Plaintiff-appellant Edward Donnelly, a high school teacher, sued defendants-appellees — the Greenburgh Central School District No. 7, his former employer, and three of his supervisors there (collectively “the District,” unless otherwise noted) — in the United States District Court for the Southern District of New York (Lewis A. Kaplan, Judge). Donnelly alleged inter alia that the District denied him tenure in retaliation for his having taken leave pursuant to the Family Medical Leave Act (“FMLA”), 29 U.S.C. §§ 2601-54.1 The District moved for summary judgment, arguing that (1) Donnelly was not eligible for FMLA leave because, as calculated under his union’s Collective Bargaining Agreement (“CBA”), he had worked only 1,247 hours — three fewer than the statutory minimum — in the preceding year, and (2) even if he were eligible, he could not show that the District acted unlawfully in denying him tenure. The magistrate judge — to whom the district court referred the case for pretrial matters — agreed, concluding that (1) Donnelly’s hours were appropriately calculated according to the provisions of the CBA, and as so calculated were insufficient to make Donnelly eligible for FMLA leave; and (2) even if he were eligible, he had not made the necessary showing that he was qualified for tenure under the standard we have applied in the *137context of university tenure denials. See Zahorik v. Cornell Univ., 729 F.2d 85, 93-94 (2d Cir.1984). The district court agreed with the magistrate judge’s conclusion that Zahorik was fatal to Donnelly’s claims, and thus found it unnecessary to address his FMLA eligibility. Donnelly v. Greenburgh Cent. Sch. Dist. No. 7, No. 08 Civ. 11031, 2011 WL 1899713 at *1 (S.D.N.Y. May 12, 2011).
On appeal, the parties raise the same arguments. Donnelly challenges Zahorik’s applicability to high school tenure decisions, and the District argues that Donnelly is not eligible for FMLA leave, and that even if he were he has failed to show that the District’s decision to deny him tenure reflected unlawful retaliation, whether or not Zahorik applies to this case. We conclude that (1) Donnelly has presented a genuine issue of material fact as to whether he qualifies for FMLA leave; (2) the standard articulated in Zahorik does not apply outside of the college or university context for which it was designed; and (3) Donnelly has presented sufficient evidence to permit a reasonable jury to find unlawful retaliation, and thus to defeat the District’s motion for summary judgment. We therefore reverse the district court’s grant of summary judgment and remand the case for trial.
BACKGROUND
I. Facts
As Donnelly is the non-moving party, we review the facts in the light most favorable to him and draw all reasonable inferences in his favor. See Terry v. Ashcroft, 336 F.3d 128, 137 (2d Cir.2003).
A. Donnelly’s Teaching Performance, 2004-2006
The District hired Donnelly on July 1, 2004, under a three-year probationary contract, as a secondary-school English teacher. When hired, he had five years’ teaching experience. During his first year in the District, Donnelly worked at Westchester Magnet Academy (“WMA”). At WMA, Donnelly was evaluated three times, receiving the highest overall rating in each review. He had perfect attendance during his year at WMA.
WMA closed at the end of Donnelly’s first year. The District thereafter transferred him to Woodlands High School for the final two years of his probationary contract. During his second year Donnelly was again evaluated three times. In one evaluation, defendant-appellee Robert Chakar, the principal at Woodlands, gave Donnelly the highest possible rating in twenty-four of the twenty-six categories. In his other two evaluations, Donnelly received the highest possible rating in every category. Each of Donnelly’s evaluations also recommended areas of improvement, as is standard practice in such evaluations, including those for tenured teachers.
Donnelly’s classroom performance included two episodes that required specific admonition from his supervisors. On one occasion, Donnelly told a student she was “acting retarded” and then wrote the word “retard” on the board. On another, he told a student to “go back to Mexico.” Chakar counseled Donnelly about both incidents, verbally and in writing.
B. Donnelly’s Medical Leave and FMLA Eligibility
In the fall of Donnelly’s third and final year of his probationary contract — during which he would be evaluated for tenure and offered or not a permanent teaching position within the District — Donnelly became ill and required gallbladder surgery. Donnelly’s surgery occurred on November *13827, 2006; he took leave from then through December 5, 2006.2
To be eligible for FMLA leave, an employee must work “at least 1,250 hours of service ... during the previous 12-month period.” 29 U.S.C. § 2611(2)(A)(ii). The parties agree that the relevant date for determining FMLA eligibility is the date of Donnelly’s surgery, November 27, 2006. During the twelve months prior to that date, Donnelly worked 172 of the expected 189 days of the school year.
