at 759-60, and,

       “As discussed infra, the holdings in Flores persuaded the Court that it
       committed clear and manifest error in its factual finding that it was
       compelled to consider the ‘panoptic results’ of LEP students in all grades
       rather than considering the achievement of primary and secondary students
       separately.”

Id., 

at 762. However, after this appeal was taken, the Supreme Court
reversed the Ninth Circuit’s Flores decision. Horne v. Flores, 
129 S. Ct. 2579 
(2009).

                                                 6
      On August 18, 2008, defendants filed a motion to further modify the
Modified Order, arguing that this court’s recent decision in Samnorwood
Independent School District v. Texas Education Agency, 
533 F.3d 258 
(5th Cir.
2008), constituted a change in law and required that the Modified Order’s
application be limited to the original school districts in the 1970 action. The
trial court granted in part, and denied in part, defendants’ motion, and amended
the Modified Order to exempt the two Samnorwood school districts and all
“school districts that: (1) were not parties to the original 1970 litigation; (2) were
unitary prior to the commencement of the 1970 litigation; and (3) have never
since been shown to have attempted to resegregate or act with segregative
intent.”     United States v. Texas (LULAC V), No. 6:71-CV-5281, 
2008 WL
5334404
, at *7 (E.D. Tex. Dec. 17, 2008). Defendants also appeal from that order.
             II. TRIAL COURT’S FINDINGS AND CONCLUSIONS
Background
      Chapter 29 of the Texas Education Code mandates bilingual education and
English as a second language (“ESL”) programs in all Texas schools. TEX. EDUC.
CODE § 29.051. Bilingual programs are offered in kindergarten through sixth
grade (primary education), while ESL programs are offered in seventh through
twelfth grade (secondary education). LULAC 
IV, 572 F. Supp. 2d at 736
.
However, local school districts may elect either bilingual or ESL programs “or
other transitional language instruction” for the seventh and eighth grade LEP
students. TEX. EDUC. CODE § 29.053(d)(2).7 These programs are significantly
different; bilingual instruction teaches course materials in both English and the
student’s native language, while ESL instruction teaches materials in modified
English for easier comprehension by LEP students. LULAC 
IV, 572 F.2d at 735
.


      7
          In the years in evidence, apparently all used ESL.

                                               7
Students’ participation in bilingual and ESL programs is subject to parental
permission, which may be, and sometimes is, withheld.
      Under the Texas Education Code, “[t]he school districts and charter
schools created in accordance with the laws of this state have the primary
responsibility for implementing the state’s system of public education and
ensuring student performance in accordance with this code.” TEX. EDUC. CODE
§ 11.002. And, “[a]n educational function not specifically delegated to [TEA] is
reserved to and shall be performed by school districts or open-enrollment charter
schools.” 
Id. § 7.003. 
Under Section 7.021(b), TEA is given diverse powers to
“administer and monitor compliance.” TEA is given certain authority respecting
monitoring compliance with various programs, including the state bilingual
education and ESL programs. 
Id. §§ 29.051-29.064. 
Under Section 7.028(a),
such compliance monitoring function extends “only as necessary to ensure . . .
compliance with federal law and regulations” (and financial integrity and certain
“data integrity”); and under section 7.028(b), “[t]he board of trustees of a school
district or the governing body of an open-enrollment charter school has primary
responsibility for ensuring that the district or school complies with all applicable
requirements of state educational programs.” TEA is required to “evaluate the
effectiveness of [bilingual and ESL] programs” as “compared to state established
standards.” 
Id. §§ 29.062(a), 
39.053(c). Factors to be considered include, among
other things, drop-out rates, graduation rates, and standardized test passing
rates. 
Id. § 39.053(c).
      In 2003, the TEA’s previous monitoring program to evaluate program
compliance was replaced with the Performance Based Monitoring Analysis
System (PBMAS). PBMAS was first implemented in the 2004–2005 school year.
Annually, the PBMAS generates a set of performance indicators on the district-
level, based mostly on student passage rates on the Texas Assessment of

                                         8
Knowledge and Skills (TAKS), a statewide achievement test. Each year the
state sets standards for student passage rates in five subject areas and compares
districts' performance to these standards. TEA assigns each district a
performance level based upon the deviation of the district's passage rates from
the state established standards. And a district’s performance level determines
whether the district will be subject to intervention and to what degree. In terms
of LEP students, PBMAS evaluates the bilingual and ESL program using TAKS
scores of LEP students, as well as other indicators, including the LEP annual
drop-out rate, the LEP graduation rate, and LEP reading proficiency. At the
time of the trial court’s evidentiary hearing, only two years’ worth of PBMAS
data was available—the 2004–2005 school year and the 2005–2006 school year.
Problems With PBMAS
      The district court identified two overall problems with TEA’s monitoring
procedures: (1) the kind of data collected by PBMAS inaccurately reflects how
well a district is performing and cannot adequately detect deficiencies at the
campus level; and (2) TEA’s inadequate response to problems flagged by PBMAS
contributes to the denial of educational opportunity.
      In particular, the court determined that data collected by PBMAS show
that some school districts are likely under-reporting the number of LEP
students, and TEA has done nothing to verify these numbers. The court also
found that although TEA compared the bilingual-ESL indicators to state
standards for all students, TEA has no procedure for comparing the performance
of LEP students to non-LEP students directly. The court further found that
PBMAS’s aggregation of scores for multiple grade levels as well as for entire
school districts distorted the performance indicators and masked problems at
specific schools.
      With respect to on-site visits, the court determined that TEA had not

