dissenting:
The majority holds that a teacher’s speech in selecting, producing, and directing a school play deserves “no First Amendment protection.” Ante at 369.1 cannot agree and therefore respectfully dissent.. In my judgment, the district court erred in dismissing Margaret Boring’s complaint for failure to state a claim upon which relief can be granted.
School administrators must and do have final authority over curriculum decisions. But that authority is not wholly unfettered. Like all other state officials, they must obey the Constitution. The Supreme Court' has long recognized that the Constitution, specifically the First Amendment, “does not tolerate laws that cast a pall of orthodoxy over the classroom.” Keyishian v. Board of Regents, 385 U.S. 589, 603, 87 S.Ct. 675, 683, 17 L.Ed.2d 629 (1967). See also Tinker v. Des Moines Indep. Sch. Dist., 393 U.S. 503, 506, 89 S.Ct. 733, 736, 21 L.Ed.2d 731 (1969) (“teachers” no less than “students” do not “shed their constitutional rights to freedom of speech or expression at the schoolhouse gate”). Thus, teachers’ in-class speech retains some, albeit limited, First Amendment protection, as is explained in detail in the panel opinion in this case. See Boring v. Buncombe County Bd. of Educ., 98 F.3d 1474, 1475-85 (4th Cir.1996). To that opinion, I add only a few thoughts.
I.
First, although the majority notes the correct standard of review, ante at 367 (“[w]e review a dismissal for failure to state a claim de novo, drawing all reasonable inferences in favor of the plaintiff and accepting the allegations that- are stated in the complaint as true”), it fails to apply this standard.
Examination of Margaret Boring’s complaint is instructive. Boring alleges that she is a “tenured teacher” in the Buncombe County public schools, that she has “built a national reputation for excellence in teaching drama and directing and producing theater,” and that “[h]er plays have won numerous awards and, each year, many of her students have gone on to college on theater-related scholarships.” She further alleges that in the fall of 1991 — twelve years after her initial employment by the Board, after her plays had been “regularly” entered in competition and won “numerous awards”- — she notified her principal “as she did every year ” of the name of the play' she had chosen for the annual competition, and the principal “did not comment or react.” (Emphasis added.)
True, as the: majority observes, Boring “does not allege that she gave the principal any information about the play other than the name.” Ante at 366. But a fair reading of the above allegations, let alone a reading that gives Boring “all reasonable inferences” from them, reveals that Boring provided her principal with precisely the same advance notice of the chosen play that she had in every previous year — notice that, until 1991, had been sufficient. According to Boring, the principal did not question her choice, ask for further information, or in any way “comment or react” to her choice. It may be that Boring pulled a fast one on the principal, choosing a more controversial play than in previous years and giving him just its name to preclude negative reaction from him. Nothing in the complaint supports this inference, however, and the School Board has not so asserted.
Boring further alleges that no violation of the “controversial materials policy” provided a basis for her transfer and that the play was performed in the state competition only after the principal insisted that she delete certain portions of the play. Taking all of these allegations together, a fair reading of them is that Boring complied with the school admin*376istration’s requirements and policies in every respect, but was nonetheless disciplined “to punish and retaliate against her for expressing an unpopular point of view through the production of the play” in violation of her First Amendment rights. Thus, as the district court recognized, Boring “does not ask the Court to find that a unilateral selection and production of the play ‘Independence,’ without prior approval, would have been protected First Amendment speech”; rather, she “argues that having passed on the play prior to its production and performance, the school does not have a right to discipline [her] in retaliation for its use in the curriculum.” It is this complaint that Boring presents to us, and this complaint that we must assess to determine whether it states a claim upon which relief can be granted.
