This is an appeal from the district court’s denial of a post-judgment motion to intervene for purposes of appeal. Having found that the at-large voting system used by Dade County, Florida (“Dade County”), to elect the members of its County Commission dilutes black and hispanic voting power in violation of section 2 of the Voting Rights Act of 1965, 42 U.S.C. § 1973, the district court barred further elections. *1475The intervenors, who were denied leave to intervene as parties below, contend that the district court erred in condemning the at-large system under section 2 as construed in Thornburg v. Gingles, 478 U.S. 30, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986). We reverse the denial of intervention and review the merits of the injunction. On the merits, we affirm the district court’s finding that the black and hispanic appellees proved their vote dilution claims that the present at-large system in Dade County, Florida, does not comply with the mandate of section 2 of the Voting Rights Act as amended in 1982.
PROCEDURAL BACKGROUND
On August 22, 1986, the black and his-panic appellees filed this action to challenge Dade County’s electoral system for county commissioners. The disputed system is part of a “federated plan” that provides for the allocation of authority between Dade County and the metropolitan governments. The County Commission is composed of a mayor and eight commissioners. Although the mayor may live anywhere in Dade County, each of the commissioners must reside in a different one of the eight residence districts. Each commissioner, however, must run for election county-wide. Thus, although Dade County is divided into various residence districts, the elections are conducted at-large and county-wide. See generally Meek v. Metropolitan Dade County, 908 F.2d 1540, 1542 (11th Cir.1990) (Meek I), cert. denied, — U.S. -, 111 S.Ct. 1108, 113 L.Ed.2d 217 (1991).
In Meek I, this court affirmed the district court’s conclusion that the appellees satisfied their burden as to the first Gingles prong, and also noted that the second Gin-gles prong was not in dispute. Meek I, 908 F.2d at 1549. After concluding that the district court granted summary judgment in favor of Dade County by applying erroneous legal principles, this court in Meek I reversed and remanded the case to the district court for the limited purpose of determining whether the appellees satisfied the third Gingles prong under the proper legal standards. Meek I, 908 F.2d at 1549 (specifically directing the district court to “decide whether Blacks or Hispanics have thus far usually elected preferred representatives” and “whether Blacks or Hispanics are impaired in their ability to elect representatives of their choice by the manner in which the voting districts are now drawn”).
On remand, the district court again found a genuine factual dispute over whether hispanics constitute a minority of eligible electors and are capable of being victimized by racial bloc voting, citing evidence that they represent 50.50 percent of Dade County’s population. The court also found evidence of persistent black electoral success which, in its view, also precluded summary judgment for the black plaintiffs on the third Gingles prong.
On December 11, 1991, intervenors R.H. Swann and David Sampson moved to intervene as defendants under Rule 24 of the Federal Rules of Civil Procedure, either permissively or as of right.1 Swann and Sampson are residents of Dade County and registered voters. The district court denied the motions on January 14, 1992. The court ruled that intervention as of right was inappropriate because the movants’ interests were “identical” to those of the existing official defendants, who therefore could be relied upon to represent them adequately. On the issue of permissive intervention, the court found that “[t]he duplicative nature of Applicants’ interest makes undue delay in the resolution of this case likely, and sheds no new light on the issues before the Court.” The court did, however, permit the movants “to intervene as amici.”
The Miami-Dade Chapter of the A. Phillip Randolph Institute (“Randolph”) and Northeast Citizens of Dade County *1476(“Northeast”) also moved on December 11, 1991, for leave to participate as amici curiae, eschewing party status “because they are not registered voters in Dade County nor entities with personal political rights that will be directly affected in any way by the determinations of the .Court in this case.” They did allege, however, that they were “comprised of registered electors in Dade County whose personal political rights and voting strength will be affected by the outcome of this litigation.” Randolph and Northeast were permitted to participate as amici.
On May 21, 1992, Swann and Sampson filed a renewed motion to intervene as parties “in-order to preserve their ability to take an appeal in the event that the Court renders a decision adverse to the Defendants and the County Defendants fail to take an appeal.” The next day, on May 22, 1992, the district court entered partial summary judgment for the plaintiffs as to the first two prongs of the Gingles test. On May 27, 1992, the district court denied the renewed motion to intervene without specifying the grounds for the denial.
After a bench trial lasting approximately three weeks, the district court entered an August 14, 1992 order, as amended on September 11, 1992, concluding that the black and hispanic appellees met their burden under the third Gingles prong, and established their vote dilution claim against Dade County’s at-large system for electing persons to the County Commission. Meek v. Metropolitan Dade County, Florida, 805 F.Supp. 967, 987, 994 (S.D.Fla.1992). On August 21, 1992, the County Commission held a special session and decided unanimously not to pursue an appeal. All four intervenors — Swann, Sampson, Randolph, and Northeast — filed motions to intervene as parties for purposes of appeal on September 1. That same day the district court denied the motions without explanation, although the court’s order did refer to the January 14, 1992, denial of intervention, which it attached as an exhibit. In the September 1, 1992 motion to intervene, the intervenors stated that “[t]he sole purpose of the requested leave to intervene is to pursue the appeal which has not been taken by Dade County” and represented that they were “willing to accept the County’s pleadings for such an appeal.”
All four intervenors filed notice of appeal on September 11, 1992. Appellees have filed a series of motions to strike and to dismiss the appeal. We ordered those motions carried with the case.
