Appellants, the School Board of Hillsborough County, Florida, and its officials, appeal two orders of the district court which subject them to continued supervision under a federal desegregation decree. See Manning v. Sch. Bd. of Hillsborough County, Fla., 24 F. Supp. 2d 1277 (M.D.Fla.), mot. to alter or amend den., mot. for clarification granted in part, 28 F. Supp. 2d 1353 (M.D.Fla.1998). Appellants argue that they have eliminated the vestiges of past discrimination to the extent practicable and have fully complied in good faith with the desegregation decree. Accordingly, Appellants claim their school district should be declared unitary and federal judicial supervision should cease. Conversely, Appellees, a class of African-American schoolchildren, contend the school district is not unitary and federal judicial oversight of Appellants remains necessary. We hold that Appellants have achieved unitary status. We reverse and remand for the district court to enter judgment, in accordance with this opinion, declaring the Hillsborough County school system to be unitary.

Appellants for many years operated a racially-segregated, dual school system. As a result of the Supreme Court's landmark decision in Brown v. Board of Education of Topeka, 347 U.S. 483, 74 S. Ct. 686, 98 L. Ed. 873 (1954) (Brown I), Appellees in 1958 filed this class-action lawsuit on behalf of all "minor Negro children and their parents" residing in Appellants' school district.1 In 1962, the district court found that Appellants, by operating a segregated school system, had violated the Fourteenth Amendment. For the next eight and half years, the district court issued various orders as part of its efforts to remedy the harm caused by Appellants' unconstitutional conduct. See, e.g., Mannings v. Bd. of Pub. Instruction of Hillsborough County, Fla., 306 F. Supp. 497 (M.D.Fla.1969).

In 1970, our predecessor court examined whether Appellants had sufficiently eradicated the illegal dual school system such that it could be found "unitary." See Mannings v. Bd. of Pub. Instruction of Hillsborough County, Fla., 427 F.2d 874 (5th Cir.1970). Relying upon the six so-called Green2 factors, the former Fifth Circuit concluded that, with regard to three factors (transportation, extracurricular activities, and facilities), Appellants had indeed achieved a unitary school district. See Mannings, 427 F.2d at 878. Nonetheless, based on its examination of three other factors (faculty desegregation, staff desegregation, and student assignments), the court found Appellants had fallen short and had not attained unitary status. See id. The case was remanded to the district court with instructions to remedy the deficiencies. See id.

After remand, the Supreme Court, in Swann v. Charlotte-Mecklenburg Board of Education, 402 U.S. 1, 91 S. Ct. 1267, 28 L. Ed. 2d 554 (1971), gave firm guidance on a district court's equitable power to remedy illegal segregation. On May 11, 1971, just 21 days after Swann was decided, the district court directed Appellants to submit a comprehensive desegregation plan that conformed with the requirements of Swann. Thereafter, Appellants submitted such a plan, and the district court adopted the plan in its order dated July 2, 1971 (the July 1971 Order). From 1971 to 1991, the district court's supervision of Appellants was governed, with some minor modifications, exclusively by the July 1971 Order.3

In 1991, Appellants and Appellees entered into a consent decree (1991 Consent Order). The primary reason for the 1991 Consent Order was to enable Appellants to reorganize the school district, so as to eliminate single grade centers and to create middle schools. The 1991 Consent Order, which was to be implemented over a 7-year period, did not annul the July 1971 Order, but merely modified it.

Appellee moved in 1994 to enforce the 1991 Consent Order. The matter was referred to the magistrate judge who recommended denying the motion. The district judge, however, deferred ruling on the motion and sua sponte recommitted the matter to the magistrate judge to consider whether the school district had become unitary, thereby removing the need for federal judicial oversight.

In October 1996, the magistrate judge conducted a 7-day hearing, at which both sides presented considerable evidence. In August 1997, the magistrate judge issued a detailed report and recommendation wherein she recommended the district court find that Appellants had achieved unitary status and thus should be released from federal judicial supervision. Without holding an evidentiary hearing, the district judge in a 110-page order dated October 26, 1998, rejected in part and adopted in part the magistrate judge's report and recommendation. See Manning, 24 F. Supp. 2d at 1277-1335. The district judge concluded that Appellants had not attained unitary status and therefore federal judicial supervision was still warranted.4See Manning, 24 F. Supp. 2d at 1335. Within ten days of the order dated October 26, 1998, Appellants filed a motion to alter or amend judgment pursuant to Fed.R.Civ.P. 59(e). The district court, in a 13-page order, denied the motion on December 4, 1998. See Manning, 28 F. Supp. 2d at 1361. Within 30 days, Appellants filed a notice of appeal as to the district judge's orders of October 26, 1998, and December 4, 1998.

To analyze this case that has endured for over 40 years, we first summarize the contents of the July 1971 Order and the 1991 Consent Order, which, with minor modifications, have served as the guideposts for Appellants' journey toward a unitary school district. Then, we set forth the district court's most recent factual findings with respect to Appellants' unitary status.

As previously discussed, the district court ordered Appellants in May of 1971 to submit a comprehensive desegregation plan. The district court stated that the "primary objective" of the plan should be the abolition of segregation. In particular, the district court sought to eliminate, at every school, racial balances where black students comprised more than 50% of the student population. The district court further stated that the "most acceptable and desirable" result would be a white/black ratio of 86%/14% in senior high schools, 80%/20% in junior high schools, and 79%/21% in elementary schools.

The July 1971 Order, which ratified Appellants' proposed desegregation plan, dealt primarily with student assignments in grades one through twelve. For elementary schools, the plan clustered each predominately black school with two to five predominantly white schools. The white schools would be used for grades one through five and would be integrated with black children from "satellite zones"5 who had previously attended the black school in the cluster. The black school in each cluster would be used as a sixth-grade center and would be integrated with white children who had previously attended the white schools in the cluster.

