Robinson v. Ardoin

Case: 22-30333 Document: 00516353574 Page: 1 Date Filed: 06/12/2022 United States Court of Appeals for the Fifth Circuit United States Court of Appeals Fifth Circuit FILED June 12, 2022 No. 22-30333 Lyle W. Cayce Clerk Press Robinson; Edgar Cage; Dorothy Nairne; Edwin Rene Soule; Alice Washington; Clee Earnest Lowe; Davante Lewis; Martha Davis; Ambrose Sims; National Association for the Advancement of Colored People Louisiana State Conference, also known as NAACP; Power Coalition for Equity and Justice, Plaintiffs—Appellees, versus Kyle Ardoin, in his official capacity as Secretary of State for Louisiana, Defendant—Appellant, Clay Schexnayder; Patrick Page Cortez; Louisiana Attorney General Jeff Landry, Intervenor Defendants—Appellants, ______________________________ Edward Galmon, Sr.; Ciara Hart; Norris Henderson; Tramelle Howard, Plaintiffs—Appellees, Case: 22-30333 Document: 00516353574 Page: 2 Date Filed: 06/12/2022 No. 22-30333 versus Kyle Ardoin, in his official capacity as Secretary of State for Louisiana, Defendant —Appellant, Clay Schexnayder; Patrick Page Cortez; Louisiana Attorney General Jeff Landry, Movants—Appellants. Appeal from the United States District Court for the Middle District of Louisiana USDC Nos. 3:22-CV-211 & 3:22-CV-214 Before Smith, Higginson, and Willett, Circuit Judges. Per Curiam: Before the court are three emergency motions to stay, pending appeal, an order of the district court that requires the Louisiana Legislature to enact a new congressional map with a second black-majority district. Although we must acknowledge that this appeal’s exigency has left us little time to review the record, we conclude that, though the plaintiffs’ arguments and the district court’s analysis are not without weaknesses, the defendants have not met their burden of making a “strong showing” of likely success on the merits. Nor do we conclude that the cautionary principle from Purcell v. Gonzalez, 549 U.S. 1 (2006) (per curiam), prevents the ordered remedy from taking ef- fect. So we vacate the administrative stay and deny the motion for stay pend- ing appeal. Nevertheless, we expedite this appeal to the next available merits panel, to be selected at random from the regular merits panels already 2 Case: 22-30333 Document: 00516353574 Page: 3 Date Filed: 06/12/2022 No. 22-30333 scheduled to hear cases the week of July 4, 2022. Either before or after argu- ment that week, that merits panel may, in its discretion, opt to reimpose a stay, and its more comprehensive review may well lead it to rule in the de- fendants’ favor on the merits. The plaintiffs have prevailed at this prelimi- nary stage given the record as the parties have developed it and the arguments presented (and not presented). But they have much to prove when the merits are ultimately decided. I. A fuller account of this case’s factual background and procedural his- tory can be found in the district court’s thorough opinion. Robinson v. Ardoin, No. 22-CV-211, 2022 WL 2012389 (M.D. La. June 6, 2022). For purposes of this expedited decision, we summarize only the salient points. This case arises from Louisiana’s congressional redistricting process. After the 2020 census, the state was apportioned six seats , the same number as during the previous redistricting cycle. The Louisiana Legislature thus enacted a map that, like the one in force during the last decade, created just one black-ma- jority district, in the state’s southeast. The Governor vetoed the map, but the Legislature overrode his veto on March 30, 2022. Later that day, the plaintiffs brought this action. The plaintiffs claim that, under the Voting Rights Act (“VRA”) as interpreted by the Supreme Court in Thornburg v. Gingles, 478 U.S. 30 (1986), Louisiana was required to create a second black-majority district. They sought a preliminary injunction to require the Legislature to do so in time for the 2022 election. After a five-day evidentiary hearing, the district court issued a 152- page ruling and order granting the plaintiffs’ motion. The district court con- cluded that the plaintiffs had carried their burden under Gingles. That ruling meant that the plaintiffs had shown that (1) Louisiana’s black population is 3 Case: 22-30333 Document: 00516353574 Page: 4 Date Filed: 06/12/2022 No. 22-30333 sufficiently large and compact to form a majority in a second district, (2) the black population votes cohesively, and (3) whites tend to vote as a bloc usually to defeat black voters’ preferred candidates. Id. at 50–51. The district court gave the Legislature until June 20 to enact a remedial plan that would then be used in the November primary election. 1 The defendant, along with two intervenors (collectively “the defend- ants”), appealed that decision, and that appeal will be decided in due course by a merits panel of this court. Today, as a motions (“administrative”) panel, we consider only the defendants’ emergency motions for stay pending appeal. To decide those motions, we consider “(1) whether the stay applicant has made a strong showing that he is likely to succeed on the merits; (2) whether the applicant will be irreparably injured absent a stay; (3) whether issuance of the stay will substantially injure the other parties interested in the proceeding; and (4) where the public interest lies.” Nken v. Holder, 556 U.S. 418, 434 (2009) (quotation omitted). We review the district court’s legal conclusions de novo and its factual findings for clear error. NAACP v. Fordice, 252 F.3d 361, 364–65 (5th Cir. 2001). A finding is clearly erroneous where, after reviewing the entire record, we are “left with the definite and firm conviction” that the district court erred. Id. at 365 (quotation omitted). 1 We take judicial notice that on June 7, 2022, in response to the order a quo, the Governor called a special session of the Legislature to begin June 15. By letter to the legis- lative leadership dated June 10, partly in response to this panel’s administrative stay, the Governor expressed hope that that stay would be lifted but concluded by stating, “Should the [Fifth Circuit] retain a stay over [the district court’s] decision, I agree that further action of the legislature should be delayed until the Fifth Circuit can review the merits of [that] decision.” 4 Case: 22-30333 Document: 00516353574 Page: 5 Date Filed: 06/12/2022 No. 22-30333 II. We begin with the defendants’ likelihood of success on the merits. The defendants posit four ways the district court erred. First, they say the court used an unduly expansive measure of the black voting-age population (BVAP). Landry Mot. at 16–17. Second, they claim the plaintiffs’ illustrative plans relied on insufficiently compact districts. Ardoin Mot. at 8; Schex- nayder Mot. at 12–15; Landry Mot. at 17–22. Third, they aver that if the state had implemented the plaintiffs’ illustrative plans, it would have engaged in an unconstitutional racial gerrymander. Ardoin Mot. at 5–6; Schexnayder Mot. at 12–15; Landry Mot. at 23–24. Fourth, they contend that the plaintiffs failed to show white bloc voting in light of evidence indicating substantial white crossover voting. Ardoin Mot. at 7; Schexnayder Mot. at 8–12; Landry Mot. at 24–27. A. The first Gingles precondition requires plaintiffs to show that a minor- ity group “is sufficiently large and geographically compact to constitute a ma- jority in a single-member district.” Gingles, 478 U.S. at 50. To do that, plain- tiffs must first define the minority group. The plaintiffs defined Louisiana’s black population to include anyone who identifies as at least partially black. Robinson, 2022 WL 2012389, at *9. That metric, which the parties call “Any Part Black,” would count as black a potential voter who identifies, for example, as both black and American In- dian. The parties discussed two alternative metrics. One is “DOJ Black,” which counts as black a voter who identifies as either solely black or as both black and white. Id. at *20. The “DOJ Black” metric would not count as black a voter who identifies, for example, as both black and Asian. The other alternative, which the parties call “Single-Race Black,” counts a voter as black only when the voter identifies as black and no other race. Id. at *34. 