WISDOM, Senior Circuit Judge:
The plaintiffs in this case allege that the at-large scheme for electing the municipal government for the city of Opelika, Alabama violates the Fourteenth and Fifteenth Amendments and section 2 of the Voting Rights Act of 1965, as amended, 42 U.S. C.A. § 1973 (West Supp.1984). The case was tried under the law as pronounced in Nevett v. Sides, 5 Cir.1978, 571 F.2d 209, cert. denied, 1980, 446 U.S. 951, 100 S.Ct. 2916, 64 L.Ed.2d 807. After trial but before the district court had issued its decision in this case, the Supreme Court issued two decisions affecting the standard of proof in constitutional discrimination cases. Rogers v. Lodge, 1982, 458 U.S. 613, 102 S.Ct. 3272, 73 L.Ed.2d 1012; City of Mobile v. Bolden, 1980, 446 U.S. 55, 100 S.Ct. 1490, 64 L.Ed.2d 47. Moreover, in 1982 Congress amended section 2 of the Voting Rights Act, in effect overruling Mobile and restoring the legal standard in the Fifth and Eleventh Circuits governing voter discrimination decisions before that case was decided.1 The district court deferred its ruling pending the decision in Mobile, then issued an opinion denying relief to the plaintiffs under the principles stated in that ease. In a later brief opinion, the court denied the plaintiffs’ motion to alter or amend the judgment, or for a new trial, and held that the plaintiffs were not entitled to relief under either section 2 of the Voting Rights Act or under the factors enumerated by the Supreme Court in Rogers. While this case was on appeal, two decisions were issued that affect our disposition of this ease. In Escambia County v. McMillan, 1984, — U.S. —, 104 S.Ct. 1577, 80 L.Ed.2d 36, the Supreme Court vacated a finding that an at-large election scheme violated the Fourteenth Amendment and remanded the case for consideration of whether the scheme was unlawful under section 2 of the Voting Rights Act. In United States v. Marengo County Commission, 11 Cir.1984, 731 F.2d 1546, this Court explained in detail how the “results” test of section 2 is to be applied to an allegation that an at-large system unlawfully dilutes minority votes.
We hold that Escambia requires that the plaintiffs’ section 2 claim be decided first and that Marengo sets forth the legal standard governing that claim. We remand this case to the district court to allow the parties an opportunity to update the record and to present evidence on the question whether Opelika city elections have exhibited racially polarized voting, a key consideration under the Marengo standard.
I. FACTS
Opelika, Alabama is a city of approximately 22,000, of which about 31 percent are black. The city is governed by a three-person commission whose members are elected at-large for a three-year term. The elections are staggered; one new commissioner is elected each summer. If no candidate receives a majority of votes, the two candidates receiving the most votes participate in a run-off election. After each election, the city commissioners select from among themselves a president who serves as mayor.
Although pervasive de jure discrimination in Opelika ended in about 1970, residential patterns in the city at the time of trial were segregated. As of 1978, Opelika was divided into six voting wards or “boxes”. Witnesses at trial testified that the majority of the population in boxes 1, 2, and 6 were white and the majority of those in boxes 3 and 5 were black. The remaining box was generally considered a “white” box. There was no direct evidence in the record of voter registration by race in Opelika.2 The plaintiffs introduced evidence suggesting that a disparity existed be[1476]*1476tween levels of voter registration among blacks and whites.
No black has ever been elected to Opeli-ka’s city commission. On five occasions between 1969 and 1978, a black candidate ran for and lost election to the commission. In 1970, no blacks were employed by the city in either managerial or clerical positions. In 1978, three of thirty-three clerical positions were filled by blacks. Three of the eleven administrative positions with the City Water Works Board and four of the forty-four city managerial positions were filled by blacks. Since 1972 the Opelika school system has been fully integrated at a ratio of approximately 60 percent white and 40 percent black in each school throughout the system.