The Board may indeed have “legitimate pedagogical concerns” that are “reasonably related” to its disciplinary decision. See Hazelwood Sch. Dist. v. Kuhlmeier, 484 U.S. 260, 273, 108 S.Ct. 562, 571, 98 L.Ed.2d 592 (1988). But, of course, Boring alleges no such concerns and the Board has not yet stated any. Hence, nothing in the record before us, at this early stage in the proceedings, allows us to draw such a conclusion. Prior to today, every court to consider the matter has required that school administrators offer some evidence — if only an affidavit — to establish the legitimacy of the pedagogical concerns purportedly related to their actions. See Boring, 98 F.3d at 1479. The majority, however, concludes that even this slight evidentiary showing is unnecessary.1 The majority maintains that because “pedagogical” is defined as “educational,” any and every curriculum decision made by school administrators is “by definition a legitimate pedagogical concern” and thus constitutionally acceptable. Ante at 370.
The Supreme Court’s careful reasoning in Hazelwood, an opinion authored by Justice White and joined by all members of the present Court then sitting (the Chief Justice, and Justices Stevens, O’Connor, and Sealia), offers no support for this astonishing conclusion.2 Rather, in Hazelwood the Court held that school administrators’ curriculum choices did not offend the First Amendment “so long as their actions are reasonably related to legitimate pedagogical concerns.” Hazelwood, 484 U.S. at 273, 108 S.Ct. at 571 (emphasis added). Indeed, the Court went on to recognize that, on occasion, a particular curriculum decision may have “no valid educational purpose” and that in such an instance “the First Amendment is so directly and sharply implieate[d] as to require judicial intervention.” Id. (citation and internal quotation omitted; alteration in original). Thus, the Supreme Court in Hazelwood clearly did not hold, as the majority does here, that each and every curriculum decision is “by definition a legitimate pedagogical concern.” Ante at 370 (emphasis added). Instead, the Court meticulously analyzed the speech before it and concluded that the school administrators had demonstrated — -through the testimony of several witnesses — the legitimacy of their pedagogical concerns and that for this reason “no violation of First Amendment rights occurred.” Id. at 275-76, 108 S.Ct. at 572.
Nor do the two cases upon which the majority relies, ante at 370, support its holding that each and every curriculum decision of a school administration is “by definition a legitimate pedagogical concern.” In neither Kirkland v. Northside Indep. Sch. Dist., 890 F.2d 794 (5th Cir.1989); cert. denied, 496 U.S. 926, 110 S.Ct. 2620, 110 L.Ed.2d 641 (1990), nor Searcey v. Harris, 888 F.2d 1314 (11th Cir.1989), did the courts hold that the plaintiffs failed to state a claim upon which *377relief could be granted or that school administrators’ decisions were motivated by legitimate pedagogical concerns simply because those decisions concerned the curriculum.
In Kirkland, the Fifth Circuit did conclude that the teacher “suffered no impairment of his First Amendment rights.” Kirkland, 890 F.2d at 795. But that teacher, Timothy Kirkland, unlike Boring, admitted that he refused to follow the school’s well-established rules. For example, he admitted using a “nonapproved reading list.” Kirkland, 890 F.2d at 795 (emphasis added). Boring, by contrast, alleges that her principal initially acquiesced in her choice and production of Independence. Moreover, Kirkland did not concede, as Boring does, that the school authorities were entitled to the broad discretion vested in them under the Hazelwood standard. Rather, Kirkland contended that “his control -of the world history class curriculum [was] unlimited.” Kirkland, 890 F.2d at 801 (emphasis added). The Kirkland court properly rejected this argument. Id. But the Fifth Circuit’s reasoning in Kirkland does not foreclose Boring’s quite different and far more modest contention that although administrators may discipline a teacher even when the teacher does follow the school’s rules, they may do so only “so long as [administrators’] actions are reasonably related to legitimate pedagogical concerns.” Hazelwood, 484 U.S. at 273, 108 S.Ct. at 571. Actually, rather than foreclosing this reasoning, the Kirkland court seemed to embrace it: “Our decision should not be misconstrued ... to suggest that public school teachers foster free debate in their classrooms only at their own risk or that their classrooms must be ‘east "with a pall of orthodoxy.’ ” Kirkland, 890 F.2d at 801-02.
