dissenting.
The issue for the District Court was whether racial considerations were predominant in the design of North Carolina’s Congressional District 12. The issue for this Court is simply whether the District Court’s factual finding— that racial considerations did predominate — was clearly erroneous. Because I do not believe the court below committed clear error, I respectfully dissent.
I
The District Court’s conclusion that race was the predominant factor motivating the North Carolina Legislature is a factual finding. See Hunt v. Cromartie, 526 U.S. 541, 549 (1999); Lawyer v. Department of Justice, 521 U.S. 567, 580 (1997); Shaw v. Hunt, 517 U.S. 899, 905 (1996); Miller v. Johnson, 515 U.S. 900, 910 (1995). See also Anderson v. Bessemer City, 470 U.S. 564, 573 (1985) (“[Intentional discrimination is a finding of fact...”). Accordingly, we should not overturn the District Court’s determination unless it is clearly erroneous. See Lawyer, swpra, at 580; Shaw, supra, at 910; Miller, supra, at 917. We are not permitted to reverse the court’s finding “simply because [we are] convinced that [we] would have decided the case differently.” Anderson, supra, at 573. “Where there are two permissible views of the evidence, the factfinder’s choice between them cannot be clearly erroneous.” 470 U.S., at 574. We should upset the District Court’s finding only if we are “‘left with the definite and firm conviction that a mistake has been committed.’ ” Id., at 573 (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395 (1948)).
The Court does cite cases that address the correct standard of review, see ante, at 242, and does couch its conclusion in “clearly erroneous” terms, see ante, at 257-258. But these incantations of the correct standard are empty gestures, contradicted by the Court’s conclusion that it must *260engage in “extensive review.” See ante, at 243. In several ways, the Court ignores its role as a reviewing court and engages in its own factfinding enterprise.1 First, the Court suggests that there is some significance to the absence of an intermediate court in this action. See ante, at 242-243. This cannot be a legitimate consideration. If it were legitimate, we would have mentioned it in prior redistricting cases. After all, in Miller and Shaw, we also did not have the benefit of intermediate appellate review. See also United States v. Oregon State Medical Soc., 343 U.S. 326, 330, 332 (1952) (engaging in clear error review of factual findings in a Sherman Act ease where there was no intermediate appellate review). In these cases, we stated that the standard was simply “clearly erroneous.” Moreover, the implication of the Court’s argument is that intermediate courts, because they are the first reviewers of the factfinder’s conclusions, should engage in a level of review more rigorous than clear error review. This suggestion is not supported by law. See Fed. Rule Civ. Proc. 52(a) ('‘Findings of fact, whether based on oral or documentary evidence, shall not be set aside unless clearly erroneous . . .”). In fact, the very ease the Court cited to articulate clear error review discussed the standard as it applied to an intermediate appellate court, which obviously did not have the benefit of another layer of review. See ante, at 242 (citing Anderson, supra, at 573).
Second, the Court appears to discount clear error review here because the trial was “not lengthy.” Ante, at 243. Even if considerations such as the length of the trial were relevant in deciding how to review factual findings, an as*261sumption about which I have my doubts,2 these considerations would not counsel against deference in this action. The trial was not “just a few hours” long, Bose Corp. v. Consumers Union of United States, Inc., 466 U. S. 485, 500 (1984); it lasted for three days in which the court heard the testimony of 12 witnesses. And quite apart from the total trial time, the District Court sifted through hundreds of pages of deposition testimony and expert analysis, including statistical analysis. It also should not be forgotten that one member of the panel has reviewed the iterations of District 12 since 1992. If one were to calibrate clear error review according to the trier of fact’s familiarity with the case, there is simply no question that the court here gained a working knowledge of the facts of this litigation in myriad ways over a period far longer than three days.
