This section 2 voting rights appeal raises one issue: Did the district court clearly err in finding legally significant white bloc voting in elections involving the Thirteenth Court of Appeals for the State of Texas? For the reasons discussed below, we conclude that the district court did commit clear error in finding--essentially on the basis of one election--that whites vote sufficiently as a bloc so as to usually defeat the preferred candidate of Hispanics in Thirteenth Court elections. Accordingly, we reverse the judgment of the district court.I. Background

In 1988, two Hispanic registered voters ("Plaintiffs") filed suit against various officials of Texas ("State Defendants"). They alleged that the manner in which Texas elects judges to the Thirteenth Court of Appeals violates section 2 of the Voting Rights Act of 1965, as amended, 42 U.S.C. § 1973. In particular, the Plaintiffs contended that the current practice of electing the six judges of the Thirteenth Court from an at-large election district, which covers some twenty counties, impermissibly dilutes the voting strength of Hispanics.

The section 2 liability issue was tried to the district court in April 1989. Thereafter, on July 28, 1989, the district court entered findings of fact and conclusions of law. The district court first found that, in 1988, Hispanics comprised 46% of the registered voters1 in the twenty-county area constituting the Thirteenth Judicial District. The court went on to find that the Plaintiffs had satisfied the three threshold requirements set forth in Thornburg v. Gingles, 478 U.S. 30, 106 S. Ct. 2752, 92 L. Ed. 2d 25 (1986). Specifically, the court found: (a) that four single-member districts could be drawn in which Hispanics would constitute 63.7% of the total population; (b) that Hispanics in the Thirteenth Judicial District are politically cohesive; and (c) that there is legally significant white bloc voting in the Thirteenth Judicial District. The district court then analyzed the totality of the circumstances or Zimmer factors--specifically, the factors listed in the Senate Report accompanying the 1982 amendments to section 2, see S.Rep. No. 417, 97th Cong., 2d Sess. (1982), reprinted in 1982 U.S.C.C.A.N. 177 (citing Zimmer v. McKeithen, 485 F.2d 1297 (5th Cir.1973) (en banc), aff'd per curiam sub nom. East Carroll Parish School Board v. Marshall, 424 U.S. 636, 96 S. Ct. 1083, 47 L. Ed. 2d 296 (1976)). It concluded that the following Zimmer factors weighed in favor of a vote dilution finding: (a) the history of "some discrimination in the 20-county area that touched the rights of Hispanics to participate in the political process"; (b) the "high degree of racial polarization within a majority of the counties in the 20-county area"; (c) the unusually large size and population of the Thirteenth Judicial District; and (d) socioeconomic disparities between Hispanics and Anglos.

Based on these findings, the district court concluded that the Plaintiffs had proven a section 2 violation. It specifically held that "the at-large nature of the election system used to elect judges to the Thirteenth Court of Appeals makes it more difficult for Hispanics to elect representatives of their choice, thus making the present process violative of law." The district court gave the parties thirty days "to meet and negotiate on a proposed remedy." The district court further instructed the parties, in the event they could not reach an agreement concerning the remedy, to separately submit their proposed remedies to the court within forty-five days.

The State Defendants immediately filed a notice of appeal challenging the district court's section 2 liability finding. In an "abundance of caution," the State Defendants further requested the district court to certify its liability determination pursuant to 28 U.S.C. § 1292(b). The district court declined to do so and instead entered a judgment adopting the Plaintiffs' proposed remedy on November 3, 1989. In this judgment, the district court ordered an "interim plan" to be implemented in "all future elections." This plan calls for, among other things, dividing the current Thirteenth Judicial District into six single-member districts. The State Defendants also filed a notice of appeal from this judgment, again indicating their intent to contest the district court's liability determination.2II. Analysis

The State Defendants argue that the district court's section 2 liability determination must be reversed. They argue specifically that the district court clearly erred in finding--on the basis of one election--that whites vote sufficiently as a bloc in the Thirteenth Judicial District so as usually to defeat the Hispanic-preferred candidate. We agree.

