Brenda M. JOHNSON, William Guice, et al., Plaintiffs-Appellants, v. DeSOTO COUNTY BOARD OF COMMISSIONERS, R.V. Griffin, in his official capacity as chairperson of the DeSoto County Board of Commissioners, et al., Defendants-Appellees, 204 F.3d 1335

EDMONDSON, Circuit Judge: Plaintiffs, black citizens of DeSoto County, brought suit, alleging that the current at-large method of electing the county school board and county commission unlawfully dilutes black-minority voting strength, under

section 2 of the Voting Rights Act and the Fourteenth and Fifteenth Amendments. After a trial, the district court found that Plaintiffs had failed to prove vote dilution and entered judgment for Defendants. We affirm the judgment. BACKGROUND The DeSoto County commission and school board, pursuant to Florida law, are each composed of five members. The members of each board, who serve four-year staggered terms, are required to live in five separate residency districts but are elected by an at-large, countywide vote.

The elections are partisan, with a majority requirement in the primaries but not in the general election. No black person has ever run for a seat on the commission; only one has run (unsuccessfully) for the school board. At the time of the 1990 census, blacks comprised 15.6 percent of the county’s total population and 13.7 percent of the total voting age population.

The county, however, contains a substantial nonvoting, mostly nonresident population, housed in a state prison and a state mental institution: few of the mental institution patients are county residents; and the inmates, convicted felons, cannot vote under Florida law. See Fla. Const. art. 6, § 4; Fla. Stat. § 97.041(2)(b). Removing these institutionalized members of the population from the total voting age population, blacks — in 1990 — comprised only 11.8 percent of the potential voters in the county.

At trial, Plaintiffs’ experts testified that, using 1990 census data, Plaintiffs could produce election plans for the county, consisting of five single-member districts for each board with blacks constituting a majority of the noninstitutionalized voting population in one of the districts. But Defendants introduced evidence that, because of changes in the black and white populations since 1990, the creation of a majority-black district was no longer possible in 1998.

One of Defendants’ experts compared the 1990 census data with 1991 voter registration data and calculated ratios of registered voters to voting age population in each proposed district; he then extrapolated, from 1998 voter registration data, the voting age population in 1998. From these calculations, he testified that blacks in 1998 could constitute only about 46 percent of the voting age population of Plaintiffs’ proposed black-majority district.

Another defense expert testified that considerable growth had occurred in the county since 1990, but not in the black population of the proposed black-majority district. Defendants also offered other evidence (not based on voter registration data) of the county’s population growth. For example, a member of the county commission testified that, based on the commission’s approval of new subdivisions, the southwest corner of the county was the major growth area: according to the witness, this area was not one with a high black population.

The district court entered judgment for Defendants, finding that Plaintiffs failed to establish their vote dilution claims. In particular, the district court found that Plaintiffs failed to show “discriminatory effects”: failed to show that the county’s at-large election system resulted in blacks having less opportunity to participate in the political process and elect candidates of their choice. Plaintiffs appeal. THE VOTING RIGHTS ACT CLAIM An electoral system violates

section 2 of the Voting Rights Act if the system causes the members of a distinct racial group to “have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.” 42 U.S.C. § 1973(b). The Supreme Court has said that, to satisfy

section 2’s standard in a vote dilution case, plaintiffs must show (at a minimum) 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 white majority votes as a bloc sufficiently to defeat the minority group’s preferred candidates. Thornburg v. Gingles, 478 U.S. 30, 106 S.Ct. 2752, 2766, 92 L.Ed.2d 25 (1986).

The district court, in this case, found that Plaintiffs failed to establish the first Gin-gles factor: “[Wjhile the Plaintiffs demonstrated the existence of the first Gingles precondition as of 1990, the Defendants have established by a preponderance of the evidence that as of the date of trial it is no longer possible to create a minority-controlled district in DeSoto County.” The district court, therefore, rejected Plaintiffs’ statutory claim.

Plaintiffs contend that the district court’s finding that Plaintiffs failed to establish the first Cin-gles factor was error because the district court should have never considered Defendants’ evidence of post-1990 population changes. We cannot accept Plaintiffs’ contention. A. Plaintiffs first contend that the district court should have excluded Defendants’ evidence of post-1990 population changes because the evidence contradicted a stipulation and several admissions agreed to by Defendants before trial.