DISCUSSION
I. INTERVENTION
We have jurisdiction to review an order denying intervention as a matter of right because such a determination is a “final decision” under 28 U.S.C. § 1291 that “ends the litigation on the merits,” Catlin v. United States, 324 U.S. 229, 233, 65 S.Ct. 631, 633, 89 L.Ed. 911 (1945), for the intervenor, Brotherhood of RR. Trainmen v. Baltimore & O. RR., 331 U.S. 519, 524, 67 S.Ct. 1387, 1389-90, 91 L.Ed. 1646 (1947). Standing alone, an order denying permissive intervention is neither a final decision nor an appealable interlocutory order because such an order does not substantially affect the movant’s rights. Brotherhood of RR. Trainmen, 331 U.S. at 524, 67 S.Ct. at 1389-90. See also Stringfellow v. Concerned Neighbors in Action, 480 U.S. 370, 374-76, 107 S.Ct. 1177, 1181-82, 94 L.Ed.2d 389 (1987); United States v. Dallas County Comm’n, 850 F.2d 1433, 1442-43 (11th Cir.1988); cert. denied, 490 U.S. 1030, 109 S.Ct. 1768, 104 L.Ed.2d 203 (1989).
Even though the district court’s September 11, 1992 injunction order is appealable by the parties under 28 U.S.C. § 1292(a)(1), we would have no jurisdiction to review it in the absence of our conclusion that the district court improperly denied the intervenors’ motions to intervene. EEOC v. Eastern Airlines, Inc., 736 F.2d 635, 637 (11th Cir.1984) (explaining this court’s anomalous rule); United States v. Jefferson County, 720 F.2d 1511, 1515 (11th Cir.1983) (same). Before proceeding to the merits of the appeal, therefore, we first discuss our jurisdiction by reviewing the district court’s intervention orders un*1477der rule 24(a) or (b). FSLIC v. Falls Chase Special Taxing Dist., 983 F.2d 211, 215 (11th Cir.1993); Stallworth v. Monsanto Co., 558 F.2d 257 (5th Cir.1977).
A. Motion to Dismiss
Appellees contend that Swann and Sampson could have appealed immediately from the January 14 and May 27, 1992, orders denying intervention of right and urge that the unexplained failure to file notice of appeal until September 11, 1992, deprives us of jurisdiction to review the September 1, 1992, order. While we agree that the time for appeal from the January and May denials of intervention of right had expired, a fact that deprives us of authority to review them now, we reject appellees’ argument that the September 1 motions were merely subterfuges to reinstate the expired time for filing notice of appeal. Dade County’s subsequent decision to forego its right to appeal the district court’s injunction was a sufficient change in circumstances to justify a renewed motion for intervention with a new thirty-day time for filing notice of appeal under Federal Rule of Appellate Procedure 4(a). See United States EPA v. City of Green Forest, 921 F.2d 1394, 1401 (8th Cir.1990), cert. denied sub nom. Work v. Tyson Foods, — U.S. -, 112 S.Ct. 414, 116 L.Ed.2d 435 (1991); Hodgson v. United Mine Workers, 473 F.2d 118, 125-27 (D.C.Cir.1972). Accordingly, we deny the appellees’ motions to dismiss the appeal of the September 1 intervention order as untimely.
B. Standard of Review
Absent an express statutory grant of right to intervene, intervention as of right is appropriate whenever a “timely” applicant, having a direct, substantial and legally protectible interest in the subject matter of the suit, demonstrates that his ability to protect that interest may “as a practical matter” be affected by the litigation and that his interest may not be represented adequately by existing parties. Fed. R.Civ.P. 24(a); Falls Chase, 983 F.2d at 215; ManaSota-88, Inc. v. Tidwell, 896 F.2d 1318, 1321 (11th Cir.1990); Chiles v. Thornburgh, 865 F.2d 1197, 1213 (11th Cir.1989). A timely application for permissive intervention may be granted in the district court’s discretion, guided by considerations of undue delay or prejudice to the adjudication of the rights of existing parties. Fed. R.Civ.P. 24(b).
We have said that denial of intervention as a matter of right under rule 24(a) is reviewed “for error.” Falls Chase, 983 F.2d at 214-15; Walters v. City of Atlanta, 803 F.2d 1135, 1151 n. 16 (11th Cir.1986). Orders denying permissive intervention under rule 24(b) are reviewed for abuse of discretion. United States v. Dallas County Comm’n, 850 F.2d at 1443. Although we generally review denial of intervention under rule 24(a) for error, with subsidiary factual findings subject to review for clear error, our review of the district court’s determination of timeliness under both rule 24(a) and 24(b) is conducted under the abuse of discretion standard, Walters, 803 F.2d at 1151 n. 16. See also Howard v. McLucas, 782 F.2d 956, 960 (11th Cir.1986).
C.Adequate Representation
When applicants for intervention seek to achieve the same objectives as an existing party in the case, that party is presumed to represent the applicants’ interests adequately. Falls Chase, 983 F.2d at 215; Athens Lumber Co. v. Federal Election Comm’n, 690 F.2d 1364, 1366-67 (11th Cir.1982). This presumption, like any, serves to guide a court’s analysis of facts. It is not a substitute for facts, nor is it to be given any weight if the facts tend to contradict the presumed result. Rather, like the great majority of presumptions, at most it merely requires the presumed result unless some evidence is placed before the court tending to rebut it. See generally, e.g., County Court v. Allen, 442 U.S. 140, 99 S.Ct. 2213, 60 L.Ed.2d 777 (1979). When such evidence exists, resort to the presumption is inappropriate, and a court is obligated to make its own determination of whether the requirements of the rule are met.
*1478Representation is adequate “if no collusion is shown between the representative and an opposing party, if the representative does not have or represent an interest adverse to the proposed intervenor, and if the representative does not fail in the fulfillment of his duty.” Falls Chase, 983 F.2d at 215; United States v. United States Steel Corp., 548 F.2d 1232, 1236 (5th Cir.1977); Martin v. Kalvar Corp., 411 F.2d 552, 553 (5th Cir.1969). A finding that one of these circumstances exists should be made in light of the Supreme Court’s holding that “[t]he requirement of the Rule is satisfied if the applicant shows that representation of his interest ‘may be’ inadequate; and the burden of making that showing should be treated as minimal,” Trbovich v. United Mine Workers, 404 U.S. 528, 538 n. 10, 92 S.Ct. 630, 636 n. 10, 30 L.Ed.2d 686 (1972). See Falls Chase, 983 F.2d at 216 (citing Chiles, 865 F.2d at 1214). “Any doubt concerning the propriety of allowing intervention should be resolved in favor of the proposed intervenors because it allows the court to resolve all related disputes in a single action.” Falls Chase, 983 F.2d at 216.
