MILWAUKEE BRANCH OF THE N.A.A.C.P., et al., Plaintiffs-Appellants, v. Tommy THOMPSON, Governor of Wisconsin, et al., Defendants-Appellees, Wisconsin Association of Trial Judges, et al., Intervenors-Appellees.
Milwaukee County in Wisconsin elects 46 judges to the circuit court, its trial court of general jurisdiction, and 4 judges to the state's court of appeals. Elections are held at large. Each candidate for the circuit court runs for a numbered post, called a branch, though every trial judge has county-wide jurisdiction. When this suit began, 3 of the 46 seats on the circuit court, and none of the 4 on the appellate court, were held by black lawyers. Approximately 20 percent of the population, 16 percent of the voting age population, and 3 percent of the lawyers in Milwaukee County are black. The percentage of judges who are black is substantially below the first two numbers, but twice the percentage of blacks in the pool of persons eligible for election. (Wisconsin limits judicial office to persons who have been members of the bar for at least five years.)
Plaintiffs, registered voters in Milwaukee County and organizations to which they belong, contend that § 2 of the Voting Rights Act of 1965, 42 U.S.C. § 1973, requires Wisconsin to replace county-wide elections with smaller districts, which could be drawn so that some districts contain majorities of black voters. Chisom v. Roemer, 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d 348 (1991), and Houston Lawyers' Ass'n v. Attorney General of Texas, 501 U.S. 419, 111 S.Ct. 2376, 115 L.Ed.2d 379 (1991), hold that § 2 of the Voting Rights Act applies to judicial elections, and Thornburg v. Gingles, 478 U.S. 30, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986), establishes the framework for analyzing multi-member districts under § 2:
Plaintiffs must show three threshold conditions: first, the minority group “is sufficiently large and geographically compact to constitute a majority in a single-member district”; second, the minority group is “politically cohesive”; and third, the majority “votes sufficiently as a bloc to enable it ․ to defeat the minority's preferred candidate.” 478 U.S. at 50-51, 106 S.Ct. at 2765-2766. Once plaintiffs establish these conditions, the court considers whether, “on the totality of circumstances,” minorities have been denied an “equal opportunity” to “participate in the political process and to elect representatives of their choice.” 42 U.S.C. § 1973(b).
Abrams v. Johnson, 521 U.S. 74, ---- - ----, 117 S.Ct. 1925, 1935, 138 L.Ed.2d 285 (1997) (ellipsis in original). The possibility of increasing minority representation does not compel a jurisdiction to achieve that outcome, unless the three conditions have been met and the judge is satisfied that minority voters have lacked an equal opportunity to participate in the political process. Johnson v. De Grandy, 512 U.S. 997, 114 S.Ct. 2647, 129 L.Ed.2d 775 (1994).
After a trial, the district court concluded that plaintiffs did not establish the third threshold condition, because white voters in Milwaukee County have not consistently voted against black judicial candidates. 935 F.Supp. 1419, 1427-30 (E.D.Wis.1996). Moreover, the court explained, it would not have resolved the case in plaintiffs' favor even if it had found all three conditions satisfied. The court wrote that the state has a substantial interest in electing judges from the same jurisdiction in which they exercise jurisdiction. 935 F.Supp. at 1430-31, citing Southern Christian Leadership Conference v. Sessions, 56 F.3d 1281, 1296-97 (11th Cir.1995) (en banc) (SCLC); see also Nipper v. Smith, 39 F.3d 1494 (11th Cir.1994) (en banc). Judges do not “represent” voters, and their work affects all litigants and potential litigants, not just their constituents. The district court also thought it significant that, although the pool of persons eligible to be judges is only 3 percent black, about 6.5 percent of judges serving in Milwaukee County are black. District Judge Gordon recognized that the Voting Rights Act entitles voters to equal influence and is not (directly) concerned with the interests of persons who seek public office, but the outcome of the process is entitled to weight, see Baird v. Indianapolis, 976 F.2d 357, 360 (7th Cir.1992)-just as a shortfall of black judges compared with the pool of attorneys eligible for election would have been a mark in plaintiffs' favor.