In 1991, Defendants admitted that Plaintiffs’ proffered plans indeed created five single-member districts with one majority-black district. And, in an April 1998 pretrial statement, the parties stipulated that Plaintiffs had drawn two electoral schemes with a black-majority district. Defendants never amended these admissions or the stipulation. Based on the admissions and stipulation, Plaintiffs contend that Defendants’ evidence of post-census changes is barred because Defendants conclusively admitted that Plaintiffs could establish the required majority-minority district.

Before filing the April pretrial statement, Defendants disclosed that they expected to call two expert witnesses at trial. On 1 May 1998, pursuant to the preexisting pretrial order, Defendants informed Plaintiffs that, given new 1998 voter registration data, Defendants’ experts would challenge the continued validity of the 1990 census figures in their testimony. Defendants explained that their experts would testify that blacks, by 1998, were no longer sufficiently geographically concentrated to permit the creation of a black-majority district.

After this disclosure, Plaintiffs filed a motion in limine to exclude Defendants’ proffered evidence; the district court — after a hearing — denied the motion. Plaintiffs never moved for a continuance. Defendants did not seek to amend or to withdraw the admissions or stipulation. Instead, Defendants argued that their evidence of population changes did not contradict Defendants’ earlier admissions and stipulation because the admissions and stipulation were tied to and, thus, limited to the 1990 census figures. The district court agreed with Defendants.

The district court found that the admissions and stipulation inherently were based on the use of 1990 census data. The district court’s conclusion that the pretrial stipulation was defined by the 1990 census data was no abuse of discretion. See generally Pulliam v. Tallapoosa County Jail, 185 F.3d 1182, 1185 (11th Cir.1999) (noting that trial court’s

interpretation of pretrial order is reviewed for abuse of discretion). The trial court is in the best position to interpret the scope of stipulations in the pretrial statement. See Morrison v. Genuine Parts Co., 828 F.2d 708, 709 (11th Cir.1987) (noting that district court has broad discretion to decide whether to hold party to stipulation). We note that, in the April pretrial statement, Defendants expressly contended that Plaintiffs had not established the first Gin-gles factor because the 1990 census might not reflect the county’s population in 1998.

Therefore, in the light of the pretrial statement as a whole, the district court’s decision that the stipulation was based on and limited to the 1990 census figures was not error. Nor do we think the district court erred in construing the admissions as limited to the 1990 census. We think, at the very least, ambiguity did exist about whether the admissions were absolute or limited to the 1990 census figures, figures which might or might not accurately describe the county’s population at the time of trial. See Woods v. Robb, 171 F.2d 589, 541-42 (5th Cir.1948).

For example, one admission stated: “[I]n making [this] admission Defendants assume that Plaintiffs have accurately reported the figures for each of the districts.... If the figures shown in [the exhibit] turnout to be other than those Plaintiffs have shown, then Defendants reserve the right to supplement this answer accordingly.” And, Plaintiffs’ requests for admission specifically relied on exhibits created with figures from the 1990 census. See Rolscreen Co. v. Pella Prods. of St.

Louis, Inc., 64 F.3d 1202, 1210 (8th Cir.1995) (noting that conclusive effect of admission “may not be appropriate where requests for admissions or the responses to them are subject to more than one

interpretation” and that “[ijssues change as a case develops, and the relevance of discovery responses is related to their context in the litigation”). The scope and effect of admissions (like the scope and effect of stipulations) is a matter for determination by the trial court, in the exercise of its broad discretion. Given the circumstances of this case, the district court did not abuse its discretion in interpreting the admissions and stipulation as limited to the 1990 census figures. B.

Plaintiffs also contend that, even if the admissions and stipulation did not bar the introduction of Defendants’ post-census evidence, the district court erred in allowing Defendants’ evidence to oppose the census figures. No one challenges the initial accuracy of the 1990 census; the trial, however, was in 1998. At trial, Defendants pointed to the lapse of time since the census and to the changed circumstances. The presumption is that census figures are continually accurate. See Valdespino v. Alamo Heights Indep. Sch. Dist., 168 F.3d 848, 853-54 (5th Cir.1999). But this presumption is not irrebuttable.