We disagree with the district court’s conclusion that the county defendants were adequate representatives of the intervenors because they had identical interests. The intervenors sought to advance their own interests in achieving the greatest possible participation in the political process. Dade County, on the other hand, was required to balance a range of interests likely to diverge from those of the intervenors. For example, the County Commissioners had to consider the overall fairness of the election system to be employed in the future, the expense of litigation to defend the existing system, and the social and political divisiveness of the election issue. In addition, the County Commissioners were likely to be influenced by their own desires to remain politically popular and effective leaders. These divergent interests created a risk that Dade County might not adequately represent the applicants. E.g., Nash v. Blunt, 140 F.R.D. 400, 402-04 (W.D.Mo.1992) (three judge court). Therefore, Dade County was an inadequate representative for the interve-nors once it decided not to appeal, in spite of their shared litigation objective to uphold the at-large system.2
D. Timeliness
The district court did not discuss whether any of the intervention motions were untimely. Although we review a district court’s determination of timeliness only for abuse of discretion, we have no need to remand the case for the district court to further consider it because the record clearly shows that only one result would satisfy that standard of review. See In re Legel, Braswell Gov’t Sec. Corp. (Westchester County Sav. & Loan Ass’n v. Legel, Braswell Gov’t Sec. Corp.), 648 F.2d 321, 326 n. 8 (5th Cir. Unit B June 1981).
The timeliness analysis concerns not only the chronology leading up to the motion for intervention but also “all the circumstances,” including
1. The length of time during which the would-be intervenor actually knew or reasonably should have known of his interest in the case before he petitioned for leave to intervene ... 2. The extent of the prejudice that the existing parties to the litigation may suffer as a result of the would-be intervenor’s failure to apply for intervention as soon as he actually knew or reasonably should have known of his interest in the case ... 3. The extent of the prejudice that the would-be intervenor may suffer if his petition for leave to intervene is denied ... [and] 4. *1479The existence of unusual circumstances militating either for or against a determination that the application is timely.
Stallworth, 558 F.2d 257, 263-66 (5th Cir.1977). Measured by these standards, none of the motions to intervene could be considered untimely.
Stallworth makes plain that “absolute measures of timeliness,” such as “how far the litigation has progressed when intervention is sought” and “the amount of time that may have elapsed since the institution of the action” are not to be relied upon: “the mere fact that judgment already has been entered should not by itself require an application for intervention to be denied.” 558 F.2d at 266 (quotations omitted). Although our decisions have tended to disfavor intervention following final judgment, this is not such a case. See United States Steel, 548 F.2d at 1235 (recognizing that post-judgment motions to intervene are “ ‘ordinarily looked upon with a jaundiced eye’ ” because of their “strong tendency to prejudice existing parties to the litigation or to interfere substantially with the orderly process of the court”). As in Liddell v. Caldwell, 546 F.2d 768 (8th Cir.1976), cert. denied sub nom. St. Louis Bd. of Education v. Caldwell, 433 U.S. 914, 97 S.Ct. 2987, 53 L.Ed.2d 1100 (1977), in which the district court’s determination of untimeliness was reversed, “the district court, even as of this late date [September 1, 1992], [had] only partially approved specific plans for [remedying discrimination],” 546 F.2d at 771.
Addressing the issue of prejudice, which is the essence of the timeliness inquiry, Stallworth held that mere delay in determining the rights of existing parties is not a relevant consideration. Instead, the court must focus on any additional prejudice arising from the applicant’s delay in seeking intervention.
[T]he prejudice to the original parties to the litigation that is relevant to the question of timeliness is only that prejudice which would result from the would-be intervenor’s failure to request intervention as soon as he knew or reasonably should have known about his interest in the action. Although it is sometimes suggested that any prejudice that would result by intervention is relevant, this is incorrect. Whether allowing intervention will delay the progress of the case or prejudice the rights of the original parties is a factor which the court must consider under section (b) of Rule 24 [which provides that the court should consider “whether the intervention will unduly delay or prejudice the adjudication of the rights of the original parties”]. Since a similar provision is not included in section (a) of the Rule, it is apparent that prejudice to the existing parties other than that caused by the would-be intervenor’s failure to act promptly was not a factor meant to be considered where intervention was sought under section (a).
558 F.2d at 265.
The delay occasioned by this appeal is no different from the similar delay that would have arisen if Dade County itself had perfected an appeal. See Triax Co. v. TRW, Inc., 724 F.2d 1224, 1228 (6th Cir.1984).
On the other side of the equation is the fact that no other parties remain in the case to pursue the objective of defending the at-large system, with the result that denial of intervention irrevocably condemns that system, to the intervenors’ prejudice. “[T]he thrust of the inquiry must be the extent to which a final judgment in the case may bind the movant even though he is not adequately represented by an existing party.” Jefferson County, 720 F.2d at 1517. No doubt exists that the intervenors were “bound” by the order enjoining further elections under the at-large plan, even though they were neither parties nor (as we have held) adequately represented by parties. See Black & White Children v. School Dist., 464 F.2d 1030 (6th Cir.1972) (per curiam), cited in Jefferson County, 720 F.2d at 1518 n. 17.
The substantial public interest at stake in the case is an unusual circumstance militating in favor of intervention. Howard v. McLucas, 782 F.2d at 960 (finding denial of intervention to be an abuse of discretion in part because “intervenor-appellants raise *1480an important question regarding the legality of the race-conscious promotional remedy”). Cf. Swanson Mining Corp. v. Federal Energy Regulatory Comm’n, 790 F.2d 96, 105 (D.C.Cir.1986).
