R.B., a minor, by and through her Guardian Ad Litem, F.B., and F.B. (“appellants”) appeal the district court’s entry of summary judgment in favor of the Napa Valley Unified School District (“appellee” or “District”). The district court, in turn, upheld the decision by the California special education hearing officer (“SEHO”) that R.B. is not entitled to special education protection and services under the Individuals with Disabilities Education Act (IDEA), 20 U.S.C. § 1400 et seq., and corresponding provisions of the California Education Code. Because R.B. did not qualify for special education services, appellants were ineligible for reimbursement of the expenses they incurred in placing R.B. at Intermountain Children’s Home and Services (“Intermountain”), a private school in Helena, Montana.

Appellants challenge the SEHO’s decision on procedural and substantive grounds. Appellants claim that R.B.’s individual education program (IEP) team should have included her teacher or therapist from Intermountain and that this procedural violation, in and of itself, denied R.B. a free appropriate public education (FAPE). Appellants further claim that the SEHO and district court erred in finding that R.B. did not have a “serious emotional disturbance” under the criteria enumerated in 34 C.F.R. § 300.7 (c)(4) (2003) and CaLCode Regs. tit. 5 § 3030(f). Appellants contend R.B. could not form satisfactory relationships with peers and teachers, manifested inappropriate behavior under normal circumstances, and was pervasively depressed.

We have jurisdiction pursuant to 28 U.S.C. § 1291 . We affirm.

BACKGROUND

R.B. was born in 1991 to a mother who abused cocaine, alcohol, and heroin. In infancy, R.B. demonstrated symptoms of exposure to illegal drugs in útero (including irritability, delayed visual maturation, and delayed motor skills). Both of R.B.’s birth parents were incarcerated. F.B., a single parent and schoolteacher, adopted R.B. at eighteen months of age. F.B.’s mother assisted in caring for R.B.

R.B. was molested by her natural father when she was two. Afterward, she required a year of play therapy because of her self-mutilation and inappropriate displays of affection. A psychologist diagnosed R.B. with Attention Deficit Hyperactivity Disorder (ADHD) and began prescribing medication when she was three. Other diagnoses included Reactive Attachment Disorder and Post Traumatic Stress Disorder.

R.B. was expelled from three preschool programs because of her classroom misconduct. F.B. then placed R.B. with the District, which determined R.B. was eligible for special educational services and developed an IEP program for her. R.B. transitioned into a regular kindergarten class with resource support, taught by Jams Sparks.

During R.B.’s first grade year at Donaldson Way Elementary School, the District concluded that R.B. no longer qual ified for special educational services. Instead, the District found R.B. was a “qualified handicapped individual” under Rehabilitation Act § 504 and developed a behavioral intervention plan. F.B. acquiesced in these changes only after the District agreed to a neutral psychological evaluation. Dr. Emily Jordan conducted the evaluation and confirmed the District’s conclusion that R.B. was no longer a “child with a disability.”

R.B.’s elementary school history includes a series of disturbing incidents. In second grade, R.B. banged a classmate’s head against a computer monitor for refusing to give up the computer at recess. R.B. was suspended in third grade for throwing chairs and running off campus until law enforcement restrained her. R.B. was suspended again in fourth grade when she refused to take her ADHD medication, yelled at her teacher, and was again restrained by law enforcement.

R.B.’s behavior reached an extreme point during the second trimester of fifth grade. She was suspended twice in the span of just over a month. First, R.B. twisted a child’s arm during recess and said she hoped her music teacher would die. Then, R.B. poked another student with a mechanical pencil while refusing to turn in her work. At the time, R.B. was alternately refusing to take her ADHD medications and receiving occasional double dosages from F.B. Working with R.B. and F.B., the District adopted a behavior management plan, which largely remedied R.B.’s misconduct. Throughout elementary school, R.B. excelled in her classes, scored high marks on achievement tests, and frequently made the honor roll.

In the spring of 2002, F.B. met with an educational consultant who referred R.B. to Dr. Paula Solomon for a psychological evaluation. Without observing R.B. in the classroom, Dr. Solomon recommended treatment in a residential placement program. Therefore, on July 15, 2002, F.B. wrote to the District that R.B. had “reached a crises [sic] point.” F.B. said that she would place R.B. in a residential treatment facility within ten (10) days and expected the District to reimburse her for the placement.

F.B. placed R.B. with Intermountain. Tina Morrison, the Intermountain staff psychologist, was R.B.’s therapist. Morrison observed that R.B. engaged in controlling and physically aggressive behavior toward staff and fellow students, to the point that R.B. was “derailed cognitively” at times. R.B.’s teacher at Intermountain was Kathy Brandt. R.B. took almost twice as long as the average Intermoun-tain student to transition into Brandt’s classroom. From November to March 2003, Brandt observed R.B. intimidating other students almost daily.

On August 6, 2002, F.B. requested an impartial due process hearing, pursuant to 20 U.S.C. § 1415©- Therefore, the District arranged for its psychologist, Denise Struven, to travel to Intermountain to conduct an evaluation. Struven concluded that R.B. did not qualify for special education benefits under the IDEA.