5 Case: 22-30333 Document: 00516353574 Page: 6 Date Filed: 06/12/2022 No. 22-30333 The district court adopted the “Any Part Black” metric. Ibid. The defendants claim that decision “contorted” the first Gingles precondition. Landry Mot. at 16. They observe that the “Any Part Black” metric “includes persons who may be 1/7th Black and who also self-identify as both Black and Hispanic.” Landry Mot. at 17. True. But we do not appreciate that observation’s significance. As the district court noted, the Supreme Court has confronted this question before. 2 It explained that the DOJ Black metric “may have more relevance if the case involves a comparison of different minority groups.” Ibid. But where “the case involves an examination of only one minority group’s” voting strength, the Court considered it “proper to look at all individuals who identify them- selves as black.” Ibid. We have no reason to part from that holding. This case, like Georgia v. Ashcroft, presents no need for comparing minority groups. The plaintiffs seek another BVAP-majority district at the expense of a white-majority dis- trict. So the district court did not err by using the “Any Part Black” metric to calculate BVAP. The defendants are unlikely to succeed on that basis. B. The defendants’ next claim also relates to the first Gingles precondition—specifically, its requirement that the minority group be “reasonably compact.” LULAC v. Perry, 548 U.S. 399, 430 (2006). They say that the population of black voters in the plaintiffs’ new majority-minority district cannot satisfy that precondition. Landry Mot. at 15–24; Ardoin Mot. 2 Georgia v. Ashcroft, 539 U.S. 461, 473 n.1 (2003), superseded by statute on other grounds, Voting Rights Act Reauthorization and Amendments Act of 2006, Pub. L. No. 109-246, 120 Stat. 577, as recognized in Ala. Legis. Black Caucus v. Alabama, 575 U.S. 254, 276–77 (2015). 6 Case: 22-30333 Document: 00516353574 Page: 7 Date Filed: 06/12/2022 No. 22-30333 at 8; see also Schexnayder Mot. at 12–15. That new district is Congressional District 5 (“CD 5”). Although its exact borders vary, 3 CD 5 stretches from Louisiana’s northern border down to Baton Rouge and Lafayette. See Robinson, 2022 WL 2012389, at *10, *12. The plaintiffs’ showing of compactness is not airtight. But to warrant a stay, the defendants must make a “strong showing” that they are likely to succeed on the merits. Nken, 556 U.S. at 434. And from the record before us, we cannot conclude that the district court erred in holding that the plaintiffs satisfied Gingles’s compactness requirement. As the court observed, the “[d]efendants did not meaningfully refute or challenge Plaintiffs’ evidence on compactness.” Robinson, 2022 WL 2012389, at *36. Instead, they put all their eggs in the basket of racial gerrymandering, which we discuss below. That tactical choice has consequences. It leaves the plaintiffs’ evi- dence of compactness largely uncontested. And based on that evidence, we hold that the defendants have not shown that they are likely to succeed on the merits. Before explaining why, we should first relate the law governing Gingles’s compactness requirement. Importantly, that requirement relates to the compactness of the minority population in the proposed district, not the proposed district itself. LULAC, 548 U.S. at 433. Although Gingles itself described the precondition as a requirement that the minority population be “geographically compact,” 478 U.S. at 50, there is more to compactness than geography. Unfortunately, the Supreme Court has not developed a “precise rule” for evaluating all facets of that requirement. LULAC, 548 U.S. at 433. 3 The plaintiffs have introduced six illustrative maps. 7 Case: 22-30333 Document: 00516353574 Page: 8 Date Filed: 06/12/2022 No. 22-30333 But it has identified a few factors. Beyond geography, plaintiffs must also show that putting the minority population into one district is consistent with “traditional districting princi- ples such as maintaining communities of interest and traditional boundaries.” Ibid. (quotation omitted); see also id. at 432 (noting the importance of the district’s population having similar “needs and interests”). Thus, combining “discrete communities of interest”—with “differences in socio-economic status, education, employment, health, and other characteristics”—is impermissible. Id. at 432 (quotation omitted). Finally, compactness must be shown on a district-by-district basis, for a “generalized conclusion” cannot adequately answer “the relevant local question whether the precondition[ ] would be satisfied as to each district.” Wis. Legislature v. Wis. Elections Comm’n, 142 S. Ct. 1245, 1250 (2022) (per curiam) (quotation omitted). The plaintiffs introduced evidence sufficient to show that the population of black voters in their illustrative CD 5 likely satisfied the first Gingles precondition. First, like the district court, we think the illustrative CD 5 appears geographically compact upon a visual inspection. See Robinson, 2022 WL 2012389, at *39. To assess geographical compactness, we may examine the shape of proposed districts. See Bush v. Vera, 517 U.S. 952, 980–81 (1996). And the illustrative versions of CD 5 largely appear compact to the naked eye. They all have their rectangular core in the parishes in the northeastern region of Louisiana between its border with Arkansas and Baton Rouge. Robinson, 2022 WL 2012389, at *10, *12. Indeed, the illustrative CD 5 typi- cally appears just as compact as the benchmark CD 5, if not more so. All have their core in the delta parishes of northeast Louisiana. See Robinson, 2022 WL 2012389, at *2, *10, *12. And although the illustrative versions of CD 5 have small tendrils that jut into parts of central Louisiana, they also 8 Case: 22-30333 Document: 00516353574 Page: 9 Date Filed: 06/12/2022 No. 22-30333 eliminate part of a tendril in the benchmark CD 5 that extends deep into southeastern Louisiana, capturing all but one parish that borders Mississippi. Compare id. at *10, *12, with id. at *2. The district court, however, also assessed geographic compactness with mathematical measures provided by the plaintiffs’ map-drawing experts, William Cooper and Anthony Fairfax. See id. at *35–36. Those experts showed that the districts in their illustrative plans had better Reock, Polsby- Popper, and Convex Hull scores on average than the districts in the benchmark plan. See id. at *36. The problem with that analysis is that it addresses compactness on a plan-wide basis, not a district-by-district basis— as the first Gingles precondition requires. Wis. Legislature, 142 S. Ct. at 1250. Thus, we cannot rely on that evidence to conclude that the minority population in the plaintiffs’ proposed district is geographically compact. Even so, our visual inspection of the proposed CD 5 leads us to agree with the district court that the plaintiffs likely showed that it was geographically compact. Second, as the district court concluded, the illustrative maps respect traditional redistricting criteria. Both map-drawers testified that they took criteria such as “political subdivision lines, contiguity” and “the Legisla- ture’s Joint Rule 21” into account when drawing their maps. Robinson, 2022 WL 2012389, at *10, *13. Fairfax also said he grouped populations with similar economic demographics together and attempted to keep census designated places together when possible. Id. at *13–14. And Cooper stated that he had declined to draw maps for plaintiffs in the past when doing so would require him to violate traditional redistricting criteria. Id. at *11. The district court found both of those experts credible based on their extensive experience in this area, the analytical quality of their reports, their perceived candor, and their ability to respond to cross-examination persuasively. Id. at *38–39. Thus, their testimony indicates that the districts they drew— 9 Case: 22-30333 Document: 00516353574 Page: 10 Date Filed: 06/12/2022 No. 22-30333 including CD 5—are likely consistent with traditional redistricting criteria. Accordingly, the