Whenever the outcome depends on an assessment of the totality of the circumstances, appellate review is deferential. Abrams holds this for § 2 in particular. A totality-of-circumstances approach means that there is no entitlement to a particular outcome if one proves specific facts. There are only shadings and nuance, which different people estimate differently. Litigants are entitled to an honest application of the legal framework, not to a particular result. Inevitable differences in balancing incommensurables do not justify appellate substitution of judgment; when assessments differ, the trial court's prevails because there is no rule of law by which its decision may be called “wrong.” One could say that three judges' gestalts are better than one, because less likely to be idiosyncratic, but this observation has not carried the day with the Supreme Court. E.g., Cooter & Gell v. Hartmarx Corp., 496 U.S. 384, 399-405, 110 S.Ct. 2447, 2457-60, 110 L.Ed.2d 359 (1990). Appellate judges' reactions are less likely to be informed by an accurate appreciation of the picture painted by the evidence at trial, and having three judges duplicate the work of one, when it is not possible to identify right answers with any confidence, is an imprudent investment. So we must accept the district court's characterization unless it is clearly erroneous. That the judge considered legal questions on the way to resolving the ultimate issue does not authorize the court of appeals to depart from the usual deferential approach. Pullman-Standard v. Swint, 456 U.S. 273, 285-90, 102 S.Ct. 1781, 1788-90, 72 L.Ed.2d 66 (1982). If a legal error influences a finding, we may if necessary remand the case for a new trial, but we do not think that any of the judge's conclusions can be attributed to a mistake of law. To the contrary, the district judge followed the sinuous path of § 2 law with great care.
Take for example the finding that white voters in Milwaukee County do not vote so cohesively that they regularly defeat the candidates preferred by black voters. Between 1972 and the trial, 16 judicial contests (6 primaries and 10 general elections) pitted black against white candidates; in another 6 campaigns black candidates ran unopposed. In the 6 contested primaries, the black candidates received about 24 percent of the votes cast by white voters. All of the primaries had at least 3 candidates; in none did any candidate receive more than 50 percent of the votes cast by whites. Black candidates advanced to the general election in 5 of these 6 primaries. (Wisconsin uses a “nonpartisan” system of elections to the judiciary, so the top 2 candidates in the primary advance to the general election. If only 2 candidates enter the contest there is no primary.) In the 10 contested general elections, the black candidates received on average 40 percent of the votes cast by whites. In 2 of the 10 elections, the black candidates prevailed. Of course, the black candidate prevailed all 6 times in uncontested elections, and the district court inferred from the lack of contest that white voters would have been willing to support these candidates; otherwise a white candidate could have entered and won. Since 1980 black judges running for reelection have been opposed only 3 times, and they have prevailed in 2 of these races. Meanwhile 4 white judges lost or withdrew from races for reelection. These results, the district judge wrote, show not only substantial willingness among white voters to support black candidates, but also the ability of these candidates to gain and keep office. Bloc voting among white voters does not “consistently defeat minority candidates”, Gingles, 478 U.S. at 100, 106 S.Ct. at 2791 (O'Connor, J., concurring) and therefore does not establish the third of the necessary conditions, the judge concluded.
Plaintiffs might have replied by challenging the understanding that the third component of the threshold showing requires proof that white bloc voting “consistently” defeats black candidates. Justice O'Connor's concurring opinion in Gingles is not entirely harmonious with Justice Brennan's plurality opinion on this score, and later cases have not directed sustained attention to the issue. But plaintiffs do not contest the district court's reliance on Justice O'Connor's view. Nor do they contend that, when black voters are as small a minority as they are in Milwaukee County, the fact that black candidates receive significant support from white voters says more about the candidates than about the voters. Drawing inferences from voting patterns is hazardous, because a combination of voters and candidates decides elections-and candidates' choices may mask the inclinations of the electorate. Suppose most white voters in Milwaukee are strongly disinclined to vote for a black candidate for judge. Unless black attorneys want to tilt at windmills, they would not run for judicial office. Those who do would have some quality-unusually favorable public prominence, copious financing, demonstrated accomplishments that ensure endorsements from newspapers and bar associations-that helps them overcome the white electorate's 'druthers. Outside observers would see substantial white support for these black candidates, but this would be deceiving; in races between equally qualified (equally well known, etc.) white and black candidates, the whites would win every time. Strategic decisions by would-be candidates make inferences based on results at the polls unreliable; reluctance of white voters to support black candidates then would be manifest in the scarcity of black candidates, not in the distribution of the votes when blacks do run.