To establish legally significant white bloc voting under the Gingles threshold inquiry, minority plaintiffs "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--usually to defeat the minority's preferred candidate." Gingles, 478 U.S. at 51, 106 S.Ct. at 2766 (emphasis added). Said another way, to prove legally significant white bloc voting, minority plaintiffs must present evidence of "a white bloc vote that normally will defeat the combined strength of minority support plus white 'crossover' votes." Id. at 56, 106 S.Ct. at 2769. It is the "usual predictability of the majority's success [that] distinguishes structural dilution from the mere loss of an occasional election." Id. at 51, 106 S.Ct. at 2767.

The amount of white bloc voting that can generally cancel out minority voting strength will, of course, "vary from district to district according to a number of factors." Id. at 56, 106 S.Ct. at 2769. Among the factors affecting this inquiry is the percentage of registered voters in the district who are members of the minority group. Id. Where, as in the Thirteenth Judicial District, the minority group borders on constituting a majority of registered voters, it will probably be more difficult to establish a white bloc vote that will usually defeat the minority group's preferred candidate. Conversely, if the minority group constitutes only a small fraction of the total number of registered voters, it may be, relatively speaking, easier for the members of that group to establish their effective submergence in a white majority.

Because a district court's finding of legally significant white bloc voting is a question of fact, we review it for clear error. See Gingles, 478 U.S. at 77-80, 106 S.Ct. at 2780-2782; Westwego Citizens for Better Gov't v. City of Westwego, 872 F.2d 1201, 1203 (5th Cir.1989) ("Westwego I "); Campos v. City of Baytown, 840 F.2d 1240, 1243 (5th Cir.1988), cert. denied, 492 U.S. 905, 109 S. Ct. 3213, 106 L. Ed. 2d 564 (1989). That is, as long as the district court applies the appropriate legal standards, we will not reverse its finding of legally significant white bloc voting unless, based upon the entire record, we are "left with the definite and firm conviction that a mistake has been committed." Anderson v. City of Bessemer City, 470 U.S. 564, 573, 105 S. Ct. 1504, 1511, 84 L. Ed. 2d 518 (1985). If the district court's account of the evidence is plausible in light of the record viewed in its entirety, we will not reverse it--even if convinced that had we been sitting as trier of fact, we would have weighed the evidence differently. Id. at 573-74, 105 S.Ct. at 1511-12.

In this case, the district court found that there was legally significant white bloc voting in judicial elections involving the Thirteenth Court of Appeals by relying "heavily" on the 1984 Democratic primary race for the Thirteenth Court between Justice Horace Young ("Young") and Homer Salinas ("Salinas")--the only race for that court pitting an Anglo candidate against a Hispanic candidate. In that race, Salinas, who was clearly the preferred candidate of Hispanic voters, was defeated by a margin of 57% to 43%.3 Obviously, this race is highly probative. See, e.g., Magnolia Bar Ass'n, Inc. v. Lee, 994 F.2d 1143, 1149 (5th Cir.1993) (recapitulating that "elections involving the particular office at issue will be more relevant than elections involving other offices").4

However, since the focus of the third Gingles factor is upon the usual predictability of the majority's success, evidence of one or two elections may not give a complete picture as to voting patterns within the district generally. Thus, where, as here, there is evidence of only one election on all fours with the challenged process, that evidence must be viewed together with available evidence of other elections "that encompass more geographic area than just [the election district at issue]: e.g., ... state-wide or national elections,"5 to determine whether significant white bloc voting and the other Gingles elements are met. See Citizens for a Better Gretna v. City of Gretna, 834 F.2d 496, 502 & n. 13 (5th Cir.1987) (citing Gingles, 478 U.S. at 57 n. 25, 106 S. Ct. at 2769 n. 25), cert. denied, 492 U.S. 905, 109 S. Ct. 3213, 106 L. Ed. 2d 564 (1989); Westwego I, 872 F.2d at 1208.