On January 31, 2003, the following individuals met as part of the District’s IEP team: Laura Miller, a special education teacher and Director of Special Education for the district; Sparks, then a principal of Donaldson Way Elementary; Struven and Donna Poninski, District psychologists; Sally Dutcher, attorney for the District; Jane F. Reid, then-counsel for appellants; and F.B. No one from Intermountain attended, although Struven reported her observations of R.B. there. The IEP team concluded R.B. was not eligible for special education benefits.

F.B. appealed to the California State Education Agency, pursuant to 20 U.S.C. § 1415 (g). Hearing Officer Jessica Katz of the California Special Education Hear ing Office conducted the hearing over six days in June and August 2003. The SEHO found for the District, concluding that R.B. did not meet the IDEA’S standard for a child with a “severe emotional disturbance” for either the 2001-02 school year (R.B.’s fifth grade year at Donaldson Way Elementary) or the 2002-03 school year (R.B.’s first year at Intermountain). The SEHO also found that any procedural violation in the composition of the IEP team did not result in a lost educational opportunity for R.B.

On January 5, 2005, F.B. filed a complaint for violation of the IDEA in the Northern District of California. The district court granted the District’s motion for summary judgment, and denied appellants’ cross-motion. After independently reviewing the record and giving due deference to the SEHO, the district court agreed that R.B. did not have a “serious emotional disturbance” under the IDEA and that any procedural violation did not result in a lost educational opportunity. Appellants timely appealed on July 5, 2005.

This court has jurisdiction under 28 U.S.C. § 1291 .

II.

STANDARD OF REVIEW

The court reviews findings of fact for clear error, even if those findings are based on the administrative record. Amanda J. ex rel. Annette J. v. Clark County Sch. Dist., 267 F.3d 877, 887 (9th Cir.2001). A finding of fact is clearly erroneous if “ 'the reviewing court is left with a definite and firm conviction that a mistake has been committed.’ ” Id. (quoting Burlington N., Inc. v. Weyerhaeuser Co., 719 F.2d 304, 307 (9th Cir.1983)). Mixed questions of fact and law are reviewed de novo unless, as here, the question is primarily factual. Id.; Gregory K. v. Longview Sch. Dist., 811 F.2d 1307, 1310 (9th Cir.1987).

When a party challenges the outcome of an IDEA due process hearing, the reviewing court receives the administrative record, hears any additional evidence, and, “basing its decision on the preponderance of the evidence, shall grant such relief as the court determines is appropriate.” 20 U.S.C. § 1415 (i)(2)(B) (2003). Courts give “ ‘due weight’ ” to the state administrative proceedings, Van Duyn ex rel. Van Duyn v. Baker Sch. Dist. 5J, 481 F.3d 770, 775 (9th Cir.2007) (quoting Bd. of Educ. v. Rowley, 458 U.S. 176, 206 , 102 S.Ct. 3034 , 73 L.Ed.2d 690 (1982)), and, at a minimum, “ ‘must consider the findings carefully,’ ” Ojai Unified Sch. Dist. v. Jackson, 4 F.3d 1467 , 1474 (9th Cir.1993) (quoting Gregory K, 811 F.2d at 1311 ). The court gives particular deference where the hearing officer’s administrative findings are “thorough and careful.” Union Sch. Dist. v. Smith, 15 F.3d 1519 , 1524 (9th Cir.1994).

III.

DISCUSSION

A. Procedural violation

The purpose of the IDEA is to provide special education services for children with qualifying disabilities. 20 U.S.C. § 1400 (d)(1)(A) (2003). In drafting the IDEA, “Congress placed every bit as much emphasis upon compliance with procedures ... as it did upon the measurement of the resulting IEP against a substantive standard.” Bd. of Educ. of the Hendrick Hudson Cent. Sch. Dist. v. Rowley, 458 U.S. 176, 205-06 , 102 S.Ct. 3034 , 73 L.Ed.2d 690 (1982). Procedural compliance “would in most cases assure much if not all of what Congress wished in the way of substantive content in an IEP.” Id. at 206 , 102 S.Ct. 3034 . Therefore, a reviewing court first considers a school district’s procedural compliance before reaching the IEP’s substance. Id.

The IDEA requires “the provision of a free appropriate public education [PAPE] to [each] child.” 20 U.S.C. § 1415 (b). A child is denied a FAPE only when the procedural violation “result[s] in the loss of educational opportunity or seriously infringe[s] the parents’ opportunity to participate in the IEP formation process.” W.G. v. Bd. of Trustees of Target Range Sch. Dist. No. 23, 960 F.2d 1479, 1484 (9th Cir.1992); accord Park v. Anaheim Union High Sch. Dist., 464 F.3d 1025, 1031 (9th Cir.2006). Where a school district improperly constitutes an IEP team, “IDEA procedural error may be held harmless[.]” M.L. v. Fed. Way Sch. Dist., 394 F.3d 634, 652 (9th Cir.2005) (Gould, J., concurring). Therefore, not all procedural violations deny the child a FAPE. Park, 464 F.3d at 1033 n. 3; Ford ex rel. Ford v. Long Beach Unified Sch. Dist., 291 F.3d 1086, 1089 (9th Cir.2002).

