In this case we review plaintiff-appellants’ challenge to two at-large judicial election districts in Florida under Section Two of the Voting Rights Act, 42 U.S.C. § 1973 (“Section Two”). Although the district court found that racially polarized voting plagued the electoral systems at issue, it granted judgment to the defendant-appellees- on the ground that Florida’s interest in maintaining its current system of selecting judges outweighs the plaintiff-appellants’ interest in their proposed remedy. In addition, the district court ruled that it could not accept plaintiff-appellants’ remedial plan because it would involve racially-eonscious redistricting without a compelling state purpose. Because we believe that our Section Two precedents foreclose any significant restructuring of a state’s judicial election system, we affirm.

I. BACKGROUND

In this class action, plaintiff-appellants Anita Davis, et al. (“Davis”) attack two at-large judicial election systems in Florida on the grounds that black voters within these systems suffer from illegal vote dilution. Although the two districts at issue differ in size and jurisdiction, they share similar electoral systems and demographies. First, the Second Judicial Circuit (“Second Circuit”) comprises the six counties of Franklin, Gadsen, Jefferson, Leon, Liberty, and Wakulla. All eleven judges on the Second Circuit are *1417elected in at-large, circuit-wide voting for six-year terms. Within the Second Circuit, blacks constitute 28.9% of the overall population, 26.1% of the voting age population, and 26.1% of registered voters. Much like the judges on the Second Circuit, the four judges on the Leon County Court are elected in at-large, countywide voting for four-year terms. In Leon County, blacks make up 24.2% of the overall population, 22.2% of the voting age population, and 21.8% of registered voters. In both election districts, the black population is concentrated in a few areas, with many black voters residing either within Gadsen County or a few precincts of Tallahassee. Further, the non-partisan election systems in both the Second Circuit and Leon County include majority vote requirements, post-numbered systems,1 and staggered terms. In both circuits, the Governor may fill any mid-term vacancies through appointment of candidates recommended by a Judicial Nominating Commission. Finally, the Second Circuit Court is a trial court of general jurisdiction, see Fla. Const, art. V § 6; Fla. Stat. § 26.012, while the Leon County Court is a trial court of limited jurisdiction, see Fla. Const, art. V § 5; Fla. Stat. § 34.01.

Beyond these structural similarities, the two judicial districts also share a history of racially polarized voting. In the few elections in which black candidates have competed against white candidates (prior to Davis’s initiation of this litigation), no black lawyer has ever won election to either the Second Circuit or Leon County Courts.2 In each of these black-versus-white elections, the overwhelming majority of black voters supported the black candidates.3 Notwithstanding this political cohesion among black voters, however, white voters did not supply enough crossover votes for the black candidates to prevail, but instead provided overwhelming support to the white candidates.4 In 1992, for example, black voters in Leon County gave approximately 98% of their support to a black candidate, but a white candidate who received 68% of the white vote still won the election. As a result of this dynamic, racial block voting has become “a well-known political reality” in elections between black and white candidates for the Second Circuit and Leon County Courts. R5r-146 at 18.

This description of the voting patterns in the Second Circuit and Leon County receives further support from a review of “split-preference” elections, in which black and white voters have preferred different white candidates.5 In the eleven split-preference elections in the record involving either the Second Circuit or Leon County Courts, black voters have never succeeded in electing their first choice candidate. In nine of the split-preference elections, the black-preferred candidate lost outright to the white-*1418preferred candidate. In the tenth split-preference election, the black-preferred candidate won a primary election over the white-preferred candidate, but the black-preferred candidate then lost the general election to a white-preferred candidate. In the eleventh case, black voters’ first choice was a black candidate who lost to a white candidate in the primary election; only after this defeat did a majority of black voters settle on the white candidate who ultimately defeated the white-preferred candidate in the general election. Thus, black voters lack the ability to play even a “swing” role within the two election districts, whatever the race of the candidates.