For junior high schools, each black school would be clustered with one to three white schools. The black schools would be used for seventh grade, and the white schools would be used for eighth and ninth grades. Again, the schools would be integrated with students from satellite zones. For senior high schools, the two black schools would be closed, but their facilities would be used for junior high schools. Through a series of re-zoning and satellite busing measures, the white high schools would be integrated.

With regard to faculty and staff, the district court had previously entered an order on August 25, 1970. In the July 1971 Order, the court found that no additional measures were necessary, and it directed Appellants to continue to abide by the August 25, 1970, order. With respect to transportation, facilities, and extracurricular activities, the July 1971 Order directed Appellants to comply with all previous orders. The order also mandated that all operations relating to these areas should be conducted in a non-discriminatory manner and should be regularly re-examined by Appellants.

The July 1971 Order also remarked on four other topics: (1) majority to minority transfers, (2) other transfer rules, (3) the bi-racial committee, and (4) approval of site locations. Specifically, the district court stated, "Each of these has been required by previous orders of this Court. Some of them may not be required if [Appellants'] plan is effectuated and accomplished." Nevertheless, the district court stated that it was "retaining jurisdiction and [would] require the continuation of all of these procedures to be available and used as necessary."

Attached to the July 1971 Order were two exhibits which explained the bi-racial committee, majority-to-minority transfers, and other transfer rules. The bi-racial committee was to serve as an advisory body on a number of issues and was to consist of ten members, with Appellants and Appellees each selecting five members. In some instances, the committee was to receive reports about students transferring from their assigned schools. In other instances, Appellants were barred from approving student transfers before considering a recommendation from the committee.

Concerning majority-to-minority transfers, the attached exhibits stated the following:

Majority to minority transfer-Any student shall be permitted to transfer from a school in which his race is in the majority in order to attend the closest school to his residence in which his race is in the minority.

. . . .

The transfer forms shall be available at each public school in Hillsborough County and the County School offices.

. . . .

The transferee is to be given priority for space and thus the transfer is not to be dependent on space being available.

Transportation will be provided by the School Board in service or in kind to the school to which the transfer is made if that school is more than two miles from the home.

In conclusion, the July 1971 Order stated that Appellants' desegregation plan "fully complie[d] with the [district] [c]ourt's order of May 11, 1971," and that it would "result in the establishment of a unitary school system in Hillsborough County, Florida." The July 1971 Order noted, however, that Appellants had a "continuing responsibility" to ensure the plan would be effectuated. Moreover, each year following the implementation of the July 1971 Order, Appellants filed two reports per year with the district court and provided copies to Appellees. Lastly, the July 1971 Order retained jurisdiction in the district court for "such further action as may be necessary and required."

Between 1971 and 1991, the modifications to the July 1971 Order were minor.6 By 1991, however, Appellants had determined, based upon a comprehensive study, that the school district would benefit if middle schools (grades 6-8) were established. To accomplish this, the July 1971 Order had to be modified, because that order had relied extensively on single-grade centers (grades 6 and 7) to desegregate the school district. Thus, Appellants entered into negotiations with Appellees, and the product of those negotiations was the 1991 Consent Order.

The centerpiece of the 1991 Consent Order was the Middle School Task Force Report 3 of July 1991 (hereinafter "Task Force Report"), which the district court incorporated into the consent order.7The Task Force Report was presented to the school board by a 12-person committee. This committee included Mr. Henry Carley, President of the Tampa Branch of the NAACP. In addition, a legal committee, consisting of, inter alia, Mr. Al Davis of the NAACP, reviewed the Task Force Report, and Appellees' own desegregation expert, Dr. Leonard Stevens, was involved in shaping the Task Force Report. Furthermore, the Task Force Report included two separate position statements prepared by various African-American community groups (Howard W. Blake Alumni Group, Coalition of African American Organizations, Greater Tampa Urban League, Inc., and Beta Sigma Zeta Chapter of Zeta Phi Beta Sorority). In sum, as the 1991 Consent Order explained, the Task Force Report was the result of "extensive discussions" between Appellants and Appellees, and the discussions were designed to "to ensure ... that plans for the implementation of the [Task Force Report] were formulated in a manner that addressed the interests and concerns of [Appellees]."

The Task Force Report proposed substantial changes to the structure of the school system. Under the July 1971 Order, the school system generally had consisted of five tiers: elementary schools (grades K-5), sixth-grade centers, seventh-grade centers, junior high schools (grades 8-9), and high schools (grades 10-12). In contrast, the Task Force Report called for a three-tier school system: elementary schools (grades K-5); middle schools (grades 6-8); and high schools (grades 9-12). This new structure would be achieved through a so-called "cluster model," under which 17 clusters would be formed. For each cluster, the high school would serve as the "basic unit" and its students would be drawn from "feeder" middle and elementary schools. The Task Force Report estimated that it would take five to seven years to implement the cluster model.

The cluster model proposed by Task Force Report was undoubtedly the most substantial change to the structure of the school system since the July 1971 Order. The objective of July 1971 Order had been to desegregate the school system. By contrast, one of the primary objectives of the Task Force Report was "to maintain a desegregated school system." (emphasis added). In an apparent attempt to meet this objective, the 1991 Consent Order directed Appellants "to minimize (to the extent practicable) the number of schools which deviate from the system-wide student enrollment [race] ratios." Moreover, the 1991 Consent Order assumed that intervening demographic changes might necessitate future modifications in student assignments in order to maintain a desegregated school district.8

To further the goal of maintaining desired race ratios, the Task Force Report attempted to project the race ratios at each school after the cluster plan's implementation. In summary, the Task Force Report predicted:

The number of schools reflecting a ten or higher percent race ratio variance [from the recommended ratio of 20/80%] will increase from 36 to 46. The plan also increases the number of schools from 56 to 72 that will have an almost perfectly balanced race ratio with a student variance of five percent or less from the recommended ratio of 20/80% ratio.