Under the CBA between the teachers’ union — of which Donnelly was a member— and the District, “[t]he regular working day for all K-12 teachers shall be the equivalent of up to one (1) hour in excess of the pupil’s regular school day ... but in no case ... shall the working day exceed seven (7) hours and fifteen (15) minutes.” The CBA also acknowledges the District’s and the Union’s
recognition] that teachers have responsibilities which they readily and willingly perform that extend beyond the pupil’s regular school day. Among these responsibilities are classroom preparation, correction of papers, clerical work, record keeping, tutoring, parent-teacher and student-teacher conferences, staff meetings, curriculum planning and development, and in-service training.... To the extent practicable, the time between the end of the pupil’s regular school day and the close of the working day shall be devoted to the said responsibilities.
The parties agree that, as per the terms of the CBA, Donnelly worked at least 1,247 hours — that is, seven hours and fifteen minutes per day for 172 days — during the twelve-month period prior to his leave. That total falls three hours short of the statutory FMLA eligibility requirement.
Donnelly asserted in a declaration in opposition to the District’s motion for summary judgment that he and “most teachers regularly work in excess of a total of one hour before and after class,” and that he “typically worked a total of 1.5 hours before and after class every day.” In an evaluation dated June 6, 2006, and thus within the relevant FMLA eligibility period, Chakar commented that Donnelly “arrives to work in a professional manner; early, on time and often. He often stays late into the afternoon working with his kids to ensure their success.” The record provides no further information regarding any hours Donnelly may have worked beyond the maximum specified in the CBA, nor does it clarify what, exactly, Donnelly did during any such additional hours.
C. Donnelly’s Post-Leave Evaluations
Shortly after Donnelly returned from leave on December 15, 2006, Chakar conducted an in-class evaluation. Chakar concluded that Donnelly’s teaching was “Satisfactory” — the highest rating3 — but gave *139the lower rating of “Needs Improvement” for Professional Responsibility due in part to the absences Donnelly had accumulated during his leave. As with previous evaluations, Chakar suggested ways in which Donnelly could improve his performance.
Donnelly was evaluated again on January 8, 2007, this time by defendant-appellee Will Washington, an assistant principal at Woodlands. For the first time during his two-and-a-half years as a teacher in the District, Donnelly received an overall rating of “Unsatisfactory,” the lowest rating possible. Among other significant criticisms of Donnelly’s pedagogy and classroom management, Washington criticized Donnelly’s excessive absences, including those taken during Donnelly’s medical leave. Donnelly contested the results of his evaluation and complained that he was being mistreated for taking leave for surgery.
On January 12, 2007, Chakar criticized Donnelly orally and in writing for his absences, including those taken during his medical leave. Because of concerns regarding teacher absenteeism, every teacher evaluation conducted during 2006-2007 included reference to that teacher’s number of absences, irrespective of the reason for those absences. During that same period, the District terminated the employment of a tenured teacher for excessive absenteeism.
On March 20, 2007, Chakar evaluated Donnelly again. According to Donnelly’s recollection, which Chakar does not dispute, immediately after the evaluated lesson Chakar praised Donnelly’s performance and offered no suggested improvements. Nevertheless, Chakar’s written evaluation gave Donnelly an overall rating of “Needs Improvement.” Chakar issued this rating notwithstanding his assessment that Donnelly was “Satisfactory” on three of the five evaluation categories: Classroom Management, Knowledge of Content, and Demonstration of Effective Teaching Strategies. Among other critiques, Chakar again noted Donnelly’s excessive absences, including those that occurred during his medical leave.
D. The District’s Tenure Decision
Under the District’s tenure evaluation process, a teacher is evaluated throughout the three-year probationary period. Near the conclusion of that period, the teacher’s head principal — in Donnelly’s case, Chakar — reviews all the evaluations, and speaks to the teacher’s colleagues and supervisors. The principal then makes a recommendation to the assistant superintendent as to whether tenure should be granted. The assistant superintendent then recommends a decision to the Superintendent, who in turn recommends a decision to the Board of Education. See N.Y. Educ. Law § 3012(2). (“At the expiration of the probationary term ... the superintendent of schools shall make a written report to the board of education ... recommending for appointment on tenure those persons who have been found competent, efficient, and satisfactory....”).