Third, the Court downplays deference to the District Court’s finding by highlighting that the key evidence was expert testimony requiring no traditional credibility determinations. See ante, at 248. As a factual matter, the Court overlooks the District Court’s express assessment of the legislative redistricting leader’s credibility. See Cromartie v. Hunt, 133 F. Supp. 2d 407, 419, 420, n. 8 (EDNC 2000). It is also likely that the court’s interpretation of the e-mail written by Gerry Cohen, the primary drafter of District 12, was influenced by its evaluation of Cohen as a witness. See id., at 420, n. 8. See also App. 261-268. And, as a legal matter, the Court’s emphasis on the technical nature of the *262evidence misses the mark. Although we have recognized that particular weight should be given to a trial court’s credibility determinations, we have never held that factual findings based on documentary evidence and expert testimony justify “extensive review,” ante, at 243. On the contrary, we explained in Anderson that “[t]he rationale for deference ... is not limited to the superiority of the trial judge’s position to make determinations of credibility.” 470 U.S., at 574. See also Fed. Rule Civ. Proe. 52(a) (specifically referring to oral and documentary evidence). Instead, the rationale for deference extends to all determinations of fact because of the trial judge’s “expertise” in making such determinations. 470 U.S., at 574. Accordingly, deference to the factfinder “is the rule, not the exception,” id., at 575, and I see no reason to depart from this rule in the case before us now.
Finally, perhaps the best evidence that the Court has emptied clear error review of meaningful content in the redistricting context (and the strongest testament to the fact that the District Court was dealing with a complex fact pattern) is the Court’s foray into the minutiae of the record. I do not doubt this Court’s ability to sift through volumes of facts or to argue its interpretation of those faets persuasively. But I do doubt the wisdom, efficiency, increased accuracy, and legitimacy of an extensive review that is any more searching than dear error review. See id., 574-575 (“Duplication of the trial judge’s efforts ... would very likely contribute only negligibly to the accuracy of fact determination at a huge cost in diversion of judicial resources”). Thus, I would follow our precedents and simply review the District Court’s finding for clear error.
II
Reviewing for clear error, I cannot say that the District Court’s view of the evidence was impermissible.3 First, the *263court relied on objective measures of compactness, which show that District 12 is the most geographically scattered district in North Carolina, to support its conclusion that the district’s design was not dictated by traditional districting concerns. 133 F. Supp. 2d, at 419. Although this evidence was available when we held that summary judgment was inappropriate, we certainly did not hold that it was irrelevant in determining whether racial gerrymandering occurred. On the contrary, we determined that there was a triable issue of fact. Moreover, although we acknowledged "that a district’s unusual shape can give rise to an inference of political motivation,” we "doubtfed] that a bizarre shape equally supports a political inference and a racial one.” Hunt, 526 U.S., at 547, n. 3. As we explained, "[s]ome districts . . . are ‘so highly irregular that [they] rationally cannot be understood as anything other than an effort to segregate]... voters’ on the basis of race.” Ibid, (internal quotation marks omitted).
Second, the court relied on the expert opinion of Dr. Weber, who interpreted statistical data to conclude that there were Democratic precincts with low black populations excluded from District 12, which would have created a more compact district had they been included.4 133 F. Supp. 2d, at 419. And contrary to the Court’s assertion, Dr. Weber did not merely examine the registration data in reaching his conclusions. Dr. Weber explained that he refo-*264eused his analysis on 'performance. He did so in response to our concerns, when we reversed the District Court’s summary judgment finding, that voter registration might not be the best measure of the Democratic nature of a precinct. See ibid, (citing Trial Tr., which appears at App. 90-92, 105-107, 156-157). This fact was not lost on the District Court, which specifically referred to those pages of the record covering Dr. Weber’s analysis of performance.