On January 25, 1978, the Lee County Branch of the NAACP, the Lee County Voters League, and several of the members of these organizations filed suit against the City of Opelika and the three members of its city commission, alleging that the at-large commission form of government impermissibly dilutes the votes of black citizens in violation of the Fourteenth and Fifteenth Amendments and section 2 of the Voting Rights Act, as amended, 42 U.S.C. § 1973 (1976). The case was tried before the district court in the summer of 1978 under the law as set forth in Nevett v. Sides, 5 Cir.1978, 571 F.2d 209, cert. denied, 1980, 446 U.S. 951, 100 S.Ct. 2916, 64 L.Ed.2d 807. Nevett held that a claim of unconstitutional voting dilution could be established by proof of the factors outlined in Zimmer v. McKeithen, 5 Cir. 1973, 485 F.2d 1297 (en banc), aff'd per curiam sub nom. East Carroll Parish School Board v. Marshall, 1976, 424 U.S. 636, 96 S.Ct. 1083, 47 L.Ed.2d 296. Proof of these factors “raises an inference of intentional discrimination”. Nevett, 571 F.2d at 225. The focus of the evidence at trial was on the two Zimmer factors that the Nevett Court stated were of special relevance in voting dilution cases — whether the plaintiffs had access to the political process and whether government officials were responsive to the interests of the plaintiff racial group.3 Nevett, 571 F.2d at 223 n. 19.
After trial, but before a decision in this case had been rendered, the Supreme Court decided City of Mobile v. Bolden, 1980, 446 U.S. 55, 100 S.Ct. 1490, 64 L.Ed.2d 47. A plurality of the Court held that intentional discrimination must be shown to establish a claim of unconstitutional voting dilution. The plurality rejected the holding of Nevett that proof of an aggregate of the Zimmer factors is sufficient to establish such a claim. 446 U.S. at 73, 100 S.Ct. at 1502-[1477]*14771503. Five justices in Mobile also ruled that section 2 of the Voting Rights Act incorporates the constitutional intent standard. Id. at 60-61, 105 n. 2, 100 S.Ct. at 1495-1496, 1520 n. 2. After the decision in Mobile, the plaintiffs moved to reopen the record in this case to submit evidence bearing more directly upon intent.
Free access — add to your briefcase to read the full text and ask questions with AI
WISDOM, Senior Circuit Judge:
The plaintiffs in this case allege that the at-large scheme for electing the municipal government for the city of Opelika, Alabama violates the Fourteenth and Fifteenth Amendments and section 2 of the Voting Rights Act of 1965, as amended, 42 U.S. C.A. § 1973 (West Supp.1984). The case was tried under the law as pronounced in Nevett v. Sides, 5 Cir.1978, 571 F.2d 209, cert. denied, 1980, 446 U.S. 951, 100 S.Ct. 2916, 64 L.Ed.2d 807. After trial but before the district court had issued its decision in this case, the Supreme Court issued two decisions affecting the standard of proof in constitutional discrimination cases. Rogers v. Lodge, 1982, 458 U.S. 613, 102 S.Ct. 3272, 73 L.Ed.2d 1012; City of Mobile v. Bolden, 1980, 446 U.S. 55, 100 S.Ct. 1490, 64 L.Ed.2d 47. Moreover, in 1982 Congress amended section 2 of the Voting Rights Act, in effect overruling Mobile and restoring the legal standard in the Fifth and Eleventh Circuits governing voter discrimination decisions before that case was decided.1 The district court deferred its ruling pending the decision in Mobile, then issued an opinion denying relief to the plaintiffs under the principles stated in that ease. In a later brief opinion, the court denied the plaintiffs’ motion to alter or amend the judgment, or for a new trial, and held that the plaintiffs were not entitled to relief under either section 2 of the Voting Rights Act or under the factors enumerated by the Supreme Court in Rogers. While this case was on appeal, two decisions were issued that affect our disposition of this ease. In Escambia County v. McMillan, 1984, — U.S. —, 104 S.Ct. 1577, 80 L.Ed.2d 36, the Supreme Court vacated a finding that an at-large election scheme violated the Fourteenth Amendment and remanded the case for consideration of whether the scheme was unlawful under section 2 of the Voting Rights Act. In United States v. Marengo County Commission, 11 Cir.1984, 731 F.2d 1546, this Court explained in detail how the “results” test of section 2 is to be applied to an allegation that an at-large system unlawfully dilutes minority votes.
We hold that Escambia requires that the plaintiffs’ section 2 claim be decided first and that Marengo sets forth the legal standard governing that claim. We remand this case to the district court to allow the parties an opportunity to update the record and to present evidence on the question whether Opelika city elections have exhibited racially polarized voting, a key consideration under the Marengo standard.
