ESTES ET AL. v. METROPOLITAN BRANCHES OF THE DALLAS NAACP ET AL.
CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FIFTH CIRCUIT.
Argued October 29, 1979.
Decided January 21, 1980. *
Certiorari dismissed. Reported below: 572 F.2d 1010.
[ Footnote * ] Together with No. 78-282, Curry et al. v. Metropolitan Branches of the Dallas NAACP et al.; and No. 78-283, Brinegar et al. v. Metropolitan Branches of the Dallas NAACP et al., also on certiorari to the same court.
Warren Whitham argued the cause for petitioners in No. 78-253. With him on the brief was Mark Martin. Robert L. Blumenthal argued the cause for petitioners in No. 78-282. With him on the briefs was Robert H. Mow, Jr. James A. Donohoe argued the cause and filed a brief for petitioners in No. 78-283.
E. Brice Cunningham argued the cause for respondents Metropolitan Branches of the Dallas NAACP et al. With him on the brief were Nathaniel R. Jones, Merle W. Loper, and Louis R. Lucas. Edward B. Cloutman III argued the cause for respondents Tasby et al. With him on the brief were Jack Greenberg, James M. Nabrit III, and Bill Lann Lee.
Deputy Solicitor General Wallace argued the cause for the United States as amicus curiae urging affirmance. With him on the brief were Solicitor General McCree, Assistant Attorney General Days, Sara Sun Beale, Brian K. Landsberg, and Mildred M. Matesich.Fn
Fn [444 U.S. 437, 437] H. Ron White filed a brief for the Dallas Alliance et al. as amici curiae urging reversal in all cases.
The writs of certiorari are dismissed as improvidently granted. [444 U.S. 437, 438]
MR. JUSTICE MARSHALL took no part in the consideration or decision of these cases.
MR. JUSTICE POWELL, with whom MR. JUSTICE STEWART and MR. JUSTICE REHNQUIST join, dissenting.
The Court today dismisses the writs previously granted in this litigation and thereby reinstates the ruling of the Court of Appeals. The suit now will be returned to the District Court for elaboration of that court's conclusions on the feasibility of extensive busing to achieve racial balance in the Dallas public schools. The Court of Appeals directed the trial court to supplement the record with formal studies of the anticipated times and distances of likely bus routes, and to make additional findings on desegregation in the city's high schools.
Although the remand is narrow, aimed solely at the sufficiency of the record on which the District Court based its desegregation order, I do not think it is justified. After studying the schools of the Dallas Independent School District through the many years of this litigation, the trial judge drew on his familiarity with Dallas and its schools, and on the advice of many community groups, to fashion an effective and fair desegregation order. The Court of Appeals failed to accord proper deference to the District Court's conscientious execution of this delicate task.
In addition, this case presents a long-needed opportunity to re-examine the considerations relevant to framing a remedy in a desegregation suit. It is increasingly evident that use of the busing remedy to achieve racial balance can conflict with the goals of equal educational opportunity and quality schools. In all too many cities, well-intentioned court decrees have had the primary effect of stimulating resegregation. The experience in Dallas during this litigation presents a striking illustration of this problem. If the District Court orders substantial additional busing, as the Court of Appeals apparently [444 U.S. 437, 439] thinks it should, recent history suggests that the Dallas school district will be well [BAD TEXT] the road to the "separate but equal" conditions mistakenly approved in Plessy v. Ferguson, 163 U.S. 537 (1896). Such an outcome is no less real or less regrettable when caused by courts with benign motives. The promise of Brown v. Board of Education, 347 U.S. 483 (1954), cannot be fulfilled by continued imposition of self-defeating remedies.
The Dallas Independent School District (School District) has been in desegregation litigation since 1955, although the present case is not part of the original suit. During this quarter of a century, the School District has grown into the eighth largest school district in the country, covering 351 square miles and spanning 35 miles at its widest point. Since the present action first was tried in 1971, the student population of the district has changed dramatically. Total enrollment has dropped from 163,000 to 133,000, while the racial distribution of students has shifted from 69% Anglo in 1971 to 33.5% Anglo, 49.1% black, and 16.3% Mexican-American in 1979. There were 112,000 Anglo students in the School District in 1970; there are now fewer than 45,000.