At the same time, black voters cannot rely on the appointment process to offset the effects of racially polarized voting. Prior to 1992, when the district court first ruled that racial polarization existed in the districts at issue, no black person had ever received an appointment to either the Second Circuit or the Leon County Court.6 Moreover, while the appointment process has been a significant route to the bench in Leon County, election rather than appointment has been the primary path to judicial office for the Second Circuit.

On June 5, 1990, Davis brought a Section Two suit in the district court against defendant-appellees Chiles, et al. (“Chiles”) to challenge the legality of the at-large election systems for the Second Circuit and Leon County Courts. Specifically, Davis alleged that illegal vote dilution tainted elections for judgeships on the two courts. As her proposed remedy, Davis asked the court to impose a modified subdistricting plan.7 Under this proposed system, the two current at-large districts would be split into a combination of single- and multi-member subdistricts. In each of the new, smaller districts, voters would choose individual judges in competitive, post-numbered elections.8 Then, each successful subdistrict candidate would face a circuit- or county-wide retention vote by all of the citizens over whom they would exercise jurisdiction.9 Should any candidate chosen by a subdistrict fail to receive majority support in a retention vote, the Governor would have the power to fill the empty judgeship as he would any mid-term vacancy.

After conducting a bench trial, the district court ruled on September 3, 1992, that the judicial election systems in both the Second Circuit and Leon County violated Section Two.10 Specifically, the district court held *1419that Davis had proven each of the three Gingles factors that the Supreme Court has held are required to establish a prima facie case of vote dilution regarding a multi-mem-ber district: (1) that the black population in the two systems is sufficiently large and geographically compact to constitute a majority in a single-member district; (2) that black voters in the two systems are politically cohesive; and (8) that whites in the at-large districts vote sufficiently as a block to enable them usually to defeat the black voters’ preferred candidate.11 See Gingles, 478 U.S. at 50-51, 106 S.Ct. at 2766-67.12

Shortly after its 1992 ruling, however, the district court set aside its judgment and stayed farther proceedings during the 1993 session of the Florida legislature in order to allow the state to develop a remedy.13 Subsequently, the district court extended its stay while it awaited this court’s en banc decision in Nipper. After we delivered our opinions in Nipper and also in Southern Christian Leadership Conference v. Sessions, 56 F.3d 1281 (11th Cir.1995) (en banc) (“SCLC”), the district court conducted further hearings specifically directed at the efficacy and propriety of Davis’s proposed remedy. Then, on July 21, 1996, the district court rejected Davis’s remedial plan and granted judgment for Chiles. Although the district court noted that its second set of hearings had only reinforced its earlier finding of racially polarized voting, it now held that Davis had not met her prima facie burden of proposing an appropriate remedy under the first Gingles factor as our en banc court had recently interpreted in Nipper and SCLC. Specifically, the district court held that Florida’s interests in (1) maintaining the judicial model established by its Constitution, (2) preserving the territorial link between its judges’ electoral districts and jurisdictions, and (3) preventing the racial stigmatization of its judiciary collectively outweighed Davis’s interest in adopting her proposed remedy to ameliorate the effects of racially polarized voting. Then, the district •court ruled that it could not accept Davis’s modified subdistricting plan because her proposal constituted racially-conseious redistricting that was not justified by any compelling interest.

II. DISCUSSION

Davis challenges both the district court’s holding that Florida’s interest in preserving its judicial election system outweighs her interest in a remedy for racially polarized voting and its ruling that her proposed sub-districting remedy is impermissibly race-conscious. We discuss each issue in turn.