The continuing accuracy of census figures is presumed only until the party challenging the census data overcomes the presumption with competent evidence to the contrary. See id. Although a burden rests on the party challenging the continuing accuracy of the census to introduce evidence to the contrary, we stress that the burden to establish the first Gingles factor remains, throughout the case, with the plaintiff. See Gingles, 106 S.Ct. at 2764. We conclude that the district court did not err in considering non-census data.

Plaintiffs claim that the district court erred in considering non-census evidence based on voter registration figures because, Plaintiffs say, registration data is an inherently unreliable measure of voting age population and cannot be used to contradict census figures. First, we note that there is no per se rule against the use of voter registration data in voting rights cases. Although the Supreme Court has written that voter registration data may be less probative than pure population data in voting cases, the Court has treated voter registration evidence as credible and as reliable. See Burns v.

Richardson, 384 U.S. 73, 86 S.Ct. 1286, 1297, 16 L.Ed.2d 376 (1966). We also have recognized the competence of this kind of data. See Wyche v. Madison Parish Police Jury, 635 F.2d 1151, 1161-62 (5th Cir.1981). Like most evidence presented by expert testimony, we think its admissibility has to be determined on a case-by-case basis by the district court. See generally Daubert v. Merrell Dow Pharmaceuticals Inc., 509 U.S. 579, 113 S.Ct. 2786, 2796-98, 125 L.Ed.2d 469 (1993).

And, this court has previously said, in a voting rights case, that statistical evidence derived from a sampling method, using reliable statistical techniques, is admissible on the question of determining the relevant population. See Negron v. City of Miami Beach, Fla., 113 F.3d 1563, 1570 (11th Cir.1997). We see no reason why the evidence presented in this ease — calculations of the county’s population in 1998 derived from voter registration information — should be subject to a different analysis.

Whether evidence derived from voter registration figures is' sufficiently reliable to be admitted and considered is a determination in the discretion of the district court. See generally Allison v. McGhan Medical Corp., 184 F.3d 1300, 1306 (11th Cir.1999) (citing General Elec. Co. v. Joiner, 522 U.S. 136, 118 S.Ct. 512, 517, 139 L.Ed.2d 508 (1997)) (discussing review of admissibility of expert testimony).

If the evidence is admissible, that voter registration data might not be as reliable as some other measures of population goes to the weight of the evidence, but does not preclude use of the figures by the district court. Here, the district judge was presented with expert testimony, from both sides, on the reliability of Defendants’ evidence. Defendants’ expert spoke to the reliability of the registration data for the county, pointing, out the lack of obstacles to registration.

He stated that he believed that registration data would not underrepre-sent the black population in 1998 because the passage of the Motor Voter law, if anything, would increase registration rates since 1991. The district court’s receipt and consideration of evidence on the demographic changes in. the county was no abuse of discretion. The kinds of evidence introduced in this case are not unfit for the purpose of challenging the continuing accuracy of census data. C. Next, we inquire whether the district court erred in finding no vote dilution for the

section 2 claim in this case. We review the district court’s finding of no vote dilution for clear error only. See Gingles, 106 S.Ct. at 2781. In the present case, Plaintiffs, using the 1990 census, proffered (at best) majority-minority districts with a black voting age majority of only about 54-57 percent. No one disputes that there has been substantial population growth in the county since the 1990 census. Defendants showed that a considerable amount of that growth has been within the white population.

Based on the post-census population data derived from voter registration figures, Defendants’ experts concluded that the creation of a black-majority district was not feasible in 1998: (1) movement of the district lines would include more whites in the district, thus decreasing the black majority; and (2) the black population growth was geographically distant from the proposed district and could not be included to make a black-majority district. Defendants also offered other evidence, not derived from voter registration figures, of the population changes and growth in the county.

In Valdespino, the Fifth Circuit upheld a district court’s finding that the plaintiffs had failed to establish the first Gingles factor. 168 F.3d at 856. There, even though, according to the 1990 census, the plaintiffs could create a majority-minority district, the Valdespino defendants presented evidence, at the 1997 trial, that demographic changes since the 1990 census had made the creation of such a district impossible.

The defendants’ presentation included evidence that a large apartment complex in the district had closed and reopened with fewer residents, while, at the same time, residential development outside the district increased. See id. at 850-51. The Fifth Circuit concluded that the district court did not err in deciding that the defendants’ figures demonstrated sufficient post-census demographic changes to raise considerable doubt that a majority-minority district could still be created and in deciding that the plaintiffs had not carried their burden of proof. See id. at 854, 856.