In Searcey, the Eleventh Circuit recognized, as I do, that curricular programs by nature have pedagogical purposes. See 888 F.2d at 1319. But it did not hold, as the majority does here, that each and every curricular decision is “by definition a legitimate pedagogical concern.” Ante at 370 (emphasis added). In fact, the Eleventh Circuit’s holding stands in stark contrast to that set forth by the majority. The Searcey court upheld a judgment against a school board precisely because the board offered “no evidence” to support its challenged requirement. Searcey, 888 F.2d at 1322 (emphasis added). The court reasoned “[w]e cannot infer the reasonableness of a regulation from a vacant record.” Id. (citing Hazelwood, 484 U.S. at 275 & n. 8, 108 S.Ct. at 572 & n. 8). Moreover, Searcey expressly rejected the school board’s argument that even though it failed to offer any evidentiary support, a court must defer to its decision; the court concluded that this would “overstate[ ] the deference a court must pay to School Board decisions.” Id. at 1321. The majority erroneously relies on Searcey to do precisely what the Searcey court itself would not do — overstate the deference due school board decisions and infer the reasonableness of such decisions from a vacant record.
The Buncombe County Board of Education may possess legitimate pedagogical concerns reasonably related to its discipline of Boring. But, to date, the Board has not even attempted to state those concerns, let alone offered a scintilla of evidence establishing them. On this record, I do not see how a court can conclude, as the majority does, that “the school administrative authorities had a legitimate pedagogical interest” justifying discipline of Boring and dismissal of her complaint. Ante at 370.
II.
Like the district court, Boring, and the two associations that filed amici briefs on behalf of the School Board (the National School Boards Association and the Virginia School Board Association Council-, of School Board Attorneys), I believe that the standard articulated in Hazelwood, not that set forth in Connick v. Myers, 461 U.S. 138, 103 S.Ct. 1684, 75 L.Ed.2d 708 (1983), provides the appropriate test for analyzing the speech at issue in this case. But, contrary to the majority’s suggestion, even if Connick were applicable here, it would fail to provide an alternative basis on which to dismiss Boring’s complaint.
A.
Connick held that a court generally determines whether a public employee’s speech is *378constitutionally protected by balancing the employee’s interest, “ ‘as a citizen, in commenting upon matters of public concern and the interest of the State, as an employer, in promoting the efficiency of the public services it performs through its employees.’” Id. at 142, 103 S.Ct. at 1687 (quoting Pickering v. Board of Educ., 391 U.S. 563, 568, 88 S.Ct. 1731, 1734-1735, 20 L.Ed.2d 811 (1968)). Under Connick, Boring must first establish that her speech related to “a matter of public concern.” See Connick, 461 U.S. at 146, 103 S.Ct. at 1689. If she does so, school authorities may still prevail, provided they can show that their interest in workplace efficiency and harmony outweighs Boring’s interest in commenting upon the matters of public concern. See Board of County Comm’rs v. Umbehr, 518 U.S. 668,-- -, 116 S.Ct. 2342, 2347-2348, 135 L.Ed.2d 843 (1996); Connick, 461 U.S. at 146-147, 103 S.Ct. at 1689-1690; Pickering, 391 U.S. at 568, 88 S.Ct. at 1734-1735.
In my view, the Connick framework does not provide a workable formula for analyzing whether the First Amendment protects a teacher’s in-class speech. Neither element of the Connick balancing test provides much assistance in assessing whether this speech is entitled to constitutional protection.
The public concern element articulated in Connick fails to account adequately for the unique character of a teacher’s in-class speech.
When a teacher steps into the classroom she assumes a position of extraordinary public trust and confidence: she is charged with educating our youth. Her speech is neither ordinary employee workplace speech nor common public debate. Any attempt to force it into either of these categories ignores the essence of teaching — to educate, to enlighten, to inspire — and the importance of free speech to this most critical endeavor. As the Supreme Court proclaimed more than forty years ago: “Teachers and students must always remain free to inquire, to study and to evaluate, to gain new maturity and understanding; otherwise our civilization will stagnate and die.” Sweezy v. New Hampshire, 354 U.S. 234, 250, 77 S.Ct. 1203, 1212, 1 L.Ed.2d 1311 (1957).