population of black voters in those districts is likely to be reasonably compact as well. Unfortunately, the district court also made the same mistake here that it did in analyzing geographical compactness—namely, analyzing consistency with traditional redistricting criteria on a plan-wide basis. Specifically, the court corroborated the experts’ representations by comparing the number of split political subdivisions in the illustrative and benchmark plans. See id. at *39. But once again, the Gingles inquiry relates to specific districts—not redistricting plans as a whole. Wis. Legislature, 142 S. Ct. at 1250. The district court thus erred by failing to focus on the compactness of the black population in the plaintiffs’ specific proposed districts. Even so, the rest of its analysis is enough to show that the plaintiffs are likely to succeed in showing the first Gingles precondition. We thus do not disturb the district court’s finding on this point. Finally, as the district court concluded, the illustrative CD 5 preserves communities of interest. The plaintiffs introduced extensive lay testimony supporting their claim that the black populations in the illustrative CD 5 were culturally compact. Those witnesses testified that the black populations in those regions share family, culture, religion, sports teams, and the media they consume. Robinson, 2022 WL 2012389, at *15. They also emphasized the educational ties between northeastern Louisiana and the Baton Rouge area, including the fact that many residents of the delta parishes attend college at Southern University in Baton Rouge. Ibid. Likewise, they noted that the black voters in those regions share the same economic interests in the petrol- eum and sugarcane industries. Id. at *16. And all this testimony went unrebutted: The “[d]efendants did not call any witnesses to testify about communities of interest.” Id. at *40. Accordingly, we must agree with the district court that the plaintiffs showed that their proposed CD 5 respected 10 Case: 22-30333 Document: 00516353574 Page: 11 Date Filed: 06/12/2022 No. 22-30333 communities of interest. Granted, the plaintiffs’ evidence has weaknesses. But at this pre- merits stage, it is stronger than the evidence produced by the defendants. Again, as the district court observed, the “[d]efendants did not meaningfully refute or challenge Plaintiffs’ evidence on compactness”; they instead tried to show racial gerrymandering. Id. at *36. Indeed, actions speak louder than words, and the defendants mention very little of what they introduced before the district court in connection with the compactness inquiry in their motions for a stay. Although that would be grounds enough for us to reject the defendants’ position in this posture, we discuss what little evidence the defendants introduced in the interest of showing that the district court’s conclusion on compactness was not erroneous despite its analytical errors. That’s because the testimony the defendants introduced in the district court only obliquely and unpersuasively supports their claim that CD 5’s black population is not compact. First, the defendants’ expert Dr. Thomas Bryan observed that the illustrative districting exercised “surgical” precision in splitting Baton Rouge and Lafayette between congressional districts such that the black neighborhoods were included in CD 5. Id. at *17. Those split political divisions tend to show that CD 5 breached a traditional redistricting criterion in those locations and raise the possibility that CD 5 divides communities of interest based in a single municipality. But providing evidence of a minor departure in one area of the district has only limited probative value with respect to the compliance of the district with traditional redistricting criteria on the whole. And any implication that the proposed CD 5 splits up communities of interest in Baton Rouge and Lafayette is outweighed by the plaintiffs’ direct testimony that the black populations in CD 5 are culturally compact. 11 Case: 22-30333 Document: 00516353574 Page: 12 Date Filed: 06/12/2022 No. 22-30333 Second, the defendants’ expert Dr. Christopher Blunt introduced evidence relating to simulations of redistricting. Dr. Blunt ran 10,000 simula- tions of redistricting in Louisiana and concluded that his simulated districts never had a majority of black voters and were more compact than those in the illustrative plans. Id. at *18–19. By his own admission, however, he did not take communities of interest, previous district boundaries, or municipal boundaries into account when programming his simulations. Id. at *19. And as the district court observed, “Dr. Blunt has no experience, skill, training or specialized knowledge in the simulation analysis methodology that he employed to reach his conclusions.” Id. at *37. Thus, because of Dr. Blunt’s shortcomings as a witness and the fact that his simulations “did not incorporate the traditional principles of redistricting required by law,” the district court concluded that “his opinions merit little weight.” Ibid. In accord with that finding of fact, we discount his opinion as well for whatever purpose it could serve in showing the compactness (or lack thereof ) among the black voting population. Third, the defendants’ expert Dr. M.V. Hood III analyzed the core retention of the districts in the illustrative and benchmark plans. Id. at *19– 20. He testified that the districts in the plaintiffs’ illustrative plans— including CD 5—had lower core retention on average than the districts in the enacted plan. Ibid. But that analysis has little value, for the defendants have not explained why Louisiana’s previous districting should be used as a measuring stick for compactness. Accordingly, Dr. Hood’s analysis has little value in evaluating whether the plaintiffs satisfied the compactness requirement. Finally, the defendants also introduced the testimony of their expert Dr. Alan Murray, who analyzed the spatial distribution of the black voting age population and the white voting age population in Louisiana. Id. at *20. He concluded that “the Black and White populations in Louisiana are 12 Case: 22-30333 Document: 00516353574 Page: 13 Date Filed: 06/12/2022 No. 22-30333 heterogeneously distributed” across the state. Id. at *38. But that statewide analysis has limited probative value with respect to the compactness of the black voting population that would reside in plaintiffs’ proposed district— especially in light of the plaintiffs’ direct evidence supporting compactness. Therefore, the district court did not err in giving that analysis little weight. The arguments that the defendants make on appeal fare no better— especially since they have the burden to make a “strong showing” that the district court erred. Nken, 556 U.S. at 434. First, they say that CD 5 spans long distances. Landry Mot. at 21–22; Ardoin Mot. at 8; Schexnayder Mot. at 13. But they do not explain why those distances are too great—especially for rural regions such as the delta parishes included in CD 5. Indeed, it is not unusual for districts in rural parts of Louisiana to span such distances. Accordingly, that observation does not displace the district court’s conclusion that plaintiffs had satisfied the compactness inquiry. Second, the defendants say that the plaintiffs’ proposal combines populations of voters that are not culturally compact. See LULAC, 548 U.S. at 430–35. The Attorney General maintains that the plaintiffs “reach[ed] out to grab small and apparently isolated minority communities” to pack into CD 5 by stretching some of their illustrative districts down to Lafayette and Baton Rouge, splitting those cities and including only black neighborhoods in CD 5. See Landry Mot. at 17–21 (quoting LULAC, 548 U.S. at 433). The Secretary of State also observes that the illustrative CD 5 combines rural populations in northern Louisiana with urban populations in Baton Rouge, which have distinct interests. Ardoin Mot. at 8. But Dr. Bryan made the same observations before the district court, and we reject these arguments here for the same reasons. That evidence only moderately weighs against a finding of compactness, and it is outweighed by the evidence plaintiffs introduced in favor of that finding. 