Therefore, based on our review of the totality of circumstances in this case, we conclude that the denial of intervention on timeliness grounds would have been an abuse of discretion.
E. Legally Protectible Interest
Appellees have moved to strike and dismiss the appeal on the ground that the intervenors lack sufficiently substantial legally protectible interests. The motions have no merit. In this circuit, a movant who shows standing is deemed to have a sufficiently substantial interest to intervene. Howard v. McLucas, 782 F.2d at 959. The intervenors sought to vindicate important personal interests in maintaining the election system that governed their exercise of political power, a democratically established system that the district court’s order had altered. As such, they alleged a tangible actual or prospective injury and did not merely challenge unlawful conduct in the abstract. See generally, e.g., Lujan v. Defenders of Wildlife, — U.S. -, -, 112 S.Ct. 2130, 2144, 119 L.Ed.2d 351 (1992). Moreover, we reject appellees’ contention that the intervenors had only non-justiciable generalized grievances simply because they asserted interests widely shared by others. Allen v. Wright, 468 U.S. 737, 756-60, 104 S.Ct. 3315, 3327-29, 82 L.Ed.2d 556 (1984). If we accepted such an argument, we would be forced to conclude that most of the plaintiffs also lack standing, a conclusion foreclosed by the many cases in which individual voters have been permitted to challenge election practices. See, e.g., Whitcomb v. Chavis, 403 U.S. 124, 91 S.Ct. 1858, 29 L.Ed.2d 363 (1971); Baker v. Carr, 369 U.S. 186, 82 S.Ct. 691, 7 L.Ed.2d 663 (1962).
Appellees also argue that Randolph and Northeast have conceded they lack sufficient interest by admitting that they have no personal political rights. Looking to the intervention motions on their face, however, we find a sufficiently clear statement that these organizations asserted standing as representatives of members who do have such rights. No formulaic incantation of the requisites for representational standing is required, Church of Scientology v. Cazares, 638 F.2d 1272, 1276-78 (5th Cir. Mar. 1981), because we construe the pleadings liberally in favor of standing, Pennell v. City of San Jose, 485 U.S. 1, 7, 108 S.Ct. 849, 855, 99 L.Ed.2d 1 (1988). The contention that these entities somehow conceded a lack of standing by admitting that they themselves cannot vote is specious. See generally Hunt v. Washington State Apple Advtg. Comm’n, 432 U.S. 333, 343, 97 S.Ct. 2434, 2441, 53 L.Ed.2d 383 (1977).
For the foregoing reasons, the appellees’ motions to strike and dismiss the appeal are DENIED. The district court’s September 1, 1992 order denying the intervenors’ motion to intervene as a matter of right is REVERSED. We therefore retain jurisdiction and review the merits.3
II. THE MERITS
In challenging an at-large system under the 1982 amendment to section 2 of the Voting Rights Act, a plaintiff must prove that: (1) the minority group is “sufficiently large and geographically compact to constitute a majority in a single-member district”; (2) the minority group is “politically cohesive”; and (3) the majority votes “sufficiently as a bloc to enable it — in the absence of special circumstances, such as the minority candidate running unopposed— usually to defeat the minority’s preferred candidate.” Thornburg v. Gingles, 478 U.S. 30, 50-51, 106 S.Ct. 2752, 2766-67, 92 L.Ed.2d 25 (1986).
*1481On appeal, the intervenors make the following claims that the district court erred in determining whether the appellees established their vote dilution claim: (1) the district court clearly erred in determining that a “keen hostility” existed between blacks and hispanic voters of Dade County; (2) the district court misapplied the applicable law in determining racially polarized voting based on analysis of only seven elections over three election years, encompassing a four-year period; (3) the district court erred in not also considering county school board elections in determining the existence of racially-polarized-voting; (4) the district court erred in concluding that the election of incumbents who were first appointed constituted a “special circumstance” which explains the evidence of some successful black candidates; (5) the district court analyzed the third Gingles prong under the erroneous notion that at-large systems are per se illegal and that section 2 assures racial minorities proportional representation; (6) the district court clearly erred in relying on the hispanic appellees’ expert regression and homogenous precinct analyses which were based on data that classified only foreign-born hispanics as hispanics; (7) the district court abused its discretion in rejecting data derived from survey results included in a pretrial stipulation, and thus clearly erred in determining that hispanics did not constitute a majority of registered voters; (8) the district court clearly erred in rejecting Dade County’s lack of racial bias affirmative defense; and (9) the district court erred in not requiring the hispanic appel-lees to demonstrate that racially polarized voting would continue to deny their effective participation in the future.4 We find the intervenors’ contention that a plaintiff must show that a section 2 violation continues into the future clearly without merit and warrants no further discussion. We address the intervenors’ remaining contentions separately.
This court reviews the district court’s ultimate factual finding of vote dilution under the clearly erroneous standard. Hall v. Holder, 955 F.2d 1563, 1566 (11th Cir.1992). This court, however, may correct errors of law or findings of fact based on misconceptions of the law. Hall, 955 F.2d at 1566.
A. Keen Hostility Between Blacks and Hispanics
The intervenors argue that the district court clearly erred in finding that a “keen hostility” exists between blacks and hispanics in Dade County. The district court found that a “keen hostility” exists between blacks and hispanics in Dade County based on its independent examination of the trial evidence. Meek, 805 F.Supp. at 981 & n. 30 (citing Meek I); see Meek I, 908 F.2d at 1545 (concluding that “the social reality is that Black and Hispanic voters are hostile toward each other in the electoral arena” based on the district court’s findings in its October 5, 1988 Order, as amended in its Order of November 6, 1988, granting summary judgment in favor of Dade County).
This court will not disturb a district court’s findings of fact under the clearly erroneous standard unless it is left “with the definite and firm conviction that a mistake has been made” after making all credibility choices in favor of the fact-finder’s choice, in light of the record as a whole. Maddox v. Claytor, 764 F.2d 1539, 1545 (11th Cir.1985).