This suit was brought by several parents, acting on behalf of their children, against the superintendent and the Board of Trustees of the School District (Board). Other parents groups have intervened in the suit. 1 In the summer of 1971, the District Court found that "elements" of a segregated school system "still remain" in the Dallas schools. Tasby v. Estes, 342 F. Supp. 945, 947 (ND Tex. 1971). The court imposed a number of remedies, including the busing of approximately [444 U.S. 437, 440] 15,000 students. The original plaintiffs appealed to the Court of Appeals for the Fifth Circuit for more extensive reassignment and transportation of pupils. That court declared that "nothing less than the elimination of predominantly one-race schools is constitutionally required. . . ." Tasby v. Estes, 517 F.2d 92, 103, cert. denied, 423 U.S. 939 (1975). Finding that the District Court's decree failed to satisfy this standard, the Court of Appeals remanded for the formulation of a new desegregation plan.
In a month-long trial on remand, the District Court considered in detail six plans submitted by the various parties and a court-appointed expert. It heard nearly 50 witnesses, including numerous experts, and produced a trial transcript of some 4,000 pages. The court also conferred with concerned community groups, the most prominent of which was the Educational Task Force of the Dallas Alliance (Alliance), a multiracial, nonpartisan organization. 2 In a thorough opinion, Tasby v. Estes, 412 F. Supp. 1192 (ND Tex. 1976), the District Court found that the decline in Anglo enrollment between 1971 and 1976 was not the result of actions taken by the Board. In fact, the court noted the Board's continuing "good faith" efforts to establish a unitary school system. Id., at 1207. See Pasadena City Bd. of Education v. Spangler, 427 U.S. 424, 435 -436 (1976). The court's duty, it asserted, was to adopt a plan that would "realistically and effectively" [444 U.S. 437, 441] achieve desegregation in light of demographic changes in the School District. 412 F. Supp., at 1207.
With careful attention to the special characteristics and history of the School District, the District Court promulgated a progressive and comprehensive plan that drew heavily on the proposals of the Alliance. 3 For purposes of student assignment, the plan divides the School District into six subdistricts. In integrated areas within each subdistrict, present student assignments are retained wherever possible. In other areas, children in grades K-3 remain in neighborhood schools; those in grades 4-8 are assigned to central schools in each subdistrict; and high school students are assigned to schools in their own subdistricts on the basis of geographical attendance zones. The plan provides for a number of "magnet high schools" that offer enriched educational programs. 4 The pupil assignment plan is supplemented by majority-to-minority transfers upon request. 5
This case now focuses on student assignment and busing. 6 [444 U.S. 437, 442] Each of the six plans considered by the District Court provided for substantial busing. 7 Although the court did not estimate the number of pupils to be bused under its decree, the Alliance proposal which paralleled the court's plan anticipated the busing of some 20,000 students. The court concluded that the racial composition of the student population in each subdistrict "will approximate the racial makeup of the [district] as a whole, with the exception of [East] Oak Cliff," a black neighborhood. Id., at 1204.
Respondents argue that the District Court's plan leaves 62 schools, about one-third of the 176 schools in the district, with "one-race" student bodies (defined as those where more than 75% of the students are of one race). 8 Fifty-two of these schools would be predominantly black, nine Anglo, and one Mexican-American. A majority of these one-race schools result from the District Court's refusal to bus very young children. 9 This Court has recognized that concern for the health and welfare of younger children may dictate their exclusion from student transportation plans, see Swann v. Charlotte-Mecklenburg Bd. of Education, 402 U.S. 1, 31 (1971), and respondents do not dispute that feature of the District Court order. In addition, several high schools have one-race student bodies because the District Court declined to order the busing of high school students. It noted that "of approximately 1,000 Anglos ordered to be transported to formerly all-black high schools under this Court's 1971 student assignment plan, fewer than 50 Anglo students attend those schools today." 412 F. Supp., at 1205. The court [444 U.S. 437, 443] also concluded that the establishment of magnet high schools was the "most realistic, feasible, and effective method for eliminating the remaining vestiges of a dual [school] system on the 9-12 level. . . ." Ibid.