Appendix 1 of the Task Force Report provided more detail as to the projected racial balance at each school. Fourteen schools were projected to have a 40% or greater black population,9 and an additional four schools were projected to have a 39% black population.10 These projections, as part of the Task Force Report, were incorporated into the 1991 Consent Order-a consent decree to which Appellees were parties.

Along with projecting race ratios, the Task Force Report instituted a variety of programs designed to maintain a desegregated school system. For example, the report implemented magnet programs,11 which were designed in part "to reduce minority isolation" and "to promote desegregation in schools."12 The report also suggested a multi-cultural curriculum for students in all grades and instructional sessions to enable staff to deal with diverse populations.

One omission in the Task Force Report, however, is particularly noteworthy in light of subsequent events: majority-to-minority transfers. Although they had been discussed briefly in the July 1971 Order, majority-to-minority transfers were not mentioned once in the 37-page Task Force Report, nor in the 8 attached appendices, nor in the 1991 Consent Order. Dr. Stevens-Appellees' desegregation expert who helped shape the Task Force Report-admitted at the 1996 evidentiary hearing that majority-to-minority transfers would have been insignificant in alleviating the racial imbalances in the school district.

The magistrate judge, in her 1997 report and recommendation, structured her factual findings around the six Green factors,13 plus a seventh factor, quality of education, which has been used by some courts to evaluate a school district's unitary status. See, e.g., Mills v. Freeman, 942 F. Supp. 1449, 1460 (N.D.Ga.1996), aff'd, 118 F.3d 727 (11th Cir.1997). In addition, the magistrate judge made factual findings regarding Appellants' good-faith compliance with past desegregation decrees, as required by Board of Education of Oklahoma City v. Dowell, 498 U.S. 237, 250, 111 S. Ct. 630, 638, 112 L. Ed. 2d 715 (1991), and its progeny. Based on all of these findings, the magistrate judge recommended the school district be declared unitary in all respects and that federal judicial supervision of the school district be terminated.

The district judge found that the magistrate judge's report and recommendation was "comprehensive" and "agree[d] with a majority of the [m]agistrate [j]udge's analysis." Manning, 24 F. Supp. 2d at 1287. For six of the seven factors (faculty, staff, transportation, extracurricular activities, facilities and resource allocation, and quality of education), the district judge adopted most, if not all, of the magistrate judge's findings.14 See id. at 1316-34. In their brief to this Court, Appellees do not contest the findings related to these six factors, and thus further elaboration is unnecessary except to emphasize that the findings support a declaration of unitary status.

With regard to the remaining Green factor (student assignments), the district judge expressed some disagreement with the magistrate judge. The district judge also differed with the magistrate judge's finding on Appellants' good-faith compliance. As we discuss in Part II, the district judge used an incorrect legal standard. As a result, the district judge's legal conclusion, though based mostly upon the magistrate judge's factual findings, was erroneous. To help illuminate our discussion in Part II, we narrate below the magistrate judge's factual findings related to student assignments and Appellants' good faith, and we highlight, when necessary, particular findings adopted by the district judge. We also note those few areas where the district judge rejected the factual findings of the magistrate judge.

a. Student Assignments

As of the 1995-96 school year, the Hillsborough County school district consisted of approximately 120,000 students, 108 elementary schools, 27 junior high schools, and 15 senior high schools. Relying on a standard proposed by Appellees' and Appellants' experts, the magistrate judge designated any school with a black/white ratio varying plus or minus 20 points from a 20/80 ratio as being "racially identifiable" or "racially imbalanced."15 It was undisputed that, by the 1971-72 school year, all schools were desegregated. Furthermore, the evidence showed that, at the time of the 1996 evidentiary hearing, approximately 90% of the schools were not racially identifiable. Appellees, however, identified 17 schools as racially identifiable and made these schools the focus of the 1996 evidentiary hearing.16 Ironically, of these 17 schools, 9 were among the 14 schools that the Task Force Report projected would have a 40% or greater black population, and two were among the four schools that the Task Force Report projected would have a 39% black population.17 Immediately after the July 1971 Order was implemented, all of these schools became racially balanced.18

The critical factual question at the evidentiary hearing was whether the racial imbalances at the 17 schools were caused by Appellants' past de jure segregation (or other discriminatory conduct), or whether, instead, the racial imbalances were caused by nondiscriminatory factors and circumstances. In a joint statement submitted prior to the evidentiary hearing, Appellants contended that demographic shifts had caused the schools to become racially identifiable. In the same joint statement, Appellees did not directly rebut Appellants' contention, but instead argued that "[d]emographic change alone does not account for the racial identifiability in the Hillsborough County school system." (emphasis added).

To resolve their factual dispute, the parties presented considerable evidence to the magistrate judge, including, inter alia, reports on attendance boundaries, demographic reports, and expert testimony. We need not repeat all this evidence here, as it is fully set forth in the opinion of the district court. See Manning, 24 F. Supp. 2d at 1290-1312. Instead, we compare the findings of the magistrate judge to those of the district judge.

The magistrate judge, in her report and recommendation, recounted the expert testimony of Dr. David Armor, one of Appellants' experts-testimony that the magistrate judge personally observed. Dr. Armor explained that he reviewed the extensive demographic data contained in a report by Dr. W.A.V. Clark, another of Appellants' experts. Based on this evidence, Dr. Armor opined on the witness stand that "none of the schools that were currently imbalanced ... were caused by [Appellants'] action[s,]" but rather such imbalances "were caused by demographics." The magistrate judge also reviewed the testimony of Dr. Fred Shelley, Appellees' demographic expert-once again, testimony that the magistrate judge personally observed. Dr. Shelley did not conduct his own demographic study, but rather relied on the data supplied by Dr. Clark (Appellants' expert). Dr. Shelley opined "that it is perhaps difficult to conclude that [the] increase in racial imbalance is attributable solely to the processes of natural demographic change." In the end, the magistrate judge agreed with Dr. Armor and found that demographic change is the "most likely explanation" for the racial imbalances. The magistrate judge faulted Dr. Shelley for not providing an alternative explanation for the racial imbalances.19

The district judge, though not observing any of the testimony, agreed for the most part with the findings of the magistrate judge. In conformity with Dr. Armor's opinion (and the magistrate judge's finding), the district judge found that "[t]here [was] no indication that the racial identity of the schools in Hillsborough County has been deliberately caused by segregative policies or practices by [Appellants]" and that "based on the totality of the evidence, a shift in demographics [was] a substantial cause of the racial identifiability in Hillsborough County's schools." Manning, 24 F. Supp. 2d at 1303, 1310. The district judge likewise was unable to cite, in either of her two orders, any independent demographic evidence put forward by Appellees or an alternative explanation proffered by Dr. Shelley as to the cause of the racial imbalances.