13 Case: 22-30333 Document: 00516353574 Page: 14 Date Filed: 06/12/2022 No. 22-30333 Finally, the defendants claim that the district court analyzed only the compactness of the plaintiffs’ proposed districts when it should have analyzed the compactness of the black population instead. Landry Mot. at 22–23; Ardoin Mot. at 8. The district court, the defendants observe, credited the plaintiffs’ expert testimony that their districts were more compact on average throughout the state. Landry Mot. at 22–23; Ardoin Mot. at 8. As we have explained, we agree that was error (although for a different reason). 4 But once again, we conclude that that error is not fatal to the district court’s overall finding that plaintiffs have shown that the black voting population in CD 5 is likely to be compact. 5 In sum, the plaintiffs have much to prove when the merits are ultimately decided. But our review is limited by the evidence and arguments that defendants chose to present in the district court and on appeal, with the burden on the defendants to show that a stay is appropriate. Nken, 556 U.S. at 434. When we consider the record as the parties have developed it, the defendants have not shown they are likely to succeed on the merits of their appeal. 4 It is a correct statement of law to say that the compactness of the minority population—not the proposed district—is what matters for the first Gingles precondition. LULAC, 548 U.S. at 433. But the geographic compactness of a district is a reasonable proxy for the geographic compactness of the minority population within that district, which is one factor in the compactness inquiry. 5 The Attorney General also complains that the plaintiffs ran their calculations using an incorrect measure of the size of the black population and that their proposed districts barely qualify as majority-black districts. See Landry Mot. at 16–17. But we have already explained why the plaintiffs’ measure is consistent with Supreme Court precedent. And if their measure is accurate, then the fact that their proposed districts have only small majorities of black voters does not prevent them from satisfying the first Gingles precondition. 14 Case: 22-30333 Document: 00516353574 Page: 15 Date Filed: 06/12/2022 No. 22-30333 C. The defendants further suggest that they will succeed on the merits because the “Plaintiffs’ illustrative plans are plainly racial gerrymanders.” Ardoin Mot. at 5; see also Schexnayder Mot. at 13, Landry Mot. at 23. Race was undoubtedly a factor in the drawing of the illustrative maps. But, as the district court noted, racial consciousness in the drawing of illustrative maps does not defeat a Gingles claim. And even if it did, the defendants have not shown that the plaintiffs’ maps prioritized race so highly as to commit racial gerrymandering, or that complying with the district court’s order would re- quire the Legislature to adopt a predominant racial purpose. Racial gerrymandering is prohibited by the Equal Protection Clause of the Fourteenth Amendment. Shaw v. Reno, 509 U.S. 630, 642 (1993). A state racially gerrymanders when it assigns its citizens to legislative districts based on their race, such that “one district [contains] individuals who belong to the same race, but who are otherwise widely separated by geographical and polit- ical boundaries, and who may have little in common with one another but the color of their skin.” Id. at 647. The Supreme Court has, however, recognized high bars to challenging supposed racial gerrymanders. For a legislative map to constitute a racial gerrymander, a challenger must show that race was the “predominant factor” in its design, such that “the legislature subordinated traditional race-neutral districting principles, including but not limited to compactness, contiguity, and respect for political subdivisions or communi- ties defined by actual shared interests, to racial considerations.” Miller v. Johnson, 515 U.S. 900, 916 (1995). The defendants point out that the illustrative maps presented by the plaintiffs were drawn with race in mind. Cooper, a key expert relied on by plaintiffs to meet the first prong of Gingles, freely admitted that the plaintiffs had “specifically asked” him to draw maps with two minority-majority 15 Case: 22-30333 Document: 00516353574 Page: 16 Date Filed: 06/12/2022 No. 22-30333 districts. 6 Robinson, 2022 WL 2012389, at *47. And as noted above, the maps proposed by the plaintiffs featured districts that, the defendants say, split cit- ies and encompass geographically divergent communities. The defendants also point to the work of their own experts, including Dr. Blunt, who ran thou- sands of random simulations but was unable to produce any black-majority districts. Id. at *18. But despite that evidence, the defendants have not overcome the dis- trict court’s factual findings indicating that the illustrative maps are not racial gerrymanders. Cooper and the plaintiffs’ other key expert, Anthony Fairfax, both testified that, while they considered race, they did not subordinate race to other redistricting criteria, and the district court deemed that testimony credible. Id. at *47. As explained above, both experts weighed racial consid- erations alongside traditional factors such as communities of interest and re- spect for political subdivisions. On the other hand, the defendants’ experts often ignored those same traditional factors. That omission, along with other shortcomings of expertise and demeanor, led the district court to deem the testimony of the defendants’ experts on the question of predominant racial purpose to be “poorly supported,” id. at *36, “merit[ing] little weight,” id. at *37, and “unilluminating,” id. at *38. Neither are the plaintiffs’ proposed maps so bizarrely shaped as to be “unexplainable on grounds other than race.” Shaw, 509 U.S. at 643 (quoting Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 266 (1977)). As explained above, other factual findings by the district court, based on ex- pert and lay testimony presented by the plaintiffs, indicate that the bounda- ries of the illustrative maps have at least some basis in traditional districting 6 Cooper’s “admission” is unsurprising because determining whether another majority-minority district can be drawn consistent with traditional districting principles is the purpose of a Gingles claim. 16 Case: 22-30333 Document: 00516353574 Page: 17 Date Filed: 06/12/2022 No. 22-30333 principles such as communities of interest. The proposed districts also tend to be as geographically compact as the current map, and neither our visual inspection nor the defendants’ analysis indicates that any districts are partic- ularly unnatural. Though the plaintiffs considered race, the defendants have not shown that that consideration predominated over more traditional redis- tricting principles. The inference of racial intent is an intensely factual pro- cess, see Arlington Heights, 429 U.S. at 266, and the unchallenged findings of the district court foreclose the defendants’ contention that the plaintiffs’ il- lustrative maps are racial gerrymanders. Moreover, even if the plaintiffs had engaged in racial gerrymandering as they drew their hypothetical maps, it would not follow that the Legislature is required to do the same to comply with the district court’s order. Illustra- tive maps are just that—illustrative. The Legislature need not enact any of them. For similar reasons, we have rejected the proposition that a plaintiff’s attempt to satisfy the first Gingles precondition is invalid if the plaintiff acts with a racial purpose. See Clark v. Calhoun Cnty., 88 F.3d 1393, 1406–07 (5th Cir. 1996). 7 The plaintiffs have proposed several alternative maps, and the Legis- lature has previously considered maps that would create two minority-major- ity districts. Robinson, 2022 WL 2012389, at *5. The Legislature will be free to consider all those proposals or come up with new ones and to weigh what- ever factors it chooses alongside the requirements of Gingles. The task will no doubt be difficult, but the Legislature will benefit from a strong 7 Contrary to the Attorney General’s position, that holding has not been overruled by the Supreme Court’s observation that Gingles plaintiffs must demonstrate that their proposed districts will perform to elect minority-preferred candidates. See Abbott v. Perez, 138 S. Ct. 2305, 2332 (2018); Harding v. Cnty. of Dallas, 948 F.3d 302, 309–11 (5th Cir. 2020). 