A. THE BALANCE OF INTERESTS REGARDING DAVIS’S PROPOSED REMEDY

As part of any prima facie case under Section Two, a plaintiff must demonstrate the existence of a proper remedy. See SCLC, 56 F.3d at 1289, 1294-97; Nipper, 39 F.3d at 1530-31 (plurality opinion), 1547 (Edmondson, J., concurring).14 In assessing a plaintiffs proposed remedy, a court must look to the totality of the circumstances, weighing both the state’s interest in maintaining its election system and the plaintiffs *1420interest in the adoption of his suggested remedial plan. See Houston Lawyers’ Ass’n. v. Attorney General of Tex., 501 U.S. 419, 426, 111 S.Ct. 2376, 2381, 115 L.Ed.2d 379 (1991). Although the district court found that black voters in the Second Circuit and Leon County suffered from racially polarized voting, it concluded that Florida’s interests, as previously described, in (1) protecting, the judicial model established by its Constitution, (2) preserving linkage between its judges’ jurisdictions and electoral bases, and (3) avoiding racial stigmatization of its judicial system outweighed Davis’s interest in her proposed remedy. As a result, the court held that Davis had not demonstrated the existence of an appropriate remedy and therefore had failed to set forth a Section Two violation under our circuit’s jurisprudence. We review the district court’s factual findings regarding Davis’s proposed remedy for clear error and its analysis of law de novo. See Gingles, 478 U.S. at 79, 106 S.Ct. at 2780-81; SCLC, 56 F.3d at 1291.

1. Interference with Florida’s Constitution

The district court detailed in its opinion a number of ways in which Davis’s proposed remedy would contravene the Florida Constitution. Although Davis does not now contest any of these constitutional problems under Florida law, we review the ramifications of Davis’s proposed remedy to establish the extent to which Davis’s plan would affect Florida’s interest in maintaining its judicial model.

First, Davis’s plan would require changes to Article V, Section 10 of the Florida Constitution, which directs that circuit and county judges be elected “by a vote of the qualified electors within the territorial jurisdiction of their respective courts.” Fla. Const, art. V § 10. Under Florida law, there is a difference between an “election,” which allows for competing candidates, and a “retention” vote for judicial office, which does not. See Fla. Stat. § 105.011(2) (distinguishing between an “election” and a “retention” in defining a “judicial office”). ■ Although Davis’s plan would permit citizens residing outside a sub-district to vote in retentions, it would deny them the right to take part in the “elections,” in contravention of Florida’s provision that they can participate in both. See Fla. Stat. §§ 105.061,105.051.

Second, the retention language of Article V, Section 10 includes only justices of the Florida Supreme Court or judges of a district court of appeal, as do Florida’s statutes. See Fla. Const, art. V § 10(a); Fla. Stat. §§ 105.051(2), 105.061. Accordingly, there is no constitutional or statutory basis in Florida law for the retention system Davis proposes; as the district court observed, Florida’s Constitution and statutes would have to change to allow for Davis’s proposed move from jurisdiction-wide, competitive elections to competitive, subdistrict elections followed by jurisdiction-wide retentions.

Third, Article V, Section 9 of the Florida Constitution provides that either the Supreme Court of Florida or the Florida legislature shall define the territories of Florida’s judicial circuits. See Fla. Const, art. V § 9. If a court were to require Florida to accept one of the subdistrict design plans contained in Davis’s subdistricting proposal, it would, by necessity, contravene this provision. This last constitutional problem is of lesser import, though, because it goes only to Florida’s manner of implementing its judicial model (ie., how it defines the borders of its circuits) -rather than to the nature of Florida’s judicial model itself (which Davis would dramatically change by requiring subdistrict-ing and circuit and county retentions, above).15

-In Nipper, we explained that a state has an interest in maintaining the judicial selection model established by its constitution. See Nipper, 39 F.3d at 1531 (plurality opinion), 1547 (Edmondson, J., concurring). In rejecting a plan to replace some of Florida’s at-large judicial election districts with single-member subdistricts, a majority of this *1421court joined then-Chief Judge Tjoflat’s holding that:

Implicit in this first Gingles requirement is a limitation on the ability of a federal court to abolish a particular form of government and to use its imagination to fashion a new system. Nothing in the Voting Rights Act suggests an intent on the part of Congress to permit the federal judiciary to force on the states a new model of government; moreover, from a pragmatic standpoint, federal courts simply lack legal standards for choosing among alternatives. Accordingly, we read the first threshold factor of Gingles to require that there must be a remedy within the confines of the state’s judicial model that does not undermine the administration of justice.