The district judge seemed to have adopted in toto Appellants' theory of the case (and the magistrate judge's finding). As the district judge stated time and time again, a shift in demographics was a substantial or significant cause of the racial imbalances, and Appellants did not deliberately cause the racial imbalances. See id. at 1293 (stating "that the racial imbalances in the schools [were] not the result of a deliberate attempt by [Appellants] to affect or alter demographic patterns to affect the racial composition of the schools"); id. at 1311 (finding "the evidence presented by the parties establishes that a shift in demographics played a significant role in the racial compositions of the schools" and that Appellants had "not 'affirmatively' exacerbated racial imbalances"); Manning, 28 F. Supp. 2d at 1356 (stating that "demographics have played a significant role in Hillsborough County"). The district judge even agreed that "[i]t [was] probable that" the 17 challenged "schools would have become racially imbalanced regardless of [Appellants'] efforts." Manning, 24 F. Supp. 2d at 1303. Even Appellees' own expert, according to the district judge, "agreed that demographics have played a significant role in Hillsborough County." Id. Despite all of the evidence showing that demographics, and not de jure segregation, caused the racial imbalances, the district judge ultimately declined to find the school district unitary as to student assignments. The district judge declined to so find because she could not conclude that demographics were the sole cause of the racial imbalances. See id. at 1302. In the district judge's view, therefore, the legal presumption remained that any racial imbalances were the result of the prior de jure segregation. See Manning, 28 F. Supp. 2d at 1357.

Finally, we note that the district judge's findings were inconsistent and difficult to follow. For example, the district judge was concerned by Dr. Clark's data, especially with regard to its reliability.20 See Manning, 24 F. Supp. 2d at 1298-1300. Nevertheless, the district judge never stated that the data-which both sides relied upon-was so unreliable so as to undermine her finding that "a shift in demographics [was] a substantial cause of the racial identifiability in Hillsborough County's schools." Id. at 1303. Additionally, the district judge remarked that she was "not convinced that a shift in demographics explain[ed] the racial imbalance in the Hillsborough County school system." Manning, 28 F. Supp. 2d at 1356; see also Manning, 24 F. Supp. 2d at 1293. Yet, in the very same sentence, the district judge also stated that "demographics have played a significant role in Hillsborough County." Manning, 28 F. Supp. 2d at 1356. In spite of these inconsistencies, we are convinced that the district judge agreed with the magistrate judge and found that shifting demographics was a substantial cause of the racial imbalances in Appellants' student assignments and that Appellants did not deliberately cause the racial imbalances.

b. Good-Faith Compliance

The magistrate judge concluded that Appellants had complied in good faith with past federal desegregation decrees, as required by Dowell and its progeny. 498 U.S. at 249-50, 111 S.Ct. at 638. The magistrate judge's conclusion was based largely on the testimony given during the 7-day evidentiary hearing by a variety of witnesses, including inter alia: Dr. Walter Sickles, the school superintendent from 1989 to 1996; Dr. Earl Lennard, the school superintendent as of 1996; five members of the school board; Doris Reddick, the chair of the school board; Andrew Manning, the lead plaintiff in the case since its inception; and Joanna Tokley, the president and CEO of the Urban League.

Dr. Sickles, who had been an employee of the school board since 1969, testified that he believed the school district was unitary in the early 1990's. Dr. Sickles did not seek a declaration of unitary status at that time because he believed Appellees would have opposed and impeded his effort to establish middle schools, which was his top priority. Instead, Dr. Sickles sought Appellees' cooperation in establishing middle schools via the Task Force Report and the 1991 Consent Order. Dr. Lennard, an employee of the school board since 1964, testified that he felt a "moral and legal obligation on the school system to continue a desegregated school system." Five school board members "expressed no misgivings about ... the intent and ability of the School Board to continue a desegregated school system while receiving input from all members of the community."

Some witnesses "voice[d] concern about what might happen in the future if [c]ourt supervision ended." For instance, Ms. Reddick, the chair of the school board, testified against a finding of unitary status. Nevertheless, when asked by Appellees' counsel what would happen if the school district were found unitary, Ms. Reddick asserted, "[W]e're going to guarantee that students will receive equal education."

The magistrate judge found that for more than 25 years Appellants had complied with the court's orders to desegregate and that not once had Appellants been found in violation of any court order. In addition, the magistrate judge noted that Appellants had "regularly conferred [with Appellees] ... to ensure that the school system was moving forwards, not backwards, toward compliance with the [c]ourt's orders." Accordingly, "the long history of compliance with the [c]ourt's orders" outweighed the "opinions and the anecdotal evidence offered by [Appellees]." In sum, the magistrate judge concluded, "The testimony of ... most School Board members, as well as the current superintendent and those responsible for various facets of school operation[,] demonstrates that [Appellants] have accepted the principle of racial equality and will not revert back to a dual school system."

The district judge, however, did not adopt this finding of good faith. Since the district judge did not observe any of the testimony from the evidentiary hearing, naturally she could not evaluate the credibility of the witnesses. Instead, the district judge's finding of bad faith centered upon two interrelated areas of concern: Appellants' "apathy" and the lack of a majority-to-minority (MTM) transfer program.21 See Manning, 24 F. Supp. 2d at 1293, 1312-14.