The record, in this case, presents ample evidence of population growth in the county since 1990, particularly outside of Plaintiffs’ proposed district. And, even using the 1990 census, Plaintiffs’ proffered black-majority district was not one that was overwhelmingly black. Thus, we cannot say that the district court erred in this finding that Defendants’ evidence undercut the presumption that the 1990 census reflected the truth about the county’s population and population distribution in 1998.

On this record, the district court (considering all the evidence including the 1990 census) did not clearly err in finding and in concluding that Plaintiffs failed to show the existence of the black-majority district needed to establish their prima facie case of vote dilution. THE CONSTITUTIONAL CLAIMS Plaintiffs contend that, even if their statutory claim fails, the district court erred in rejecting their constitutional vote dilution claims. Plaintiffs point out that the district court found that a discriminatory purpose underlies the county’s current at-large voting scheme.

And, Plaintiffs assert that the record shows that the county’s black population lacks an equal opportunity to participate in the political process and elect candidates of its choice. Plaintiffs argue that they, therefore, have sufficiently established claim's under the Fourteenth and Fifteenth Amendments. We disagree. As an initial matter, we doubt that any plaintiff, challenging an electoral system like DeSoto County’s, can establish a constitutional vote dilution claim where his

section 2 claim has failed. Plaintiffs say that, after a claimant has proved discriminatory intent, he need only produce minimal evidence of injury resulting from the challenged electoral scheme to prevail under the Constitution. But, the Supreme Court, historically, has articulated the same general standard, governing the proof of injury, in both

section 2 and constitutional vote dilution cases; plaintiffs, in both cases, must show that “there is evidence that excluded groups have ‘less opportunity to participate in the political process and to elect candidates of their choice.’ ” Compare Davis v. Bandemer, 478 U.S. 109, 106 S.Ct. 2797, 2809, 92 L.Ed.2d 85 (1986) (quoting Rogers v. Lodge, 458 U.S. 613, 102 S.Ct. 3272, 3279, 73 L.Ed.2d 1012 (1982)), with Gingles, 106 S.Ct. at 2763, and 42 U.S.C. § 1973(b). The pertinent issues seem the same (or almost the same) in both cases. And, even if the standards are not completely identical in application, we know that

section 2 was intended to be more permissive than the constitutional standard. See Sobmon, 899 F.2d at 1015 (Kravitch, J., specially concurring); see also Lee County Branch of NAACP v. City of Opelika, 748 F.2d 1473, 1478 n. 7 (11th Cir.1984) (“[I]f the plaintiffs cannot prevail under the generally more easily proved ‘results’ standard of

section 2, it is unlikely that they could prevail on their constitutional claims in any event.”). The parties have cited (and we have found) -no case in which a circuit court has concluded that an at-large or multi-member-district electoral system, although not in violation of

section 2, unconstitutionally dilutes minority voting strength. In the absence of Supreme Court direction, therefore, we question, as a legal proposition, whether vote dilution can be established under the Constitution when the pertinent record has not proved vote dilution under the more permissive

section 2. But, we need not resolve this question today. Even if we assume that it is possible, as a matter of law, to prevail on a constitutional claim where no

section 2 violation can be in fact established, Plaintiffs here have not proved their constitutional claim. Briefly stated, to establish a constitutional vote dilution claim, Plaintiffs must show that: (1) the county’s black population lacks an equal opportunity to participate in the political process and elect candidates of its choice; (2) this inequality of opportunity results from the county’s at-large voting scheme; and (3) a racially discriminatory purpose underlies the county’s voting scheme. Lucas v. Townsend, 967 F.2d 549, 551 (11th Cir.1992); see also Bandemer, 106 S.Ct. at 2809.

We will accept, for the purposes of this appeal, the district court’s finding that Plaintiffs have shown discriminatory intent. And, we will assume, for the sake of argument, that Plaintiffs’ evidence demonstrates the absence of equal opportunity. Plaintiffs, nonetheless, failed to establish their constitutional claims because the record fails to show that the inequality of opportunity results from the county’s current electoral system. In other words, Plaintiffs have failed to establish causation.