There are countervailing effects. Suppose the income of black lawyers in Milwaukee County is less than that of white lawyers. This would make low-paying jobs in the public sector more attractive to black attorneys-and because their opportunity costs of seeking office (the income foregone during the campaign) would be lower, black attorneys would be willing to run despite a low chance of success at the polls. A lower success rate for black candidates then would reflect their economic circumstances rather than anything about the electorate.
Candidates' decisions could make evidence of bloc voting in the black electorate similarly unedifying. One strategy would be to run a campaign designed to attract 90 percent support from black voters and 40 percent support from white voters; the success of such a campaign would be more a tribute to the candidate than a prediction about how voters would behave if randomly selected candidates appeared. Yet the statistical models used by both sides in this litigation, and in similar suits under the Voting Rights Act, suppose that candidates appear randomly, and that the only difference among candidates is race. That premise is untrue, devaluing the inferences from the statistical analysis. Maybe it would be more useful to define racial polarization of the electorate as evidence that white support for a candidate falls as black support for the candidate rises. But, as we say, the plaintiffs make nothing of the problems in the statistics for Milwaukee County.
Taking the figures at face value, as the parties do, leads straight to the conclusion that the district court's findings are not clearly erroneous. During the last 25 years, black candidates have prevailed in 6 uncontested races, qualified for the general election in 5 of 6 contested primaries, and won 2 of 10 contested general elections. Black candidates did not lose “consistently.” Plaintiffs observe that both of the black candidates who prevailed in contested general elections had been appointed to the bench by the Governor and enjoyed the advantage of incumbency, but the fact remains that these judges' color did not lead the voters to turn them out. Black candidates, whether or not incumbent, attracted substantial support from white voters-greater than the level of support that led the Supreme Court in Abrams to hold that a finding adverse to the plaintiffs was not clearly erroneous. White support for black candidates in Georgia (the state involved in Abrams) ranged from a low of 22 percent to a high of 38 percent, depending on the office, see 517 U.S. 952, ----, 116 S.Ct. 1941, 2001-02, 135 L.Ed.2d 248, enabling black candidates to win a few contests. Not very many, as Justice Breyer's dissent in Abrams points out, 521 U.S. at ---- - ----, 117 S.Ct. at 1949 but the majority concluded that a few is enough to negate the third element of the threshold showing under Gingles. Levels of white support for black-preferred candidates in Milwaukee County judicial elections exceed the levels in Georgia, and black candidates are more likely to prevail in Milwaukee than in Georgia.
Notwithstanding all of this, plaintiffs (with the support of the United States as amicus curiae) insist that two legal errors by the district court vitiate its findings. They contend first that the district court should have ignored the uncontested elections, and second that the court should not have investigated potential non-racial reasons why black candidates lost in contested elections. On the first of these issues, the district court was on solid ground-and for the reason it gave. One good measure of white voters' willingness to support black candidates is the failure of white candidates to present themselves for election even when a majority of the electorate is white. Potential opponents concede the election only when they face certain defeat. That 6 black candidates ran without opposition therefore is highly informative. No decision of the Supreme Court suggests otherwise.
As for the second issue: we agree with plaintiffs to the extent they observe that some of the reasons the district judge considered are not necessarily independent of race, but this does not undercut the court's ultimate conclusion. The district judge analyzed 4 of the 8 general elections in which black candidates were defeated, concluding that in each of these 4 the white candidate enjoyed a big advantage-better name recognition, success in raising funds, ability to accuse his opponent of domestic abuse, and so on-seemingly unrelated to race. 935 F.Supp. at 1429. What turns out to be related to race is not easy to know, however. Maybe the victorious candidates were able to amass a larger war chest precisely because donors expected them to prevail, given the electorate's preference for white candidates. Name recognition, too, may be a proxy for antecedent preferences of the voters. A district judge therefore should not assign to plaintiffs the burden of showing why the candidates preferred by black voters lost; it is enough to show that they lost, if white voters disapproved these candidates en masse. Proving discriminatory intent is not part of the plaintiffs' case under § 2. Voinovich v. Quilter, 507 U.S. 146, 113 S.Ct. 1149, 122 L.Ed.2d 500 (1993); Barnett v. Daley, 32 F.3d 1196, 1202 (7th Cir.1994); Baird, 976 F.2d at 359-60.