Viewed on a geographic basis, the order left 28 "predominantly black" schools in East Oak Cliff, which is bounded by the Trinity River bottom on one side and by Interstate 35 on the other. The court found that the "practicalities of time and distance" prevent the effective integration of the schools in this neighborhood through busing. Id., at 1204. In contrast, the Seagoville subdistrict remains predominantly white. Seagoville, however, is "geographically isolated" from the rest of the city, Tasby v. Estes, 572 F.2d 1010, 1013 (CA5 1978), and its school population represents less than 2% of the School District's student body.
The Court of Appeals was not impressed by the District Court's carefully structured plan. It concentrated almost exclusively on the "large number of one-race schools" remaining in Dallas. Id., at 1012.
The duration and complexity of this litigation demonstrate the difficulty of providing effective relief in a school desegregation case. The school board and the court must consider many economic, social, and educational factors, and those factors vary widely from community to community. Courts frequently are caught between the constitutional prohibition against segregation and the severe limitations on their ability to implement an effective plan with public support. See Columbus Bd. of Education v. Penick, 443 U.S. 449, 486 -488 (1979) (POWELL, J., dissenting). Consequently, this Court has been reluctant to give more than general instructions for desegregation orders, and those instructions have not always been completely consistent. 10 The result in too many instances has been confusion in the lower courts. See infra, at 449-450.
I believe that two rules provide the basic outline for responsible exercise of the courts' equitable powers in school desegregation cases. First, "the nature of the desegregation remedy is to be determined by the nature and scope of the constitutional violation." Milliken v. Bradley, 433 U.S. 267, 280 (1977) (Milliken II). The constitutional deprivation must be identified accurately, and the remedy must be related closely to that deprivation. Otherwise, a desegregation order may exceed both the power and the competence of courts. Second, "[t]he measure of any desegregation plan is its effectiveness." Davis v. School Comm'rs of Mobile County, 402 U.S. 33, 37 (1971). A court must act decisively to remove purposeful segregation, but it also must avoid the danger of inciting resegregation by unduly disrupting the public schools. [444 U.S. 437, 445]
Much of the confusion that has plagued this litigation derives from neglect of these principles. The District Court failed to identify the link between the constitutional violation and the desegregation remedy, and the Court of Appeals showed little concern for either that problem or the question of effectiveness. Unless courts carefully consider those issues, judicial school desegregation will continue to be a haphazard exercise of equitable power that can, "like a loose cannon, . . . inflict indiscriminate damage" on our schools and communities. 11
The opinion of the Court of Appeals focuses almost entirely on the one-race schools remaining in the School District. This preoccupation apparently derives from the oft-repeated language in Green v. County School Board, 391 U.S. 430, 442 (1968), that desegregation must create "a system without a `white' school [or] a `Negro' school." As I have noted before, this language was suitable to the small rural county before the Court in that case, where there were only two schools and 1,300 schoolchildren of both races scattered throughout the county. But it makes no sense to apply that statement to the Dallas Independent School District or any major metropolitan school district. In large cities, the principal cause of segregation in the schools is residential segregation, which results largely from demographic and economic conditions over which school authorities have no control. E. g., Pasadena City Bd. of Education v. Spangler, 427 U.S., at 435 -437; see Keyes v. School Dist. No. 1, Denver, Colo., 413 U.S. 189, 222 -223 (1973) (POWELL, J., concurring in part and dissenting in part). 12 In cases since Green, the Court has stated explicitly [444 U.S. 437, 446] that the existence of "predominantly white or predominantly black [schools,] without more, . . . does not offend the Constitution." Dayton Bd. of Education v. Brinkman, 433 U.S. 406, 417 (1977); Milliken II, supra, at 280, n. 14; Swann v. Charlotte-Mecklenburg Bd. of Education, 402 U.S., at 26 . 13 It is puzzling that many trial and appellate courts continue to misapply Green and largely to ignore more recent statements on this issue.