That a plaintiff, claiming a violation of his voting rights under the Fourteenth and Fifteenth Amendments, must show that an injury is caused by the government conduct he seeks to challenge is hardly a novel proposition. See Kirksey v. Bd. of Supervisors of Hinds County, 554 F.2d 139, 148 (5th Cir.1977) (inquiring whether reapportionment plan “will in fact have the effect of perpetuating the denial of access to the political process that was proved by plaintiffs to exist”); Personnel Adm’r of Mass. v.

Feeney, 442 U.S. 256, 99 S.Ct. 2282, 2293, 60 L.Ed.2d 870 (1979) (requiring that claimant show discriminatory impact traceable to discriminatory purpose); see also Allen v. Wright, 468 U.S. 737, 104 S.Ct. 3315, 3327, 82 L.Ed.2d 556 (1984) (requiring equal protection plaintiff to show causation as element of standing). To show that inequality of opportunity is caused by a particular electoral system, a plaintiff must establish that “an alternative election scheme exists that would provide better access to the political process.” Burton v. City of Belle Glade, 178 F.3d 1175, 1199 (11th Cir.1999); see also Reno v.

Bossier Parish Sch. Bd., 520 U.S. 471, 117 S.Ct. 1491, 1498, 137 L.Ed.2d 730 (1997); Holder v. Hall, 512 U.S. 874, 114 S.Ct. 2581, 2585, 129 L.Ed.2d 687 (1994); Nipper v. Smith, 39 F.3d 1494, 1533 (11th Cir.1994) (en banc).

As we have explained, “[I]f a minority cannot establish that an alternative election scheme exists that would provide better access to the political process, then the challenged voting practice is not responsible for the claimed injury.” Burton, 178 F.3d at 1199; see also Gingles, 106 S.Ct. at 2766 (explaining if first Gingles factor is not shown, then “the multi-member form of the district cannot be responsible for minority voters’ inability to elect its candidates”).

Here, Plaintiffs failed to make the requisite showing of causation: Plaintiffs did not establish that an alternative system of districting could exist whereby the black-minority vote could elect its preferred candidates. The district court found that the creation of a black-majority district, in 1998, was not feasible.

Plaintiffs, on appeal, argue that a black-majority district is not required; according to Plaintiffs, a “black-influence district,” where a substantial black minority is coupled with sufficient white cross-over voting so that the black minority in fact can elect candidates of its choice, is sufficient. We, however, need not decide whether Plaintiffs’ “influence district” theory is correct. Plaintiffs never argued this theory to the district court. Plaintiffs have failed to establish causation. Plaintiffs’ contention that the district court erred in rejecting their constitutional claims, therefore, must fail.

CONCLUSION The district court dealt with the evidence without error. The case was fully tried. The district court was the finder of fact. The district court did not err in determining that Plaintiffs showed no lack of equal political opportunity that was caused by the county’s electoral system! Plaintiffs failed to meet their burden. The judgment of the district court is AFFIRMED. . See Fla. Const. art. 8, § 1(e); Fla. Stat. § 100 et seq. . The parties do not dispute that, for the purposes of this litigation, the appropriate category of voters is blacks. .

This expert plotted on a map the location of each black registered voter and found that, since 1991, more black registered voters were living outside the proposed district, showing a dispersion of the county's black population. . For example, one admission stated: "Defendants admit that the plaintiffs five single-member district plan ... is one which has a maximum deviation of 6.2% and that the plan contains one district in which African Americans make up a majority of the voting age population.” .

In the April pretrial statement, the parties stipulated: Plaintiffs have drawn an election plan for De Soto County containing 5 single member districts, which includes one district in which African-Americans are 54.37 percent of the population 18 years of age and over and which has a total deviation of 0.16% from the "ideal” district, (i.e. a district containing exactly one fifth of the county’s total population).

Plaintiffs have also drawn an election plan for De Soto County containing 5 single member districts, which includes one district in which African-Americans are 57.33 percent of the population 18 years of age and over and which has a total deviation of 6.2 % from the "ideal” district. The pretrial statement, however, also stated: "Defendants do not concede that plaintiffs have met or will be able to meet the first Gingles precondition....

The census figures are now eight years old and may no longer accurately reflect the present population percentages.” .Plaintiffs argue that the district court's construction of the stipulation amounts to an amendment of the pretrial order and, therefore, is subject to the manifest injustice standard of Fed.R.Civ.P. 16(e). We disagree. The district court did not amend the pretrial statement, but construed it. Also, Defendants’ disclosure of their experts' new testimony was submitted pursuant to the pretrial order. .