It does not follow that the circumstances of individual elections must be banished from the analysis. They matter to the totality-of-circumstances inquiry even if not to the threshold case, and in principle they are relevant to the threshold case as well. Suppose black candidates prevailed in 2 of 10 contested general elections, and the only victories could be attributed to shortcomings of the white candidate rather than to the electorate holding the black candidates in high esteem. Perhaps one white candidate was brought down by revelations of ethical improprieties, and the other was deemed incompetent by all the bar associations and newspapers in the county. A judge then might properly conclude that white bloc voting has consistently defeated the black voters' preferred candidates in genuinely contested races, even though the candidates preferred by the black electorate won some races. And if analysis of this kind can in principle cut in favor of plaintiffs, it can cut in favor of defendants. As a rule, however, it would be best for district judges to postpone this kind of inquiry to their consideration of the totality of the circumstances, the second part of the Gingles inquiry. Yet although the district court did not discuss the potential link between race and funding levels (for example), neither did the judge assign to plaintiffs the burden of showing causation. Victories in other elections of candidates preferred by black voters, coupled with white voters' significant support for black candidates, relegates this part of the district court's analysis to the periphery.
Having concluded that the district judge did not commit clear error in finding that defendants prevail on the third element of the Gingles threshold showing, we could stop and affirm the judgment. But the possibility that plaintiffs will seek review in a higher tribunal, coupled with the recurring nature of the questions, leads us to address plaintiffs' contention that the district court erred in analyzing the totality of the circumstances. Plaintiffs, supported by the United States, contend that the district judge erred in giving any weight to defendants' contention that judges should be elected from districts coterminous with the court's geographic jurisdiction. They also submit that the judge erred in believing that the ability of black attorneys to gain judicial office out of proportion to the number of black attorneys in the county is legally significant.
The second of these arguments essentially asks us to overrule Baird, which held that a judge may consider the race of victorious candidates, as well as the voting patterns of the electorate, when conducting the second stage of the Gingles inquiry. See also Johnson v. De Grandy, 512 U.S. at 1012 n. 10, 114 S.Ct. at 2657 n.10. Gingles did not yield a majority opinion. We concluded in Baird that Justice White's separate opinion in Gingles had the support of five Justices, to the extent Justice White believed that courts may take account of prevailing candidates' race. Section 2(b) affords strong support: “The extent to which members of a protected class have been elected to the office in the State or political subdivision is one circumstance which may be considered.” 42 U.S.C. § 1973(b). A case such as ours shows why this consideration helps. If this were a Title VII case, and the defendant were an employer, then the fact that the work force at the employer had twice as many blacks as would have been produced by random draws from the applicant pool would be a powerful, though not dispositive, see Connecticut v. Teal, 457 U.S. 440, 102 S.Ct. 2525, 73 L.Ed.2d 130 (1982), argument against a claim that the employer discriminated against blacks-just as the failure of the employer to hire blacks in proportion to the applicant population would be a telling arrow in the plaintiffs' quiver. Milwaukee County hires judges, just as a large law firm hires lawyers. That hiring decisions are made by voting does not disable a court from using statistics to conclude that these hiring decisions are not discriminatory-and if the outcome does not discriminate against black attorneys (who, plaintiffs insist, are the candidates preferred by black voters), this is the basis for a rational inference that the rules for conducting elections (including the drawing of district lines) do not provide minority voters “less opportunity” to “participate in the political process and to elect representatives of their choice.” 42 U.S.C. § 1973(b). Plaintiffs want to use “input” statistics (whom black and white voters support) to build their case; defendants are entitled to use “output” statistics to undermine the inferences plaintiffs propose.
Let us return, then, to the question whether the court may give weight to a state's asserted interest in matching the boundaries of a court's jurisdiction to the boundaries of the judges' electoral base. Houston Lawyers' Ass'n, 501 U.S. at 427, 111 S.Ct. at 2381 tells us that “the State's interest in maintaining an at-large, district-wide electoral scheme for single-member [judicial] offices ․ does not automatically, and in every case, outweigh proof of racial vote dilution.” Judge Gordon did not treat this interest as “automatically” outweighing proof of vote dilution; he said no more than that the interest is weighty. In asking us to hold that this was error, plaintiffs really want us to say that the only proper weight is zero (or, if not nil, so low that it can never tip the balance). Such a conclusion would put us in conflict with en banc decisions of the fifth and eleventh circuits, which hold that this consideration is powerful, indeed dispositive unless the plaintiffs show gross racial vote dilution. League of United Latin American Citizens v. Clements, 999 F.2d 831, 872 (5th Cir.1993) (en banc) (LULAC); SCLC, 56 F.3d at 1294-95; Nipper, 39 F.3d at 1543; accord, Cousin v. McWherter, 46 F.3d 568, 577 (6th Cir.1995).