In judicial cases ... single-member districts may run counter to the state’s judicial model.

Id. at 1531 (plurality opinion) (emphasis added); see id. at 1547 (Edmondson, J., concurring); see also Holder v. Hall, 512 U.S. 874, 880, 114 S.Ct. 2581, 2585, 129 L.Ed.2d 687 (1994) (“In a § 2 vote dilution suit, along with determining whether the Gingles preconditions are met and whether the totality of the circumstances supports a finding of liability, a court must find a reasonable alternative practice as a benchmark against which to measure the existing voting practice.”) (quoted in Nipper, 39 F.3d at 1531-32 (plurality opinion)). Under Nipper, therefore, this court must carefully consider the impact that any remedial proposal would have on the judicial model enshrined in a state’s constitution or statutes.

Responding to Chiles’ reliance on Nipper, Davis contends that any viable remedy for racially polarized voting must necessarily effect some change in established electoral practices. Given both the Nipper precedent and the extent of the interference with Florida’s judicial model that Davis’s proposed remedy would require, however, Davis’s argument cannot dispel our duty to give weight to Florida’s right to maintain the integrity of its constitutional system. Although we are troubled by the apparent presumption in favor of status-quo polarization Nipper suggests, precedent requires that we consider Florida’s interest in maintaining its Constitution’s judicial selection system in determining whether Davis has proposed a permissible remedy. We therefore weigh this factor against imposition of Davis’s modified subdis-tricting plan.

2. Linkage Between Judges’ Jurisdictions and Electoral Bases

Territorial linkage between a trial judge’s jurisdiction and electoral base serves Florida’s interest in judicial accountability. See SCLC, 56 F.3d at 1296-97; Nipper, 39 F.3d at 1543-45 (plurality opinion). Were a judge to be answerable to an electorate smaller than his jurisdiction, the judge would have an incentive, however unethical, to engage in “home cooking,” favoring litigants from his election district over others. See SCLC, 56 F.3d at 1297. Thus, as the Supreme Court has observed, “the State’s interest in maintaining ... the link between a district judge’s jurisdiction and the area of residency of his or her voters ... is a legitimate factor to be considered by courts among the ‘totality of the circumstances’ in determining whether a § 2 violation has occurred.” Houston Lawyers’, 501 U.S. at 426, 111 S.Ct. at 2381. When, as in this case, there is no evidence that a state is administering its judicial election system in a racially discriminatory manner, the state’s interest in preserving linkage between judges’ jurisdictions and electoral bases is even weightier. See Nipper, 39 F.3d at 1544 (plurality opinion). Moreover, we have suggested that Florida has an interest in avoiding even the appearance that its judges may harbor “home cooking” biases. See id.16

Well aware of these precedents, Davis argues that her modified subdistricting plan would protect Florida’s linkage interests because each judge elected at the subdistrict level would face a retention vote by all of the citizens within his jurisdiction. Although we appreciate Davis’s creativity in attempting to *1422surmount the challenges that Nipper and SCLC pose to her suit, we must conclude that her proposed remedy would substantially break Florida’s linkage between its judges’ jurisdictions and electoral bases. First, as a practical matter, Davis’s proposed retention votes would place no real check on judges on the Second Circuit or Leon County Courts. Based on the history of elections involving incumbents on these two courts, the district court found that “in election systems limiting non-subdistrict voters to a right to vote for or against retention or for jurisdiction-wide approval or disapproval after an initial election, the powerful effect of incumbency in judicial elections would render that right virtually worthless.” R8-277 at 43. Since Davis has not challenged the district court’s assessment, and we have found no reason in the record to disagree with the district court’s factual findings concerning incumbency, we agree with the district court that imposition of Davis’s plan “would be akin to compelling the state to disenfranchise every voter residing in the two jurisdictions, but outside the subdistriet.”17 See id. at 44. Second, precedent requires us to recognize the risk that judges under Davis’s proposal would prove unaccountable to many voters even within their subdistricts because of continued racial block voting:

[I]n the judge’s own subdistrict, voters would be disenfranchised: In white subdis-tricts the voting power of blacks would be diluted; in black subdistricts the voting power of whites would be diluted. The likely effects of the loss of minority influence would be more pronounced in this context of a lone decisionmaker, a trial judge, who would lack input from the colleagues elected by the rest of the citizenry of the jurisdiction.

SCLC, 56 F.3d at 1297.18 In sum, Davis’s proposed remedy would substantially vitiate Florida’s linkage interest, another significant factor that we must weigh against imposing Davis’s proposal.

3. The Appearance of Justice

In Nipper, a plurality of this court insisted that any remedy for racially polarized voting in judicial elections must not undermine “the administration of justice.” Nipper, 39 F.3d at 1546 (plurality opinion). “By altering the current electoral schemes for the express purpose of electing more black judges,” the plurality wrote, the plaintiffs in Nipper risked “proclaiming that race matters in the administration of justice.” Id. at 1546 (plurality opinion). The plurality thus posed what it saw as a remedial impossibility:

The case at hand, therefore, presents a remedial paradox: A remedy designed to foster a perception of fairness in the administration of justice would likely create, by the public policy statement it would make, perceptions that undermine that very ideal. In the eyes of the public and litigants, at least, justice would not remain colorblind.

Id. Based on this language in Nipper, the district court held that Davis’s proposed remedy would improperly inject race into the administration of justice in the Second Circuit and Leon County.

Although we, too, are concerned that racial politics should not appear to taint Florida’s judicial system, we agree with Davis that her proposed remedy would be no worse in this regard than a judgment preserving the status quo. Today, voting in judicial elections for the Second Circuit and Leon County Courts is racially polarized, giving black candidates little hope of achieving judicial office. Whether or not we adopt Davis’s plan, therefore, race would “matter” within these jurisdictions; Davis’s scheme would simply exchange present misgivings about whites’ successes in at-large judicial *1423elections for new qualms from those who would view lawyers elected from Davis’s new subdistricts as representatives of racial groups rather than as neutral jurists.19 Further, we note that a majority of our court chose not to join the Nipper plurality’s discussion of this issue, so we are not bound by the plurality’s concept of a “remedial paradox.” In this case, at least, we do not think that fear of injecting race into judicial administration favors either side, so we do not weigh it as an interest for or against Davis’s proposed remedy.20

4. Weighing the Interests

As we observed above, a plaintiff must propose a viable and proper remedy in order to establish a prima facie case under Section Two. See SCLC, 56 F.3d at 1294-97; Nipper, 39 F.3d at 1530-31 (plurality opinion), 1547 (Edmondson, J., concurring). Before determining whether Chiles is violating Section Two, therefore, we must consider Florida’s interest in maintaining the challenged electoral system. See Houston Lawyers’, 501 U.S. at 426-27, 111 S.Ct. at 2381; SCLC, 56 F.3d at 1294-97; Nipper, 39 F.3d at 1530-31 (plurality opinion), 1547 (Edmondson, J., concurring). Of primary importance in this case, our adoption of Davis’s plan would require us to contravene Florida’s Constitution and to substantially break the link between the affected judges’ jurisdictions and electoral bases. In Nipper and SCLC, we ruled that a state’s interest in maintaining its judicial model and in preserving such linkage outweighed the plaintiffs’ interest in ameliorating the effects of racial polarization in at-large judicial elections. See SCLC, 56 F.3d at 1296-97; Nipper, 39 F.3d at 1543-45 (plurality opinion). Based on these precedents, we hold that Davis’s modified subdistricting plan would not be a proper remedy for the racial block voting that exists in the Second Circuit and Leon County-