With regard to Appellants' "apathy," the district judge faulted Appellants for not "utiliz[ing] all available techniques [to desegregate] to the maximum extent practicable" and for not demonstrating "that they [were] willing to aggressively desegregate the school district to the maximum extent practicable." Id. at 1312, 1335. Additionally, the district judge spoke of missed opportunities and a failure to take "affirmative steps ... to be released from the [c]ourt's supervision." Id. at 1312; accord Manning 28 F. Supp. 2d at 1359 (citing Lockett v. Bd. of Educ. of Muscogee Sch. Dist., 92 F.3d 1092, 1099 (11th Cir.1996) (Lockett I)). Nevertheless, after a request by Appellants to clarify exactly what steps should be taken to desegregate to the maximum extent practicable, see id. at 1355, the district judge declined to give any specifics.

Related to the issue of apathy is the inaction demonstrated by Appellants in their MTM program. As previously noted, the July 1971 Order directed the establishment of an MTM program. See supra Part I.B.1. Under this program, a student who attended a school where his race was the majority would be permitted to transfer to a school where his race was the minority. At the evidentiary hearing, Dr. John Miliziano, who was Appellants' "in-house desegregation expert," expressed ignorance about the requirement to implement an MTM program:

I dealt with the Biracial Committee for many years, and no member of the committee ever brought that up. And to tell you the truth, I thought it was one of those things that was considered by the court and it was never meant to apply. It wasn't until recently that I, a person who has spent many, many years in dealing with the court order, maybe it's stupidity on my part, but I didn't even know that this hidden clause-this clause, not hidden, but this clause in the [July 1971 Order] meant really anything.

Manning, 24 F. Supp. 2d at 1313. The evidence also showed that, from 1977 to 1996, no student had ever applied for an MTM transfer, and Appellants had not made any effort to publicize or market the MTM program. See id. at 1314; Manning, 28 F. Supp. 2d at 1359.

While faulting Appellants on the MTM program, the district judge acknowledged that the July 1971 Order "did not specifically direct [Appellants] to market the [MTM] program." Manning, 24 F. Supp. 2d at 1314. Furthermore, the district judge noted (and did not dispute) that no member of the Bi-Racial Committee and no representative of Appellees had complained, prior to the mid-1990's, about the non-availability of an MTM program. See id. at 1313 (quoting testimony of Dr. Miliziano). The district judge also found that when Appellees did raise the lack of an effective MTM program, Appellants announced the program "to the [school] district as a whole and ... took action to publish information about the program." Id. at 1312.

Although the district judge stated that an MTM program is an "indispensable remedy" and a "useful part of every desegregation plan,"22 id. at 1314, the district judge did not explain why this particular MTM program was needed to desegregate the Hillsborough County school district. In fact, as previously mentioned, Dr. Stevens-Appellees' expert who was involved in the negotiations that led to the 1991 Consent Order-conceded that an MTM program would not significantly impact the race ratios at the schools Appellees were challenging as racially imbalanced. Furthermore, the Task Force Report and the 1991 Consent Order (both of which were approved by the district judge) failed to even mention the MTM program. Lastly, notwithstanding the criticism of the MTM program, the district judge spoke favorably of Appellants' overall desegregation efforts. See, e.g., id. at 1287 (stating that "[u]ndoubtedly, [Appellants'] desegregation efforts demonstrate significant success"); id. at 1311 (stating Appellants "deserve acknowledgment for their desegregation efforts thus far"); id. at 1312 (noting Appellants "have been relatively successful in implementing desegregation techniques"); id. at 1325 (finding Appellants had acted in good faith with respect to faculty desegregation and commending Appellants for taking recommendations from the Minority Recruitment Task Force).

Where the relief sought in the district court is the dissolution of an injunction, the order of the district court is subject to a mixed standard of review.23 We review for abuse of discretion the failure to dissolve an injunction as required by law. See Wilson v. Minor, 220 F.3d 1297, 1301 (11th Cir.2000). The district court's application of law is subject to de novo review, while its findings of fact are subject to a clearly erroneous standard of review under Fed.R.Civ.P. 52(a). See Sierra Club v. Martin, 110 F.3d 1551, 1554 (11th Cir.1997) (quoting SunAmerica Corp. v. Sun Life Assurance Co. of Can., 77 F.3d 1325, 1333 (11th Cir.1996)). A declaration of a school system's unitary status is a finding of fact and thus falls under the clearly erroneous standard of Rule 52(a). See Lockett v. Bd. of Educ. of Muscogee County Sch. Dist., 111 F.3d 839, 841-42 (11th Cir.1997) (Lockett II); Jacksonville Branch, NAACP v. Duval County Sch. Bd., 883 F.2d 945, 952 n. 3 (11th Cir.1989). Under this standard of review, "[w]here there are two permissible views of the evidence, the [district court]'s choice between them cannot be clearly erroneous." Lockett II, 111 F.3d at 842; accord Anderson v. City of Bessemer City, 470 U.S. 564, 574, 105 S. Ct. 1504, 1511, 84 L. Ed. 2d 518 (1985). But where a district court applies an incorrect legal standard which "taints" or "infects" its findings of facts, such findings "lose the insulation of [Rule] 52(a) and judgment based thereon cannot stand." Corley v. Jackson Police Dept., 566 F.2d 994, 1001 (5th Cir.1978) (internal quotations and citation omitted);24 accord Bigge v. Albertsons, Inc., 894 F.2d 1497, 1502-03 (11th Cir.1990); Harris v. Birmingham Bd. of Educ., 712 F.2d 1377, 1381 (11th Cir.1983); Lincoln v. Bd. of Regents of Univ. Sys. of Ga., 697 F.2d 928, 938-39 (11th Cir.1983); see also Solomon v. Liberty County Comm'rs, 221 F.3d 1218, 1227 (11th Cir.2000) (en banc) (stating "Rule 52(a) does not inhibit an appellate court's power to correct errors of law").