In evaluating Donnelly for tenure, Chakar reviewed only Donnelly’s teaching evaluations at Woodlands and excluded Donnelly’s uniformly positive reviews at WMA. The District admits that Chakar failed to include Donnelly’s evaluations from WMA “because he could not locate them.” Appellee’s Br. 10. In March 2007, Chakar notified the assistant superintendent, defendant-appellee Hasna Muhammad, that he did not recommend Donnelly for tenure. Muhammad, after reviewing only the evaluations from Woodlands High School, accepted Chakar’s assessment of *140Donnelly’s tenure candidacy and made no independent inquiry beyond reading those evaluations and speaking with Chakar and Washington. Muhammad knew at the time she made the decision that many of Donnelly’s significant absences were related to gallbladder surgery.
Donnelly learned that he had been denied tenure in a meeting with Chakar, Muhammad, and Donnelly’s union representative. When Donnelly protested the reliance on Washington’s singularly negative evaluation, Muhammad and Chakar decided to allow Washington to conduct an additional teaching evaluation, the outcome of which would determine Muhammad’s final recommendation to the Board of Education, via the District superintendent. Donnelly prepared for that final evaluation in part by working with and providing his final lesson plans to two master teachers. Both teachers approved his lesson.
Nevertheless, Washington’s evaluation, on April 18, 2007, gave Donnelly an overall rating of “Unsatisfactory.” That evaluation specifically and emphatically criticized Donnelly’s absences, including absences during his leave, as “unacceptable ” (emphasis in original). Washington later testified that the source of Donnelly’s absences, whether personal or protected under the FMLA, was not relevant to his critique. Donnelly strongly disagreed with Washington’s evaluation, but, on the advice of the union representative, he nevertheless submitted a letter of resignation in lieu of the termination that otherwise would have been forthcoming.
II. Procedural History
Donnelly sued, alleging that the District denied him tenure in retaliation for his taking leave pursuant to the FMLA.
The district court referred the case to the magistrate judge for pretrial supervision and for a report and recommendation (“R & R”) on dispositive motions. The District moved for summary judgment, arguing inter alia that Donnelly had not worked the hours required to be eligible for FMLA leave, and even if he were eligible, he could not make a prima facie case of retaliation under McDonnell Douglas Corp. v. Green, 411 U.S. 792, 800-06, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973). Donnelly argued that “[a]t a minimum, a material issue of fact exists concerning the number of hours he actually worked,” and that he had made a sufficient showing of unlawful retaliation to survive summary judgment.4
The magistrate judge reached two conclusions relevant to this appeal. First, calculating his hours worked exclusively by reference to the CBA, the magistrate judge concluded that Donnelly was not eligible for FMLA leave as a matter of law. More particularly, the magistrate judge concluded that Donnelly’s assertion that he worked more than the hours required under the CBA failed because Donnelly “ha[d] not provided any evidence of exactly what activities he accomplished before or after the designated work day, or of how such activities were necessary.” R & R at 25. Accordingly, the magistrate judge determined that Donnelly could not “show that these activities were an ‘integral part of and indispensable to’ [his] work.” Id., quoting Mitchell v. King Packing Co., 350 U.S. 260, 263, 76 S.Ct. 337, 100 L.Ed. 282 (1956).
Second, the magistrate judge concluded that, even if Donnelly were eligible for *141FMLA leave, he had not shown that he was qualified for tenure, under the standard that we applied to plaintiffs complaining of a discriminatory denial of tenure in the context of colleges and universities. See Zahorik, 729 F.2d at 93.
The district court agreed that Donnelly had failed to demonstrate his qualification for tenure under Zahorik. Considering that issue “dispositive,” the district court did not address Donnelly’s arguable FMLA eligibility. Donnelly, 2011 WL 1899713, at *1. The district court deemed Donnelly’s argument that Zahorik did not provide any guidance in the high school context “overstated” and concluded that “[w]hile it no doubt is true that there are differences between tenure determinations in high school and collegiate settings, Zahorik nonetheless is apposite despite the fact that certain considerations there cited by the Second Circuit — like the emphasis on published scholarship — are not relevant here.” Id. Applying Zahorik, the district court concluded that, given the lack of disagreement among those who ultimately passed on Donnelly’s tenure decision, Donnelly had failed to present a triable issue of fact regarding his FMLA-retaliation claim. The district court therefore granted the District’s motion for summary judgment. Id. at *2.