I. FACTS
Opelika, Alabama is a city of approximately 22,000, of which about 31 percent are black. The city is governed by a three-person commission whose members are elected at-large for a three-year term. The elections are staggered; one new commissioner is elected each summer. If no candidate receives a majority of votes, the two candidates receiving the most votes participate in a run-off election. After each election, the city commissioners select from among themselves a president who serves as mayor.
Although pervasive de jure discrimination in Opelika ended in about 1970, residential patterns in the city at the time of trial were segregated. As of 1978, Opelika was divided into six voting wards or “boxes”. Witnesses at trial testified that the majority of the population in boxes 1, 2, and 6 were white and the majority of those in boxes 3 and 5 were black. The remaining box was generally considered a “white” box. There was no direct evidence in the record of voter registration by race in Opelika.2 The plaintiffs introduced evidence suggesting that a disparity existed be[1476]*1476tween levels of voter registration among blacks and whites.
No black has ever been elected to Opeli-ka’s city commission. On five occasions between 1969 and 1978, a black candidate ran for and lost election to the commission. In 1970, no blacks were employed by the city in either managerial or clerical positions. In 1978, three of thirty-three clerical positions were filled by blacks. Three of the eleven administrative positions with the City Water Works Board and four of the forty-four city managerial positions were filled by blacks. Since 1972 the Opelika school system has been fully integrated at a ratio of approximately 60 percent white and 40 percent black in each school throughout the system.
On January 25, 1978, the Lee County Branch of the NAACP, the Lee County Voters League, and several of the members of these organizations filed suit against the City of Opelika and the three members of its city commission, alleging that the at-large commission form of government impermissibly dilutes the votes of black citizens in violation of the Fourteenth and Fifteenth Amendments and section 2 of the Voting Rights Act, as amended, 42 U.S.C. § 1973 (1976). The case was tried before the district court in the summer of 1978 under the law as set forth in Nevett v. Sides, 5 Cir.1978, 571 F.2d 209, cert. denied, 1980, 446 U.S. 951, 100 S.Ct. 2916, 64 L.Ed.2d 807. Nevett held that a claim of unconstitutional voting dilution could be established by proof of the factors outlined in Zimmer v. McKeithen, 5 Cir. 1973, 485 F.2d 1297 (en banc), aff'd per curiam sub nom. East Carroll Parish School Board v. Marshall, 1976, 424 U.S. 636, 96 S.Ct. 1083, 47 L.Ed.2d 296. Proof of these factors “raises an inference of intentional discrimination”. Nevett, 571 F.2d at 225. The focus of the evidence at trial was on the two Zimmer factors that the Nevett Court stated were of special relevance in voting dilution cases — whether the plaintiffs had access to the political process and whether government officials were responsive to the interests of the plaintiff racial group.3 Nevett, 571 F.2d at 223 n. 19.
After trial, but before a decision in this case had been rendered, the Supreme Court decided City of Mobile v. Bolden, 1980, 446 U.S. 55, 100 S.Ct. 1490, 64 L.Ed.2d 47. A plurality of the Court held that intentional discrimination must be shown to establish a claim of unconstitutional voting dilution. The plurality rejected the holding of Nevett that proof of an aggregate of the Zimmer factors is sufficient to establish such a claim. 446 U.S. at 73, 100 S.Ct. at 1502-[1477]*14771503. Five justices in Mobile also ruled that section 2 of the Voting Rights Act incorporates the constitutional intent standard. Id. at 60-61, 105 n. 2, 100 S.Ct. at 1495-1496, 1520 n. 2. After the decision in Mobile, the plaintiffs moved to reopen the record in this case to submit evidence bearing more directly upon intent. The district court denied the plaintiffs’ offers of additional evidence.
In August 1982, the district court issued a memorandum opinion denying the plaintiffs’ relief. Memorandum Opinion, M.D. Ala. Aug. 31, 1982, Record at 273. The court concluded that the Zimmer factors are relevant to a showing of discriminatory purpose under Mobile, although Mobile made proof of those factors alone insufficient to establish the necessary intent. The court ruled that the plaintiffs had “not proved by a preponderance of the evidence the existence of an aggregate of the factors which the Zimmer Court said would make out ‘a strong case’ ”. Record at 308. The court’s opinion emphasized evidence that Opelika’s government had been responsive to black needs and that all formal barriers to black political participation had been eliminated.