Plaintiffs argue that the district court's construction of the admissions amounts to withdrawal and is subject to the unfair prejudice standard of Fed.R.Civ.P. 36. We do not agree. Defendants did not argue for, nor did the district court allow, withdrawal; the district court just construed the scope of the admission. Plaintiffs in this case were not unfairly prejudiced by the introduction of Defendants’ evidence of post-1990 population change. Defendants’ experts were specifically identified in the 1 May disclosure: fifty days before trial.

And, Plaintiffs did depose one of Defendants' experts before trial about the new evidence. Cf. Bergemann v. United States, 820 F.2d 1117, 1121 (10th Cir.1987) (finding no prejudice where party knew issue was contested). When the 1998 pretrial statement was filed, Plaintiffs already expected to call an expert to testify about the feasibility of creating a majority-black district. And, at trial, Plaintiffs presented an expert who, in fact, challenged Defendants’ new evidence. Plaintiffs have not shown that they were unfairly prejudiced by having to respond to Defendants’ new evidence. See Smith v.

First Nat'l Bank of Atlanta, 837 F.2d 1575, 1577-78 (11th Cir.1988). Also, if Plaintiffs had been surprised, they should have moved for a continuance. This court has repeatedly said that "the remedy for coping with surprise is not to seek reversal after an unfavorable verdict, but a request for continuance at the time the surprise occurs.” United States v. Battle, 173 F.3d 1343, 1350 (11th Cir.1999) (citations omitted). Plaintiffs requested no continuance. Plaintiffs also contend that Defendants should have amended formally their admissions. Plaintiffs cite American Auto. Ass’n v.

AAA Legal Clinic of Jefferson Crooke, P.C., 930 F.2d 1117, 1119 (5th Cir.1991), and Williams v. City of Dothan, Ala., 818 F.2d 755, 762, modified on other grounds on denial of reh’g by 828 F.2d 13 (11th Cir.1987), for the proposition that the district court is not free to reject an admission because it finds more "credible evidence.” These cases are materially different from this case. In both cases, the court simply disregarded an admission; here, the court reasonably construed Defendants’ admission as limited to the 1990 census figures. .

Plaintiffs themselves characterized Defendants’ admissions as tied to and limited to the validity of the 1990 census figures: "[Djefen-dants have admitted that 'African-Americans in DeSoto county are sufficiently geographically concentrated such that utilizing 1990 census population information, a single member district plan can be drawn (3)27' " (emphasis added). And, at least one admission stated: "Defendants' answer is subject to the further qualification that Defendants are not admitting that Plaintiffs have demonstrated that they can satisfy any of the three 'preconditions' of [Gingles ].” .

We are aware that the First Circuit, in one case, reviewed de novo a district court's

interpretation of an admission. See Talley v. United States, 990 F.2d 695, 698-99 (1st Cir.1993) (stating that construction of documents is reviewed de novo). Because we review a district court's construction of stipulations and a district court's evidentiary rulings for abuse of discretion, we decline to adopt the standard of review that the First Circuit used. See, e.g., Pulliam, 185 F.3d at 1185; Ad-Vantage Tel. Directory Consultants, Inc. v. GTE Directories Corp., 37 F.3d 1460, 1463 (11th Cir.1994); see also Milton v.

Montgomery Ward & Co., 33 Cal.App.3d 133, 138, 108 Cal.Rptr. 726 (Cal.Ct.App.1973) (noting that trial court has broad discretion to determine relevancy and scope of admission). . In fact, the Supreme Court has acknowledged that census data are not perfect and are often outdated. See Abrams v. Johnson, 521 U.S. 74, 117 S.Ct. 1925, 1940, 138 L.Ed.2d 285 (1997); see also Kirkpatrick v. Preisler, 394 U.S. 526, 89 S.Ct. 1225, 1231, 22 L.Ed.2d 519 (1969) (other data may be considered in voting rights cases). .The Supreme Court has never precluded the use of voter registration data. See Rollins v. Fort Bend Indep. Sch.