Before analyzing the case before us, it is important to recall some general principles. In Brown I, the Supreme Court held that de jure racial segregation practiced by school districts violates the Fourteenth Amendment. See Lockett II, 111 F.3d at 842 (citing Brown I, 347 U.S. at 495, 74 S.Ct. at 692). To remedy the illegal conduct, the Supreme Court ordered federal district courts to supervise local officials in desegregating school systems. See id. (citing Brown v. Bd. of Educ., 349 U.S. 294, 301, 75 S. Ct. 753, 757, 99 L. Ed. 1083 (1955) (Brown II)). Federal judicial supervision of local officials, however, was intended to be a temporary measure. See id. (citing Bd. of Educ. v. Dowell, 498 U.S. 237, 247, 111 S. Ct. 630, 637, 112 L. Ed. 2d 715 (1991)).

A desegregation order is remedial in nature, for it is the means by which victims of discriminatory conduct are restored to the position they would have occupied in the absence of such conduct. See Missouri v. Jenkins, 515 U.S. 70, 87, 115 S. Ct. 2038, 2048, 132 L. Ed. 2d 63 (1995) (citing Milliken v. Bradley, 418 U.S. 717, 746- 47, 94 S. Ct. 3112, 3128, 41 L. Ed. 2d 1069 (1974)). The purpose of federal supervision is not to maintain a desired racial mix at a school. See Pasadena City Bd. of Educ. v. Spangler, 427 U.S. 424, 434-37, 96 S. Ct. 2697, 2704-05, 49 L. Ed. 2d 599 (1976); see also Freeman v. Pitts, 503 U.S. 467, 494, 112 S. Ct. 1430, 1447, 118 L. Ed. 2d 108 (1992) (noting that "[r]acial balance is not to be achieved for its own sake"). Rather, a federal court may insist upon a racially balanced school only in those situations where a constitutional violation has caused the school to become racially imbalanced. See id. at 494, 112 S.Ct. at 1447. As the Supreme Court instructed 30 years ago, "[I]n the absence of a showing that either the school authorities or some other agency of the State has deliberately attempted to fix or alter demographic patterns to affect the racial composition of the schools, further intervention by a district court should not be necessary." Swann, 402 U.S. at 31-32, 91 S.Ct. at 1283-83. Put simply, a school board has no obligation to remedy racial imbalances caused by external factors, such as demographic shifts, which are not the result of segregation and are beyond the board's control. See Jenkins, 515 U.S. at 102, 115 S.Ct. at 2055-56 (citing Spangler, 427 U.S. at 434, 96 S.Ct. at 2703-04; Swann, 402 U.S. at 22, 91 S.Ct. at 1279).

The ultimate objective of any desegregation order is the "restoration of state and local authorities to the control of a school system that is operating in compliance with the Constitution."25 Id. at 89, 115 S.Ct. at 2049 (internal quotations and citation omitted). To guide district courts in assessing when it is appropriate to restore local control, the concept of a "unitary" school system has evolved in federal jurisprudence. See generally Freeman, 503 U.S. at 485-92, 112 S.Ct. at 1443-46 (1992). In evaluating whether a school system is "unitary," a district court must carefully assess the facts and utilize its sound discretion to determine (1) whether local authorities have eliminated the vestiges of past discrimination to the extent practicable, and (2) whether local authorities have in good faith fully and satisfactorily complied with, and shown a commitment to, the desegregation plan. See Lockett II, 111 F.3d at 842 (quoting Lee v. Etowah County Bd. of Educ., 963 F.2d 1416, 1425 (11th Cir.1992) (citing Dowell, 498 U.S. at 249-50, 111 S. Ct. at 638)). For a district court to determine whether the vestiges of discrimination have been eliminated to the extent practicable, it must examine the six facets of school operation, the so-called Green factors: student assignments, faculty assignments, staff assignments, transportation, extra-curricular activities, and facilities. See id. (citing Dowell, 498 U.S. at 245, 111 S. Ct. at 636 (quoting Green, 391 U.S. at 435, 88 S.Ct. at 1693)). Using its discretion, a court may also consider other facets. See id. (citing Freeman, 503 U.S. at 492, 112 S. Ct. at 1446).

It is not uncommon for plaintiffs in school desegregation cases to allege, as Appellees did here, that racial imbalances in student assignments, are a "vestige of discrimination." Until unitary status is attained, the defendant school board has the burden of showing that any racial imbalance in the school system is not traceable, in a proximate way, to the prior de jure segregation. See Freeman, 503 U.S. at 494, 112 S. Ct. at 1447. Stated differently, once a plaintiff shows de jure segregation (as Appellees did here in 1962, see supra Part I.A), a presumption arises that all racial imbalances in a school district are the result of the de jure segregation. See Keyes v. Sch. Dist. No. 1, Denver, Colo., 413 U.S. 189, 208, 93 S. Ct. 2686, 2697, 37 L. Ed. 2d 548 (1973). To rebut this presumption, "a school board must prove that the imbalances are not the result of present or past discrimination on its part." Lockett II, 111 F.3d at 843 (citing Swann, 402 U.S. at 25, 91 S.Ct. at 1281); accord Keyes, 413 U.S. at 211, 93 S.Ct. at 2699.

a. Wrong Legal Standard

With the foregoing principles and standards in mind, we turn to the case before us. Appellants argue, inter alia,26 that the district court committed reversible error by relying upon our opinion in Lockett v. Board of Education of Muscogee County, 92 F.3d 1092 (11th Cir.1996) (Lockett I).27 In Lockett I, a panel of this Court reversed a district court's finding of unitary status and ordered the district court to "retain jurisdiction ... to monitor the progress of the school district's desegregation efforts" until the school district could show that it "ha[d] desegregated its schools to the maximum extent practicable." Id. at 1101 (emphasis added). Lockett I was subsequently vacated by the same panel. See Lockett II, 111 F.3d at 840-45.