                                       9
conducted any on-site monitoring for some years and had no bilingual-ESL
certified monitors at the time of the trial. The court also considered other ways
TEA monitors LEP students, such as monitoring that it performs under the No
Child Left Behind (NCLB) program and the Texas accountability rating system.
The court found that, although some of these monitoring programs disaggregate
student performance data, they do not overcome the deficiencies in PBMAS
because they do not target all relevant criteria for evaluating LEP students.
Achievement Of LEP Students
       The district court also found that too high a proportion of the statewide
secondary LEP student population were failing. At all grade levels and in all
subject areas, secondary LEP students are performing poorly on TAKS. In fact,
secondary LEP students have much lower passage rates on standardized tests
than the general student body—creating a significant gap in achievement that
ranges (roughly) between thirty-five and forty-five percentage points.                  Also,
secondary LEP students have higher retention and drop-out rates when
compared to the general student population. “LEP students fail to progress
through or exit LEP programs in a reasonable time.” LULAC IV, 
572 F. Supp.
2d 
at 751. And LEP students participate in advanced courses at a “far lower
rate than all students.”8 
Id. at 752.
       The trial court made no finding or determination that with respect to
primary grades (kindergarten through sixth grade) the defendants had violated
Section 1703(f).
                                   III. DISCUSSION
A. Appellate Jurisdiction


       8
       The trial court’s findings in this respect are extremely lengthy and detailed; for a
more complete recital of these, see the trial court’s opinion. 
572 F. Supp. 2d 
at 738–42,
747–53.

                                             10
      The United States and intervenors first argue that this court lacks
jurisdiction to review the interlocutory LEP Order. We disagree.
      28 U.S.C. § 1292(a)(1) provides this court with jurisdiction over
“[i]nterlocutory orders of the district courts of the United States . . . granting,
continuing, modifying, refusing or dissolving injunctions, or refusing to dissolve
or modify injunctions.” The trial court ordered that defendants “establish a
monitoring system and establish a language program that fulfill the
requirements of the Equal Education Opportunity Act, 20 U.S.C. § 1703(f)” for
the 2009–2010 academic year and thereafter. The LEP order also requires
defendants to “submit a monitoring plan addressing the failures of PBMAS and
to submit a proposed new language program for secondary LEP students by
January 31, 2009, or earlier.” 
Id. We find 
these instructions to be sufficiently
coercive to constitute a mandatory injunction appealable under Section
1292(a)(1). See Morales v. Turman, 
535 F.2d 864
, 867 n.6 (5th Cir. 1976), rev’d
on other grounds, 
97 S. Ct. 1189 
(1977); see also 
Morales, 97 S. Ct. at 1190 
(the
district court’s “judgment is reviewable on the merits in the Court of Appeals”).
      Though the above instructions do not specify the plan’s exact content,
defendants were ordered to affirmatively (and promptly) act to establish a
monitoring system and to devise and implement a new secondary LEP program.
There are over 1,200 school districts in the State of Texas. Devising a new
monitoring system and secondary LEP language program for all of these
districts would require extraordinary effort. Defendants would not only be
forced to address a wide variety of concerns, e.g., funding, staffing, and even
their own legislative authority to create and carry out such a plan, they would
also need to implement substantial changes systemwide in a very short period
of time.
      Because the district court’s order compels defendants to promptly and

                                        11
affirmatively act in a specific and extremely extensive manner, the order
constitutes a mandatory injunction over which this court may assert its
appellate jurisdiction.
B. Standard of Review
      The district court’s order granting intervenors’ motion to alter the
judgment, pursuant to Federal Rules of Civil Procedure 52(b) and 59(e), is
reviewed for abuse of discretion. See Ross v. Marshall, 
426 F.3d 745
, 763 (5th
Cir. 2005), cert. denied, 
549 U.S. 1166 
(2007). “A district court abuses its
discretion if it ‘bases its decision on an erroneous view of the law or on a clearly
erroneous assessment of the evidence.’” 
Id. (quoting Hesling 
v. CSX Transp.,
Inc., 
396 F.3d 632
, 638 (5th Cir. 2005)). In granting intervenors’ motion to alter
the judgment, the district court held that defendants violated both the Modified
Order and the EEOA. Thus, this court must independently review both holdings
for abuse of discretion.
C. Section G of the Modified Order
      The district court retained remedial jurisdiction over all actions brought
to enforce or amend the Modified Order. Modified Order, Section J. Intervenors
seek to enforce the Modified Order by invoking the broad language of Section
G(1), which requires defendants to “insure that school districts are providing
equal educational opportunities in all schools.” 
Id. at Section 
G. Thus, the
district court appropriately asserted jurisdiction over intervenors’ claim. After
reviewing the evidence, the district court held that defendants violated Section
G because PBMAS’s flaws were so significant that defendants could not
effectively insure that LEP students received equal educational opportunities in
Texas public schools. We conclude that this holding was in error.
      “The Modified Order directs Defendants to take affirmative steps to
eliminate all remaining vestiges of the former de jure segregated school system

                                        12
in Texas, to prevent the recurrence of a segregated system, and to achieve fully
integrated schools.” LULAC IV, 
572 F. Supp. 2d 
at 753 (citing United States v.
Texas, 
793 F.2d 636
, 642 (5th Cir. 1986)). Significantly, the district court retains
broad remedial jurisdiction only “over those facets of school operations [that]
represent or flow from an earlier de jure discriminatory system.” United States
v. Texas (Goodrich), 
158 F.3d 299
, 311 (5th Cir. 1998). Thus, to find a violation
of the Modified Order, the trial court must determine that the contested action
“represents or flows” from a de jure segregated or discriminatory system. As
defendants are state actors and the trial court ordered a statewide remedy, the
decision below cannot be upheld absent a showing of statewide de jure
segregation of Mexican-Americans.
       This court has repeatedly acknowledged the historical and statewide de
jure segregation of black and white students in Texas, but has held that at no
time “ha[s] Texas segregated Anglo students from Mexican-American ones by
law.” United States v. Gregory Portland Independent School District, 
654 F.2d
989
, 992 (5th Cir. 1981). Nor is the trial court entitled to presume such
statewide de jure segregation absent an evidentiary hearing. See 
id. at 998. 
The
trial court made no factual findings with regard to statewide de jure segregation
of Mexican-Americans,9 nor would the record here support such a finding.