Nonetheless, we are troubled by the analysis and the conclusion that our precedents appear to require in eases such as the one at bar. The Supreme Court has clearly and repeatedly held that Section Two applies to state judicial elections. See Chisom v. Roemer, 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d 348 (1991); Houston Lawyers’, 501 U.S. at 428, 111 S.Ct. at 2381. Moreover, the Court has explicitly stated that

[bjecause the State’s interest in maintaining an at-large, district-wide electoral scheme for single-member [judicial] offices is merely one factor to be considered in evaluating the “totality of the circumstances,” that interest does not automatically, and in every case, outweigh proof of racial vote dilution.

Houston Lawyers’, 501 U.S. at 427, 111 S.Ct. at 2381. In interpreting Chisom and Houston Lawyers’, our circuit in Nipper and SCLC has placed what now seems, in hindsight, to be an insurmountable weight on a state’s interest in preserving its constitution’s judicial selection system and in maintaining linkage between its judges’ jurisdictions and electoral bases. Together with Nipper, SCLC, and the additional case of White v. Alabama, we will with this decision have disallowed redistricting, subdistricting, modified subdistrieting, cumulative voting, limited voting, special nomination, and any conceivable variant thereof as remedies for racially polarized voting in at-large judicial elections. See Nipper, 39 F.3d at 1542-46 (plurality opinion) (rejecting subdistricting, redistriet-ing, and cumulative voting (and effectively precluding limited voting)), 1547 (Edmondson, J., concurring); SCLC, 56 F.3d at 1294-97 (rejecting redistricting and sub-districting); White v. Alabama, 74 F.3d 1058, 1072-73 (11th Cir.1996) (invalidating consent decree adding judgeships to be filled through *1424a special nomination commission). Given such rulings, neither we, nor Davis, nor Chiles have been able to envision any remedy that a court might adopt in a Section Two vote dilution challenge to a multi-member judicial election district. Thus, in this circuit, Section Two of the Voting Rights Act frankly cannot be said to apply, in any meaningful way, to at-large judicial elections. We recognize that this doctrinal development appears to conflict with the Supreme Court’s initial pronouncements on this subject in Chisom and Houston Lawyers’. This panel must, however, adhere to the reasoning of the en banc decisions of this court in Nipper and SCLC until either our circuit decides to revisit this issue en banc or we receive further guidance from the Supreme Court. See United States v. Woodard, 938 F.2d 1255, 1258 (11th Cir.1991) (per curiam) (“The law in this circuit is emphatic that ‘only a decision by this court sitting en banc or the United States Supreme Court can overrule’ ” a prior decision of this court.) (quoting United States v. Machado, 804 F.2d 1537, 1543 (11th Cir.1986)).

B. RACIALLY-CONSCIOUS SUBDIS-TRICTING

Although the district court found that Davis had failed to prove a Section Two violation because she had not proposed a permissible remedy under Nipper and SCLC, it ultimately did not rest its judgment on our Section Two precedents. Instead, the district court ruled that Davis’s subdistricting proposal would amount to unconstitutional racial gerrymandering. Because we hold that there is no statutory Section Two violation, we do not believe that a constitutional analysis of Davis’s proposed remedy should be necessary to our decision. Since the district court explicitly rested its decision on the constitutional issue, however, we think it necessary and appropriate to explain why we believe the district court to be in error.