After reviewing the trial record, we hold that the district court’s finding that a “keen hostility” exists between blacks and hispanics in Dade County is not clearly erroneous. The record shows that persons of diverse backgrounds, like black attorney *1482and longtime resident, Harold T. Smith, businessman and politician, Maurice Ferre, and Republican Party Chairman, Jeb Bush, generally agreed that effective use of subtle and overt racial and ethnic appeals is a common tactic used during campaigns in the black and hispanic communities. Smith testified that he has long observed racial and ethnic hostility between hispanics and blacks in Dade County, which has manifested itself in competition for employment as well as political offices.
Smith, Ferre, and Bush testified about the political divisiveness in the black and hispanic communities based on a candidate’s perceived affiliation with public figures such as the Reverend Jesse Jackson, African National Congress President Nelson Mandela, Cuban leader Fidel Castro, and former U.S. Ambassador to the United Nations Andrew Young. According to Smith, Ferre, and Bush, perceived affiliation with these national and international figures often results in a dramatic reduction in a candidate’s ability to raise funds within some racial and ethnic communities in Dade County. For example, Smith testified that if a candidate is perceived to be affiliated with Jesse Jackson, the hispanic community would refuse to give that candidate contributions and would give money to that candidate’s opponents.
Although the record also contains testimony tending to minimize the level of conflict between blacks and hispanics in Dade County, we conclude that the district court did not clearly err in finding that “keen hostility” exists between blacks and hispan-ics in Dade County, based on the extensive testimony supporting its finding. See Anderson v. City of Bessemer City, N.C., 470 U.S. 564, 575, 105 S.Ct. 1504, 1512, 84 L.Ed.2d 518 (1985) (recognizing that a trial court’s finding based on a decision to credit the testimony of one of two or more witnesses, which extrinsic evidence does not contradict, can virtually never be clear error).
B. Number of Elections Considered
The intervenors also challenge the district court’s conclusion that “the Black plaintiffs’ analysis of seven elections over the last three election years, which encompassed a four-year period, is sufficient on the circumstances of this case to determine that racially-polarized-voting impairs blacks in their ability to elect their preferred representatives.” See Meek, 805 F.Supp. at 983-84. The intervenors argue that the analysis of only three election years is insufficient as a matter of law under Gin-gles. The intervenors rely on that portion of the Gingles ruling constituting the judgment, not the opinion, of the Court reversing the district court’s finding of dilution in one district where evidence showed sustained minority success in six elections. See Gingles, 478 U.S. at 77, 106 S.Ct. at 2780 (Brennan, J., with whom White, J., joined) (recognizing that evidence relating to only three elections is insufficient to rebut evidence of sustained success of black candidates in six elections which resulted in persistent proportional representation for blacks in one of the six challenged districts); Gingles, 478 U.S. at 104-05, 106 S.Ct. at 2794-95 (O’Connor, J., concurring in judgment, with whom Burger, C.J., and Powell and Rehnquist, JJ., joined) (agreeing that the district court clearly erred in not giving weight to the evidence that black candidates achieved sustained success in one of the six challenged districts).
The intervenors’ argument that Gingles stands for the proposition that analysis of only three election years is insufficient as a matter of law is directly contrary to the explicit holding in the opinion of the Gingles Court. See Gingles, 478 U.S. at 57 n. 25, 106 S.Ct. at 2770 n. 25 (recognizing that “the number of elections that must be studied in order to determine whether voting is polarized will vary according to pertinent circumstances,” such as the number of elections in which the minority group has sponsored candidates). In this case, the district court determined the sufficiency of analyzing seven elections over the last three election years, after considering relevant circumstances including the recent growth in the hispanic population of Dade County and the “ ‘keen hostility’ ” between blacks and hispanics. See Meek, 805 *1483F.Supp. at 983 (finding that the hispanic population in Dade County was marginal during the 1960s and 1970s, as compared to the 1990 census data demonstrating that hispanics constituted a majority of the total population of Dade County). Because it did not clearly err in finding that the recent and tremendous growth in the hispanic population of Dade County, along with the “keen hostility” between blacks and hispan-ics, made recent elections more probative of the existence of racial bloc voting in Dade County today, we find no error in the district court’s determination that analysis of seven elections over a three-election year period is sufficient under the unusual circumstances of this case. See Gingles, 478 U.S. at 76, 106 S.Ct. at 2779 (recognizing that nothing in section 2 or its legislative history prohibits a district court “from deciding on the basis of all the relevant circumstances to accord greater weight to blacks’ relative lack of success over the course of several recent elections”).
C. Type of Elections Considered
The intervenors further contend that the district court erred in not considering Dade County School Board elections in determining the existence of racially polarized voting. The intervenors, however, fail to demonstrate that Dade County litigated this issue. We find no evidence in the record that Dade County presented the district court with evidence suggesting a legally significant correlation between voting in the partisan School Board elections and voting in the non-partisan County Commission elections. See generally Lucas v. Townsend, 967 F.2d 549, 552 (11th Cir.1992) (recognizing that “drawing valid inferences and conclusions about voter behavior from statistics requires the selection of relevant elections for analysis” and rejecting an assertion that the district court erred in refusing to equate candidate and referenda elections based on testimony indicating that the two types of elections are not readily interchangeable). As appellants for purposes of appeal, the intervenors occupy the position of Dade County, and are confined to raising those issues on appeal that Dade County presented to the district court. See Garrity v. Gallen, 697 F.2d 452, 457 (1st Cir.1983) (holding that an intervenor who seeks to step into the shoes of the defendants for purposes of appeal is confined to those issues that the defendant raised in the district court); Petition of Geisser v. United States, 554 F.2d 698, 705 n. 6 (5th Cir.1977) (recognizing that interve-nors must accept the proceedings as they find them). We find no indication in the record, especially the Revised Pre-Trial Stipulation on issues of law and facts to be litigated, that Dade County litigated the issue of the applicability of School Board elections to a racial bloc determination. Thus, we hold that the intervenors are precluded from raising this issue for the first time on appeal.