       9
        The trial court did state that: “this Court has already found [in the 1981
intervention], and the Fifth Circuit has upheld, that TEA and the State of Texas [were]
involved in intentional discrimination against Spanish speaking Mexican-American
students [in the past].” LULAC IV, 
572 F. Supp. 2d 
at 756. The district court clearly
misunderstood this court’s 1982 decision, where we reversed the district court’s findings
and held that there was not adequate “factual support for the [district] court’s finding of
historical segregation of Mexican-Americans.” LULAC 
II, 680 F.2d at 370
. The district
court also ignored this court’s determination in Gregory-Portland that no such system
existed. 654 F.2d at 992
. Thus, the lower court erred in concluding that Mexican-
Americans suffered statewide de jure discrimination in the past (although the court
disclaimed basing its presently challenged orders on any such determination, 
572 F. Supp.

                                            13
Because there is no showing of statewide de jure segregation of Mexican-
Americans, the trial court cannot enforce Section G under the facts and claims
presented.
       Further, the district court erred in relying on its broad remedial
jurisdiction under the Modified Order to assert jurisdiction over intervenors’
EEOA claim.       In determining whether it had jurisdiction, the trial court
reasoned that, because intervenors’ claims were a continuation of the 1981
intervention under the Modified Order, the “instant action [was] a successive
motion . . . lineally descending from the Fifth Circuit’s remand in [1982].”
LULAC IV, 
572 F. Supp. 2d 
at 732. Yet, the trial court had previously held that
intervenors’ 1981 Section G claim was without merit. LULAC 
I, 506 F. Supp. at
410
.   There, the trial court held that Section G only required defendants to
conduct a study and file a report proposing remedial programs for, among other
things, curricular offerings for students who primarily spoke a language other
than English. 
Id. at 411. 
And intervenors’ Section G claim was denied because
TEA timely satisfied this obligation.10 
Id. at 411.
       It is unclear to this court how intervenors’ claim could be a successive
motion when the original claim was previously denied. And even if, as the trial
court suggests, intervenors raised new facts suggesting a violation of Section G’s
broader principle of equal educational opportunities, there has been no finding
of statewide de jure segregation of Mexican-Americans in Texas. Thus, the lower
court erred in finding that intervenors could establish a Section G violation, and


2d at 758).
       10
         The trial court addressed this apparent contradiction between the present order on
appeal and the original 1981 order, reasoning that “it is abundantly clear that [it] was
referring to Section G(2), and therefore did not hold that Section G(1) [sic], which has
nothing to do with the reporting requirement, was complied with by virtue of filing the
report.”

                                            14
the district court should not have exercised jurisdiction over intervenors’
supplemental EEOA claim pursuant to its remedial jurisdiction over the
Modified Order.
      Because intervenors’ Section G claim was invalid pursuant to the district
court’s own order in 1981 and Fifth Circuit precedent, we hold that intervenors’
Section G and EEOA claims should have been severed into entirely separate
lawsuits.
D. Equal Educational Opportunities Act
      1. Absence of Local Districts
      The district court’s finding of an EEOA violation is unreliable because the
district court failed to adequately address this circuit’s 1982 remand
instructions. In 1982, this court questioned whether intervenors’ EEOA claim
could be appropriately addressed absent local school districts as parties.
Specifically, we concluded that “there exists little if any practical or logical
justification for attempting to deal on a statewide basis with the problems
presented by this case.” LULAC 
II, 680 F.2d at 373
–74. See also note 
4, supra
.
And we remanded the action to the district court with the following instructions:
      “[T]he language problems to be met will necessarily vary by district
      . . . . [W]hether the effect of a local language program,
      state-mandated or not, constitutes appropriate action to deal with
      language barriers faced by the students of a given school district
      will of necessity be an essentially local question. . . . We fail to see
      how [a question of Section 1703(f) compliance] can be properly
      resolved in the absence of the school district concerned or how [it]
      can effectively be dealt with on a statewide basis. . . . [W]e therefore
      direct the district court to determine . . . what questions—if
      any—presented by the case are subject to resolution on a statewide
      basis before proceeding further on the remand that we mandate.”
Id. at 374.
      In light of this court’s analysis of the evidence below, we conclude that the

                                        15
issues raised by intervenors’ EEOA claim have not been properly addressed in
the absence of individual school districts as parties.
      This court notes that intervenors identified fourteen LULAC members who
were parents of one or more LEP students. Those parents represent, in total,
twenty LEP students attending schools within three individual school districts.
See also note 
5, supra
. Not one of these school districts is named as a party to
this action or located within the Eastern District of Texas.
      Further, because no school district is a party to the present litigation, the
issue remains as to whether the district court constitutes an “appropriate
district court” capable of asserting jurisdiction over intervenors’ claim. See 20
U.S.C. § 1708 (“The appropriate district court of the United States shall have
and exercise jurisdiction of proceedings instituted under [this statute].”
(emphasis added)).      The district court improperly relied on its remedial
jurisdiction under the Modified Order to assert jurisdiction over intervenors’
supplemental EEOA claim; thus, it failed to adequately address whether it
constituted an “appropriate district court” pursuant to Section 1708. Congress
did not define “appropriate district court” in the statute, and there is little
legislative history or judicial precedent on the issue.
      We conclude that “an appropriate district court” is normally the district
court in which a local school district, as a party to the action, is located. And
once one or more local districts are added as parties in this litigation, the district
court should reconsider whether it constitutes an “appropriate district court” for
jurisdictional purposes or if the case should, instead, be handled in a forum that
constitutes a proper venue for such additional parties.
      2. The Evidence Presented Below
      This court’s decision regarding the necessity of a local school district is
further supported because the record below is insufficient to uphold the trial

