accept patently, objectively false findings. By my reading, the panel opinion obviously disagrees with the plaintiffs’ experts and perhaps would have given them much lеss weight than the district court did. But “an appellate court may not reverse even if it is convinced that it would have weighed the evidence differently in the first instance.” Id. When we reach down into the factual record and erroneously override the district court‘s factual
findings, we disrupt the balance of powers between trial courts and appellate courts established by Congress. As Judge Jordan mentioned in a recent case, this court has gotten into a bad habit of casting aside our clear error standard of review in certain cases. See Adams, 57 F.4th at 828 (Jordan, J., dissenting). It is far past time that we restore some discipline to our consideration of district court fаctual findings.
III.
We take cases en banc when necessary to “secure or maintain uniformity” of our decisions, or when the case presents a “question of exceptional importance.”
While these cases were certainly important in their own way, so is this case. There are just over 14 million active, registered voters in the State of Florida.9 Thirteen
Black.10 The federal district cоurt in this case thus concluded that the challenged provisions of S.B. 90 were enacted with the intent to discriminate against 1.8 million Americans because of their race. While our nation has taken great strides towards racial equality and equal access to the vote, future progress depends on our present efforts, for “[l]asting gains will not be made without persistent efforts to eradicate totally voting injustices.”11 The errors in the panel opinion‘s reasoning set our jurisprudence back. They hamper the ability of voters to seek the protections promised them by our Constitution. There can be no question that this case presented issuеs of exceptional importance for our court. We should have reheard this case en banc, and accordingly, I dissent from our refusal to do so.