D. Incumbency as a Special Circumstance
The intervenors also claim that the district court erred in recognizing incumbency as a special circumstance which explains the success of some black candidates in County Commission elections. They argue that Gingles requires that appellees prove that the incumbents who successfully ran for reelection did not represent the interest of black voters before a special circumstance is established.
The holding in Gingles is clearly contrary to the intervenors’ assertion that the special circumstance explanation is inapplicable to explain the election of incumbents who are, in fact, the preferred representative of the minority community. In explaining the third prong of the threshold analysis, the Gingles Court held that “the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it — in the absence of special circumstances, such as the minority candidate running unopposed [citation omitted] — usually to defeat the minorities’ preferred candidate.” Gingles, 478 U.S. at 51, 106 S.Ct. at 2766-67. When stated in the alternative, the Gingles Court’s formulation of the third prong of the threshold test demonstrates that the presence of special circumstances such as incumbency or lack of opposition may explain the success of *1484the minorities’ preferred candidate in some elections. Moreover, the Gingles Court recognized the incumbency of a successful black candidate in the 1978 primary in North Carolina’s Senate District 22 as a special circumstance explaining success, even where the candidate was the preferred black candidate receiving eighty-seven percent of the black votes that were cast. See Gingles, 478 U.S. at 60 n. 29, 80, 106 S.Ct. at 2771-72 n. 29, 2781. Therefore, we reject the intervenors’ contention that Gingles requires a section 2 plaintiff to show that a reelected incumbent is not their preferred candidate before a “special circumstance” explaining electoral success is properly recognized.
In this case, the district court found that incumbency, following initial appointments, explained the success of three of the four black Commissioners elected to the County Commission with a majority of the black vote. The district court noted that the fourth successful black candidate who won with a majority of the black vote was Earl Carroll, the first black to serve on the County Commission. In the six most recent County Commission elections, the black candidate who received a majority of the black vote lost in every instance except one. Barbara Carey, a two-term incumbent who was initially appointed to the County Commission, won her race. Accordingly, we hold that under the circumstances of this case the district court did not commit clear error in finding that the special circumstance of incumbency, after initial appointment, explains the evidence of success for some black candidates for the County Commission.
E. No Right to Proportional Representation
The intervenors also argue that the district court analyzed the appellees’ vote dilution claims under the erroneous assumption that at-large systems are per se illegal and that section 2 assures racial minorities proportional representation. In support of this contention, the intervenors rely on the district court’s mention of the statistics demonstrating that the black population has been unable to elect a second representative to the County Commission, based on the consistent defeats of black preferred candidates for a second seat between 1968 and 1990. See Meek, 805 F.Supp. at 982, 987. We are unconvinced that the district court analyzed the vote dilution claim under a misconception of the law merely because the district court noticed the fact that the black population has consistently been unsuccessful in winning a second seat on the County Commission. See generally Gingles, 478 U.S. at 84, 106 S.Ct. at 2784 (O’Connor, J., concurring in judgment) (recognizing that “any theory of vote dilution must necessarily rely to some extent on a measure of minority voting strength that makes some reference to the proportion between the minority group and the electorate at large”). To the contrary, we are persuaded that the district court applied the correct legal standard that “proportional representation is not and has never been required by the courts, and that at-large elections are not automatically invalidated under the ‘results’ standard.” See Meek, 805 F.Supp. at 972 (citation omitted). The district court’s opinion clearly shows an analysis of the appellees’ claims of vote dilution with detailed discussion of the Gingles threshold requirements and the totality of circumstances. Hence, we reject the intervenors’ contention that the district court assumed that at-large systems are per se illegal and that section 2 requires proportional representation.
F. Expert Analyses Based on Flawed Data
The intervenors contend that the district court clearly erred in determining that the hispanic voters established a vote dilution claim. The intervenors argue that the district court improperly relied on Dr. Henderson’s bivariate regression analysis, which is flawed because Dr. Henderson analyzed the existence of racially polarized voting using data which classified only foreign-born hispanics as “hispanics.” The intervenors argue that Dr. Henderson’s analysis is not competent to describe political cohesion among hispanics generally in Dade County, because Dr. Henderson’s *1485data was only relevant to the existence of polarization between foreign-born hispanics and others, not all hispanics and others.
At trial, Dr. Henderson specifically responded to the flaw in his data based on Dade County’s practice of identifying his-panics as only those persons born in Spanish-speaking countries and the Commonwealth of Puerto Rico. Dr. Henderson testified that
the extremes of separation in terms of the high proportion of Hispanics voting this way and the high proportion of non-Hispanics voting some other way, just that such a difference even, for example, if somehow, we were able to make an estimate of the number of persons counted as non-Hispanic who should have properly been counted as Hispanic had moved, I don’t think it would make any difference whatsoever in the conclusion reached that there was ethnically polarized voting.
Moreover, Dr. Henderson pointed out that census data did not provide a flawless alternative because of its infrequency; and that classifying hispanics based on surnames would lead to flawed data because of the variation in “hispanic” surnames and the inability to identify hispanics who have married or changed their names. Dr. Henderson concluded that “if we waited around for the perfect data set to appear, we would never, in this country, be able to decide a voting rights case. So we deal with imperfect data sets as best we can.” Dade County’s expert, Dr. Stanley, agreed that the use of census data, hispanic surname classification, and exit polls would also result in flawed data, and admitted that the search for a better data set is illusory.
Based on the concessions of Dade County’s expert regarding the illusory nature of obtaining a flawless data set in this case, and the failure of the intervenors to point to extrinsic evidence contradicting Dr. Henderson’s testimony, we hold that the district court did not clearly err in crediting the analysis and conclusions of Dr. Henderson regarding the existence of racially polarized voting between “hispanics” and “non-hispanics” in Dade County. See Anderson, 470 U.S. at 575, 105 S.Ct. at 1512 (recognizing that a trial court’s finding based on a decision to credit the testimony of one of two or more witnesses, which extrinsic evidence does not contradict, can virtually never be clear error).