Third, the court credited Dr. Weber’s testimony that the districting decisions could not be explained by political motives.5 133 F. Supp. 2d, at 419. In the first instance, I, like the Court, ante, at 246-247, might well have concluded that District 12 was not significantly “safer” than several other districts in North Carolina merely because its Democratic reliability exceeded the optimum by only 3 percent. And I might have concluded that it would make political sense for incumbents to adopt a “the more reliable the better” policy *265in districting. However, I certainly cannot say that the court’s inference from the facts was impermissible.6
Fourth, the court discredited the testimony of the State’s witness, Dr. Peterson. 133 F. Supp. 2d, at 420 (explaining that Dr. Weber testified that Dr. Peterson’s, analysis “ignored] the core,” “ha[d] not been appropriately done,” and was “unreliable”). Again, like the Court, if I were a district court judge, I might have found that Dr. Weber’s insistence that one could not ignore the core was unpersuasive.7 However, even if the core could be ignored, it seems to me that Dr. Weber’s testimony — that Dr. Peterson had failed to analyze all of the segments and thus that his analysis was incomplete, App. 119-120 — reasonably could have supported the court’s conclusion.
Finally, the court found that other evidence demonstrated that race was foremost on the legislative agenda: an e-mail *266from the drafter of the 1992 and 1997 plans to senators in charge of legislative redistricting, the computer capability to draw the district by race, and statements made by Senator Cooper that the legislature was going to be able to avoid Shaw's majority-minority trigger by ending just short of the majority.8 133 F. Supp. 2d, at 420. The e-mail, in combination with the indirect evidence, is evidence ample enough to support the District Court’s finding for purposes of clear error review. The drafter of the redistrieting plans reported in the bluntest of terms: “I have moved Greensboro Black community into the 12th [District], and now need to take ... 60,000 out of the 12th [District].” App. 369. Certainly the District Court was entitled to believe that the drafter was targeting voters and shifting district boundaries purely on the basis of race. The Court tries to belittle the import of this evidence by noting that the e-mail does not discuss why blacks were being targeted. See ante, at 254. However, the District Court was assigned the task of determining whether, not why, race predominated. As I see it, this inquiry is sufficient to answer the constitutional question because racial gerrymandering offends the Constitution whether the motivation is malicious or benign. It is not a defense that the legislature merely may have drawn the district based on the stereotype that blacks are reliable Demo-*267eratic voters. And regardless of whether the e-mail tended to show that the legislature was operating under an even stronger racial motivation when it was drawing District 1 than when it was drawing District 12, cf. ibid., I am convinced that the District Court permissibly could have accorded great weight to this e-mail as direct evidence of a racial motive. Surely, a decision can be racially motivated even if another decision was also racially motivated.
If I were the District Court, I might have reached the same conclusion that the Court does, that “[t]he evidence taken together ... does not show that racial considerations predominated in the drawing of District 12’s boundaries,” ante, at 257. But I am not the trier of fact, and it is not my role to weigh evidence in the first instance. The only question that this Court should decide is whether the District Court’s finding of racial predominance was clearly erroneous. In light of the direct evidence of racial motive and the inferences that may be drawn from the circumstantial evidence, I am satisfied that the District Court’s finding was permissible, even if not compelled by the record.
Despite its citation of Bose Corp. v. Consumers Union of United States, Inc., 466 U.S. 485 (1984), ante, at 243,1 do not read the Court’s opinion to suggest that the predominant factor inquiry, like the actual malice inquiry in Bose, should be reviewed de novo because it is a “constitutional fac[t].” 466 U.S., at 515 (Rehnquist, J., dissenting). Nor could it, given our holdings in Lawyer v. Department of Justice, 521 U.S. 567 (1997), Miller v. Johnson, 515 U.S. 900 (1995), and Shaw v. Hunt, 517 U.S. 899 (1996).
Bose, which the Court cites to support its discounting of dear error review, ante, at 243, does state that “the likelihood that the appellate court will rely on the presumption [of correctness of factual findings] tends to increase when trial judges have lived with the controversy for weeks or months instead of just a few hours.” 466 U.S., at 500. It is unclear, however, what bearing this statement of fact — that appellate courts will defer to factual findings more often when the trial was long — had on our understanding of the scope of dear error review. In Bose, we held that a lower court’s “actual malice” finding must be reviewed de novo, see id., at 514, not that dear error review must be calibrated to the length of trial.