One of the IDEA’S procedural requirements is the creation of an IEP team to determine a child’s eligibility for IDEA benefits. Cf. 20 U.S.C. §§ 1414 (b)(4)(A), (c)(4). Prior to the 1997 amendments, the IEP team required the presence of “the teacher.” 20 U.S.C. § 1401 (a)(20) (1994). By contrast, under the amended statute and implementing regulations, the IEP team must include “at least one regular education teacher of such child (if the child is, or may be, participating in the regular education environment)” and “at least one special education teacher, or where appropriate, at least one special education provider of such child.” 20 U.S.C. § 1414 (d)(l)(B)(ii)-(iii); 34 C.F.R. § 300.344 (a)(2)-(3); cf. Cal. Educ.Code § 56341(b)(2)-(3) (2001). Appellants claim the District’s IEP team failed on both counts by including Sparks, who taught R.B. in kindergarten six years before the IEP meeting, and by not including Brandt, R.B.’s special education teacher at Inter-mountain.

Whether Sparks’s participation on the IEP team was a procedural violation requires us to interpret the phrase “at least one regular education teacher of such child[.]” 20 U.S.C. § 1414 (d)(l)(B)(ii); 34 C.F.R. § 300.344 (a)(2). Although appellants acknowledge the statutory amendments in a footnote, they seek to minimize the significance of those amendments by relying on cases that interpreted the prior version of the statute to require the participation of the student’s current teacher on the IEP team. Shapiro ex rel. Shapiro v. Paradise Valley Unified Sch. Dist. No. 69, 317 F.3d 1072, 1076 (9th Cir.2003), superseded by statute on other grounds, Individuals with Disabilities Education Act Amendments of 1997, Pub.L. 105-17, § 614 (d)(1)(B), 111 Stat. 37 ; Target Range, 960 F.2d at 1484. By using Shapiro and Target Range to assert that our law requires the presence of a child’s private school teacher at an IEP meeting, appellants are essentially asking us to conclude that the statutory amendments had no effect. M.L. did not resolve this issue because the IEP team there included no regular education teacher, although dicta in two of the opinions take a position at odds with appellants’ argument.

We conclude that, after the 1997 amendments, the IDEA no longer requires the presence of the child’s current regular education teacher on the IEP team. The phrase “at least one regular education teacher of such child” gives a school district more discretion in selecting the regular education teacher than the phrase “the teacher.” As the Shapiro court explained in its interpretation of the pre-1997 IDEA, “the teacher” was Congress’s way of requiring more than simply “a teacher” on the IEP team. 317 F.3d at 1077 . Allowing “a teacher” to assume the role set aside for “the teacher” interpreted the statute too broadly. Id. This case is simply Shapiro in reverse. If Congress had wanted the child’s current regular education teacher on the IEP team, Congress would have used more specific language than “at least one regular education teacher of such child.” Indeed, the phrase “at least one” contemplates that the IEP team will include regular education teachers other than the child’s current teacher. Requiring “the current regular education teacher” to assume the role set aside for “at least one regular education teacher” would interpret the statute too narrowly.

Like the Shapiro court, we find support for our construction in the- authorities of the Office of Special Education Programs (“OSEP”). See 317 F.3d at 1077 (referencing a footnote in an OSEP regulation to show that the child’s current teacher must participate on the IEP team). OSEP’s Notice of Interpretation explains, “[t]he regular education teacher who serves as a member of a child’s IEP team should be a teacher who is, or may be, responsible for implementing a portion of the IEP, so that the teacher can participate in discussions about how best to teach the child.” 34 C.F.R. Pt. 300 App. A — Question 26 (1999). This interpretation imposes no stringent requirement that the child’s current regular education teacher attend the IEP meeting. An agency’s interpretation of its own regulations carries controlling weight unless the interpretation is plainly erroneous or is inconsistent with the regulations themselves. Stinson v. United States, 508 U.S. 36, 45 , 113 S.Ct. 1913 , 123 L.Ed.2d 598 (1993); Nat’l Ass’n of Home Builders v. Norton, 340 F.3d 835, 843 (9th Cir.2003). Because the agency’s failure to require the participation of the current regular education teacher is neither plainly erroneous nor inconsistent with the regulation, we do not impose such a requirement.

Whether the district’s failure to include Brandt on the IEP team was a procedural violation requires us to interpret the phrase “at least one special education teacher or, where appropriate, at least one special education provider of such child[J” 20 U.S.C. § 1414 (d)(l)(B)(iii) (2003); 34 C.F.R. § 300.344 (a)(3) (2003). Like the companion provision on the regular education teacher discussed swpra, we interpret this provision not to require the participation of the child’s current special education teacher. Therefore, Brandt’s exclusion from the IEP team was not a procedural violation per se.