Donnelly appealed.
DISCUSSION
We review orders granting summary judgment de novo, assessing whether the district court properly concluded that there was no genuine issue of material fact and that the moving party was entitled to judgment as a matter of law. See Miller v. Wolpoff & Abramson, L.L.P., 321 F.3d 292, 300 (2d Cir.2003). While “conelusory statements or mere allegations [are] not sufficient to defeat a summary judgment motion,” Davis v. New York, 316 F.3d 93, 100 (2d Cir.2002), “we are required to resolve all ambiguities and draw all permissible factual inferences in favor of the party against whom summary judgment is sought.” Terry, 336 F.3d at 137 (internal quotation marks omitted). Summary-judgment is appropriate only “[w]here the record taken as a whole could not lead a rational trier of fact to find for the non-moving party.” Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986).
We conclude that (1) Donnelly has presented sufficient evidence to raise an issue of fact with respect to FMLA eligibility; (2) Zahorik does not apply to high school teachers challenging allegedly unlawful tenure denials; and (3) Donnelly has presented sufficient evidence to require a trial to determine whether the District unlawfully retaliated against him for taking FMLA leave in denying him tenure.
I. FMLA Eligibility
Although the district court declined to engage the question of Donnelly’s FMLA eligibility, defendants seek affirmance primarily on the argument that Donnelly is ineligible for FMLA leave as a matter of law. Accordingly, we address that issue first.
A. Statutory Eligibility Provisions
To be eligible for FMLA leave, an employee must work “at least 1,250 hours of service ... during the previous 12-month period.” 29 U.S.C. § 2611(2)(A)(ii). In “determining whether an employee meets the hours of service requirement specified [above], the legal standards established under [the Fair Labor Standards Act or “FLSA”] shall apply.” Id. § 2611(2)(C), citing 29 U.S.C. § 207. The relevant FLSA legal standards are explicated, inter alia, under the Portal-to-Portal Act of 1947, 29 U.S.C. § 254.
*142B. Burden of Proof
Department of Labor regulations interpreting the FMLA address the burden of proof regarding eligibility. The regulations require that
[i]n the event an employer does not maintain an accurate record of hours worked by an employee, including for employees who are exempt from FLSA’s requirement that a record be kept of their hours worked ..., the employer has the burden of showing that the employee has not worked the requisite hours. An employer must be able to clearly demonstrate, for example, that full-time teachers (see § 825.800 for definition) of an elementary or secondary school system, or institution of higher education, or other educational establishment or institution (who often work outside the classroom or at their homes) did not work 1,250 hours during the previous 12 months in order to claim that the teachers are not eligible for FMLA leave.
29 C.F.R. § 825.110(c)(3).
This provision, not cited by the District, plainly provides that the District must prove FMLA ineligibility. There is no indication in the record that the District maintained records of the working hours of teachers, and the regulation makes clear that an “employer [that] does not maintain an accurate record of hours worked by an employee ... has the burden of showing that the employee has not worked the requisite hours.” Id. Moreover, the regulation expressly addresses the specific situation of teachers such as Donnelly, providing that an employer of “full-time teachers ... of an elementary or secondary school system ... who often work outside the classroom or at their homes” must show that such employees “did not work 1,250 hours during the previous 12 months in order to claim that the teachers are not eligible for FMLA leave.” Id. Thus, the burden of proof is squarely on the District to prove that Donnelly did not work 1,250 hours in the year preceding his leave.
C. Collective Bargaining Agreements and the FMLA
To the extent that the District argues that the CBA between Donnelly’s union and the District was effectively determinative of the number of hours Donnelly worked, we cannot agree. Noting that Donnelly was present on 172 of the 189 days of the preceding school year, the District multiplies that number by the 7.25 hour work day provided for by the CBA, and concludes that Donnelly worked only 1,247 hours.
The calculation, however, cannot stop there. With respect to the initial computation of the 1,250 hours, the regulations provide that
[t]he determining factor is the number of hours an employee has worked for the employer within the meaning of the FLSA. The determination is not limited by methods of recordkeeping, or by compensation agreements that do not accurately reflect all of the hours an employee has worked for or been in service to the employer. Any accurate accounting of actual hours worked under FLSA’s principles may be used.