                                         16
court’s finding of an EEOA violation. To find a violation of, and order a remedy
under, the EEOA, intervenors must establish (1) a violation of a student’s rights
under the EEOA, (2) that the violation “stem[med] from a failure to take
‘appropriate action’” on the part of the defendants, 
Horne, 129 S. Ct. at 2605
, and
(3) that any remedial order is “essential to correct [the] particular denials” of
EEOA rights found. 20 U.S.C. § 1712. We conclude that intervenors failed to
make the requisite showing.
       Under the EEOA, a state may not “deny equal educational opportunity to
an individual on account of his or her race, color, sex, or national origin, by . . .
the failure by an educational agency to take appropriate action to overcome
language barriers that impede equal participation by its students in its
instructional programs.” 20 U.S.C. § 1703(f) (emphasis added). Both state and
local educational agencies are responsible for taking “appropriate action” under
the statute,11 but the EEOA “leave[s] state and local educational authorities a
substantial amount of latitude in choosing the program and techniques they
would use to meet their obligations under the EEOA.” Castaneda v. Pickard,
648 F.2d 989
, 1009–10 (5th Cir. 1981).             See also 
Horne, 129 S. Ct. at 2589
(quoting Castaneda), 
id. at 2595 
(“. . . [T]he EEOA itself limits court-ordered
remedies to those that ‘are essential to correct particular denials of equal
educational opportunity or equal protection of the laws,’” (quoting 20 U.S.C. §




       11
          The EEOA defines “educational agency” as “a local educational agency or a ‘State
educational agency,’” and a state educational agency as “the agency primarily responsible
for the State supervision of public elementary schools and secondary schools.” 20 U.S.C.§§
1720(a), 7801(41); see Gomez v. Ill. State Bd. of Educ., 
811 F.2d 1030
, 1042–43 (7th Cir.
1987) (holding that, where powers are retained by the state or its educational agency, the
state is obligated to take appropriate action under § 1703(f)); Idaho Migrant Council v. Bd.
of Educ., 
647 F.2d 69
, 71 (9th Cir. 1981) (holding that the EEOA “imposes requirements on
the State Agency to ensure that . . . language deficiencies are addressed”).

                                             17
1712, emphasis by Supreme Court)).12
       “[T]he State of Texas has chosen a system of shared responsibilities
between state actors and local officials” in educational matters. LULAC IV, 
572
F. Supp. 2d 
at 757. Under this system, local school districts are primarily
responsible for the implementation of LEP programs, while TEA is responsible
for ensuring compliance with federal and state law and evaluating and
monitoring the effectiveness of LEP programs.             TEX. EDUC. CODE §§ 7.003,
7.021(b)(1), 7.028, 11.002, 29.062. Thus, defendants’ liability under the EEOA
is dependent upon the extent to which they have failed to take “appropriate
action” in fulfilling their legal responsibilities under state law – which in this
case requires that they monitor and evaluate LEP programs via PBMAS – which
failure has caused a deprivation of the rights of Mexican-American LEP students
under Section 1703(f).
       To determine the appropriateness of an educational agency’s action, this
court has instituted a three-prong test: (1) whether the program is based on
sound educational theory, (2) whether reasonable efforts are being made to
implement the theory (implementation prong), and (3) whether the program,
over a legitimate period of time, has achieved some success in overcoming
language barriers (results prong). 
Castaneda, 648 F.2d at 1009
–10. After



       12
          Moreover, while the No Child Left Behind Act of 2001 (NCLB) does not replace
Section 1703(f) (and compliance with the former does not necessarily constitute compliance
with the latter), nevertheless, as Horne states:
        “NCLB marked a dramatic shift in federal education policy. It reflects
        Congress’ judgment that the best way to raise the level of education
        nationwide is by granting state and local officials flexibility to develop and
        implement educational programs that address local needs, while holding
        them accountable for the results. . . . NCLB conditions the continued receipt
        of funds on demonstrations of ‘adequate yearly progress.’” 
Id. at 2601.
Title III of NCLB specifically focuses on LEP students and their acquisition of English
language proficiency. 
Id.

                                            18
applying 
the above analysis, the district court concluded that defendants’ use of
PBMAS in its current form fails both the implementation prong and the results
prong.13
       The district court’s ruling that a given defendant has or has not taken the
“appropriate action” required of it under section 1703(f) presents “a mixed
question of fact and law,” review of which requires us to determine whether the
trial court’s “conclusion was adequately supported by subsidiary findings of fact
which do not appear clearly erroneous.” 
Castaneda, 648 F.2d at 1010
. Whether
such essentially historic facts adequately support the conclusion presents a
question of law reviewable de novo. See St. Tammany Parish Sch. Bd. v.
Louisiana, 
142 F.3d 776
, 782 (5th Cir. 1998).14 To the extent that the subsidiary
findings of fact are “premised upon an erroneous view of controlling legal
principles” they “are entitled to no deference.” Johnson v. Uncle Ben’s, Inc., 
628
F.2d 419
, 422 (5th Cir. 1980).15
               a. Implementation Prong
       Under the implementation prong, a court must determine whether
defendants’ implementation of PBMAS “follow[s] through with practices,
resources and personnel necessary to transform the [educational] theory into

       13
        It is undisputed that defendants’ primary and secondary LEP programs are based
upon sound educational theory. See, e.g., 
572 F. Supp. 2d 
at 763-64 (“There is no dispute
that Defendants’ bilingual and ESL programs are sound in theory.”).
       14
         “. . . [O]ur court will ‘review de novo, as a mixed question of law and fact, a district
court’s [merits-] decision that a local school district’s IEP was or was not appropriate and
that an alternative placement was or was not inappropriate under the IDEA.’” 
Id. (quoting
Cypress 
Fairbanks ISD v. Michael F., 
118 F.3d 245
, 252 (5th Cir. 1997)).
       15
           See also, e.g., Lake Charles Stevedores v. Professor Vladimer Popov MV, 
199 F.3d
220
, 223 (5th Cir. 1999); Johnson v. Hospital Corp. of America, 
95 F.3d 383
, 395 (5th Cir.
1996); Delta S.S. Lines, Inc. v. Avondale Shipyards, Inc., 
747 F.2d 995
, 1000 (5th Cir. 1984)
(“. . . [W]hen essentially based on an incorrect legal principle, Rule 52(a) clearly erroneous
does not apply and we disregard any such possible findings.”).