Dist., 89 F.3d 1205, 1219 (5th Cir.1996) (noting that Supreme Court has not "held that voter registration is irrelevant: it is simply not the sole criterion”). And, neither have we. See, e.g., Negron v. City of Miami Beach, Fla., 113 F.3d 1563, 1568 (11th Cir.1997). Plaintiffs cite a special concurrence in Solomon v. Liberty County, Fla., 899 F.2d 1012, 1018 (11th Cir.1990) (en banc) (Kravitch, J., specially concurring), for the proposition that voter registration figures should not be considered in the instant case.

The Solomon en banc court, however, decided merely that the three Gingles factors had been established because, despite being only 45 percent of registered voters, the evidence showed that the minority population in the district was 51 percent of the voting age population. Id. at 1013 (per curiam). Moreover, Judge Kravitch never said that voter registration data was per se inadmissible evidence; she merely noted that the relevant criteria to consider is population, not the number of registered voters. See id. at 1018; see also Johnson v.

DeGrandy, 512 U.S. 997, 114 S.Ct. 2647, 2656, 129 L.Ed.2d 775 (1994) (considering population as relevant criterion). In this case, Defendants and the district court, in fact, focused on the correct criteria: population. The use of evidence derived from voter registration data to show population is not, in itself, impermissible. . The Motor Voter law was enacted in 1993, in part, to increase the voter registration rates among minorities by allowing citizens to register at more places, including at driver’s licensing facilities.

See 42 U.S.C. § 1973gg(b)(1) (stating that purpose of Act is "to establish procedures that will increase the number of eligible citizens who register to vote"). . Defendants' expert also observed the 66.1 percent registration rate among blacks within the proposed black-majority district (compared to the countywide white registration rate of 63.5 percent). Plaintiffs point out that black voter registration dropped in 1994, but Defendants' expert explained that the 1991 figures were based on inactive and active voter registration roles, whereas the 1994 figures were based purely on active registered voters. .

Plaintiffs contend that the court impermis-sibly presumed that voter registration mirrored population, in violation of Fed.R.Evid. 301. We find this argument to be without merit. Any assumption made by Defendants’ expert that voter registration mirrored voting age population went to the weight of the testimony and was freely challengeable on cross-examination. . At the motion in limine hearing, the district court, concluding that evidence of current population numbers would be relevant, said these words: "And I'm not saying we ignore the 1990 census.” The 1990 census figures are facts.

As evidence, they — even after Defendants introduced contrary evidence — retained probative force and could support inferences on the part of the fact finder. But the district court did not err in weighing the census figures against Defendants’ evidence. . When reviewing for clear error: "As long as the district court’s findings are plausible, we may not reverse the district court even if we would have decided the case differently.” United. States v. Engelhard Corp., 126 F.3d 1302, 1305 (11th Cir.1997); see generally Anderson v.

Bessemer City, 470 U.S. 564, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985) ("Where there are two permissible views of the evidence, the factfinder's choice between them cannot be clearly erroneous.”). . Whether vote dilution is cognizable at all under the Fourteenth and Fifteenth Amendments is uncertain. See Burton v. City of Belle Glade, 178 F.3d 1175, 1187 n. 9 (11th Cir.1999). .

Plaintiffs initially argue that, because the district court found a discriminatory purpose behind the county’s electoral system, they have shown the existence of a de jure segregation system and, therefore, have established a claim under United States v. Fordice, 505 U.S. 717, 112 S.Ct. 2727, 120 L.Ed.2d 575 (1992). We have noted previously that no court has applied Fordice outside of the education setting, Burton v. City of Belle Glade, 178 F.3d 1175, 1190 (11th Cir.1999), and we decline to do so for the first time in this case.

Moreover, the government's discriminatory intent alone, without a causal connection between the intent and some cognizable injury to Plaintiffs, cannot entitle Plaintiffs to relief in this case: a facially neutral law "is unconstitutional under the Equal Protection Clause only if [a discriminatory] impact can be traced to a discriminatory purpose.” Personnel Adm’r of Mass. v. Feeney, 442 U.S. 256, 99 S.Ct. 2282, 2293, 60 L.Ed.2d 870 (1979) (emphasis added).