17 Case: 22-30333 Document: 00516353574 Page: 18 Date Filed: 06/12/2022 No. 22-30333 presumption that it acts in good faith. See Miller, 515 U.S. at 915. The defendants observe that all the plaintiffs’ maps have one feature in common: They combine “East Baton Rouge [Parish] with the Delta Par- ishes.” Schexnayder Reply at 6. They claim that the first Gingles precondi- tion cannot be satisfied without that feature, but that its “racial design” is “clear.” Ibid. Yet as we have explained, the plaintiffs advanced race-neutral reasons supporting that combination, and the district court accepted them. Moreover, it does not necessarily follow that, for a Gingles claim to succeed, there must be more than one way to draw a compliant, non–racially gerry- mandered district. The plaintiffs have shown that it is possible to draw a sec- ond Gingles district while giving due weight to traditional redistricting crite- ria; that is enough. We do not rule out that a Gingles showing transparently dependent on racial gerrymandering might fail under Gingles’s totality-of-the-circum- stances assessment. Gingles, 478 U.S. at 43; Schexnayder Reply at 5–6. But where, as here, the district court’s findings suggest that racial gerrymander- ing is far from inevitable, that doctrine presents no obstacle to orders like the one issued by the district court. The defendants and their amici are not the first to point out that the doctrine of racial gerrymandering exists in some tension with Gingles. Weigh race too heavily and a legislature risks violating the Constitution; weigh it too lightly and a legislature risks violating the VRA. See, e.g., Perez, 138 S. Ct. at 2315. Legislators who are found to have racially gerrymandered often insist that they were merely seeking to comply with Gingles. See, e.g., Cooper v. Har- ris, 137 S. Ct. 1455, 1468–69 (2017). But that friction remains part of the law, and it is not for us to resolve. If the plaintiffs’ Gingles showing is invalid be- cause of racial gerrymandering, it is difficult to see how any Gingles showing could be successful. Gingles remains good law, and so the defendants have 18 Case: 22-30333 Document: 00516353574 Page: 19 Date Filed: 06/12/2022 No. 22-30333 not shown that they are likely to succeed on that basis. D. The defendants’ final merits challenge concerns the third Gingles pre- condition. Plaintiffs seeking to compel states to create more majority-minor- ity districts must show that “the white majority votes sufficiently as a bloc to enable it . . . usually to defeat the minority’s preferred candidate.” Gingles, 478 U.S. at 51. The plaintiffs must show that such bloc voting would be pre- sent in the challenged districting plan. Harris, 137 S. Ct. at 1470; LULAC, 548 U.S. at 427; Growe v. Emison, 507 U.S. 25, 40 (1993). And that conclusion must be true for voters in a particular location; recall that a “generalized con- clusion” cannot adequately answer “the relevant local question whether the preconditions would be satisfied as to each district.” Wis. Legislature, 142 S. Ct. at 1250 (quotation omitted). So the question posed by the third Gingles precondition is concrete: If the state’s districting plan takes effect, will the voting behavior of the white majority cause the relevant minority group’s preferred candidate “usually to be defeated”? Covington v. North Carolina, 316 F.R.D. 117, 171 (M.D.N.C. 2016) (three-judge court) (emphasis omitted), aff’d, 137 S. Ct. 2211 (2017) (mem.). Although the answer will likely depend in some measure on the num- ber of white voters who buck racial trends and vote for the minority-preferred candidate, the proportion of these so-called “crossover” votes is not directly relevant. Instead, white crossover voting is indirectly relevant because it in- fluences the outcome of elections and, therefore, what really matters for the third Gingles precondition: whether minority-preferred candidates would usually lose under the challenged plan. See, e.g., Westwego Citizens for a Better Gov’t v. City of Westwego, 946 F.2d 1109, 1119 (5th Cir. 1991). The district court concluded that, without a new majority-minority district, white bloc voting would prevent black voters who satisfy the first and 19 Case: 22-30333 Document: 00516353574 Page: 20 Date Filed: 06/12/2022 No. 22-30333 second Gingles preconditions from electing their preferred candidates. Rob- inson, 2022 WL 2012389, at *50–51. The court primarily relied on the plain- tiffs’ two experts, who explained that, despite some white crossover voting, “no Black-preferred candidate” had won a statewide or congressional race in the elections they examined except in CD 2, the preexisting majority-minor- ity district. Id. at *50. And it dismissed the testimony of the defendants’ experts, who pointed to some examples where whites did not vote as a bloc or where black voters would have been able to elect the candidates of their choice if the proposed maps had been in place. Id. at *50–51. It reasoned that those experts’ examples were based on a single, unusual election—the 2020 Presidential Contest—and relied on “limited data” or “outlier[s],” unlike the analyses offered by the plaintiffs’ experts. Id. Whether bloc voting will usually defeat black voters’ attempts to elect their preferred candidates is a question of fact. Rangel v. Morales, 8 F.3d 242, 245 (5th Cir. 1993). Nevertheless, we review de novo the district court’s ap- plication of the legal standard for bloc voting. The defendants challenge that application. They say the “district court failed to ask the correct legal question.” Schexnayder Mot. at 11. And they claim that the plaintiffs “failed to prove, or even address,” the question of whether white crossover voting was “legally significant,” which is to say that it would normally cause minority voters’ preferred candidates to lose. Id. at 8–9 (quotation omitted). In their telling, the plaintiffs’ experts established only that “black voters and white voters voted differently.” Id. at 9 (quotation omitted). We disagree. The district court framed the legal question correctly. Although it discussed crossover voting, it explained that “crossover voting was inherently included in” the plaintiffs’ experts’ analysis. Robinson, 2022 WL 2012389, at *51 (emphasis added). It concluded that “the levels [of 20 Case: 22-30333 Document: 00516353574 Page: 21 Date Filed: 06/12/2022 No. 22-30333 crossover voting the experts] found were insufficient to swing the election for the Black-preferred candidate in any of the contests they examined.” Id. In other words, the district court relied on the experts’ analysis to answer the right question: whether black voters’ preferred candidates could win the pro- posed district under the enacted maps. Contra Schexnayder Reply at 3. And the plaintiffs’ experts tailored their analysis to that question. They consid- ered the outcomes of elections, not the abstract behavior of voters by race. Robinson, 2022 WL 2012389, at *50. 8 Next, the defendants claim that Covington supports their position. Schexnayder Mot. at 8–12; Schexnayder Reply at 3. They correctly observe that the question under Covington is whether, without a VRA remedy, the minority voters’ preferred candidate will usually lose. 316 F.R.D. at 170–71. But the defendants then explain that this case is like Covington because all experts acknowledge that some parts of Louisiana enjoy significant white crossover voting. Schexnayder Mot. at 10–11. That contention loses the plot. As the defendants themselves have explained, crossover voting is not relevant per se; it is relevant only for its ef- fect on the outcome of elections. 