I assume, because the District Court did, that the goal of protecting incumbents is legitimate, even where, as here, individuals are incumbents *263by virtue of their election in an unconstitutional racially gerrymandered district. No doubt this assumption is a questionable proposition. Because the issue was not presented in this action, however, I do not read the Court’s opinion as addressing it,
I do not think it necessary to impose a new burden on appellees to show that districting alternatives would have brought about “significantly greater racial balance.” Ante, at 258. I cannot say that it was impermissible for the court to conclude that race predominated in this action even if only a slightly better district could be drawn absent racial considerations. The District Court may reasonably have found that racial motivations predominated in selecting one alternative over another even if the net effect on racial balance was not “significant.”
Dr. Weber admitted that, when he first concluded that race was the motivating factor, he was under the mistaken impression that the legislature's computer program provided only racial, not political, data. The Court finds that this admission undercut the validity of Dr. Weber’s conclusions. See ante, at 249-250. Although the District Court could have found that this impression was a sufficiently significant assumption in Dr. Weber’s analysis that the conclusions drawn from the analysis were suspect, it was not required to do so as a matter of logic. The court reasonably could have believed that the false impression had very little to do with the statistical analysis that was largely responsible for Dr. Weber’s conclusions.
In addition, the Court discounts Dr. Weber’s testimony because he "express[ed] disdain for a process that we have cautioned courts to respect,” ante, at 250. Dr. Weber did openly state that he believes that the best districts he had seen in the 1990’s were those drawn by judges, not by legislatures. App. 150-151. However, whether Dr. Weber was simply stating the conclusions he has reached through his experience or was expressing a feeling of contempt toward the legislature is precisely the kind of tone, demeanor, and bias determination that even the Court acknowledges should be left to the factfinder, c£ ante, at 243.
The Court also criticizes Dr. Weber's testimony that Prednct 77s split was racially motivated and his proposed alternative that all of Prednct 77 could have been moved into District 9. Apparently the Court believes that it is obvious that the Republican incumbent in District 9 would not have wanted the whole of Prednct 77 in her district. See ante, at 248. But the Court addresses only part of Dr. Weber's alternative of how the districts could have been drawn in a race-neutral fashion. Dr. Weber explained that the alternative was not simply to move Prednct 77 into District 9. The alternative would also indude moving other reliably Democratic predncts out of District 9 and into District 12, which presumably would have satisfied the incumbent. App. 157. This move would have had the result, not only of keeping Precinct 77 intact, but also of widening the corridor between the eastern and western portions of District 9 and thereby increasing the functional contiguity. The Court's other criticism, that moving all of Precinct 77 into District 12 would not work, is simply a red herring. Dr. Weber talked only of moving all of Prednct 77 into District 9, not of moving all of Prednct 77 into District 12.
Of course, considering that District 12 has never been constitutionally drawn, Dr. Weber's criticism — that the problem with the district lies not just at its edges, but at its core — is not without force.
The court also relied on the statement of legislative redistricting leader Senator Cooper to the North Carolina Legislature, see 133 F. Supp. 2d, at 419, in which the senator mentioned the goals of geographical, political, and racial balance, App. 460. In isolation, this statement does appear to support only the finding that race was a motive. Unlike this Court, however, the District Court had the advantage of listening to and watching Senator Cooper testify. I therefore am in no position to question the court’s likely analysis that, although Senator Cooper mentioned all three motives, the predominance of race was apparent. This determination was made all the more reasonable by the faet that the District Court found the senator's claim regarding the “happenstance” final composition of the district to lack credibility in light of the e-mail. 133 F. Supp. 2d, at 420, n. 8.