The court’s opinion made no mention either of Rogers v. Lodge, 1982, 458 U.S. 613, 102 S.Ct. 3272, 73 L.Ed.2d 1012 — decided two months before the district court issued its opinion — or of the Voting Rights Act Amendments of 1982, P.L. No. 97-205, 96 Stat. 131 (codified at 42 U.S.C. §§ 1971— 1974e), effective June 29, 1982. Rogers confirmed the vitality of the Zimmer factors and held that these factors can be used to support an inference of intentional discrimination. Rogers, 458 U.S. at 620-22, 624, 102 S.Ct. at 3277-78, 3279. The 1982 Amendments changed section 2 of the Voting Rights Act to prohibit any practice or procedure “imposed or applied ... in a manner which results in a denial or abridgement of the right ... to vote on account of race or color . . . .” 42 U.S.C.A. § 1973(a) (West Supp.1984) (emphasis added). The statute further provides:
A violation of subsection (a) of this section is established if, based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a) of this section in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice! The extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered: Provided, that nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population.
42 U.S.C.A. § 1973(b) (West Supp.1984).
On September 10, 1982, the plaintiffs moved to amend the judgment or for a new trial in the light of Rogers and amended section 2. They also asserted that Rogers reflected a renewed emphasis on certain evidentiary factors outlined in Zimmer that would further support their right to relief. In March 1983, the district court denied the plaintiffs’ motion. The court entered no detailed findings of fact or conclusions of law. Its three-paragraph opinion concluded that
(1) the preponderance of the evidence showing facts alleged and proved in this case do not indicate a violation of 42 U.S.C. § 1973 as last amended or of the Constitution and (2) the Opinion in this case is inclusive of the primary factors discussed in Rogers and Rogers supplies neither legal nor factual reasons to alter or amend the judgment entered in this case on August 31, 1982.
Record at 322-23.4 The plaintiffs appealed from the denial of their motion to alter or [1478]*1478amend, or for a new trial, and from the district court’s original judgment of August 1982.
II. THE SECTION 2 CLAIM SHOULD BE ADJUDICATED FIRST
Although the plaintiffs filed claims under both the Fourteenth and Fifteenth Amendments and section 2 of the Voting Rights Act, the case was tried in the district court primarily in constitutional terms under the intent standard set forth in Nevett v. Sides, 5 Cir.1978, 571 F.2d 209, cert. denied, 1980, 446 U.S. 951, 100 S.Ct. 2916, 64 L.Ed.2d 807.
There was relatively little judicial interpretation of the scope of section 2 in the reported decisions before the Supreme Court’s Mobile decision in 1980.5 The Voting Rights Act Amendments of 1982, however, made it clear that the “results” standard under section 2 was intended to replace the “intent” standard of Mobile.6 The proof of a section 2 violation, therefore, is different from the proof of a claim of unconstitutional voting dilution.
A recent Supreme Court case, decided while this case was on appeal, requires that the plaintiffs’ statutory claim be adjudicated first. In Escambia County v. McMillan, 1984, — U.S. —, 104 S.Ct. 1577, 80 L.Ed.2d 36 (per curiam), the Court vacated a finding that an at-large election scheme violated the Fourteenth Amendment and remanded the case for consideration of whether the scheme was unlawful under section 2 of the Voting Rights Act. The district court had ruled that the at-large scheme was unconstitutional because it had been maintained for discriminatory purposes. The Fifth Circuit affirmed on the standards set forth in Rogers v. Lodge, 1982, 458 U.S. 613, 102 S.Ct. 3272, 73 L.Ed.2d 1012. McMillan v. Escambia County, 5 Cir.1982, 688 F.2d 960. Although the Fifth Circuit’s opinion was issued after section 2 was amended, the Court declined to reconsider the case under the statute because further delay in county elections would cause great hardship and because a decision of the issues raised under section 2 would not affect the outcome of the. case, for the plaintiffs were entitled to relief on their Fourteenth Amendment claim. 688 F.2d at 961 n. 2.