Supplementing the evidence of the Young/Salinas race, evidence was offered by both sides of the following exogenous judicial elections pitting an Anglo candidate against a Hispanic candidate in the Democratic primary6 from which the district court could also find indications of racial voting patterns in the twenty-county area:

(1) The 1984 Democratic primary for the Texas Court of Criminal Appeals, in which George Martinez ("Martinez")--the Hispanic-preferred candidate--garnered 38% of the total vote and 11.3% of the Anglo vote in the twenty-county area, making him the top vote-getter;

(2) The 1986 Democratic primary for the Texas Court of Criminal Appeals, in which Martinez again received the majority of total vote--this time winning 47% of the total vote and 16% of the Anglo vote in the twenty-county area7;

(3) The 1986 Democratic primary runoff for the Texas Court of Criminal Appeals, in which Martinez received 60% of the total vote in the Thirteenth District area--including 22% of the white crossover vote;

(4) The 1986 Democratic primary for the Texas Supreme Court, in which Justice Raul Gonzalez ("Gonzalez")--the Hispanic-preferred candidate--was able to garner 60% of the total vote, including 28.7% of the Anglo vote, in the twenty-county area8; and

(5) The 1986 Democratic primary runoff for the Texas Supreme Court, in which Justice Gonzalez received 72% of the votes in the twenty-county area, including 40% of the Anglo vote, and defeated his Anglo opponent.9

The district court discounted these exogenous judicial races in large part because they were "state-wide." However, the exogenous character of the elections does not render them nonprobative in a case where there is only one election on all fours. The statistics showing the performance of these candidates over the entire twenty-county area is, in our view, quite indicative of district-wide voting patterns.

The trial court also considered the elections not to be highly probative because it found the winning candidate to be unique10 or because no evidence was introduced to determine the "home" area of the candidate. The court surmised that Gonzalez was "hypothetically" successful in the twenty-county area for his Supreme Court campaign because he was a native of the area. Conversely, it observed that there was no evidence as to the other candidates' home area. This "hypothetical" limitation upon the evidence is flawed because it involves speculation about a subjective factor affecting voter preference for which there is no evidence in the record.

Finally, the district court discredited the evidence involving Martinez' 1984 bid for the Court of Criminal Appeals since it was "theoretically" possible (there being no evidence on the subject) that he was not ultimately successful in the 1984 primary runoff. While that may be a basis for caution with respect to Martinez' victory in the 1984 primary, it does not undercut the fact that Martinez was clearly the top vote-getter throughout the twenty-county area in the 1986 Democratic primary, as well as in the 1986 run-off election. Moreover, the fact that Martinez may not have been successful state-wide does not militate against consideration of the overwhelmingly favorable results of the primary elections specific to the Thirteenth District.

In summary, the district court's reasons for disregarding the success of Hispanic candidates in these exogenous elections are largely conjecture, and its assumptions simply do not impeach the significant evidence that minority judicial candidates have been successful in the Thirteenth District. The evidence of these exogenous elections is demonstrative of voting patterns in the entire twenty-county area at issue since the statistics are district-wide and specific to the Thirteenth region. Moreover, the elections were for judicial positions and thus most closely resemble the elections at issue. The notable Hispanic success in these exogenous elections cuts heavily against a finding of legally significant white bloc voting. We find, therefore, that it was error for the district court to disregard this body of exogenous election evidence when it had only one indigenous election from which to consider the third Gingles factor.