The important distinction is between "desegregated" schools and "integrated" schools. There can be no legitimate claim that "racial balance" in the public schools is constitutionally required. Milliken v. Bradley, 418 U.S. 717, 740 -741 (1974) (Milliken I). Rather, the Constitution mandates that no school system be structured to segregate the races. The proposition was stated fully in Swann:
In this litigation, the District Court ordered the busing of 15,000 students in 1971, 342 F. Supp., at 956, while the current decree is likely to result in the transportation of some 20,000 students. See supra, at 442. On the record before us, we cannot determine whether the trial court made findings of constitutional violations that justified these and other remedies [444 U.S. 437, 448] that were ordered. For the purpose of deciding the cases in this Court, we may assume that such violations were duly found. In any event, since the 1975 ruling of the District Court this litigation has concentrated solely on the need to eliminate one-race schools through further busing. The petition for certiorari raised only that issue, see Pet. for Cert. in No. 78-253, pp. 2-3, and both the Board, the principal petitioner here, and the Dallas Alliance have asked only that the District Court's order be reinstated. See Brief for Petitioners in No. 78-253, pp. 71-72; Brief for Dallas Alliance as Amicus Curiae 28. Consequently, I believe this Court should reach the merits of the remedial question and review the decision of the Court of Appeals that the District Court must explain why "any" one-race schools remain in Dallas. 572 F.2d, at 1018.
Court orders to remedy constitutional deprivations in formerly segregated school systems must be drawn "in light of the circumstances present and the options available," Green v. County School Board, 391 U.S., at 439 , "taking into account the practicalities of the situation." Davis v. School Comm'rs of Mobile County, 402 U.S., at 37 . Although this Court's guidance in desegregation cases necessarily has been general, its emphasis on effectiveness and practicalities reflects an appreciation that perfect solutions may be unattainable in the context of the demographic, geographic, and sociological complexities of modern urban communities. The imperfect nature of court action in school cases is evident in the phenomenon of self-defeating "remedies," desegregation plans and continuing court oversight so unacceptable that many parents seek to avoid the reach of the court's decree. The impact of such remedies may be seen in higher enrollment in private schools, in further migration to the suburbs, or in refusals to move into the school district.
This Court has not considered seriously the relationship between the resegregation problem and desegregation decrees. [444 U.S. 437, 449] The most helpful precedent is Pasadena City Bd. of Education v. Spangler, which arose several years after a local desegregation plan had been implemented. 15 We held in that case that the Constitution does not require that a desegregation decree be modified periodically as migration patterns shift the distribution, of the races within the school district. The Court recognized that, absent further segregation by the State, there is no constitutional obligation to remedy resegregation after an approved plan is implemented. That holding accents the need for courts to consider with care the impact a remedy is likely to have on resegregation. As Pasadena establishes, once resegregation occurs without state action courts have no power to impose an additional remedy.
Surprisingly few courts, however, have understood this imperative. One exception is the decision on remand in Milliken I, 418 U.S. 717 (1974), where a desegregation plan that left many one-race schools was approved by the Court of Appeals for the Sixth Circuit. The court explained that the available alternatives would have "accelerate[d] the trend toward rendering all or nearly all of Detroit's schools so identifiably black as to represent universal school segregation. . . ." Bradley v. Milliken, 540 F.2d 229, 239 (1976), aff'd, Milliken II, supra. 16 In a case involving a school district in Alabama, [444 U.S. 437, 450] however, the Court of Appeals for the Fifth Circuit approved a plan "that will probably result in an all-black student body, where nothing in the way of desegregation is accomplished and where neither the white students nor black students are benefited.'" Lee v. Macon County Bd. of Education, 465 F.2d 369, 370 (1972). Even though the court acknowledged that the remedy was self-defeating, it ordered the plan implemented unless the local school board could come forward with a plan "equally effective" in eliminating one-race schools. Ibid. 17
The pursuit of racial balance at any cost - the unintended legacy of Green - is without constitutional or social justification. Out of zeal to remedy one evil, courts may encourage or set the stage for other evils. By acting against one-race schools, courts may produce one-race school systems. Parents with school-age children are highly motivated to seek access to schools perceived to afford quality education. A desegregation plan without community support, typically one with objectionable transportation requirements and continuing judicial oversight, accelerates the exodus to the suburbs of families able to move. The children of families remaining in the area affected by the court's decree are denied the opportunity to be part of an ethnically diverse student body. See Parents Assn. [444 U.S. 437, 451] of Andrew Jackson High School v. Ambach, 598 F.2d 705, 717 (CA2 1979). The general quality of the schools also tends to decline when substantial elements of the community abandon them.