Notwithstanding the Fifth Circuit’s explanation for its refusal to decide the plaintiffs’ section 2 claim, the Supreme Court vacated the judgment and remanded on the premise that “the Court will not decide a constitutional question if there is some other ground upon which to dispose of the case.” Escambia, — U.S. at —, 104 5. Ct. at 1579, 80 L.Ed.2d at 39. We hold that Escambia dictates that the plaintiffs’ section 2 claim should have been adjudicated first. A finding of liability under section 2 would obviate the necessity to decide the plaintiffs’ Fourteenth and Fifteenth Amendment claims.7 The defendants urge us to decide the plaintiffs’ section 2 claim based on the record before us. The plaintiffs insist that we should remand the case for an evidentiary hearing under section 2. [1479]*1479We turn now to the question whether this case should be remanded to the district court for further proceedings concerning the plaintiffs’ section 2 claim.
III. SHOULD THIS CASE BE REMANDED FOR FURTHER PROCEEDINGS ON THE SECTION 2 CLAIM?
While this case was on appeal, this Court decided United States v. Marengo County Commission, 11 Cir.1984, 731 F.2d 1546. Marengo explains in detail how the “results” test of section 2 is to be applied to an allegation that an at-large system unlawfully dilutes minority votes. Discriminatory intent need not be shown to establish a violation. Section 2 focuses “on whether minorities have an equal right to participate in the political process.” Id. at 1565. A denial of equal participation may be demonstrated by proof of various objective factors, many of which are those outlined in Zimmer v. McKeithen, 5 Cir.1973, 485 F.2d 1297 (en banc), aff'd per curiam sub nom. East Carroll Parish School Board v. Marshall, 1976, 424 U.S. 636, 96 S.Ct. 1083, 47 L.Ed.2d 296. These factors include the existence of racially polarized voting; past discrimination and its lingering effects; access to the slating process; election practices that exacerbate the deficiencies in minority participation in the political process; elections characterized by racial appeals; tenuousness of state policies underlying the at-large elections; success of minority candidates; and “enhancing factors” such as the existence of large districts, majority vote requirements, anti-single shot voting provisions and the lack of provision for at-large candidates running from particular geographic subdistricts. See Marengo, 731 F.2d at 1566-73. The factors are to be weighed under a “totality of the circumstances” approach. Id. at 1565-66.
The defendants insist that a remand is not necessary to decide the plaintiffs’ section 2 claim. They note that, although the legal theory of the plaintiffs’ case has changed from “intent” to “results”, the Zimmer factors are relevant under both theories. Because the case was tried under the intent standard articulated in Nevett v. Sides, which looks to the Zimmer factors, the defendants conclude that the necessary evidence for a ruling under the results standard is before this Court. They urge that the district court correctly concluded, on the basis of that evidence, that no section 2 violation occurred.
We reject the defendants’ arguments for several reasons. First, this Court emphasized in Marengo that “the Zimmer factors serve a different purpose in litigation under section 2 from their purpose in constitutional litigation.” Marengo, 731 F.2d at 1566. Some factors weigh more heavily under the results standard, while others weigh more heavily under the intent standard. In particular, we noted in Marengo that a showing of racially polarized voting “will ordinarily be the keystone of a dilution case” under section 2.8 Id. By contrast, although the responsiveness of elected officials to minority needs is an important factor under the intent test, it “is considerably less important under the results test.”9 Unresponsiveness is relevant under the results test only if the plaintiffs choose to make it so. Id. Moreover, Congress relegated another factor that is primary under the intent standard — tenuousness of state policies underlying an at-large scheme — to secondary importance under the results standard. Jones v. City of Lubbock, 5 Cir.1984, 727 F.2d 364, 384. It is therefore not enough for the defendants simply to note that the Zimmer factors constitute relevant evidence under both the intent and the results standard, because [1480]*1480the weighing of those factors is different under the two standards.