G. Relief From Pretrial Stipulation
The intervenors argue that the district court abused its discretion in declining to credit estimates derived from a pretrial stipulation, and thus clearly erred in determining that hispanics do not constitute a majority of registered voters in Dade County. The intervenors argue that Dade County stipulated to the accuracy of the results in the Sheskin Survey, which is based on interviews of 791 hispanic adults. Based on a mathematical derivation of data in the Sheskin Survey, Dade County alleged that hispanics represent the majority of all registered voters in Dade county. The intervenors rely on Dade County’s derived estimates to argue that the appellees stipulated that hispanic voters constitute between 51.96 percent and 53.6 percent of Dade County’s total 673,838 registered voters.
The district court found that the hispanic appellees did not agree with Dade County that the Sheskin Survey specifies that his-panics are a majority of registered voters in Dade County, even though the hispanic appellees did stipulate to the accuracy of the survey. See Meek, 805 F.Supp. at 989 n. 41. Based on the disagreement of the parties about the scope of the stipulation concerning the Sheskin Survey, the district court declined to credit Dade County’s figures indicating that hispanics constituted a majority of all registered voters in Dade County. See Meek, 805 F.Supp. at 989 n. 41. Instead, the district court found that hispanics, including U.S.-born hispanics, do not constitute a majority of all registered voters in Dade County based on the testimony of the Supervisor of Elections for Dade County. Meek, 805 F.Supp. at 989 n. 41. The Supervisor of Elections testified about his affidavit dated March 17, 1988, which stated that hispanics, including his-panics born in the United States, comprise *1486approximately 32.96 percent of the registered voters in Dade County. The Supervisor of Elections also testified that he did not believe that hispanics constituted a majority of all registered voters in Dade County at the time of trial.
In determining whether to hold a party to its stipulation, the district court has broad discretion. Morrison v. Genuine Parts Co., 828 F.2d 708, 709 (11th Cir.1987), cert. denied, 484 U.S. 1065, 108 S.Ct. 1025, 98 L.Ed.2d 990 (1988). Although we find it is a close question, based on our careful review of the disputed Revised Pre-Trial Stipulation, we find no abuse of discretion in the district court’s decision to relieve the hispanic appellees from Dade County’s derived estimates indicating that hispanics are a majority of registered voters. We note the following “reservation” in the stipulation:
According to the survey attached hereto as Exhibit 1 as of May, 1991, voting age Hispanics in Dade County had attained a citizenship level of 61.1 percent. Under Florida law, these citizen Hispanics are qualified to become voters. Removing the 38.9 percent of the voting age Hispanics in Dade County who are not citizens from the pool of eligible voters reduces the share of citizen voting age Hispanics to an electoral minority.
See “Concise Statement of Facts Which Have Been Stipulated to as to Accuracy and Authenticity, But For Which One Party Has Made a Reservation” in Revised Pre-Trial Stipulation (Feb. 21, 1992). A derivation of the data referenced in the “reservation” would result in estimates that 452,526 hispanics of voting age are citizens, thus making hispanics an electoral minority with only 30.8 percent of the total 1,469,084 voting age population in Dade County.5 Thus, the district court did not abuse its discretion in relieving the appel-lees from estimates derived from the stipulated Sheskin Survey, where the stipulation shows the appellees reserved rights regarding what the survey indicates about hispan-ics being an electoral majority.
The intervenors also argue that the district court clearly erred in crediting the testimony of the Supervisor of Elections, because he possessed no personal knowledge of the methodology which another elections department official, who died before trial, utilized in arriving at the 32.96 percent figure. In addition, the intervenors argue that the Supervisor of Elections’s testimony that hispanics constituted a minority of registered voters at the time of trial was imprecise and without an adequate foundation. Although we agree that the Supervisor of Elections’s testimony did not represent the best evidence, we find nothing to suggest that Dade County argued that it was not admissible evidence, nor do we find that Dade County presented any contrary evidence, except the extrapolated estimates from the Sheskin Survey. Because of our previous holding that the district court did not abuse its discretion in not crediting the extrapolated estimates which Dade County derived from the Shes-kin Survey, we conclude that the district court did not clearly err in crediting the Supervisor of Elections’s testimony, the only evidence presented at trial on the issue of whether hispanics represented a majority of all registered voters in Dade County-
H. Lack of Racial Bias Affirmative Defense
The intervenors contend that the district court acted under a misconception of the law in failing to consider the lack of racial bias affirmative defense in light of the totality of the circumstances. The interve-nors admit that the evidence establishes that County Commission candidates in Dade County have successfully engaged in a tactic which the district court describes as “ ‘guilt by association’ ” to defeat black candidates. See Meek, 805 F.Supp. at 981-82 & n. 32 (explaining that the use of the tactic “ ‘guilt by association’ ” results in *1487dividing the black community, which strongly supports figures like Jesse Jackson and Nelson Mandela, and some Cuban and Jewish communities who oppose them). The intervenors argue, however, that evidence of the use of the “guilt by association” tactic does not support a finding that Dade County politics currently involves subtle racial appeals. Rather, the interve-nors argue that the evidence on the extensive use of the “guilt by association” tactic demonstrates only that black candidates suffered defeat because of their indirect link to Fidel Castro, and not because Jesse Jackson and Nelson Mandela are black.