As the district court correctly observed, a court must apply strict scrutiny to predominantly race-based redistricting or reapportionment plans. See, e.g., Miller v. Johnson, 515 U.S. 900, 920, 115 S.Ct. 2475, 2490, 132 L.Ed.2d 762 (1995). In order to determine whether race is the predominant factor underlying a particular district’s design, a court must find that a district-drawer has subordinated traditional, race-neutral districting principles (such as geographical compactness, contiguity, and respect for political subdivisions) to race. See, e.g., id. at 919, 115 S.Ct. at 2489. A court may base such a finding either on circumstantial evidence regarding a district’s shape and demographics or on direct evidence of a district-drawer’s purpose. See, e.g., id. at 916, 115 S.Ct. at 2488.

Applying these rules, the district court relied on the testimony of one of Davis’s experts, Dr. E. Walter Terrie, to hold that Davis’s remedy subordinated traditional redistricting criteria to race and therefore that strict scrutiny should apply.21 Then," because the court believed that Davis could not satisfy the first Gingles factor as would be required to prove a violation of Section Two, the court held that Davis could not point to a compelling interest to justify her plan. As a result, the district court held __ that Davis’s proposal would be unconstitutional under the Equal Protection Clause of the Fourteenth Amendment.

On appeal, Davis contends that the district court’s ‘legal analysis contradicts the Supreme Court’s holding in Gingles that a Section Two plaintiff must show that it would be possible to draw a majority-black district. Davis also argues that, regardless of the legal rule applied, the district court erred in concluding that race is the predominant factor underlying Davis’s modified subdistrict-ing plan. We review the district court’s findings of fact for clear error, cf. Miller, 515 U.S. at 917, 115 S.Ct. at 2488, and its analysis and application of the law de novo, see Gingles, 478 U.S. at 79, 106 S.Ct. at 2781.

Notwithstanding the polemics regarding race-based redistricting that pervade Chiles’s brief to this court,22 we agree *1425with Davis that the district court has misread the applicable law. Of course, the district court is correct that no government may use race as a predominant factor in drawing electoral districts without a compelling interest. See Miller, 515 U.S. at 920, 115 S.Ct. at 2490. The district court’s attempt to apply authorities such as Miller to this Section Two ease, however, is unpersuasive, because the Miller and Gingles /Nipper /SCLC lines address very different contexts. In Miller, the Supreme Court analyzed bi: zarrely-drawn Congressional districts in which there was “powerful evidence” that “every [objective districting] factor that could realistically be subordinated to racial tinkering in fact suffered that fate.” Miller, 515 U.S. at 919, 115 S.Ct. at 2490 (alteration in original) (quoting Johnson v. Miller, 864 F.Supp. 1354, 1384 (S.D.Ga.1994)). In Gingles, Nipper, and SCLC, however, the Supreme Court and this circuit examined at-large voting districts that, at least on their face, did not reflect racial gerrymandering but instead were alleged to support racially-polarized voting. Within this particular context, we have sensibly required that plaintiffs claiming illegal vote dilution show that minority voters are sufficiently geographically compact to allow construction of minority-majority districts; otherwise, minority voters’ failure to elect their preferred candidates does not reflect illegal vote dilution but rather the natural result of the dispersion of the minority group across an area in which white voters constitute a majority. See Gingles, 478 U.S. at 50-51, 106 S.Ct. at 2766-67; cf. Bush v. Vera, 517 U.S. 952, 958, 116 S.Ct. 1941, 1951, 135 L.Ed.2d 248 (1996) (“Strict scrutiny does not apply merely because redistricting is performed with consciousness of race.”).23 Under Gingles, a plaintiff such as Davis must demonstrate as part of her prima facie Section Two case that the relévant “minority group ... is sufficiently large and geographically compact to constitute a majority in a single-member district.” Gingles, 478 U.S. at 50, 106 S.Ct. at 2766. In interpreting this Gingles factor in the context of at-large judicial elections, we have further held that “inquiries into remedy and liability ... cannot be separated: A district court must determine as part of the Gingles threshold inquiry whether it can fashion a permissible remedy in the particular context of the challenged system.” Nipper, 39 F.3d at 1530-31 (plurality opinion), 1547 (Edmondson, J., concurring); SCLC, 56 F.3d at 1289, 1294-97 (“[P]laintiffs must show that an appropriate remedy can be fashioned.”). Thus, contrary to the district court’s holding, our precedents require plaintiffs to show that it would be possible to design an electoral district, consistent with traditional districting principles, in which minority voters could successfully elect a minority candidate. To penalize Davis, as the district court has done, for attempting to make the very showing that Gingles, Nipper, and SCLC demand would be to make it impossible, as a matter of law, for any plaintiff to bring a successful Section Two action.