The court below also discounted evidence from elections that took place in selected cities, counties, and independent school districts within the twenty-county voting district--which we refer to as "mosaic" or "building block" elections.11 We agree with the district court that these elections were of negligible probative value in the instant case. At trial, the Plaintiffs introduced evidence of city- and county-wide elections and school district elections, as well as a host of "mosaic" elections within Bee and San Patricio Counties, as evidence, inter alia, of legally significant white bloc voting within the entire twenty-county area challenged. The district court assessed this mosaic evidence as "unimpressive." We concur with the district court's view that this evidence "does little to establish whether Anglos are able to defeat the minority's preferred candidate" in the twenty-county area. See also Carrollton Branch of NAACP v. Stallings, 829 F.2d 1547, 1558, 1560 (11th Cir.1987) (Tuttle, J.) (rejecting evidence of city-wide election contests as not probative of county-wide voting practices), cert. denied, 485 U.S. 936, 108 S. Ct. 1111, 99 L. Ed. 2d 272 (1988); cf. Mississippi State Chapter, Operation PUSH v. Mabus, 932 F.2d 400, 409-11 (5th Cir.1991) (evidence of insignificant voter registration disparities within individual towns or counties did not impeach evidence of pronounced disparity with respect to overall state registration rates).

Although we express no opinion about use of a "building block" or "mosaic" theory in general--i.e., showing legally significant white bloc voting by referring to elections from areas smaller than the election district--we agree with the district court that the Plaintiffs in this case did not present evidence from enough of the "blocks" within the twenty-county area to be probative of voting patterns in the district as a whole. Both parties agree that the demographics of the Thirteenth Judicial District vary enormously between the southern, Hispanic-concentrated counties and the northern, Anglo-concentrated ones. Most of the building block evidence introduced involved cities, independent school districts, or counties (particularly Bee and San Patricio Counties) within the twenty-county area making up the Thirteenth District, and, like the district court, we are not able to conclude that this evidence from a scattering of voting subsets is indicative of county-wide voting--or more particularly district-wide voting--as a whole.12

In conclusion, we do not find the district court's finding of legally significant white bloc voting to be plausible in light of the record viewed in its entirety. Although the district court could reasonably give more weight to the 1984 Young/Salinas Democratic primary race in assessing the white bloc voting inquiry, this single race is not sufficient, in our view, to support a determination that the white bloc vote will usually defeat the preferred candidate of Hispanics in Thirteenth Court elections. Given the limited number of Thirteenth Court elections, the district court should not have discounted several exogenous judicial elections in which the Hispanic (and Hispanic-preferred) candidates won--in one case with substantial support from Anglo voters. See Gretna, 834 F.2d at 502 (recognizing that exogenous elections become more relevant where the available data is sparse). In light of these elections, the defeat of Salinas in the 1984 Democratic primary looks much more like the loss of an "occasional election"--not like evidence of the "usual predictability of the majority's success." See Gingles, 478 U.S. at 51, 106 S.Ct. at 2767.

Our conclusion that the district court clearly erred in finding legally significant white bloc voting is reinforced by the fact that, at the time of trial, Hispanics constituted 46% of the registered voters in the Thirteenth Judicial District. The evidence at trial revealed that Hispanic voters could control election outcomes with relatively little support from Anglo voters. In the 1986 Democratic primary runoff between Martinez and Duncan for the Texas Court of Criminal Appeals, Martinez received only 22.4% of the Anglo vote, yet he won the support of 60% of the total vote. Even when Anglo crossover voting was at a low of 13.8%, as it was in the 1984 Young/Salinas race, the Hispanic-preferred candidate was able to garner 43% of the total vote.

Accordingly, we hold that the district court clearly erred in finding legally significant white bloc voting. The 1984 Young/Salinas race--whether viewed standing alone or in conjunction with the exogenous judicial races--is simply insufficient to show that the white bloc vote in the twenty-county area usually defeats the preferred candidate of Hispanic voters in elections involving the Thirteenth Court of Appeals. In short, we are left with the definite and firm conviction that the district court was mistaken in finding to the contrary.

Because the district court clearly erred in finding legally significant white bloc voting in elections involving the Thirteenth Court of Appeals, its ultimate finding of vote dilution is also erroneous. See Overton v. City of Austin, 871 F.2d 529, 538 (5th Cir.1989) (failure to establish any one of the Gingles preconditions is fatal to a vote dilution claim). The Plaintiffs did not meet their burden of establishing their submergence in a white majority. The judgment of the district court is therefore REVERSED and RENDERED in favor of the State Defendants.