Therefore, for the reasons stated in this opinion, even if Fordice applies outside of the education setting, Plaintiffs cannot prevail on a Fordice-based theory in this case. . In Holder v. Hall, the Supreme Court, finding no

section 2 violation, "remanded for consideration of respondent’s constitutional [vote dilution] claim.” 512 U.S. 874, 114 S.Ct. 2581, 2588, 129 L.Ed.2d 687 (1994). Plaintiffs cite this result for the proposition that the statute and the Constitution are not coextensive for vote dilution claims. We do not think this statement bears the weight Plaintiffs have placed on it. In Holder, the circuit court originally did not address' the constitutional claims because it concluded that

section 2 vote dilution had been proved. Id. at 2585. Because the circuit court had not addressed the issue, it was proper for the Supreme Court to remand rather than consider an issue not considered by the circuit court. See Duignan v. United States, 274 U.S. 195, 47 S.Ct. 566, 568, 71 L.Ed. 996 (1927) ("This court sits as a court of review.

It is’ only in exceptional cases coming here from the federal courts that questions not pressed or passed upon below are reviewed.”). .We recognize that, in prior cases, we have said that, to prevail on a constitutional vole dilution claim, a plaintiff must show the existence of “discriminatoiy effects” and “a racially discriminatory purpose chargeable to the state.” Lucas v. Townsend, 967 F.2d 549, 551 (11th Cir.1992). Our articulation today of the constitutional standard does not alter the plaintiffs burden. We are not changing the law; we are explaining it.

Case law makes apparent that the "discriminatory effects” requirement encompasses both inequality of opportunity and a causation element. See id. (requiring that discriminatory effect "results from” intentionally discriminatory electoral scheme); see also Kirksey v. Bd. of Supervisors of Hinds County, 554 F.2d 139, 148 (5th Cir.1977) (same). Causation is also implicit in the term "discriminatoiy effects.” See Random House Dictionaiy of the English Language 622 (2d ed. unabridged 1987) (defining effect as "something that is produced by an agency or cause; result; consequence”).

We think, however, that, to avoid confusion, viewing causation and the existence of unequal opportunity as distinct elements (instead of merging these two concepts into a single "discriminatoiy effects” requirement) is the better approach: it promotes clarity. When the district court, in this case, found no "discriminatoiy effects,” we understand the court to have found no causation. . Defendants dispute the district court's finding of discriminatory intent. Given our disposition of this appeal, however, we need not address Defendants' contention. .

Plaintiffs assert that their evidence, of (1) the black population’s inability to elect consistently candidates of its choice and (2) the deterrence of blacks from seeking office in the county, is sufficient to establish that the black population lacks equal political opportunity. We need not decide, however, whether these factors (or either factor alone) are sufficient to establish a lack of equal opportunity; we assume that Plaintiffs have shown an inequality of opportunity.

But, we do observe that Plaintiffs' argument about the deterrence of black candidacies is, in this case, nothing more than a variation of the argument that the county’s black population cannot consistently elect the candidates of its choice. If the record indeed establishes that black candidates are deterred from seeking office in DeSoto County, it is only because they (according to Plaintiffs) cannot win. And, if black candidates cannot win, it (according to Plaintiffs’ theory) is only because the black-minority vote, under the present system, is insufficient to elect them.

For all practical purposes, therefore, Plaintiffs' deterrence theory comes down to the black minority's alleged inability to elect candidates of its choice. . “The early development of our voting rights jurisprudence in [equal protection] cases provided the basis for our analysis of vote dilution under the amended § 2....” Holder, 114 S.Ct. at 2592 n. 1 (Thomas, J., concurring in judgment); see also League of United Latin American Citizens v. Clements, 999 F.2d 831, 851 (5th Cir.1993) (“[T]he 1982 amendments to § 2 were intended to 'codify' the results test as employed in White and Whitcomb.”).

Therefore, we are informed in our inquiry by prior decisions construing

section 2. . The Supreme Court continually has declined to decide whether the "influence district" theory is sound. See, e.g., Voinovich v. Quilter, 507 U.S. 146, 113 S.Ct. 1149, 1155, 122 L.Ed.2d 500 (1993). .Plaintiffs also assért that the district court erred in considering Plaintiffs’ two proposed districting plans (one with less deviation than the other, but covering similar areas of black concentration) as similar. The district court noted that, at trial, the parties focused on the district with less deviation.

The district court also found that the black voting age population constituted a district majority in neither districting plan. We And no error in the district court's analysis. See Davis v. Chiles, 139 F.3d 1414, 1418 n. 8 (11th Cir.1998) (discussing proposals together where they generally raise same issues).