Second, we are unable to tell from the district court’s opinion how the court weighed the various factors to find that the plaintiffs had not established a violation of section 2. The district court in its original opinion applied “the criteria expressed by the Zimmer Court and the Supreme Court in City of Mobile v. Bolden”. Record at 306. The court concluded that the plaintiffs had “not proved by a preponderance of the evidence the existence of an aggregate of the factors which the Zimmer Court said would make out ‘a strong case’ ” and had “failed to prove by a preponderance of the evidence that the disputed plan was conceived or operated as a purposeful device to further racial discrimination”. Id. at 308-09. In part because the court’s opinion made no mention of section 2, the plaintiffs filed a motion to alter or amend the judgment, or for a new trial, asserting that their right to relief was particularly evident under the amended section 2. The court denied the plaintiffs’ motion in a short three-paragraph opinion that gave no detailed findings of fact or conclusions of law. The court stated,
In the interest of avoiding further litigation, this Court has reviewed this case in the light [of Rogers v. Lodge and the amended section 2] and is of the opinion that ... the preponderance of the evidence showing facts alleged and proved in this case do not indicate a violation of 42 U.S.C. § 1973 as last amended or of the Constitution . . . .
Record at 322.
The court’s second opinion does not articulate the basis for its consideration of the various factors that led it to find no violation of section 2. We have only the court’s explanation for its ruling under the intent standard. But the Zimmer factors carry different weights under the results test than they do under the intent test.10 We are therefore unable to review, on the record before us, the district court’s interpretation or application of the new legal standard embodied in section 2. Nor does the court’s second opinion satisfy the requirements of Fed.R.Civ.P. 52(a), which requires the court to “find the facts specially and state separately its conclusions of law thereon” in all actions tried upon the facts without a jury. “When, because of absence of findings of fact or conclusions of law, an appellate court cannot determine whether the record supports the trial court decision, it should remand the action for entry of findings of fact and conclusions of law.” Complaint of Ithaca Corp., 5 Cir. 1978, 582 F.2d 3, 4.
Finally, the evidence compiled in this record is now at least six years old and does not necessarily reflect current conditions. Moreover, because the record was not compiled under the results standard of section 2, it is incomplete on certain important issues, especially the “keystone” issue of racially polarized voting. The evidence in the case focused on two Zimmer factors of special relevance in proving voting dilution under the intent standard — whether the plaintiffs had equal access to the political process and whether government officials were responsive to the interests of the plaintiff minority.11 Thus, neither the plaintiffs nor the defendants have had an adequate opportunity to develop a record with the results standard in mind. Fairness dictates that the case be remanded.12 [1481]*1481Moreover, a remand is consistent with the Supreme Court’s action in Cross v. Baxter, 1983, 460 U.S. 1065, 103 S.Ct. 1515, 75 L.Ed.2d 942 (mem.), vacating 688 F.2d 279, 5 Cir.1982, which vacated a finding of unconstitutional vote dilution under Mobile and Rogers and remanded “for further consideration in light of Section 2 of the Voting Rights Act ... as amended in 1982”.
IV. THE PLAINTIFFS HAVE NOT ESTABLISHED A SECTION 2 VIOLATION ON THE PRESENT RECORD
The plaintiffs urge us to instruct the district court, in remanding the case to supplement the record concerning post-trial events, that a section 2 violation has been established on the present record. We decline to do so. Although this Court did so instruct in the Marengo case, that case presented much stronger proof of several factors of special importance to a showing of a section 2 violation. With respect to the “keystone” issue of racially polarized voting, the district court found “extremely strong” evidence of polarized voting in elections before 1978, and continuing, though reduced, polarization in the 1978 election. Marengo, 731 F.2d at 1567. The plaintiffs had proved polarization through direct statistical analysis of the vote returns. In addition, we noted that evidence of racial polarization could be gleaned from the Marengo County school board’s primary concern with placating whites and from the district court’s expression of serious doubts that Marengo County whites would attend desegregated black majority schools. Such attitudes, we noted, were “strong circumstantial evidence that race continues to dominate politics in Marengo County.” Id. at 1567 n. 35.
By contrast, in the record before us the plaintiffs’ evidence of racially polarized voting is weak. Neither the city nor the county kept racial data concerning registered voters, and the plaintiffs introduced no direct evidence of the racial composition of the Opelika electorate. The plaintiffs relied instead on estimates and circumstantial evidence that the rate of voter registration among blacks was lower than that among whites. More importantly, the plaintiffs offered no direct evidence of the specific racial composition of the various voting boxes in the city.13 This information is critical to a bloc voting analysis.