We first note that this circuit remains divided on the issues of whether appellees can make out a section 2 violation simply based on their satisfaction of the three core Gingles factors and whether defendants can raise a lack of racial bias defense under the totality of circumstances after the ap-pellees have satisfied the threshold Gingles factors. See Solomon v. Liberty County, 899 F.2d 1012, 1013 (11th Cir.1990) (per curiam); Solomon, 899 F.2d at 1017, 1021 (Kravitch, J., specially concurring); Solomon, 899 F.2d at 1033 (Tjoflat, C.J., specially concurring) (evenly divided en banc court), cert. denied, 498 U.S. 1023, 111 S.Ct. 670, 112 L.Ed.2d 663 (1991); see also Hall, 955 F.2d at 1568 n. 9 (declining to resolve the issue dividing the court in Solomon); Meek I, 908 F.2d at 1544, 1549 (same). We need not decide the unresolved issue in Solomon, because we hold that the district court did not clearly err in finding racial bias in voting for County Commission candidates.
We reject the intervenors’ argument that the district court clearly erred in finding that racial bias motivates voting in County Commission elections. As noted in our discussion of the “keen hostility” that exists between blacks and hispanics in Dade County, the record contains extensive evidence that racial bias has a substantial influence on voting in County Commission elections. In light of the extensive testimony that opponents used this tactic precisely because of its effect of dividing the support of the black, hispanic, and Jewish communities along racial and religious lines, the intervenors’ argument that the extensive use of the “guilt by association” tactic in county commission elections does not demonstrate racial bias is meritless. The district court noted the testimony of one candidate in a 1990 race who ran radio advertisements on Cuban radio claiming that a black opponent was strongly supportive of Jesse Jackson, knowing that the claims would make it difficult for the black opponent to gain hispanic and Jewish support, and admittedly not knowing whether the claims of affiliation with Jesse Jackson were true. See Meek, 805 F.Supp. at 982 n. 32. In addition, the testimony of the Executive Director of Dade County’s Fair Campaign Practices Committee, Andrew Rosen-blatt, supports the finding that use of racial or éthnic appeals is common during campaigning in Dade County. Rosenblatt testified via deposition that racial or ethnic appeals during campaigning in Dade County often divides the community and that “those divisions are often — and animosities that are generated in the context of a polarizing political campaign are often times not easily forgotten.”
The district court’s consideration of the totality of circumstances also supports its finding that racial bias motivates voting in County Commission elections. Based on our review of the record, we conclude that the district court did not clearly err in finding that Dade County’s history of official discrimination, along with the presence of other Senate Report factors, supported a finding of racial bias motivating voting in Dade County. See S.Rep. No. 97-417, 97th Cong., 2d Sess. 28-29 (1982), reprinted in 1982 U.S.C.C.A.N. 177, 206-07. Moreover, as discussed previously, the district court did not clearly err in finding the existence of “keen hostility” between blacks and his-panics in Dade County, nor in finding that racially and ethnically polarized voting in Dade County usually defeats the preferred candidates of the black and hispanic communities. See United States v. Marengo County Commission, 731 F.2d 1546, 1567 (11th Cir.) (recognizing that “the surest indication of race-conscious politics is a pattern of racially polarized voting”), cert. de*1488nied, 469 U.S. 976, 105 S.Ct. 375, 83 L.Ed.2d 311 (1984); see also United States v. Dallas County Commission, 850 F.2d 1430, 1439 (11th Cir.1988). cert. denied, 490 U.S. 1030, 109 S.Ct. 1768, 104 L.Ed.2d 203 (1989) (same).
Based on the evidence of racial bloc voting, the presence of Senate Report factors, “keen hostility” between blacks and hispan-ics, and the use of overt and subtle racial or ethnic appeals during county commission campaigns, we conclude that the district court did not clearly err in finding that racial bias substantially motivates voting behavior in county commission elections. Accordingly, we affirm the district court's finding that Dade County failed as a matter of fact to establish the lack of racial bias in this case.
CONCLUSION
We deny the appellees’ motions to strike and dismiss the appeal. We reverse the district court’s September 1, 1992 order denying the intervenors’ motions to intervene as a matter of right.
Having concluded that the intervenors failed to establish that the district court clearly erred in making subsidiary findings of fact, and failed to demonstrate that the district court applied the wrong legal principles, or made findings based on misconceptions of the law, we affirm the district court’s ruling that both the black and hís-pame appellees established their claims that the present at-large system in Dade County, Florida, does not comply with the mandate of section 2 of the Voting Rights Act as amended in 1982. Accordingly, on the merits WE AFFIRM.
REVERSED IN PART, AFFIRMED IN PART.
. Two other individuals, Roland Rolle and George Berlin, also joined Swann and Sampson in the December 11, 1991 motion to intervene as well as in the May 21, 1992 renewed motion to intervene, which is discussed infra. However, Rolle and Berlin did not join in the September 1, 1992 motion to intervene the denial of which is at issue in this appeal, and they have not filed any appearance in this Court.
. We do not now hold that one who has an interest in ongoing litigation in which a public body is adequately representing that interest may simply wait until judgment is entered and intervene on appeal simply because the public body decides not to appeal. Under the unique circumstances of this case, although formal intervention had not been allowed, two of these intervenors had sought repeatedly to intervene and all of the intervenors had participated in the case as amici/ Consequently, as earlier noted, the petition to intervene on appeal was not a subterfuge to make the earlier denials of intervention appealable, but an honest attempt to continue in the litigation.
. Because we conclude that the district court improperly denied intervention, the intervenors are now parties for purposes of appealing the district court’s findings and the injunction. For consistency purposes, we will continue to refer to them as intervenors in addressing their claims on the merits.
. We note that the intervenors also contend that the district court erred in not giving more weight to analyses of elections which did not involve black candidates; and also contend that the district court erred in discounting the success of a white County Commission candidate who received a majority of black votes. The intervenors concede in their brief that these alleged errors were not dispositive in the district court's determination. Based on our review of the record and the district court opinion, we agree that these claims of error are not disposi-tive, and decline to address them further.
. The district court relied on evidence of registered voters, not evidence of voting age population, in conducting its vote dilution analysis based on guidance in this court’s previous ruling in Meek I. See Meek, 805 F.Supp. at 975 n. 10, 988 n. 38; see also Meek I, 908 F.2d at 1546 & nn. 5-6.