This is not the end of the argument, however, for nothing in the record establishes that Miller, the special education teacher on the IEP team, ever taught R.B. We conclude that Miller’s participation did not satisfy the IDEA because we interpret the statute and regulation to require a special education teacher who has actually taught the student. Although this requirement would be even clearer if the phrase “of such child” also appeared immediately after “special education teacher” and not merely after “special education provider,” we think the statute and regulation are clear enough as written. Furthermore, the OSEP Notice of Interpretation states, “[t]he requirements of [the regulation] can be met by either: (1) a special education teacher of the child; or (2) another special education provider of the ehild[.]” 34 C.F.R. Pt. 300 App. A—Question 23. OSEP’s interpretation of the regulation to require a special education teacher of the child is not plainly erroneous or inconsistent with the regulation itself. See Stinson, 508 U.S. at 45 , 113 S.Ct. 1913 ; Norton, 340 F.3d at 843 . Although the District did not have to include Brandt, it did not satisfy its legal obligations by including Miller. Therefore, the District’s failure to include a special education teacher or provider on the IEP team who actually taught R.B. was a procedural violation of the IDEA. We now address the question of whether that violation resulted in a loss of educational opportunity for R.B., or was instead harmless error.

We recognize that we have, more often than not, held that an IDEA procedural violation denied the child a FAPE. Compare M.L., 394 F.3d at 656 (Gould, J., concurring), Shapiro, 317 F.3d at 1079 , Amanda J., 267 F.3d at 894 , and Target Range, 960 F.2d at 1484-85 (all finding that the child was denied a FAPE) with Park, 464 F.3d at 1033 n. 3, and Ford, 291 F.3d at 1089 (both finding no denial of FAPE). All of these cases except Ford, however, concern students who were eligi ble under the IDEA and alleged that the procedural violations resulted in some defect in their IEP. In M.L., for example, the IEP team improperly excluded a regular education teacher and provided the student with an IEP that had few opportunities for “mainstreaming,” i.e., interaction with non-disabled students. 394 F.3d at 640 (Alarcon, J.). Judge Gould’s concurrence emphasized the statutory preference for mainstreaming and cited evidence from past school years to show that the student could be placed in regular education classrooms with non-disabled students. Id. at 656-57 . The procedural violation resulted in a lost educational opportunity, and thus denied the student a FAPE, because a properly constituted IEP team would likely have given greater consideration to mainstreaming and provided the student an IEP with more mainstreaming opportunities. Id. at 657 .

In Shapiro, the failure to include the child’s current special education teacher from her private school (under the pre-1997 IDEA) resulted in a loss of educational opportunity because the private school was the only place where the child had received special education. 317 F.3d at 1077,1079 . Without a representative from the private school at the IEP meeting, the school district refused to continue placing the student with the private school and instead offered placement in its own program. Id. at 1074-75 . The school district’s program was far inferior because, at the time of the IEP meeting, the district had not hired a teacher, the program might not continue past the upcoming school year, and the student would be the first and only enrollee. Id. at 1074-75 . The loss of educational opportunity was similar in Target Range, where, in the absence of the regular education teacher from the child’s private school, the school district proposed “a preexisting, predetermined” IEP and refused to consider alternatives. 960 F.2d at 1484 . See also Amanda J., 267 F.3d at 894 (failure to disclose student’s records, including documents that suggested the possibility of autism, denied child a FAPE by preventing parents’ full participation in the IEP meeting).

In these cases involving flaws in the IEP, the child has already jumped through a significant hoop by establishing IDEA eligibility. Once the child qualifies for special education services, the district must then develop “[a]n IEP which addresses the unique needs of the child[.]” Amanda J., 267 F.3d at 892 . If the IEP team is improperly constituted, the reviewing court is ill-situated to know what the IEP would look like if the school district had included all the required participants on the IEP team. Nor can the court substitute its judgment for educational policymakers by determining what kind of IEP will best suit the disabled child’s needs. A properly constituted IEP team is in the best position to develop an IEP that suits the peculiar needs of the individual student. See Rowley, 458 U.S. at 206 , 102 S.Ct. 3034 (explaining that procedural compliance would almost always achieve what Congress intended with respect to the substantive IEP provisions).

Like this case, Ford presented a different, more preliminary question: whether the student qualified for IDEA benefits in the first instance. In Ford, the district assessed the student and found that she was not IDEA-eligible. 291 F.3d at 1087 . The student claimed the assessment violated an OSEP regulation because it did not include classroom observation by someone other than her current regular education teacher. Id. at 1089 . We held that the procedural violation did not result in a lost educational opportunity for the student because three of the student’s former teachers testified at the administrative hearing. Id. at 1089 . In effect, the hearing cured the procedural violation by including those viewpoints which should have been considered when the IEP team met.

Similarly, in this case, the administrative hearing cured the procedural error in the composition of the IEP team. If the IEP team had included Brandt or Morrison, the District would have satisfied the requirement that a special education teacher or provider of the child be included. Although R.B.’s IEP team lacked such a person, Brandt and Morrison both testified at length during the hearing. To the extent that Brandt and Morrison believed R.B. should be IDEA-eligible, they were able to say why. The SEHO, district court, and now this court all have the benefit of their testimony in determining whether the District correctly concluded that R.B. was ineligible for special education services. We then apply the law (of IDEA eligibility) to the facts (including the new facts that Brandt and Morrison presented for the first time to the SEHO). In so doing, we apply the same eligibility criteria to all cases and are, therefore, better situated to find an error harmless than in the context of whether an IEP satisfies the unique needs of a child with particular disabilities.