The district judge, in relying on Lockett I, applied the wrong legal standard. The law does not require, as stated in Lockett I, that a school board eliminate the vestiges of past discrimination "to the maximum extent practicable." 92 F.3d at 1101 (emphasis added). Rather, the law merely requires that the vestiges of past discrimination be eliminated "to the extent practicable." Lockett II, 111 F.3d at 842 (emphasis added); accord Jenkins, 515 U.S. at 90, 115 S.Ct. at 2050;28 Freeman, 503 U.S. at 492, 112 S.Ct. at 1446; Dowell, 498 U.S. at 250, 111 S.Ct. at 638; United States v. Georgia, 171 F.3d 1344, 1347 (11th Cir.1999); Lee, 963 F.2d at 1425. The district judge incorrectly referred to the "maximum extent practicable (or possible)" standard at least 12 times in her opinion of October 24, 1998, and four times in her order of December 4, 1998. See Manning, 24 F. Supp. 2d at 1287, 1289, 1290, 1292, 1293, 1301, 1312, 1326, 1334, 1335; Manning, 28 F. Supp. 2d at 1356, 1359, 1360.

Accordingly, the critical issue is whether the district judge's repeated use of the wrong legal standard sufficiently tainted or infected the findings of fact so as to strip those findings of the insulation normally accorded under Rule 52(a). See supra Part II.A; Corley, 566 F.2d at 1001. We would not permit an inadvertent use of language by a district court to constitute reversible error. Here, however, we are persuaded that the district judge's mistake is more than mere inadvertence.29 As we discuss immediately below, the district judge's findings on student assignments and good faith were tainted (and thus stripped of Rule 52(a) protection) because the district judge held Appellants to a higher standard than the law requires.

b. Application of Wrong Legal Standard to Finding on Student Assignments

In Lockett, the critical issue was, as it is here, whether the racial imbalances in student assignments precluded a finding of unitary status. See Lockett II, 111 F.3d at 842, 843. Lockett II held that, for a school board to rebut the presumption of de jure segregation, the school board had to prove "the [racial] imbalances [were] not the result of present or past discrimination on its part." Id. at 843; see also supra Part II.B.1. The district court in Lockett II found the school board rebutted the presumption by presenting expert demographic evidence showing the imbalances to be the result of voluntary housing patterns and demographic change. See id. The school board's demographic evidence was not contradicted by plaintiffs' experts. See id. The Lockett II majority affirmed the district court's finding as not being clearly erroneous. See id. at 844.

By contrast, Lockett I staked out a position on the law under which school boards would have been held to a higher standard. Instead of affirming, the Lockett I panel would have remanded, so the district court could have continued supervising the school board "until such time as a reliable body of data exist[ed] to assure ... that the school district ha[d] desegregated its schools to the maximum extent practicable." Lockett I, 92 F.3d at 1101 (emphasis added); accord Lockett II, 111 F.3d at 844 (Barkett, J. dissenting). Under Lockett I, a school board would have been required to remedy racial imbalances even when "the imbalances [were] caused by circumstances over which the school district has no control." Lockett I, 92 F.3d at 1099 (quoted in Manning, 24 F. Supp. 2d at 1310). Moreover, according to Lockett I, to be declared unitary, it is not enough for a school board to show demographic shifts as the cause of the racial imbalances, as "demographic shifts are not necessarily independent of prior unconstitutional practices."30 Id. at 1099 (cited in Manning, 28 F. Supp. 2d at 1359).

Lockett I, however, is not the law of this circuit. Rather, the law of the circuit must be distilled from Lockett II. We reiterate that, to overcome the presumption that racial imbalances are constitutionally violative, "a school board must prove that the imbalances are not the result of present or past discrimination on its part." Lockett II, 111 F.3d at 843. Lockett II stands for the proposition that a school board overcomes this presumption when it shows that some external force, which is not the result of segregation and is beyond the school board's control, substantially caused the racial imbalances. See id. (upholding declaration of unitary status when district court found demographic shifts caused racial imbalances); see also Jenkins, 515 U.S. at 102, 115 S.Ct. at 2055-56; contra Lockett II, 111 F.3d at 845 (Barkett, J. dissenting) (advocating contrary proposition). Where a defendant school board shows that demographic shifts are a substantial cause of the racial imbalances, the defendant has overcome the presumption of de jure segregation. See Lockett II, 111 F.3d at 843. Courts shall not assume that demographic shifts are a result of the past de jure segregation. Contra Lockett I, 92 F.3d at 1099 (advocating contrary proposition); Lockett II, 111 F.3d at 845 (Barkett, J. dissenting) (same). Such an assumption is improper because "[i]t is simply not always the case that demographic forces causing population change bear any real and substantial relation to a de jure violation, and the law need not proceed on that premise."31 Lockett II, 111 F.3d at 843 (internal alterations omitted) (quoting Freeman, 503 U.S. at 496, 112 S.Ct. at 1448). Lastly, a plaintiff does not undermine the strength of a defendant's demographic evidence by merely asserting that demographics alone do not explain the racial imbalances.32 Rather, for a plaintiff to preserve the presumption of de jure segregation, the plaintiff must show that the demographic shifts are the result of the prior de jure segregation or some other discriminatory conduct.

If the district judge had applied Lockett II, rather than Lockett I, her findings of fact would have led her to the same conclusion as the magistrate judge with respect to student assignments. The district judge, like the magistrate judge, found that demographic shifts were a "substantial" or "significant" reason for the racial imbalances and that the racial imbalances were inevitable, irrespective of Appellants' efforts. See supra Part I.B.3.a; Manning, 24 F. Supp. 2d at 1303, 1311; Manning, 28 F. Supp. 2d at 1356. With this finding of fact, Appellants overcame the presumption that the racial imbalances in student assignments were the result of de jure segregation. To preserve the presumption, Appellees were required to show that the demographic shifts were the result of the past segregative practices or some other discriminatory conduct. Appellees made no such showing. The district judge never found that the racial imbalances at the 17 challenged schools were caused by the past de jure segregation or other discriminatory acts. In fact, the district judge found the exact opposite. That is, the district judge found that Appellants did not deliberately cause the racial imbalances through segregative policies or practices. See supra Part I.B.3.a; Manning, 24 F. Supp. 2d at 1310. Appellees merely persuaded the district judge that demographics alone did not account for the racial imbalances. See id. at 1302 (refusing to find that a "shift in demography [was] the sole cause [of] the [racial] imbalance[s]" in the school system). Such a finding is insufficient to deny Appellants a declaration of unitary status.