What can be said in favor of plaintiffs' position is this: (i) Judges do not “represent” the people who elect them, so a match of electoral base to geographic jurisdiction is less important for judicial than for legislative office. (ii) Much of the law judges use is established by people who live outside their jurisdiction (state and national legislatures, and the Supreme Court of Wisconsin, make rules local judges must employ), diminishing the importance of a link between electoral base and scope of jurisdiction. (iii) Many of the persons affected by judicial decisions are unconnected with the court's geographic base. Litigants may come from anywhere in the world, and none of these strangers has any legitimate interest in whether the judges were elected from the county as a whole or a smaller district. (iv) Wisconsin already has weakened the link between jurisdiction and electoral base. Four appellate judges are elected from Milwaukee County. Although they hear only appeals that originate from the local circuit court, their opinions are binding on the state's other intermediate appellate courts. If the state is content with this situation for the court of appeals, why not for the circuit court? And if the state is content to have four appellate judges elected from one county, why not one appellate judge from each of four districts within the county? (v) Other states have severed the link between electoral base and jurisdiction. In Cook County, Illinois, many judges of the court of first instance are elected from very small districts, although they exercise authority in the county as a whole. If Illinois finds this satisfactory, why does Wisconsin have a substantial interest in doing things otherwise?, one may ask.
Each of these points has some force, but individually and collectively they are not enough to overcome Wisconsin's interest in electing its trial judges from whole counties. Wisconsin created its system of at-large elections about 150 years ago, long before it had a substantial minority population, and the district court found (when rejecting a claim under the fifteenth amendment) that the system has not been maintained for discriminatory reasons. 929 F.Supp. 1150, 1159 (E.D.Wis.1996). It is a vestige of the Jacksonian era decision to elect judges. At-large elections from the whole of the judge's geographic jurisdiction are designed to balance accountability and independence. In a federal system, states are entitled to do things differently; Illinois' willingness to use subdistricts no more obliges Wisconsin to do so than the other way around. Not everyone in Illinois is enamored of its system, which by limiting election to attorneys who live in the subdistricts disqualifies many of the best candidates for these posts. The fifth circuit concluded that 25 of the 29 states that elect judges do things Wisconsin's way. LULAC, 999 F.2d at 872 & n. 33. Although judges in Milwaukee County draw much of the law they enforce from sources outside the county, and resolve disputes among people who live beyond its borders, they also make or influence much of the law they administer, and apply much of it to locals. Interstitial lawmaking is part of a common law system. See Gregory v. Ashcroft, 501 U.S. 452, 465-67, 111 S.Ct. 2395, 2403-04, 115 L.Ed.2d 410 (1991). And judges exercise much discretion even when they do not make law. Choice of electoral districts may influence that discretion, for better or for worse.
Wisconsin believes that election of judges from subdistricts would lead to a public perception (and perhaps the actuality) that judges serve the interests of constituencies defined by race or other socioeconomic conditions, rather than the interest of the whole populace. Larger jurisdictions liberate judges, to some degree, from the pressure created by the need to stand for reelection. A judge elected from a small district might fear that acquittal of a person charged with a crime against a member of that neighborhood, or a decision that harms an employer in that neighborhood, will lead to defeat at the polls. To free the judge to follow the law dispassionately, Wisconsin prefers to elect judges from larger areas, diluting the reaction to individual decisions. Perhaps the belief that judges favor those who elect them is unwarranted-though the diversity jurisdiction of the federal courts rests in part on a belief that state judges highly value the interests of that state's citizens and thus are potentially biased against citizens of other states. See Henry J. Friendly, Federal Jurisdiction: A General View 146-50 (1973). So too, perhaps, for smaller jurisdictions within a state. It is odd to find the United States, whose judges have secure tenure, arguing in this case that Wisconsin must treat its judiciary like ward heelers. At all events, the Voting Rights Act does not compel a state to disregard a belief that larger jurisdictions promote impartial administration of justice, if that belief is sincerely held-as the district judge concluded that it is in Wisconsin.
EASTERBROOK, Circuit Judge.