A procedural violation does not constitute a denial of a FAPE if the violation fails to “result[ ] in a loss of educational opportunity!.]” M.L., 394 F.3d at 651 (Gould, J., concurring); Target Range, 960 F.2d at 1485 . A child ineligible for IDEA opportunities in the first instance cannot lose those opportunities merely because a procedural violation takes place. Cf. Nack ex rel. Nack v. Orange City Sch. Dist., 454 F.3d 604, 612 (6th Cir.2006) (procedural violation denies a FAPE “only if such violation causes substantive harm to the child or his parents” (internal quotation marks and citation omitted)). In other words, a procedural violation cannot qualify an otherwise ineligible student for IDEA relief. Therefore, the omission of a special education teacher or provider from R.B.’s IEP team is harmless if R.B. is ineligible for IDEA benefits. Because we affirm the district court’s acceptance of the SEHO’s determination that R.B. does not qualify for IDEA relief, we hold that the District’s procedural violation in the composition of R.B.’s IEP team is harmless error.

B. Substantive eligibility

1. Deference to the SEHO decision

Before we reach the merits of R.B.’s IDEA eligibility, we must address appellants’ arguments concerning the degree of deference we should give to the SEHO’s underlying decision. Appellants argue for no deference because they claim the SEHO omitted or distorted certain pieces of evidence. The District argues for the particular deference that we accord to “thorough and careful” findings. See Smith, 15 F.3d at 1524.

We treat a hearing officer’s findings as “thorough and careful” when the officer participates in the questioning of witnesses and writes a decision “containing] a complete factual background as well as a discrete analysis supporting the ultimate conclusions.” See Park, 464 F.3d at 1031 . Those criteria were satisfied here, as the SEHO asked follow-up questions of many witnesses, included several pages of factual background in the decision, and discretely analyzed all the issues presented for each of the two academic years in question. To this extent, the SEHO’s findings deserve particular deference.

Therefore, we can summarily dismiss most of appellants’ objections as impermissible attempts to second-guess the SEHO’s characterization and weighing of the evidence. We find no reason for according less deference to the SEHO’s decision be cause she described R.B.’s misconduct as “episodic” or labeled the transition to In-termountain a “difficult adjustment period.” Nor do we quibble with the SEHO’s citation to R.B.’s 2002 achievement test scores as evidence that R.B. performed “at or above grade level” (even though R.B.’s 2001 achievement test scores were even higher) or the failure to compare R.B.’s grades at Donaldson Way Elementary to her grades at Intermountain. We reject appellants’ assertion that the SEHO distorted the testimony of Joanna Gardner, the mother of one of R.B.’s friends: the SEHO’s conclusions that R.B. spent time with and participated in extracurricular activities with Gardner’s daughter were properly drawn from Gardner’s testimony. We also refuse to question the SEHO’s reliance on the Struven report, which included observations of R.B. in the classroom, rather than the Solomon report, which did not. The SEHO’s weighing of the evidence was consistent with the requirement that the IEP team review “[cjurrent classroom-based assessments and observations” and “[ojbservations by teachers and related service providers!.]” 34 C.F.R. § 300.533 (a)(ii)-(iii) (2003).

We are concerned, however, by the SEHO’s failure to make any reference in her decision to the testimony of Brandt or Morrison. The absence of discussion is particularly disturbing because the omission of R.B.’s special education teacher/ provider from the IEP team was the procedural violation in this case. The testimony of Brandt and Morrison at the due process hearing helped cure the procedural violation, but the SEHO decision’s failure to cite any of their testimony conveys the impression that the SEHO did not thoroughly and carefully consider their viewpoints. The district court’s decision is better: to support its conclusion that R.B. did not meet the criteria for IDEA eligibility at Intermountain, it cites Morrison’s testimony that RJB.’s behavior eventually improved during her year there. Again, however, the district court did not discuss Brandt’s testimony, other than mentioning in passing that plaintiffs relied on it.

Therefore, in our review of the SEHO’s decision, we accord particular'deference to the SEHO’s “thorough and careful” findings, except to the extent they do not discuss Brandt’s and Morrison’s testimony. In other words, we accord deference to the SEHO’s finding that R.B. was ineligible for IDEA relief in both school years, although we independently review the testimony in the record that the SEHO failed to consider. Cf. Katherine G. ex rel. Cynthia G. v. Kentfield Sch. Dist., 261 F.Supp.2d 1159, 1175 (N.D.Cal.2003) (deferring to the hearing officer’s “well-reasoned and well-supported” findings and conclusions while independently reviewing the testimony that the hearing officer failed to consider). Because IDEA eligibility determinations are fact-intensive, we review for clear error the district court’s acceptance of the SEHO’s decision. See Hood, 486 F.3d at 1104. In applying this standard of review, we are mindful that the district court discussed some of the testimony that the SEHO did not.