The plaintiffs have presented in their brief for this appeal a statistical calculation known as an R2 coefficient, which they maintain is evidence of racially polarized voting. This coefficient correlates the percentage of a particular racial group registered in a given precinct with the percentage of the precinct vote for the candidates of that racial group. This Court has acknowledged that the R2 coefficient is relevant to the issue of racial bloc voting. See NAACP v. Gadsden County School Board, 11 Cir.1982, 691 F.2d 978, 982-83; McMillan v. Escambia County, 5 Cir.1981, 638 F.2d 1239, 1241 n. 6 (former Fifth Circuit). The plaintiffs, however, did not present this calculation at trial, and, absent extraordinary circumstances, federal appellate courts will not consider evidence that was not part of the trial record. International Business Machines v. Edelstein, 2 Cir.1975, 526 F.2d 37, 45. Moreover, the defendants have not had a fair opportunity to confront this evidence.
Because it appears likely that the parties will want to develop evidence concerning the R2 coefficient on remand, we offer the following observations on the use of this evidence. First, in the two cases cited that made use of this evidence, the evidence was introduced through an expert. See Gadsden, 691 F.2d at 983; Escambia, 638 F.2d at 1241 n. 6. Moreover, the precise racial breakdown of registered voters was apparently known in both cases. See id. The R2 coefficients were therefore based on precise and detailed factual data concerning the racial composition of registered voters and election outcomes.14 The [1482]*1482plaintiffs in this case have not established on the present record a comparable factual database upon which to found their statistical analysis.
Second, we caution against placing too much reliance solely on the R2 coefficients in ruling on the issue of racially polarized voting. We agree with the cautionary remarks of Judge Higginbotham:
Care must be taken in the factual development of the existence of polarized voting because whether polarized voting is present can pivot the legality of at-large voting districts. The inquiry is whether race or ethnicity was such a determinant of voting preference in the rejection of black or brown candidates by a white majority that the at-large district, with its components, denied minority voters effective voting opportunity. In answering the inquiry, there is a risk that a seemingly polarized voting pattern in fact is only the presence of mathematical correspondence of race to loss inevitable in such defeats of minority candidates.
Jones v. City of Lubbock, 5 Cir.1984, 730 F.2d 233, 234 (Higginbotham, J., specially concurring).15 It will often be necessary to examine factors other than race that may also correlate highly with election outcomes — campaign expenditure, party identification, income, media advertising, religion, name recognition, position on key issues, and so forth. Well-established statistical methods, such as step-wise multiple regressions, can test the relative importance of multiple factors. Such analysis can assist in the determination of whether race is the dominating factor in political outcomes.
In addition to a much stronger showing of racially polarized voting than the record in this case presents, the plaintiffs in Mar-engo also presented more evidence than was presented here of election practices exacerbating the deficiencies in black participation. There was evidence in Marengo that appointments of poll officials were racially motivated and tended toward tokenism. The plaintiffs also showed that the County Board of Registrars was open only two days a month except in election years, and that, contrary to state law, the Board met only in the county seat and failed to visit outlying areas to register rural voters, who were more black than white. Maren-go, 731 F.2d at 1569-70. In Opelika, the registrar’s office is open every day of the week. One of the plaintiffs is a voting official and another has assisted voting officials in registering voters of both races. This record therefore presents a much weaker showing on two significant factors — racially polarized voting and election practices — than did the record in Marengo. We therefore decline to instruct the district court that the plaintiffs have established a section 2 violation on the present record.
V. CONCLUSION
We conclude, as in Marengo, that this case must be remanded to the district court [1483]*1483“to allow the parties to update the record and to supplement the record with evidence that might tend to affect [a] finding of discriminatory results.” Marengo, 731 F.2d at 1574. This case is now more than six years old. We do not intend that a complete retrial of every issue be had on remand. We suggest that the parties focus on those factors, referred to in this opinion, that in Marengo were central to the section 2 analysis, particularly racially polarized voting.
We VACATE the judgment of the district court that the plaintiffs did not establish a claim of voting dilution under the Fourteenth Amendment, the Fifteenth Amendment, and section 2 of the Voting Rights Act. We REMAND for further proceedings on the section 2 claim in accordance with this opinion.