                                               19
reality.” 
Castenada, 648 F.2d at 1010
. Essentially, this Court must scrutinize
whether defendants have made a “bona fide effort[] to make the program work.”
Id.
      According to 
the district court, defendants failed the implementation prong
because PBMAS, as implemented, cannot effectively monitor the progress of LEP
programs and students. LULAC IV, 
572 F. Supp. 2d 
at 765. The trial court
identified a variety of “flaws” that rendered the system ineffective: (1) the
absence of equality based comparisons, (2) under-identification of LEP students,
and (3) the aggregation of data at the district level.
                   1. Equality Based Comparisons
      First, the trial court determined that PBMAS was flawed because it did
not make equality based comparisons between LEP and non-LEP students.
Data collected under PBMAS compares LEP student achievement to state
assigned target passage rates; a district’s deviation from state targets
determines its performance level, which in turn determines whether and to what
degree TEA will intervene. Thus, PBMAS does not directly compare LEP
student achievement with non-LEP student achievement, and absent such a
direct comparison, the district court concluded that the program could not
adequately monitor the effectiveness of LEP programs.        The only basis the
district court provided for this conclusion was that the EEOA is an equality
based statute.
      Federal law provides no instruction as to how states should analyze LEP
student achievement. But the Supreme Court did recently hold that “the EEOA
requires ‘appropriate action’ to remove language barriers, § 1703(f), not the
equalization of results between native and nonnative speakers on tests
administered in English.”      
Horne, 129 S. Ct. at 2605
.     And by state law,
defendants are only required to “evaluate the effectiveness of [bilingual and

                                        20
ESL] programs” as “compared to state-established standards.” TEX. EDUC. CODE
§§ 29.062(a), 39.053(b). We do not believe that EEOA compliance requires the
comparisons insisted on by the district court.
      Further, reasonable minds could differ over what comparative method is
most effective to determine whether language barriers are being overcome. The
State of Texas and the TEA have determined that success is best measured by
determining whether LEP students are achieving state-passage rate targets, and
there is no evidence that the district court’s preferred method of comparison is
better than that of the State’s.
      In fact, the district court’s method comes with significant faults. For
example, standardized tests are administered to secondary LEP students in
English, a language in which they, by definition, lack proficiency. 
Horne, 129
S. Ct. at 2603 
n.16; 
Castaneda, 648 F.2d at 1014
. English administration makes
it difficult for test results to accurately capture an LEP student’s knowledge of
core curriculum, and it is inevitable that such students will not perform as well
as non-LEP students. Also, once an LEP student has successfully completed a
language remediation program, he or she is no longer identified as LEP,
rendering it impossible for a direct comparison to adequately capture program
successes. Thus, a direct comparison between LEP and non-LEP students may
not adequately represent the actual achievement gap between the two groups.
And, by making equality a priority, districts with a significant gap in passage
rates will require intervention regardless of how high LEP student scores might
be, whereas no intervention will occur in districts where LEP and non-LEP
students have lower, but reasonably equivalent, scores.
      The district court failed to adequately acknowledge any of the above
criticisms and, instead, abused its discretion by supplanting defendants’
educational policy decision with its own judgment.        Moreover, there is no

                                       21
evidence that any such method of comparison deficiency caused any deprivation
of the Section 1703(f) rights of Mexican-American LEP students.
                      2. Under-identification of LEP Students
       The district court also determined that school districts are “likely” under-
identifying LEP students. LULAC IV, 
572 F. Supp. 2d 
at 767–68. TEA is tasked
with establishing criteria for identification, assessment, and classification of
LEP students. TEX. EDUC. CODE § 29.056(a). Once a student is identified as
LEP, the district must obtain parental consent for the child to take part in an
LEP program. 
Id. However, parents 
may, and sometimes do, deny consent.
       At trial, intervenors presented evidence that, while the state average of
parental denials is 4.9%, some districts have reported a significantly higher
percentage of denials.16 Based on this evidence, the trial court determined that
those districts must be under-identifying LEP students, and this under-
identification decreases the perceived achievement gap between LEP and non-
LEP students. LULAC IV, 
572 F. Supp. 2d 
at 767–68.
       We find that the district court reached this conclusion in error. The
district court based its conclusion on the belief that a parental denial altered the
identification of the student, so that the child is no longer characterized as LEP.

       16
          Intervenors also presented evidence that some districts report a significantly lower
percentage of LEP students than the county-wide percentage of households in which a
language other than English is spoken. To the extent that the district court based its
finding of under-identification on this evidence, we find that this conclusion was in error.
The percentage of households in which a language other than English is spoken does not in
any meaningful way evidence the number of LEP students residing within the county.
        The district court also observed “it appears that in at least some of these schools [in
districts with high rates of parental denials], parents may not be well informed of the
advantages of bilingual-ESL programs or may be subject of coercion” (emphasis added).
There is no substantial evidence of instances of any such “coercion” (or misinformation),
and in any event the above quoted observations are entirely too vague and speculative to
form any valid basis for relief against defendants. Finally, there is no substantial evidence
of any default by defendants in performance of their duties in this respect causing Mexican-
American LEP students’ loss of their § 1703(f) rights.

                                              22

Id. at 767–68. 
This conclusion is entirely incorrect—once students are identified
as LEP, they remain identified as such until they obtain English proficiency,
regardless of participation in the program. Thus, the district court abused its
discretion in determining that the PBMAS data was unreliable due to the under-
identification of LEP students.
                     3. Aggregation of Data
       The district court also took issue with the PBMAS monitoring system’s
aggregation of data at the district level.17 
Id. at 768. 
The district court held that
this comparative method prevents intervention on individual campuses that are
not performing as well as the district overall. In reaching this conclusion, the
court relied upon a study which showed that 277 schools, attended by a total of
54,963 LEP students, performed at a performance level lower than the stage of
intervention required by the district’s assigned performance level.18 However,
this study did not ascertain the total number of campuses in need of intervention
or the total number of campuses subject to intervention based on the district-
wide data. Thus, it did not provide what percentage of campuses in need of
intervention or subject to intervention these 277 campuses actually represented.
       Rather than questioning the significance of the report, the trial court
perfunctorily concluded that aggregation of data at the district level rendered
the PBMAS monitoring system ineffective. Further, the trial court determined


       17
         The trial court also disapproved of TEA’s aggregation of TAKS subject data across
multiple grade levels. Defendants explained that TAKS subjects are not annually
administered at each grade level; thus, to determine achievement in specific subjects,
TAKS scores for each subject are combined across the grade levels in which the subject test
was administered. The district court held that this allowed the scores of successful primary
students to overinflate the success of the entire group, thus masking the underperformance
of secondary LEP students.
       18
       These numbers represent only 3.4% of campuses within the state and 7.7% of the
LEP student population.