2. Eligibility criteria

An overarching purpose of the IDEA is to provide a FAPE to “children with disabilities.” 20 U.S.C. § 1400 (d)(1)(A); cf. § 1412(a)(1)(A) (pro viding FAPE to all children with disabilities is a condition for federal IDEA funding). The term “child with a disability” includes, inter alia, a child with a “serious emotional disturbance” “who, by reason thereof, needs special education and related services.” 20 U.S.C. § 1401 (3)(A). Under federal and California regulations, a “serious emotional disturbance” requires at least one of the following characteristics:

34 C.F.R. § 300.7 (c)(4) (2003); ef. CaLCode Regs. tit. 5 § 3030(i). The child must “exhibit! ]” the characteristic(s) “[1] over a long period of time and [2] to a marked degree [3] that adversely affects a child’s educational performance.” 34 C.F.R. § 300.7 (c)(4) (2003).

Each state determines the meaning of the language in the federal regulation. J.D. ex rel J.D. v. Pawlet Sch. Dist., 224 F.3d 60 , 66 (2d Cir.2000); Mr. I v. Me. Sch. Admin. Dist. 55, 416 F.Supp.2d 147, 157 (D.Me.2006). Rather than promulgate additional regulations, California relies on case-by-case administrative adjudication of IDEA eligibility. Decisions by the California State Educational Agency (“CSEA”) are persuasive authority for resolving the issues here — namely, appellants’ claims that R.B. was eligible for IDEA benefits because she could not build or maintain interpersonal relationships, behaved inappropriately under normal circumstances, and was pervasively depressed.

a. Interpersonal relationships

Relying on Dr. Solomon’s conclusions, appellants claim the SEHO and the district court disregarded “overwhelming” evidence of R.B.’s inability to build or maintain relationships. We disagree. The SEHO openly acknowledged the conflicting evidence regarding whether R.B. could maintain satisfactory relationships and cited testimony on both sides. The SEHO found that R.B. could maintain relationships and firmly grounded that conclusion in the record. Gardner testified about R.B.’s friendship with her daughter, and R.B.’s fifth-grade teacher and principal both testified that R.B. had friends.

The SEHO also found that R.B. developed satisfactory relationships with school personnel. This finding is important because the regulation considers relationships “with peers and teachers.” 34 C.F.R. § 300.7 (c)(4)(B) (2003); Cal.Code Regs. tit. 5 § 3030(f)(2) (emphasis added); cf. Fresno Unified Sch. Dist., 39 IDELR 28, at 7 (CA SEA Jan. 16, 2003) (explaining that the regulation sets “a high standard to meet as it requires [the student’s] difficulties to be both with peers and teachers ” (emphasis added)). The record fully supports the SEHO’s conclusion. For example, R.B. had such a good relationship with her third grade teacher that, as a fifth grader, R.B. occasionally visited the third grade classroom to tutor students in reading. R.B.’s fifth-grade teacher attended R.B.’s year-end music recital. The principal testified that she had a “great relationship” with R.B., who would read to the principal in her office and talk with her on the playground about future plans. Although Solomon opined that R.B. could not form satisfactory relationships with teachers, the SEHO reasonably discounted the weight of that testimony because Solomon did not observe R.B. at school or speak to school personnel. CSEA decisions regularly give little weight to the opinions of experts who do not consult school personnel. Ventura Unified Sch. Dist., 102 LRP 7625, at 17 (CA SEA Nov. 21, 2000); Williams Unified Sch. Dist. & Colusa County Office of Ed., 26 IDELR 1198, at 10 (CA SEA Aug. 21, 1997).

The testimony of Brandt and Morrison confirms that R.B. became able to form friendships with peers and teachers at In-termountain. Brandt testified that R.B. overcame initial hostility toward classmates and developed several peer friendships. Morrison likewise testified that R.B. began to describe one peer as a “best friend” and developed strong relationships with adult counselors.

Deferring to the SEHO’s findings, a preponderance of the evidence shows that R.B. was able to build and maintain satisfactory relationships with peers and teachers during her fifth grade year at Donaldson Way and her first year at In-termountain. Therefore, R.B. was not eligible for IDEA relief under this prong.

b. Inappropriate behavior

The inappropriateness of R.B.’s behavior during the 2001-02 and 2002-03 school years is manifest. As a fifth grader, R.B. was sent to the principal’s office for, e.g., pinching and twisting classmates’ arms on the playground on multiple occasions, tearing up classroom materials, verbalizing her hope that her music teacher would die, poking a classmate with a pencil because he would not help her cheat, and using the f-word. Morrison testified that R.B. physically attacked counselors at In-termountain and damaged property at least daily. R.B. would also attack younger children and throw food. R.B. admitted that she deliberately included grammatical errors in her written work because she enjoyed making life difficult for Brandt. When R.B. became a danger to herself or to others, Brandt would send R.B. to “day coverage”: this happened weekly during R.B.’s first four months in the classroom.

Nonetheless, the SEHO questioned whether R.B.’s inappropriate behavior took place under the requisite “normal circumstances” because R.B. was not regularly taking her ADHD medication for most of the fifth grade year. (SEHO Decision, at 9.) We similarly question whether R.B.’s first months at Intermountain were “normal circumstances” because R.B. was adapting to life at a new school away from her family.