9 Crossover voting in unspecified locations 8 As we did in the context of the first Gingles precondition, we reiterate that what matters for the third Gingles precondition is whether black voters in the proposed district could elect the candidates of their choice under the challenged districting, not whether black voters in all parts of the state could. See Wis. Legislature, 142 S. Ct. at 1250. Thus, the experts’ analysis of white bloc voting statewide was not strictly relevant. But those experts also analyzed voting behavior “in the enacted plan districts that would contribute voters to an additional Black opportunity congressional district.” Robinson, 2022 WL 2012389, at *50. Accordingly, their analysis is enough to support the district court’s conclusion that the plaintiffs were likely to succeed on their claims—especially given the defendants’ weak evidence and the deference we owe to the district court. 9 See Schexnayder Reply at 1 (“The [third precondition] question does not turn on ʻany’ crossover voting but on whether it is sufficiently robust that ʻa VRA remedy’ is unneces- sary to ensure equal opportunity.” (emphasis added)). 21 Case: 22-30333 Document: 00516353574 Page: 22 Date Filed: 06/12/2022 No. 22-30333 that can range as high as “26%,” Schexnayder Mot. at 11, is not enough to defeat the district court’s conclusions about the likely future outcomes of elections. Doing so would require more persuasive evidence that reveals the likely outcomes in elections in a particular district at issue. See Wis. Legisla- ture, 142 S. Ct. at 1250. Even less relevant is the defendants’ observation that a hypothetical district could elect black-preferred candidates with as little as 40% BVAP. Landry Mot. at 25–26; Schexnayder Mot. at 11; see Schexnayder Reply at 3. That observation fails to account for the third Gingles precondition’s focus on the actual challenged districting. Harris, 137 S. Ct. at 1470; LULAC, 548 U.S. at 427; Growe, 507 U.S. at 40. As the plaintiffs observe, it would be bizarre if a state could satisfy its VRA obligations merely by pointing out that it could have—but did not—give minority voters an opportunity to elect candidates of their choice without creating a majority-minority district. Robinson Re- sponse at 16. To the extent that the defendants intend to contest the district court’s factual findings, this observation is inadequate to show clear error, at least for the purposes of our preliminary review in deciding these motions for a stay. The defendants also claim that the court’s decision is incompatible with Harris. Ardoin Mot. at 7; Schexnayder Reply at 2–3. After Harris, the plaintiffs cannot rely, defendants say, on the “black population in [East Baton Rouge Parish], where there is substantial crossover voting.” Ardoin Mot. at 7. Because they cannot satisfy the first Gingles precondition without those voters, the argument goes, the plaintiffs cannot succeed. Ibid. That position misconstrues Harris. There, the Supreme Court con- fronted a wholly different scenario. Race predominated in the state’s dis- tricting process, Harris, 137 S. Ct. at 1468–69, and the state claimed that that predominance was necessary to comply with the VRA, id. at 1469. Part of its 22 Case: 22-30333 Document: 00516353574 Page: 23 Date Filed: 06/12/2022 No. 22-30333 stated rationale included the mistaken assumption that, to satisfy the VRA, minority groups who satisfied the first and second Gingles preconditions but could not satisfy the third precondition on account of crossover voting were nonetheless entitled to a majority-minority district. See id. at 1472; see also Bartlett v. Strickland, 556 U.S. 1, 14–15 (2009) (plurality opinion). But the Court reaffirmed the principle that the third precondition is a sine qua non of a Gingles claim. Harris, 137 S. Ct. at 1472. If a minority group can already elect its preferred candidates, it does not matter whether that ability accrues in a majority-minority or a performing crossover district. Harris means these plaintiffs could not satisfy the third Gingles pre- condition if they usually were able to elect candidates of their choice. But that is not what the district court found. Harris does not mean that the third Gingles precondition is unsatisfied if some black voters necessary to form a majority happen to reside near white voters who share their political beliefs. That fact could influence the dispositive question, but the defendants have not presented sufficient evidence for us to conclude that the district court’s factual findings were clearly erroneous. 10 Finally, the defendants say the district court improperly “shift[ed] the [plaintiffs’] burden” to prove white bloc voting onto the defendants. Schexnayder Mot. at 12. They claim the court relied on the defense’s failure to produce “sufficient data.” Ibid. (quoting Robinson, 2022 WL 2012389, 10 We do not address the related question whether the third Gingles precondition can be satisfied where a substantial portion of the minority voters included in the Gingles coalition will already be able to elect candidates of their choice under the enacted plan because they live in a majority-minority district. That could be true of East Baton Rouge Parish voters who live in the enacted CD 2, which is a majority-minority district that is likely to elect black-preferred candidates. But no party has asked us to decide that question. See United States v. Sineneng-Smith, 140 S. Ct. 1575, 1579 (2020). The defendants have instead focused on the presence of white crossover voting around those minority voters. 23 Case: 22-30333 Document: 00516353574 Page: 24 Date Filed: 06/12/2022 No. 22-30333 at *51). But the defendants mischaracterize the district court’s analysis. It said that one of the defendants’ experts failed to support his opinion with “sufficient data,” Robinson, 2022 WL 2012389, at *51, not that the defend- ants had failed to produce sufficient data to support a hypothetical burden. A court does not impose a burden of proof on a party by observing that the party’s rebuttal evidence is uncompelling. Instead, it concludes that the other party has met its burden of proof. * * * None of the defendants’ merits challenges to the district court’s order carries the day. We thus conclude that the defendants have not met their bur- den of showing likely success on the merits. Because likelihood of success is “arguably the most important factor,” that fact weighs heavily against the stay. Tesfamichael v. Gonzales, 411 F.3d 169, 176 (5th Cir. 2005). III. It is beyond dispute that the defendants would suffer irreparable harm absent a stay. “When a statute is enjoined, the State necessarily suffers the irreparable harm of denying the public interest in the enforcement of its laws.” Planned Parenthood of Greater Tex. Surgical Health Servs. v. Abbott, 734 F.3d 406, 419 (5th Cir. 2013). That harm is especially clear in voting rights cases: Wrongly enjoined maps may be restored, but “[s]etting aside an elec- tion is a drastic remedy” that courts seldom undertake. Rodriguez v. Bexar Cnty., 385 F.3d 853, 859 n.2 (5th Cir. 2004). Not using the state’s enacted maps will irreparably injure the defendants, so this prong favors the requested stay. IV. We next decide whether the balance of equities and the public interest 24 Case: 22-30333 Document: 00516353574 Page: 25 Date Filed: 06/12/2022 No. 22-30333 favor a stay. See Tex. Democratic Party v. Abbott, 961 F.3d 389, 397 (5th Cir. 2020). The equities favor a stay if it would benefit the defendants more than it would harm the nonmovants. See Nken, 556 U.S. at 434. We then must ask whether a stay would serve the public interest. Ibid. Those factors merge where the state seeks to stay an injunction against its legislative enactments. That’s because the state’s interest in enforcing its laws merges with the pub- lic’s interest in the same. E.g., Veasey v. Abbott, 870 F.3d 387, 391 (5th Cir. 2017) (per curiam). Thus, if the equities favor the nonmovants, so will the public interest. See Tex. Democratic Party, 961 F.3d at 412. The defendants offer three reasons why those factors favor a stay. First, they say that Purcell v. Gonzalez, 549 U.S. 1 (2006) (per curiam), pre- cludes an injunction months before the November primaries. Second, they contend that we should stay the case pending the outcome of Merrill v. Milli- gan, which the Supreme Court will hear next term. Third, they complain that the district court did not give the Legislature enough time to adopt remedial maps. None of those grounds supports a stay. A. The defendants first invoke the principle of election nonintervention, which they attribute to Purcell. Enjoining election laws before an election may confuse voters, and that risk, Purcell says, “will increase” as the election nears. Id. at 5. We and the Supreme Court have applied Purcell to stay in- junctions that threaten to confuse voters, unduly burden election administra- tors, or otherwise sow chaos or distrust in the electoral process. 