29 C.F.R. § 825.110(c)(1). In light of this clear instruction, we cannot conclude that, when computing FMLA eligibility, we are restricted to the hours identified in a collective bargaining agreement. The regulations specifically reject the proposition that a compensation agreement, such as the CBA in this case — the sole compensation agreement between Donnelly and the District — can conclusively determine the computation of hours required for FMLA eligibility, unless it “accurately refleet[s] all of *143the hours an employee has worked for or been in service to the employer.” Id. The regulation is clear: “all of the hours an employee has worked” must be counted, regardless of the provisions of an employment agreement. Id. (emphasis added). Moreover, in calculating the number of hours worked, “[a]ny accurate accounting of actual hours worked under FLSA’s principles may be used.” Id. The number of hours worked is thus a factual question, on which the CBA — while certainly evidence to be considered — is not determinative.
Although not a case concerning a high school teacher, the Sixth Circuit’s decision in Staunch v. Cont'l Airlines Inc., 511 F.3d 625 (6th Cir.2008) — which appears to be the only reported appellate opinion construing 29 CFR § 825.110(c)(1)5 — is helpful by analogy. Staunch, a flight attendant for Continental Airlines, alleged unlawful retaliation after she took maternity leave. Id. at 627. The district court granted summary judgment to Continental in part on the basis of Staunch’s failure to meet the 1,250 hours threshold outlined in 29 U.S.C. § 2611(2)(A). Id. at 628. Continental showed, through flight records and the CBA that accounted for covered non-flight work for before and after the flights, that Staunch worked only 1,128 hours. Id. at 630. In response, Staunch provided her own “undated list of tasks and hours she compiled based on her own recollection,” that suggested significantly longer hours. Id.
The Sixth Circuit ultimately concluded that Staunch’s assertions were insufficient to present a genuine issue for trial. That conclusion, however, was not based simply on the terms of the CBA. Rather, the court noted that the evidence Staunch presented was unreliable principally because her “general allegations regarding the additional hours she worked prove[d] to be inflated and unsupported” by the evidence. This evidence included not only the hours that flight attendants were expected to work under the CBA, but also evidence that Staunch sought (1) to double-count time she spent checking in for the flights worked, and (2) to include additional time “for customs/immigration processing in ... Mexico, [and the] Caribbean” when she had not made trips to either destination during the relevant period. Id. Thus, the defendant in Staunch carried its burden, but did so by more than a reference to the plaintiffs CBA: Continental made a tailored, individualized showing based on preexisting records of hours worked that definitively established that the plaintiff could not meet the 1,250 hour threshold.
We adopt a standard similar to that expressed by the Sixth Circuit. In cases where a plaintiff avers that a relevant compensation agreement — including, as in Staunch, a collective bargaining agreement — “do[es] not accurately reflect all of the hours an employee has worked for or been in service to the employer,” 29 CFR § 825.110(c)(1), “the employer has the burden of showing that the employee has not worked the requisite hours.” Id. § 825.110(c)(3). As in Staunch, the employer’s burden is not heavy, but it is specific. To succeed on a motion for summary judgment, a defendant must show that, in the plaintiffs specific case, either the hours alleged could not have occurred or the hours alleged are not compensable as a matter of law “according to the principles established under the [FLSA].” Id. § 825.110(c)(1). See also Kosakow v. New Rochelle Radiology Assocs., P.C., 274 F.3d 706, 716-19 (2d Cir.2001) (holding that the *144district court erred in granting defendants summary judgment where plaintiff averred coming to work fifteen minutes earlier than reported in her time sheets to accomplish office tasks that she and her coworkers testified were prerequisites to opening the office in time to receive patients).
In this case, the CBA itself implicitly acknowledges that, on occasion, teachers will work outside of the seven hours and fifteen minutes anticipated by the CBA. It provides that “[t]o the extent practicable, the time between the end of the pupil’s regular school day and the close of the working day shall be devoted to the said responsibilities.” The CBA thus clearly anticipates circumstances that will require a teacher’s time outside of the seven-hour and fifteen-minute workday, when it will not be “practicable” that the teacher’s responsibilities can be fulfilled in the hour after the students’ school day ends. The CBA is not alone in recognizing this reality. The FMLA regulations also expressly note the common understanding that “full-time teachers ... of an elementary or secondary school system ... often work outside the classroom or at their homes,” 29 C.F.R. § 825.110(c)(3), and that this reality informs the interpretation of FMLA eligibility.