Appellants contend the SEHO and district court were blind to R.B.’s long history of behavioral problems. The misconduct that prompted the behavioral support plan during the fifth grade year was allegedly just a continuation of conduct that began in previous years and persisted through the remainder of R.B’s fifth grade year and at Intermountain. Appellants make much of the behavioral support plan’s passing reference to R.B.’s “habitual history of this type of resistant behavior,” concluding that this language alone establishes the District’s knowledge that R.B.’s inappropriate behavior has continued over a long period of time.

Even if these circumstances were “normal” within the meaning of the regulation, a preponderance of the evidence establishes that R.B.’s inappropriate behavior was not to a marked degree over a long period of time. Appellants misconstrue the import of the behavioral support plan that Donaldson Way Elementary implemented during the second trimester of R.B.’s fifth grade year. The whole point of the plan was that R.B.’s “habitual history” of “isolated incidents” of misconduct reached acute levels during that trimester. Once the District implemented the support plan, R.B.’s behavior improved. In other words, while R.B. engaged in inappropriate behavior over several years of school, that behavior was “to a marked degree” only during one trimester of one grade. We accord particular deference to the SEHO’s thorough and careful finding that R.B.’s behavior was not “pervasive and ongoing,” and conclude that R.B. cannot establish IDEA eligibility on the basis of inappropriate behavior during her fifth-grade year.

For the same reasons, R.B. likewise cannot establish IDEA eligibility based on her inappropriate behavior at Intermountain. In a passage cited by the district court, Morrison testified that R.B. began to develop sympathy for others and react to anxious situations without becoming violent. Similarly, Brandt saw “a big improvement” in R.B.’s classroom attitude by April 2003 and began to allow R.B. to participate in class field trips.

R.B.’s inappropriate behavior further does not amount to a “severe emotional disturbance” because it did not adversely affect her educational performance. California primarily gauges educational performance through academic measures. See Fresno Unified Sch. Dist., 39 IDELR 28, at 3, 5, 13 (CA SEA Jan. 16, 2003); Ventura Unified Sch. Dist., 102 LRP 7625, at 23-25 (CA SEA Nov. 21, 2000). Here, all grades on R.B.’s fifth-grade report card were “4” or “5,” indicating work at or above grade level. During the year at Intermountain, a majority of her grades were “A” or “B,” with only one “D.” R.B.’s achievement test scores were similarly average or better. Even the 2002 spelling score on which appellants so heavily rely placed R.B. in the 57th percentile, above more than half of her peers. A preponderance of the evidence shows that any of R.B.’s exhibited characteristics did not adversely affect her educational performance.

Appellants claim the SEHO erred by using R.B.’s grades as a “litmus test” instead of considering that R.B.’s educational performance was below her ability. They rely on out-of-circuit cases requiring analysis of the individual student’s “potential” to determine IDEA eligibility. Our rule, however, is that the IDEA does not guarantee “the absolutely best or ‘potential-maximizing’ education for the individual child.” Gregory K., 811 F.2d at 1314 (quoting Rowley, 458 U.S. at 197 n. 21, 102 S.Ct. 3034 ); accord Drew P. v. Clarke County Sch. Dist., 877 F.2d 927, 930 (11th Cir.1989). While the Supreme Court has cautioned that merely advancing from grade to grade does not satisfy the IDEA, Rowley, 458 U.S. at 203 n. 25, 102 S.Ct. 3034 , it also limited the IDEA’S guarantees to “the basic floor of opportunity”, id. at 201 , 102 S.Ct. 3034 .

Appellants cite the District’s development of Rehabilitation Act § 504 plans and behavioral support plans as further evidence that R.B.’s behavioral problems adversely affected her educational performance. The Rehabilitation Act is, however, a separate statutory scheme with different qualifying criteria, and R.B.’s satisfaction of those criteria do not automatically make her eligible under the IDEA. Muller ex rel. Muller v. Comm. on Special Ed. of the East Islip Union Free Sch. Dist., 145 F.3d 95 , 99 n. 2 (2d Cir.1998); Yankton Sch. Dist. v. Schramm, 93 F.3d 1369, 1376 (8th Cir.1996). Furthermore, California school districts commonly turn to behavioral support plans as alternative remedies for students who do not satisfy the IDEA’S criteria for a “severe emotional disturbance.” See Ventura Unified Sch. Dist., 102 LRP 7625, at 5, 10 (CA SEA Nov. 21, 2000).

In summary, by a preponderance of the evidence and with deference to the SEHO’s thorough and careful findings, we conclude that R.B.’s inappropriate behavior was not to a marked degree over a long period of time and did not adversely affect her educational performance. Therefore, R.B. was not eligible for IDEA relief under this prong.

c. Pervasive unhappiness or depression

Acknowledging a long history of depression diagnoses, the SEHO nonetheless concluded that R.B. was not depressed during her fifth grade year because school personnel testified that R.B. generally seemed happy. The SEHO also relied on the District’s failure to receive any documentation of depression diagnosis (including the Solomon report) until R.B. finished the fifth grade. However, licensed professionals who examined R.B. during the 2001-03 time frame—including Solomon, Morrison, and Struven—all diagnosed R.B. with depression.