11 In one See, e.g., Tex. All. for Retired Ams. v. Hughs, 976 F.3d 564, 566–67 (5th Cir. 2020); 11 Republican Nat’l Comm. v. Democratic Nat’l Comm., 140 S. Ct. 1205, 1207 (2020) (per curiam); Benisek v. Lamone, 138 S. Ct. 1942, 1945 (2018) (per curiam); Moore v. Harper, 142 25 Case: 22-30333 Document: 00516353574 Page: 26 Date Filed: 06/12/2022 No. 22-30333 formulation, Purcell asks whether obeying the district court’s injunction would “be feasible without significant cost, confusion, or hardship.” Merrill, 142 S. Ct. at 881 (Kavanaugh, J., concurring). If not, the defendants might be entitled to a stay. But the defendants have not identified a comparable case where we or the Supreme Court has applied Purcell’s principle. Here, the primary elec- tions are five months away. The earliest impending deadline by which candi- dates must qualify for the primaries is June 22. Robinson, 2022 WL 2012389, at *60. Most candidates qualify for the primaries later by paying a small filing fee; the deadline for that is more than a month away. Ibid. Overseas absentee ballots need not be mailed until late September, and early voting begins in October. Ibid. The classic Purcell case is different. It concerns an injunction entered days or weeks before an election—when the election is already underway. In Veasey v. Perry, 769 F.3d 890, 892 (5th Cir. 2014), we stayed an injunction entered nine days before the start of early voting. In Texas Alliance, we stayed an injunction entered eighteen days before the start of early voting. 976 F.3d at 567. In Texas Democratic Party, we stayed an injunction entered “weeks” before the start of in-person voting. 961 F.3d at 411. Purcell itself stayed an order changing election laws twenty-nine days before an election. Tex. All., 976 F.3d at 567. And the Supreme Court has blocked injunctions entered five, 12 thirty-three, 13 and sixty days 14 before Election Day. Even Merrill, an S. Ct. 1089, 1089 (2022) (Kavanaugh, J., concurring); Merrill v. Milligan, 142 S. Ct. 879, 880–81 (2022) (Kavanaugh, J., concurring). 12 RNC, 140 S. Ct. at 1206–07. 13 Tex. All., 976 F.3d at 566 (citing North Carolina v. League of Women Voters of N.C., 574 U.S. 927 (2014)). 14 Id. at 567 (citing Husted v. Ohio State Conf. of the NAACP, 573 U.S. 988 (2014)). 26 Case: 22-30333 Document: 00516353574 Page: 27 Date Filed: 06/12/2022 No. 22-30333 outlier cited by the defendants, Schexnayder Reply at 10, stayed an election less than four months away, where absentee voting would start in about two months, 142 S. Ct. at 888 (Kagan, J., dissenting). That is not to say that Purcell is just a tallying exercise. It is not. Even where an election is many months away, the movant’s showing a likelihood of success on the merits, for example, may counsel in favor of staying a district court’s injunction. 15 But previous applications of Purcell differ enough from this case that we must inquire further. In hopes of showing that the district court’s injunction implicates Purcell, the defendants highlight the testimony of Sherri Hadskey, the state elections commissioner. Ardoin Mot. at 14–15. According to the defendants, Hadskey stressed three injuries that might result from the injunction. First, Hadskey represented that “a new congressional plan,” id. at 14, would require the state “to reassign voters who are in new congressional districts” under the enjoined maps to the remedial districts required by the district court, id. at 15. About 250,000 of those voters already have received notice of their districts under the enacted maps, and the defendants say informing those voters of yet another change to their districts could confuse them. Ibid. We don’t doubt that multiple mailings could confuse some voters. But at this early stage, any confusion would be minimal. More than enough time remains for the state to assuage any uncertainty before the primary elections. This is not a case, for example, where many voters already have cast ballots 15 Cf. Merrill, 142 S. Ct. at 881 (Kavanaugh, J., concurring) (noting that Purcell’s application reflects “ordinary stay principles”); see also Nken, 556 U.S. at 434 (“The first two factors of the traditional standard” for evaluating a stay—irreparable injury and the likelihood of success on the merits—“are the most critical.”). 27 Case: 22-30333 Document: 00516353574 Page: 28 Date Filed: 06/12/2022 No. 22-30333 or submitted ballot applications, such that conducting an election with new lines would throw into doubt whether those votes would count or whether voters should request new ballots. See, e.g., LULAC v. Abbott, No. 1:21-CV- 991, 2022 WL 1410729, at *31 (W.D. Tex. May 4, 2022) (three-judge court). Here, weeks remain before the earliest candidate filing deadline, and months remain before the primary elections. Second, Hadskey noted that the June 22 deadline for candidates to qualify for office by petition is fast approaching. Ibid. “If congressional candidates do not meet” that deadline, the defendants state, “the candidates will have to pay a filing fee and qualify by” late July. Ardoin Mot. at 15. But the defendants have not shown that those deadlines implicate the Purcell principle. The June 22 deadline applies only to the few candidates who choose to qualify by nominating petition, and the record suggests that adjusting that deadline would not impact voters. Robinson, 2022 WL 2012389, at *60. It merits mention that even this June 22 deadline was ex- tended by the district court to July 8. Robinson, 2022 WL 2012389, at *63. On that score, we also remind the parties and the district court that as this litigation progresses, “[i]f time presses too seriously, the District Court has the power appropriately to extend” that deadline and other “time limitations imposed by state law.” Sixty-Seventh Minn. State Senate v. Beens, 406 U.S. 187, 201 n.11 (1972). And we agree with the district court that the State has enough time to implement new maps without having to change the more pop- ular July filing deadline. See Robinson, 2022 WL 2012389, at *59. After all, as the district court recounted, Hadskey herself testified that after the en- acted map became law, her office updated their records and notified affected voters in less than three weeks. Ibid. Yet almost six weeks remain before the July filing deadline. Those facts also discredit the defendants’ assertion that the district court’s injunction will rush election administrators, causing them to make more mistakes. See Ardoin Mot. at 17. The risk of mistakes is 28 Case: 22-30333 Document: 00516353574 Page: 29 Date Filed: 06/12/2022 No. 22-30333 relevant under Purcell, but we agree with the district court that the injunction does not meaningfully increase that risk. Third, Hadskey identified other administrative burdens that an injunc- tion would cause. The defendants highlight several of those burdens, includ- ing the need to “conduct[ ] yearly maintenance on scanners and voting equip- ment” and to review the voter rolls for accuracy—a process that the defend- ants say began on May 23. Id. at 15. Hadskey also noted the risk posed by a national paper shortage, which could threaten the state’s ability to produce enough ballot envelopes. Robinson, 2022 WL 2012389, at *32. We agree with the district court: The defendants have not shown that bearing those administrative burdens while complying with the challenged injunction would inflict more than ordinary “bureaucratic strain” on state election