                                            23
that it would be more reasonable for TEA to analyze data at the campus level to
avoid this “masking effect.”19 This conclusion ignores the allocation of shared
responsibilities within the Texas education system. Under Texas’s system, local
districts are primarily responsible for implementing LEP programs, and
addressing and remedying under-performing campuses within their own district.
On the other hand, TEA is responsible for intervening in failing districts to
ensure overall compliance.20
       There is no finding nor sufficient evidence that individual districts are
ignoring LEP under-performance on individual campuses or sufficient evidence
of resulting actual harm to such campuses and their students. We accordingly
find that the district court abused its discretion in supplanting defendants’ policy
choice with its own preferred method of comparison.21
       On remand, the district court should reconsider the evidence, in light of
this court’s opinion, to properly determine whether PBMAS, coupled with other
sources of information (such as NCLB), is capable of effectively monitoring the
success of LEP programs and whether any such monitoring deficiencies of
defendants (together with any others so found) amount to a failure to take

       19
        The district court felt a campus-level analysis was reasonable because TEA
already collects both district-wide and campus data.
       20
        In holding that the evidence adduced thus far fails to establish that the PBMAS
monitoring system does not discharge the State’s responsibilities under the EEOA, we do
not necessarily foreclose a finding upon more factual development that district-wide
monitoring may be inappropriate for larger school districts because of its potential for
masking the deficiencies on individual campuses (nor do we now give either legal blessing
or condemnation to any such finding on further factual development).
       21
         The district court also explained that defendants should monitor at the campus
level because the Modified Order was originally purposed to remedy segregation on the
campus level and Section G requires that equal opportunities be provided in “all schools.”
To the extent the trial court relied in any way on the Modified Order to reach the above
conclusions, we find it committed error because there has been no showing of statewide de
jure segregation against Mexican-Americans.

                                            24
appropriate action under Section 1703(f) which caused a denial of the rights of
Mexican-American LEP students thereunder.
             b. Results Prong
      Under the results prong, this court must determine whether defendants’
program, “after being employed for a period of time sufficient to give the plan a
legitimate trial, [failed] to produce results indicating that the language barriers
confronting students are actually being overcome,” and thus, “no longer
constitute[s] appropriate action.” 
Castaneda, 648 F.2d at 1010
. Under this
prong, the district court determined that the evidence presented—including
drop-out rates, retention rates, and the achievement scores of LEP secondary
students—indicated widespread program failure. Further, the court held that
the totality of this data conclusively proved that defendants’ failure to effectively
monitor LEP programs caused such LEP secondary student failure at the
ground level. We disagree.
      To begin with, we note that, statewide, the total secondary school LEP
population is only some twenty percent of the total LEP student population.
      This court does not dispute that secondary LEP student performance data
is alarming. It is clear that LEP students at all secondary grades are not
performing satisfactorily on TAKS, and the performance data indicate
significant achievement gaps as compared to passage rate targets—roughly
falling anywhere between a thirty-five to forty-five percent gap in achievement.
This court is further concerned that there is little evidence that these gaps are
steadily decreasing over time or that secondary LEP students are showing signs
of steady improvement on standardized tests.
      But in spite of these real concerns, we find that the district court over-
emphasized the significance of student achievement scores. As mentioned
previously, it is difficult for standardized tests administered in English to

                                         25
accurately capture an LEP student’s knowledge of core curriculum. And, “[i]t is
inevitable that [LEP] students (who by definition are not yet proficient in
English) will underperform as compared to native speakers.” 
Horne, 129 S. Ct.
at 2603 
n. 16. See also 
Castaneda, 648 F.2d at 1014
. Last, the system is
incapable of crediting high-performance by non-native speakers who have
already completed the program because they are no longer classified as LEP.22
Thus, it is unsurprising that the achievement scores of secondary LEP students
illustrate a significant gap in achievement.
       Further, the district court’s analysis of student achievement data was not
restricted to the longitudinal data derived from the PBMAS system. Instead, the
district court reviewed data over a period of four years, from the 2003–2004
school year to the 2006–2007 school year, even though the PBMAS system was
not implemented until the 2004–2005 school year, and data from the 2006–2007
school year was not in evidence. Thus, only two-years worth of PBMAS data was
properly in evidence and before the court.
       We find this period of time insufficient to show whether defendants’ use
of PBMAS enables them to effectively monitor LEP programs and ensure EEOA
compliance. See 
Castaneda, 648 F.2d at 1010 
(explaining that LEP programs
must be given a “sufficient” period of time to work). The court failed to analyze
longitudinal data provided during the time when PBMAS was in effect; thus, the
comparisons made, and conclusions reached in making them, are unreliable.
Horne, 129 S. Ct. at 2603 
n. 16. ( “[An] absence of longitudinal data in the record


       22
          Yearly in every grade (at least at and/or above grade 3) LEP students are tested in
English language proficiency tests and those who are scored “advanced high” are thereafter
no longer classified as LEP. Thus, it is virtually inevitable that the higher grades will tend
to have a lower proportion of students whose natural talent is above average at least in the
ability to learn a language other than that in which they communicated prior to entering
school.