Accordingly, by applying the correct legal standard from Lockett II to the district judge's findings of fact, we reach the same conclusion as the magistrate judge: Appellants have achieved unitary status with respect to student assignments.

c. Application of Wrong Legal Standard to Finding on Good Faith

As discussed previously, the district judge had two interrelated areas of concern which precluded, in her view, a finding of good faith: Appellants' apathy and the lack of an effective MTM program. See supra Part I.B.3.b. The district judge's reliance on Lockett I undoubtedly infected her finding with regard to Appellants' alleged apathy. The district judge, as noted above, expected Appellants to "desegregate the public schools in Hillsborough County to the maximum extent practicable." E.g., Manning, 24 F. Supp. 2d at 1287; supra Part I.B.3.b. Such an expectation was erroneous, for the law does not require a defendant school board to take every conceivable step in attempting to desegregate. See, e.g., Freeman, 503 U.S. at 493, 112 S.Ct. at 1447 (expressly rejecting premise that "heroic measures must be taken to ensure racial balance"). Thus, the district judge's concern about Appellants' alleged "apathy" was an error of law directly traceable to her reliance upon Lockett I.

Further, Appellants' lack of an effective MTM program does not necessarily amount to a finding of bad faith. For instance, in Lockett, the school board did not implement its MTM program until years after the desegregation decree was entered and ignored other requirements of the desegregation decree. See Lockett II, 111 F.3d at 844; Lockett I, 92 F.3d at 1100, 1101. Nonetheless, we affirmed a finding of good faith based in part on the district court's finding that the school board had never violated a court order, had never been enjoined or sanctioned, and had consulted with African-American members of community before modifying the student assignment plan. See Lockett II, 111 F.3d at 843-44; contra Lockett I, 92 F.3d at 1100, 1101. Likewise, in the case sub judice, Appellants never violated a court order, never were sanctioned, and consulted extensively with the African-American community, including Appellees, prior to implementing new student assignments under the 1991 Task Force Report. See supra Parts I.B.2 & I.B.3.b. Therefore, based on Lockett II, the district judge clearly could have found the Appellants acted in good faith, notwithstanding the lack of a viable MTM program.

Additionally, we are persuaded that, if the district judge had followed the law as set forth in Lockett II, she would have concluded that Appellants have acted in good faith. As we explained in Lockett II, in determining whether a school board has acted in good faith, a court should not dwell on isolated discrepancies, but rather should "consider whether the school board's policies form a consistent pattern of lawful conduct directed to eliminating earlier violations."33 Lockett II, 111 F.3d at 843 (internal quotations omitted); see also Freeman, 503 U.S. at 491, 112 S.Ct. at 1446 (holding that "a court should give particular attention to the school system's record of compliance"). Repeatedly in the October 26, 1998, opinion, the district judge commended Appellants for their desegregation efforts. See, e.g., Manning, 24 F. Supp. 2d at 1287, 1311, 1312, 1325; supra Part I.B.3.b. At the end of the opinion, the district judge further commented:

After evaluating the voluminous record in this case, the Court is convinced that [Appellants] have a short road to travel [to attain unitary status]. Essentially, Defendants need to demonstrate that they are willing to aggressively desegregate the school district to the maximum extent practicable.34

Manning, 24 F. Supp. 2d at 1335. Based on the foregoing statement, we are convinced that, if the district judge had applied the correct standard, she would have found (as the magistrate judge did) that Appellants acted in good faith, notwithstanding the absence of an effective MTM program and other possible discrepancies.

Our conclusion is buttressed by two other points. First, the MTM program, as conceded by Appellees' expert, would have been ineffective in desegregating the 17 challenged schools.35 As such, the MTM program should have had only marginal relevance in analyzing whether Appellants' "policies form[ed] a consistent pattern of lawful conduct directed to eliminating earlier violations." Lockett II, 111 F.3d at 843 (emphasis added) (internal quotations and citation omitted).

Second, discerning a school board's good faith is in some respects a subjective finding. Thus, such a finding depends in part on the judge's personal observation of the witnesses. The magistrate judge, not the district judge, observed all of the witnesses at the evidentiary hearing. Granted, the district judge was free under 28 U.S.C. 636 to make a de novo determination of the magistrate judge's findings. In other contexts, however, we have cautioned district judges from overruling a magistrate judge's finding where credibility determinations are dispositive. See, e.g., Proffitt v. Wainwright, 685 F.2d 1227, 1237 (11th Cir.1982) (noting that, in criminal or habeas corpus cases, a district judge's reasons for rejecting a report and recommendation must be consistent with the credibility choices made by a magistrate). In addition, the rationale for deferring to a district court's finding of fact is that a trial judge is aware of the variations in demeanor which bear so heavily in making a subjective determination. See Anderson, 470 U.S. at 575, 105 S.Ct. at 1512. Where, as here, a district judge does not personally observe the witnesses in making a subjective finding of fact, we view such a finding with skepticism, especially where, as here, the finding is contrary to the one recommended by the judicial official who observed the witnesses.

The district judge's finding that Appellants have not achieved unitary status was tainted and infected by reliance on an incorrect legal standard. As such, we reverse the district judge's orders of October 26, 1998, and December 4, 1998. Upon remand, the district court shall enter judgment declaring the Hillsborough County school system unitary. Therefore, federal judicial supervision of the Hillsborough County school system shall cease.

REVERSED and REMANDED.