Further, a review of the record reveals that Davis’s proposed subdistricts are not based predominantly on race. Significantly, Chiles has not been able to identify a single traditional redistricting principle which Davis’s subdistricting scheme would violate. Davis’s subdistricts are compact; they are contiguous; and they respect precinct borders. Cf. Shaw v. Reno, 509 U.S. 630, 647, 113 S.Ct. 2816, 2827, 125 L.Ed.2d 511 (1993) (discussing traditional districting principles) (“Shaw I ”). To refute the seeming inoffensiveness of Davis’s plan, Chiles and the district court point only to testimony by Terrie, an architect of Davis’s subdistricts, that “it was his charge to draw black majority subdistricts in the two ‘nucleuses within the circuit ... in which black voters tend to be concentrated.’” R8-277 at 34 (district court opinion). Although we agree with the district court that direct evidence that an election district was designed to dis*1426criminate against a particular racial group should trigger strict scrutiny, we do not believe that the record supports the conclusion that such a purpose motivated Terrie’s sub-districting plan. Certainly, race was a factor in Terrie’s process of designing the proposed subdistricts; under Gingles, Nipper, and SCLC, we require plaintiffs to show that it is possible to draw majority-minority voting districts, and plaintiff Davis and her expert Terrie wished to meet this burden. Throughout his testimony, though, Terrie insisted that race was not the predominant factor motivating his design process.24 Further supporting Terrie’s characterization of his work, he testified that he did not “begin in the majority black area and work out,” see R16-121, nor did he maximize the number of majority-minority subdistricts, see id. at 144. In fact, Teme testified that it would have been difficult for him to have drawn subdis-tricts for the Second Circuit and Leon County Courts without creating at least two majority-minority districts. See id. at 146. Absent some evidence belying Terrie’s characterization of his design process, Chiles cannot rely solely on criticism of Terrie’s motivations to block Davis’s proposed remedy. Given Terrie’s testimony, togethér with the unchallenged adherence of Davis’s proposed plan to traditional redistricting criteria, we conclude that the district court committed clear error in finding that Davis’s proposed remedy constitutes a racial gerrymander. Thus, we hold both that the district court misinterpreted the law regarding the role of race in assessing permissible remedies for violations of Section Two and that the district court incorrectly assessed the role that race played in the drawing of Davis’s proposed subdistricts.

III. CONCLUSION

In this case, Davis has presented persuasive evidence of racially polarized voting in elections for judgeships on the Second Circuit and Leon County Courts. Nonetheless, Davis has failed to propose a permissible remedy under our precedents. We agree with Davis that the district court erred in its holdings that Davis’s modified subdistricting plan would involve unconstitutional racial gerrymandering and inject race into Florida’s judicial administration. Nonetheless, our precedents compel us to conclude that Florida’s interests in maintaining its Constitution’s judicial election model and preserving linkage between its judges’ jurisdictions and electoral bases, considered together, outweigh Davis’s interest in the adoption of her proposed remedy. As a result, we hold that Davis has not proven a violation of Section Two. Therefore, we AFFIRM the district court’s judgment in Chiles’s favor.

concurring specially:

I concur in sections I, IIA, 1 throiigh 4, of the opinion for the court. It seems to me that Section IIB is simply unnecessary and therefore dicta with which I disagree but find no need to discuss. I do concur in footnote 22.