officials. Id. at *60. Notably, the district court credited the testi- mony of the Governor’s executive counsel, who explained that Louisiana has significant experience adjusting the time, place, and manner of elections and has the administrative capacity to draw a new map before this election. See id. at *31, *60. On the other hand, the district court found unconvincing the aforementioned testimony of Hadskey. Ultimately, the district court found that “although the administrative tasks that would be necessitated by a new congressional map would challenge the Secretary of State’s office, the effort required would not be a heroic undertaking.” Id. at *59. The court further explained that it did not perceive any specific reasons why voter notices could not be sent out in time. Ibid.; see also Purcell, 549 U.S. at 5 (finding error for court of appeals not to defer to discretion of district court). It is axiomatic that injunctions in voting-rights cases burden the de- fendants. But the question, under Purcell, is not whether an injunction would burden the defendants, but whether that burden is intolerable—that is, whether the defendants cannot bear it “without significant cost, confusion, 29 Case: 22-30333 Document: 00516353574 Page: 30 Date Filed: 06/12/2022 No. 22-30333 or hardship.” Merrill, 142 S. Ct. at 881 (Kavanaugh, J., concurring). Here, the burdens threatened by the injunction are, as far as the defendants have shown, entirely ordinary. Take, for example, the national paper shortage that the defendants in- voked at the district court. Though we can imagine a case where a paper shortage would augment the hardship of an injunction, this is not that case. No ballots have been printed for the November primaries, and the number of ballots needed for the elections will not change if district lines are altered. Robinson, 2022 WL 2012389, at *59. Changing the lines would mean that the defendants must mail new notices to many voters. But the district court doubted that a paper shortage, even if it complicated matters, could prevent the State from notifying voters of their districts before the elections months away. Moreover, the district court found that the State’s digital voter out- reach “can also provide information about any district changes.” Ibid. The defendants cite no case applying Purcell to stay an injunction this far from an election. Nor have they shown that the risks of chaos, distrust, or voter confusion at the heart of Purcell are present here. As Justice Kavanaugh made clear, the Purcell doctrine is about voter confusion and infeasibility, not administrative convenience. So we will not stay the order on that ground. B. The defendants next maintain that this proceeding should have been stayed pending the Supreme Court’s decision in Merrill. The district court denied a similar motion, Dkt. 135, but that decision is not before us. Here, we decide only whether Merrill’s pendency justifies staying the injunction, and it does not. It is true that Merrill (Sup. Ct. 21-1086), concerns many of the same issues as this case: The Merrill plaintiffs sued under Gingles, claiming that the VRA required the state of Alabama to create an additional minority- 30 Case: 22-30333 Document: 00516353574 Page: 31 Date Filed: 06/12/2022 No. 22-30333 majority district. The Court’s resolution of that case might, or might not, shed light on this one. But the Court plans to consider Merrill during October Term 2022. That means that any decision likely will come long after the 2022 elections, which are the subject of this appeal, have taken place. In that context, staying these proceedings would not promote judicial economy, and the defendants do not explain how a stay would serve the parties’ interests. See Landis v. N. Am. Co., 299 U.S. 248, 254–55 (1936). 16 We do not grant the defendants’ requested stay on this ground. C. The defendants also urge us to stay the district court’s order to give the Louisiana Legislature more time to enact a remedial plan. Schexnayder Mot. at 18. But they have not explained why they cannot enact a new plan in the time that the district court allotted, so we will not stay the injunction on that ground. The defendants complain that the district court gave the Louisiana Legislature only fourteen days—until June 20—“to enact a remedial plan.” Robinson, 2022 WL 2012389, at *1. Because the Legislature’s regular session has ended, Schexnayder Mot. at 18, the defendants say that any redistricting effort would have to proceed in a special session, La. Const. art. III, § 2(B). But a special session requires seven days’ notice, ibid., and the 16 See also Merrill, 142 S. Ct. at 879 (Kavanaugh, J., concurring in grant of stay) (“[T]he principal dissent is wrong to claim that the Court’s stay order makes any new law regarding the Voting Rights Act. The stay order does not make or signal any change to voting rights law.”); id. at 882 (Roberts, C.J., dissenting from grant of stay) (“I respectfully dissent from the stays granted in these cases because, in my view, the District Court properly applied existing law in an extensive opinion with no apparent errors for our correction.”). 31 Case: 22-30333 Document: 00516353574 Page: 32 Date Filed: 06/12/2022 No. 22-30333 Legislature cannot enact a bill without reading it “at least by title on three separate days in each house” and holding a public hearing, id. art. III, § 15(D). Those requisites would leave the Legislature only five working days to craft new redistricting maps. Schexnayder Mot. at 18. So the defendants conclude that “[t]he district court set the Legislature up to fail.” Ibid. In theory, that complaint could justify narrowing the district court’s remedy. A stay pending appeal “suspends judicial alteration of the status quo,” Veasey, 870 F.3d at 392 (cleaned up); the Supreme Court has stressed that courts should afford legislatures “a reasonable opportunity” to fix constitutionally defective maps, Wise v. Lipscomb, 437 U.S. 535, 540 (1978); and unduly shortening the time to enact curative maps could rob a legislature of that opportunity. That lost chance would burden the defendants, and “injunctive relief should be no more burdensome to the defendant than necessary to provide complete relief to the plaintiffs.” Califano v. Yamasaki, 442 U.S. 682, 702 (1979). But the defendants request a stay—not more limited relief. And while five legislative days is not much time, the defendants do not explain, beyond bare assertion, how or why that period is too short. And the record suggests that period would suffice. Before enacting the maps contested here, the Legislature considered “alternative maps with two majority-minority districts.” Robinson, 2022 WL 2012389, at *5. Thus, the special session would not start from scratch. Id. at *31. We conclude that a stay is not necessary. This is especially so because, as the district court stressed in refusing to stay its order pending appeal, “[i]f Defendants need more time” to draw a new map, the district court would “favorably consider a Motion to extend the time to allow the Legislature to complete its work.” Dkt. 182 at 3 (emphasis omitted). 32 Case: 22-30333 Document: 00516353574 Page: 33 Date Filed: 06/12/2022 No. 22-30333 V. For the foregoing reasons, the administrative stay is VACATED, and the motions for stay pending appeal are DENIED. This appeal is sua sponte EXPEDITED. We direct the Clerk to issue an expedited briefing schedule and to calendar this matter for argument before the next available randomly selected merits panel that is already scheduled to hear arguments during the week of July 4, 2022. Our ruling here concerns only the motions for stay pending appeal; “our determinations are for that purpose” only “and do not bind the merits panel.” Veasey, 870 F.3d at 392. At this preliminary, non-merits stage, the defendants have merely fallen short of carrying their burden. That said, neither the plaintiffs’ arguments nor the district court’s analysis is entirely watertight. And it is feasible that the merits panel, conducting a less-rushed examination of the record in the light of differently framed arguments, may well side with the defendants. 33