                                             26
precludes useful comparisons.”). Due to PBMAS’s nascent development and the
absence of longitudinal data, we find that the district court abused its discretion
in concluding that the PBMAS system had not achieved satisfactory results.
      We also take issue with the district court’s conclusion that the “totality of
[the] data establishes causation.”23 LULAC IV, 
572 F. Supp. 2d 
at 779, 779–80.
Included in the “totality of data” were comparisons of the achievement scores
discussed above as well as drop-out rates, retention rates, and participation
rates in advanced courses.          For example, the district court found that in
2003–2004, 2.0% of secondary LEP students dropped out of school, while only
0.9% of secondary non-LEP students dropped out. The district court also pointed
out that in 2003–2004, the retention rate of secondary LEP students was 13.8%,
while only 6.3% of non-LEP students were retained.                     And the trial court
determined that LEP students are far less likely to participate in a dual
enrollment, advanced placement, or international baccalaureate program, e.g.,
in 2004, 8.5% of LEP students, compared to 19.9% of all students, completed a



      23
        The district court held that its previous order erred in reading “extraneous
causation” into Castaneda’s three-part test and erroneously determined that aggregate
student performance data, without more, could not evidence the requisite degree of
causation:

      “This Court . . . misstated the factual record because Intervenors presented
      much more than ‘aggregate student performance data.’ Secondary LEP students
      across the board not only failed to perform at the level of their non-LEP peers
      on achievement tests, but also dropped out of school at significantly higher
      rates; had significantly higher retention rates; and remained in LEP programs
      for four or more years, without making adequate yearly progress. That the
      Court did not consider the multitude of indicators in its finding of fact was error
      in itself; that the Court amplified this error by misstating that Intervenors had
      presented only a paucity of evidence was . . . clear and manifest reversible error
      . . . . Examining such nebulous factors as social and economic background as
      potential primary causes of LEP student failure is a task fraught with hazard.”

LULAC IV, 
572 F. Supp. 2d 
at 773.

                                              27
dual enrollment course.
      The above comparisons have little significance. First, these comparative
factors suffer the same faults as student achievement scores—the data was
collected prior to PBMAS implementation, and because the system was
implemented only for the 2004-2005 school year, the comparative factors are not
based on longitudinal data.     Second, a 1.1% difference in the drop out rate
between LEP and non-LEP students is not significantly probative. And though
the retention rates show a more significant gap, it is foreseeable that LEP
students would have greater difficulty in mastering core curriculum while
simultaneously trying to achieve English proficiency. Last, the rate of LEP
secondary student participation in advanced courses is of little probative
value—it is to be expected that those students still attempting to achieve
English proficiency are not as likely to take part in advanced courses.
      More significantly, the totality of the data does not explain how the
“failure” of LEP students was caused by errors or omissions on the part of TEA.
To find a violation of the EEOA, a court must make sufficient findings of fact to
support a conclusion that student failures “stem from [defendants’] failure to
take ‘appropriate action.’” 
Horne, 129 S. Ct. at 2605
. The evidence in this case,
which essentially is only comprised of statewide aggregate student performance
data and participation rates, cannot support such a causal connection.
      Under Texas law the school districts are the ones having “the primary
responsibility for implementing the state’s” bilingual and ESL programs, TEX.
EDUC. CODE § 11.002 (emphasis added), and the boards of trustees of such
districts have “primary responsibility for ensuring the district . . . complies with
all applicable requirements of state educational programs.” 
Id. § 7.028(b).
Further, each school district determines whether ESL or bilingual programs (or
other transitional language instruction) are used in the seventh and eighth

                                        28
grades. 
Id. § 29.053(d)(2). 
There is essentially no evidence, findings or analysis
of the performance of any individual school districts or any groups of districts
(whether grouped geographically, or by either overall and/or LEP student
population size, or by some socio-economic criteria, or on any other basis), or of
the presence or absence or the extent or nature of differences in performance of
LEP students (or students as a whole), or the nature of the bilingual or ESL
programs and the implementation thereof, as between different districts, or
groups of districts, or the like.24 There is simply no substantial evidence that
diverse implementation deficiencies in numerous different school districts did
not primarily cause the poor performance of so many secondary LEP students.
       Moreover, the district court erred in failing to consider other possible
causes for the secondary LEP student failure. See, e.g., note 
22, supra
. The
district court did not explore alternative causes because it concluded that
primary and secondary LEP students suffer similar socio-economic conditions,
but show significantly different levels of achievement; thus, “social and economic
factors . . . are not the culprit.” LULAC IV, 
572 F. Supp. 2d 
at 780. The trial
court provided no additional evidence to support its assumption.
       While we agree that primary and secondary LEP students in a given
school district likely exist in similar social and economic situations, we do not
believe they are exposed to the same challenges. For example, the Supreme
Court recently recognized that secondary LEP students face a greater number
of obstacles:
       “There are many possible causes for the [under-]performance of
       students in [LEP programs]. These include the difficulty of teaching
       English to older students (many of whom, presumably, were not in


       24
          Apart from that relating to the percentage rate of parental denials in certain
districts. See opinion below, 
572 F. Supp. 2d 
at 737-38.

                                              29
      English-speaking schools as younger students) and problems, such
      as drug use and the prevalence of gangs.”
Horne, 129 S. Ct. at 2605 
n. 20.25 Thus, any comparison between LEP and non-
LEP students “must take into account other variables that may explain [their
differences in achievement]” in order to accurately determine the true cause of
student failure. 
Id. The district 
court abused its discretion in failing to address
other possible causes of student failure.
      In light of the above errors, we hold that the evidence relied upon by the
district court does not establish that TEA has failed to take “appropriate action”
to overcome language barriers nor does it establish that TEA has somehow
abdicated responsibility in monitoring the secondary LEP program.
Nonetheless, we recognize that LEP student performance is alarming, and we
encourage the district court and the parties involved to reconsider whether one
or more individual school districts should be added to this litigation in order for
it to proceed (with transfer to the proper district and division as appropriate).
By adding individual districts, the court can better examine the circumstances
of specific students, schools, and districts, which will be invaluable evidence for
determining the cause of LEP student failure and how best to remedy it.
      We do not suggest that state defendants cannot be held liable under the
EEOA. Instead, we merely hold that an appropriate analysis of an EEOA claim
should be conducted with regard to a particular district or districts, with state
educational agencies serving as additional parties.26