Moreover, the governmental interest element as set forth in Connick fails to give school administrators the necessary and appropriate control over a teacher’s in-class speech. School administrators should be free to specify curriculum and to curtail classroom speech for any legitimate pedagogical reason. They should not be required to demonstrate that a restriction on in-class speech is necessitated by workplace efficiency or harmony. Accordingly, as more fully set forth in the panel opinion, I believe the simpler and more rigorous Hazelwood analysis should apply to a teacher’s in-class speech, as well as a student’s in-class speech. See Boring, 98 F.3d at 1482-83.
B.
However, even if the Connick test applied to a teacher’s in-class speech, we would be required to conclude that the district court erred in dismissing Boring’s complaint. Although Boring’s in-class speech does not itself constitute pure public debate, obviously it does “relate to” matters of overwhelming public concern — family life, divorce, motherhood, and illegitimacy, among others. Thus, if the Connick analysis did apply to in-class speech, then Boring’s choice and production of a play that raises a number of important social issues obviously falls within the Supreme Court’s broad definition of “public concern,” which includes speech “relating to any matter of political, social, or other concern to the community.” Connick, 461 U.S. at 146, 103 S.Ct. at 1690 (emphasis added). See also id. at 149, 103 S.Ct. at 1691 (speech need only “touch” on a matter of public concern in order to trigger an employer’s burden to justify restrictions on it).
I have trouble understanding the basis for the majority’s contrary holding that Boring’s selection and production of the play amounts to “nothing more than an ordinary employment dispute” and does not involve a matter of public concern under Connick. Ante at 369. (The majority apparently does not contest one portion of the panel’s (now vacated) holding, see Boring, 98 F.3d at 1478-1479, i.e., that Boring’s selection and production of the play constitute speech for purposes of the *379First Amendment. Thus, the majority notes, “[tjhis case concerns itself exclusively with employee speech.” See ante at 371 n. 2 (emphasis added)).
Conceivably, the majority’s holding is grounded in misreading Connick to make the role in which a public employee speaks determinative of whether her speech merits First Amendment protection. Connick does distinguish between an employee speaking “as a citizen on matters of public concern” and an employee speaking “as an employee upon matters of personal interest.” Connick, 461 U.S. at 147, 103 S.Ct. at 1690. But Connick never holds that a public employee automatically loses all First Amendment protection whenever she speaks in her role as employee on a matter of public concern. Indeed, the Connick Court implicitly repudiates such a conclusion, by directing that factors other than the role of the speaker are critical to determining when an employee speaks on a matter of public concern — “[wjhether an employee’s speech addresses a matter of public concern must be determined by the content, form, and context of a given statement.” Id. at 147-148, 103 S.Ct. at 1690. Moreover, Connick itself held that one of the statements of the government employee before it, although in the same form and context as her other statements (a questionnaire circulated by the employee only within her office) and although quintessentially speech by an employee in her role as an employee, did “touch[ ] upon a matter of public concern.” Id. at 149, 103 S.Ct. at 1691.
Furthermore, to read Connick as holding that any speech by a public employee in her role as an employee fails to merit First Amendment protection would contravene other controlling precedent. Both the Supreme Court and this court have held that a teacher’s in-class speech, which by definition involves speech by a teacher in her role as an employee, warrants protection. See, e.g., Keyishian, 385 U.S. at 602-603, 87 S.Ct. at 683-684 (regulations inter alia restricting teachers’ in-class speech held to violate the First Amendment); Piver v. Pender County Bd. of Educ., 835 F.2d 1076, 1081 (4th Cir.1987), cert. denied, 487 U.S. 1206, 108 S.Ct. 2847, 101 L.Ed.2d 885 (1988). Because the majority does not explicitly hold that the role in which an employee speaks is determinative or attempt to distinguish Keyishian or overrule Piver, this reasoning must not be the basis for its conclusion that Boring’s speech does not relate to a matter of public concern.