The effects of resegregation can be even broader, reaching beyond the quality of education in the inner city to the life of the entire community. When the more economically advantaged citizens leave the city, the tax base shrinks and all city services suffer. And students whose parents elect to live beyond the reach of the court decree lose the benefits of attending ethnically diverse schools, an experience that prepares a child for citizenship in our pluralistic society. 18
The District Court in this litigation was properly concerned over resegregation and community support for the Dallas schools. 19 The facts before the court made that concern unavoidable. In the five years following the 1971 desegregation decree, the proportion of Anglo students in the Dallas public schools had dropped by almost half. That destabilizing [444 U.S. 437, 452] trend has continued in the School District, as reflected by the following figures:
A desegregation remedy that does not take account of the social and educational consequences of extensive student transportation can be neither fair nor effective. The District Court's plan is properly sensitive both to existing demographic realities and to the likely consequences of increased busing. The Court of Appeals seriously erred when it remanded this case with a mandate that seems certain to accelerate the destructive trend toward resegregation.
As this Court should not tolerate this error, even by silence that might give rise to an inference of approval, I dissent from the Court's failure to decide the case and reinstate the District Court's plan - a plan that does have promise for success.
[ Footnote 2 ] The Task Force consisted of seven Mexican-Americans, seven Anglos, six blacks, and one American Indian. The Dallas Alliance comprises 77 cooperating organizations, including local branches of the AFL-CIO and the Chamber of Commerce, religious groups, civic organizations, and several local chapters of the NAACP. The Alliance Task Force proposed a comprehensive desegregation plan that largely was adopted by the District Court. Of course, a group's participation in the Alliance need not signify approval of the desegregation plan proposed by the Educational Task Force, but the Alliance clearly has many ties to the entire Dallas community.
[ Footnote 3 ] In a brief amicus curiae to the Court, the Alliance stated that the plan submitted by its Educational Task Force "reflects compromise" and was reached by consensus. Brief for Dallas Alliance as Amicus Curiae 21.
[ Footnote 4 ] The District Court ordered the School District to establish seven new magnet high schools by 1979. Each must offer special career training, and the racial makeup of each school must be within 10% of the racial distribution of the School District's high school population. Tasby v. Estes, 412 F. Supp. 1192, 1215-1216 (ND Tex. 1976). Free transportation is available for students at magnet schools.
[ Footnote 5 ] The Court of Appeals correctly determined that the District Court's plan is deficient in not explicitly providing transportation at public expense to children who exercise this option. Tasby v. Estes, 572 F.2d 1010, 1015 (CA5 1978).
[ Footnote 6 ] The initial District Court ruling also dealt, apparently to the satisfaction of the plaintiffs, with staff desegregation, school construction, bilingual education for Mexican-Americans, and other programs designed to supplement the opportunities of minority students. Tasby v. Estes, 342 F. Supp. 945, 953 (ND Tex. 1971). These matters are no longer issues in this litigation.
[ Footnote 7 ] The estimates of the number of students to be bused ranged from 14,000 under the plan originally proposed by the Board to 69,000 under the NAACP plaintiffs' Plan A.
[ Footnote 8 ] Tr. of Oral Arg. 46.
[ Footnote 9 ] In their brief to this Court, the NAACP respondents concede that more than half of the one-race schools are in the elementary grade, K-3 category. Brief for Respondents NAACP et al. 9.
[ Footnote 10 ] See Yudof, School Desegregation: Legal Realism, Reasoned Elaboration, and Social Science Research in the Supreme Court, 42 Law & Contemp. Prob. 57, 87-102 (Autumn 1978).
[ Footnote 12 ] See Coleman, New Incentives for Desegregation, 7 Human Rights 10, 11 (Fall 1978); Farley, Residential Segregation and Its Implications for School Integration, 39 Law & Contemp. Prob. 164 (Winter 1975).
[ Footnote 13 ] Some federal courts continue to read Swann v. Charlotte-Mecklenburg Bd. of Education as requiring extensive transportation because of its language endorsing the need "to achieve the greatest possible degree of actual desegregation." 402 U.S., at 26 . Swann, however, simply laid down a broad rule of reason under which desegregation remedies must remain flexible and due consideration must be given to other values and interests. In Swann, we recognized that special difficulties arise when extensive busing is used in metropolitan areas "with dense and shifting population[s], numerous schools, [and] congested and complex traffic patterns." Id., at 14. Although Swann approved pupil transportation as a remedial device, the Court said that transportation orders would be suspect "when the time or distance of travel is so great as to either risk the health of the children or significantly impinge on the educational process." Id., at 30-31. The Court's more recent decisions have dispelled any doubt that may have existed as to whether Swann mandates busing to establish racial balance. In this regard, one should note that the Court of Appeals in this case failed to mention Milliken v. Bradley, 418 U.S. 717 (1974) (Milliken I), Dayton Bd. of Education v. Brinkman, or Milliken II, while it relied heavily, and mistakenly, upon Green. Tasby v. Estes, 517 F.2d 92, 103 (1975).