      25
          This court notes that a significant difference between primary and secondary
students is their ability to obtain gainful employment. Secondary students who are able to
alter their economic position through employment will invariably have less time to devote
to their studies and may feel added pressure to drop out of school altogether.
      26
      Castaneda offers just such an example, where plaintiffs appropriately initiated an
EEOA action against a local school district and later named the TEA as a defendant. See

                                            30
E. Motion to Modify the Modified Order
      This court reviews the district court’s order granting in part, and denying
in part, defendants’ motion to modify the order, pursuant to Federal Rule of Civil
Procedure 60(b)(5), for abuse of discretion. Frazar v. Ladd, 
457 F.3d 432
, 435
(5th Cir. 2006). Federal Rule of Civil Procedure 60(b)(5) provides that a party
may obtain relief from a court order when “applying it prospectively is no longer
equitable.” The party seeking to modify an injunction bears the burden of
establishing that a significant change in factual conditions or the law warrants
revision of the injunction. See Rufo v. Inmates of the Suffolk County Jail, 
502
U.S. 367
, 383–84 (1992).
      Defendants argue that this court’s decision in Samnorwood Independent
School District v. Texas Education Agency, 
533 F.3d 258 
(5th Cir. 2008),
constitutes a significant change in the law and requires this court to limit the
reach of the Modified Order to the nine all-black school district defendants in the
original lawsuit.    In Samnorwood, two school districts that voluntarily
desegregated in the 1960s and were not party to the original 1970 litigation
challenged whether the Modified Order could be appropriately applied to them.
This court held as follows:
      “[T]he prophylactic provisions created by the Modified Order to
      remedy the segregative conduct on the part of TEA and all-black
      schools in East Texas should not be imposed on these two panhandle
      school districts that had long previously already desegregated and
      have never since been found to have acted with segregative intent.”
Id. at 269. 
Central to the court’s analysis was the fact that there had been no
showing of a constitutional violation by the school districts themselves, and “the
modified order and TEA’s actions pursuant to it constitute a remedy that 
must



648 F.2d at 992
.

                                        31
flow from a constitutional violation.” 
Id. at 268, 
267–68.
       The trial court read Samnorwood very narrowly, determining that this
Court’s decision was largely based on the fact that the two school districts
involved had been unitary prior to the implementation of the Modified Order.
LULAC V, 
2008 WL 5334404 
at *5. Ultimately, the district court modified the
order in a very restrictive manner, exempting only those districts situated
almost identically to the Samnorwood districts. We find that this modification
was overly restrictive.
       The Modified Order was issued for the purpose of eliminating the diverse
continued local practices and vestiges of de jure racially segregated public
education.27 Since its issuance, nearly forty years have elapsed, and “the racial
composition of public schools in Texas has changed drastically.” See United
States v. Texas, 
457 F.3d 472
, 475 (5th Cir. 2006).                 And while this court
recognizes that some local vestiges of discrimination and segregation might still
remain, it is clear that the Modified Order certainly has, at best, “dwindling
relevance.” 
Id.
       In light 
of the above, and this court’s supervisory powers, we hold that the
Modified Order’s reach should be further limited. Related concerns led us to
modify the original order on direct appeal to provide that “[no]thing herein shall
be deemed to affect the jurisdiction of any other district court with respect to any
presently pending or future school desegregation suits.” United States v. Texas,



       27
         This court, of course, recognizes that de jure racial discrimination existed in Texas.
We also note that in 1955 the Texas Supreme Court declared invalid, under the United
States Constitution, the provisions of the Texas Constitution and statutes requiring school
racial segregation, McKinney v. Blankenship, 
282 S.W.2d 691
, 695 (Tex. 1955), and that in
1969, the Texas Legislature formally repealed all of the State’s segregation laws. Acts
1969, 61st Leg., p. 3230, H.J.R. 3. Thus, the original 1970 litigation was brought at a time
when segregation was invalid under both federal and state law.

                                              32

447 F.2d 441
, 442 (5th Cir. 1971). See also, e.g., United States v. Texas, 
466 F.2d
518
, 519 (5th Cir. 1972) (transferring case as to San Felipe Del Rio School
District to Western District of Texas). Time has shown that further restriction
of the modified order’s scope is necessary and appropriate.                         See, e.g.,
Samnorwood, 533 F.3d at 269
; 
Gregory-Portland, 654 F.2d at 993-94
. The
district court was correct to exempt school districts similarly situated to the
Samnorwood districts. In addition, we find that the same modification must be
made for local school districts declared unitary in federal cases; school districts
currently under federal court desegregation orders or decrees or which are
parties to pending federal court desegregation suits; and any other specific
school district not a party to the case when the Modified Order was issued which
requests, or for which the State requests, exemption from the Modified Order
unless a plaintiff shows the district is not then unitary. As to such districts
that are thus now or subsequently exempted from the Modified Order, the
district court may no longer direct TEA with regard to such districts pursuant
to its remedial jurisdiction over the Modified Order.28
                                   IV. CONCLUSION
       We hold that the district court abused its discretion in finding that
defendants’ violated Section G of the Modified Order because there was no
sufficient evidence or finding of statewide de jure segregation against Mexican-
Americans. We also hold that the district court abused its discretion in holding
that defendants violated Section 1703(f) of the EEOA because the evidence
presented does not establish that a student’s right has been violated or that
defendants’ acts or omissions caused any claimed violation. On remand, we



       28
         And such districts shall no longer be subject to the district court’s continuing
jurisdiction under the Modified Order.

                                              33
instruct the court below to reconsider whether local school districts should be
added to this action, and upon the inclusion of a local school district, we instruct
the court below to reconsider whether it constitutes an “appropriate district
court” for jurisdictional purposes, or if, instead, the litigation should proceed in
a different forum appropriate to such school districts under normal venue
principles.
                         REVERSED and REMANDED.




                                        34