The only other possible basis that I can see for the majority’s holding is a mistake as to the nature of Boring’s claim. The speech for which Boring seeks First Amendment protection does not constitute a private personnel grievance. Boring does not allege that she selected and produced Independence after being instructed not to choose that play, nor does she allege that school administrators disciplined her because she publicized her dissatisfaction with théir treatment of her. Such a claim, much like Kirkland’s insistence on using a disapproved reading list, would constitute an “ordinary employment dispute” rather than speech relating to a matter of public concern. Boring instead alleges that the school administration disciplined her for the selection and production of the play itself. This is not an employment dispute. Rather, it is a challenge to a restriction on classroom speech, which involves matters of public' concern.
Thus, if the Connick analysis applied here, a court could only conclude that Boring has alleged that she was disciplined for speech “relating to a matter of public concern.” We would then have to proceed to the other half of the Connick balancing test and examine the school administration’s evidence as to the necessity of its restriction of her speech. But, on this record, we could hardly conclude that school administrators had demonstrated that they reasonably believed Boring’s speech would disrupt the workplace. They have not even asserted, let alone offered evidence of, disruption. Cf. Connick, 461 U.S. at 151-154, 103 S.Ct. at 1692-1694. (Supreme Court relies on trial testimony of employer that employee’s speech caused office disruption). Accordingly, even under Connick, we would be required to hold that the district court erred in dismissing Boring’s complaint.
III.
As recognized at the outset of this dissent and in the panel opinion, school administra*380tors must “be permitted to have the final say in setting the appropriate curriculum so that students are not exposed to material that detracts from or impedes the school’s pedagogical mission.” Boring, 98 F.3d at 1483. Yet, the First Amendment lives in the classroom as it does elsewhere. Indeed, as the Supreme Court stated several decades ago:
The vigilant protection of constitutional freedoms is nowhere more vital than in the community of American schools. By limiting the power of the States to interfere with freedom of speech and freedom of inquiry and freedom of association, the Fourteenth Amendment protects all persons, no matter what their calling, But, in view of the nature of the teacher’s relation to the effective exercise of the rights which are safeguarded by the Bill of Rights and by the Fourteenth Amendment, inhibition of freedom of thought, and of action upon thought, in the ease of teachers brings the safeguards of those amendments vividly into operation.
Shelton v. Tucker, 364 U.S. 479, 487, 81 S.Ct. 247, 251, 5 L.Ed.2d 231 (1960) (internal quotation omitted). Justice Stewart wrote these words in the course of holding that the First Amendment prevented public schools from compelling teachers to list all organizations to which they had belonged or contributed in the recent past. But the words apply with equal force here. Rather than “vigilant[ly] proteeti[ng] ... constitutional freedoms ... in the community of American schools,” the majority eliminates all constitutional protection for the in-class speech of teachers. By holding that public school administrators can constitutionally discipline a teacher for in-class speech without demonstrating, or even articulating, some legitimate pedagogical concern related to that discipline, the majority extinguishes First Amendment rights in an arena where the Supreme Court has directed they should be brought “vividly into operation.” For these reasons, I must respectfully dissent.
Judges Hall, Murnaghan, Ervin, Hamilton, and Michael join in this dissent.
. Requiring school authorities to submit a single affidavit demonstrating the factual basis for their pedagogical concerns hardly imposes a "crushing burden” on them. Ante at 372 (Luttig, J. concurring).
. Of course, as noted in the panel opinion, "Ha-zelwood directly addresses the free speech rights of students, not teachers.” . Boring, 98 F.3d at 1482. Hazelwood, however, is the Supreme Court's most recent discussion of the protection due curricular speech and the majority and concurrences apparently believe that it is the curricular nature of Boring's speech that deprives it of all First Amendment protection. For this and the other reasons discussed in the panel opinion, I continue to believe that the Hazelwood analysis provides the "best means” of evaluation of the speech at issue here. Id.