[ Footnote 14 ] Last Term, the Court decided school desegregation cases from Columbus and Dayton, Ohio. Columbus Bd. of Education v. Penick, 443 U.S. 449 (1979); Dayton Bd. of Education v. Brinkman, 443 U.S. 526 (1979) (Dayton II). As the dissent by MR. JUSTICE REHNQUIST argued, the opinions in those cases appeared to depart from prior precedents of this Court. But instead of criticizing either the reasoning or the holdings of any of those cases, the Court simply avoided their force by accepting uncritically the findings made by the courts below. In Columbus, it emphasized that the District Court had found "purposefully segregative practices with current, systemwide impact." 443 U.S., at 466 , citing 429 F. Supp. 229, 252, 259-260, 264, 266 (SD Ohio 1977). Although the District Court had made no such findings in Dayton II, the Court of Appeals for the Sixth Circuit tailored its own factual findings to accord with those in Columbus, and these again were adopted by this Court. 443 U.S., at 538 -540.
In the instant case, neither the District Court nor the Court of Appeals made comparable findings.
[ Footnote 15 ] In United States v. Scotland Neck Bd. of Education, 407 U.S. 484, 491 (1972), we disapproved an attempt to create a small new school district within a county conceded to have been operating a dual school system. In a concluding sentence, the Court stated that while the possibility of resegregation "may be cause for deep concern to the respondents, it cannot . . . be accepted as a reason for achieving anything less than complete uprooting of the dual public school system." The Court concluded that the new school district would "interfer[e] with the desegregation of the . . . [c]ounty [s]chool system." Id., at 489. Given the context of the Court's passing reference to resegregation, Scotland Neck affords no guidance for the more usual desegregation case.
[ Footnote 16 ] See Mapp v. Chattanooga Bd. of Education, 525 F.2d 169 (CA6 1975), cert. denied, 427 U.S. 911 (1976). The California Supreme Court [444 U.S. 437, 450] has expressly authorized the consideration of resegregation patterns in designing decrees for school litigation under the State Constitution. Crawford v. Los Angeles Bd. of Education, 17 Cal. 3d 280, 308-309, 551 P.2d 28, 47 (1976).
[ Footnote 17 ] The position taken by counsel for one group of respondents in these cases is identical to that of the court in Lee v. Macon County Bd. of Education. At oral argument, counsel was asked if he still would support the remand ordered by the Court of Appeals if he were certain that additional busing "would result in these black children next year or the year afterwards . . . going to an all-black school because there wouldn't be any whites or any people of any other color to go to school with [in the District]." Counsel replied that his clients' position would be no different in that situation. Tr. of Oral Arg. 49.
[ Footnote 18 ] As I noted in dissent in the Columbus case, courts are the branch of government least competent to provide long-range solutions to the resegregation problem. Because the causes of segregation in residential housing are usually beyond judicial correction, wider solutions that will be acceptable to concerned parents must be sought by legislators and executive officials. See 443 U.S., at 480 -481 (POWELL, J., dissenting). See also Coleman, supra n. 12, at 48-49; Willie, Racial Balance or Quality Education?, in School Desegregation, Shadow and Substance 7 (Levinsohn & Wright eds. 1976).
[ Footnote 19 ] The District Court appears to have succeeded in enlisting active support from much of the Dallas community for the desegregation plan. The voters have approved an $80 million school bond issue that will assist in implementing the court's decree. In addition, business and civic organizations have "adopted" 144 schools in a community-wide effort to channel volunteers, equipment, and private money to those schools, and to provide part-time and full-time job opportunities for students in those schools. See Brief for Dallas Alliance as Amicus Curiae 25.
[ Footnote 20 ] The Anglo population in the School District is likely to fall off even more since current Anglo enrollment is highest in the high schools and declines steadily through the lower grades. [444 U.S. 437, 453]