Even if we accepted the opinion of the licensed professionals and rejected the SEHO’s finding that R.B. was not depressed during the 2001-02 and 2002-03 school years, appellants would still fail to establish IDEA eligibility because they could not prove that R.B.’s depression was “to a marked degree.” The Struven report, which the SEHO found most persuasive, concluded that R.B. only had mild depression below the level required to establish a “severe emotional disturbance.” The SEHO heard the testimony of all the experts and was familiar with the various methodologies employed in each report. Therefore, we defer to the SEHO’s thorough and careful analysis of the expert reports and accept the conclusion that R.B.’s depression was not to a marked degree.

Finally, for all the reasons stated in Part 111(B)(2)(b) of this opinion, R.B.’s depression did not adversely affect her educational performance. Therefore, R.B. was not eligible for IDEA relief under this prong.

IV.

CONCLUSION

The District violated the procedural requirements of the IDEA by not including a special education teacher or provider of the child on the IEP team. After reviewing the record and giving proper deference to the SEHO’s thorough and careful findings, we hold that R.B. did not qualify as a “child with a disability” because she did not meet any of the criteria for a “severe emotional disturbance.” Because R.B. is substantively ineligible for IDEA relief, we hold that the procedural error in the composition of her IEP team was harmless.

AFFIRMED.

. Appellants argue for de novo review. However, as we recently stated in another case concerning IDEA eligibility, "the fact-intensive nature of a special education eligibility determination coupled with considerations of judicial economy render a more deferential approach appropriate.” Hood v. Encinitas Union Sch. Dist., 486 F.3d 1099 , 1104 n. 4 (9th Cir.2007). . The IDEA was subsequently amended by the Individuals with Disabilities Education Improvement Act of 2004, Pub.L. 108-446, 118 Stat. 2647, which took effect on July 1, 2005. Here, we apply the statute and regulations in effect at the time of the events in question. See Adams v. Oregon, 195 F.3d 1141 , 1148 n. 2 (9th Cir. 1999). . Appellants do not claim that the alleged procedural violations infringed F.B.’s opportunity to participate in the IEP formation process. The sole question is whether the procedural violations resulted in a loss of educational opportunity for R.B. . Although each member of the M.L. panel wrote separately, that case did not alter our standard for reviewing procedural errors in IDEA cases. Two members of the panel analyzed whether the procedural violation resulted in a lost educational opportunity. M.L., 394 F.3d at 652 (Gould, J., concurring), 658 (Clifton, J., dissenting). As the narrower opinion joining in the judgment for the M.L. appellants, Judge Gould's concurrence is the “controlling opinion”. See Center for Fair Public Policy v. Maricopa County, 336 F.3d 1153, 1161 (9th Cir.2003) (citing Marks v. United States, 430 U.S. 188 , 193, 97 S.Ct. 990 , 51 L.Ed.2d 260 (1976)). Judge Gould's concurrence merely clarifies that, where a procedural violation does not result in a lost educational opportunity for the student, the violation is “harmless error” because it does not deny the student a FAPE. M.L., 394 F.3d at 651-52 (Gould, J., concurring). . See 394 F.3d at 649 (Alarcon, J.) ("Indeed, any regular education teacher would have contributed his or her knowledge of the ability of a disabled student to benefit from being placed in a regular classroom.”), 657 n. 12 (Gould, J., concurring) ("The IDEA mandates that a regular classroom teacher be a member of the IEP team.") . We also reject appellants’ argument that Sparks could not have been a teacher "responsible for implementing a portion of the IEP” (if R.B. had been entitled to an IEP). When the IEP team met, Sparks was principal of Donaldson Way Elementary and, thus, was involved in the education of all students in the school. Appellants fail to carry their burden of showing that Sparks would have no responsibility for implementing the IEP by merely pointing out that R.B. would soon enroll in middle school. See Schaffer ex rel. Schaffer v. Weast, 546 U.S. 49 , 126 S.Ct. 528, 535-37 , 163 L.Ed.2d 387 (2005) (party seeking IDEA relief bears burden of persuasion). . Appellants also argue that Morrison should have been included on the IEP team, presumably as a "special education provider of such child.” Our analysis of whether the IEP team should have included Brandt applies with equal force to Morrison. . Under California law, the IEP team must include "[a]t least one special education teacher of the pupil, or, if appropriate, at least one special education provider of the pupil.” Cal. Educ.Code § 56341(b)(3). Therefore, even if we did not interpret the IDEA to require the participation of a special education teacher who has actually taught the child, the composition of R.B.’s IEP team would still have been procedural error under California law, and we would still proceed to the harmless-error analysis infra. . The SEHO supported her conclusion that R.B. was performing at grade level academically with the results of an achievement test that Struven administered at Intermoun-tain. The SEHO’s reliance on achievement test scores rather than grades to measure R.B.’s academic progress was a reasonable choice because of the very different grading systems used by Donaldson Way and Inter-mountain.