Donnelly, however, does not rely simply on generalizations about the work life of high school teachers. He avers that “most teachers” in the District “regularly work in excess” of the one hour beyond the students’ day provided for in the CBA, and that he himself “typically worked a total of 1.5 hours before and after class every day.” This claim, standing alone, would suffice to shift the burden to the District, but it is further corroborated by Chakar’s evaluation report, which acknowledged in writing that Donnelly regularly arrived at work early, and “often stays late into the afternoon working with his kids to ensure their success.” If Donnelly did indeed work an extra half hour per day beyond what the CBA required, over the 172 work days acknowledged by the District he worked far more than the three additional hours he needed to demonstrate FMLA eligibility. Moreover, since the discrepancy between the CBA calculation and the eligibility threshold is only three hours, even a jury that believed that Donnelly’s claim of an extra half hour per day was vastly exaggerated could still rationally and indeed without difficulty find, informed by the common understanding embodied in the CBA and the regulations, that in the course of a year Donnelly worked at least three hours beyond what the CBA set as a maximum.
D. The Portah-to-Portal Act
Despite its frequent invocation of the CBA, the District effectively concedes that the question is not whether the CBA provides the exclusive basis for FMLA eligibility. The District primarily argues that even if the alleged hours Donnelly worked beyond those required by the CBA were spent on tasks somehow related to teaching, Donnelly has not presented sufficient evidence to establish that the missing three hours he alleges to have worked were, as they must be, compensable under the principles of the FLSA. In making this argument, the District relies on portions of the FLSA enacted as part of the Portal-to-Portal Act of 1947, 29 U.S.C. § 254.
Under the Act, whether hours worked “count” for purposes of the FLSA (and thus for FMLA eligibility) is determined by whether those hours are devoted to activities that constitute “an integral and indispensable part of the principal activity of the employment.” Steiner v. Mitchell, 350 U.S. 247, 256, 76 S.Ct. 330, 100 L.Ed. 267 (1956); see also Kosakow, 274 F.3d at 717-18. Those activities that *145are not integral and indispensable are not compensable under the FLSA. Gorman v. Consol. Edison Co., 488 F.3d 586, 591-95 (2d Cir.2007). The Act amended the FLSA to clarify that employees need not be compensated for
activities which are preliminary to or postliminary to said principal activity or activities, which occur either prior to the time on any particular workday at which such employee commences, or subsequent to the time on any particular workday at which he ceases, such principal activity or activities.
29 U.S.C. § 254(a)(2).
To be sure, the line separating activities that are integral and indispensable to the principal activity of employment, from those that are not so integral or indispensable, is not always clear. For example, employees at a battery plant who must suit up before work and shower after can count such activities as compensable, Steiner, 350 U.S. at 256, 76 S.Ct. 330, but employees at a nuclear power facility who change clothes when they arrive and depart the plant cannot, Gorman, 488 F.3d at 595. Moreover, the shadowy distinction has apparently never been clarified by case law specific to the context of teachers. The parties do not cite, and our research has not discovered, a single case applying the Portal-to-Portal Act to hours worked by teachers.
Nonetheless, whatever the exact line separating compensable from noncompensable work, it is clear that the three hours Donnelly alleges he worked must conform to the principles established by the Act. We need not, however, definitely resolve the precise limits of what sort of work teachers do at home or after hours that is integral to teaching, and which work is merely preliminary or postliminary to that principal activity. It is sufficiently clear, as is recognized in the CBA and the regulations, that the principal work activity of teachers must extend beyond the hours spent in front of a class, and will include such basic activities as preparing lesson plans, helping students who have difficulty absorbing the class material, and preparing and grading tests and homework. The District’s argument does not turn on the precise location of the boundary between principal and preliminary activities; rather, the District argues that because Donnelly provided no details at all on what he did during the extra hours he claims to have worked, he cannot be held to have raised a genuine dispute as to a material fact about whether he worked three additional compensable hours beyond the CBA’s requirements.
That argument is unavailing. It is certainly correct that Donnelly’s testimony is extremely thin and does not state the specific tasks he performed outside of the CBA-sanctioned workday. This is not a case, however, in which an employee with a 9-to-5 job with defined office tasks claims to engage in mysterious unspecified additional activities off-site. The CBA acknowledges that job-required tasks that are part of a teacher’s primary work responsibility are regularly performed by teachers outside the hours in which the students are in the classroom. It may reasonably be inferred from Donnelly’s declaration that he claims simply that these same tasks sometimes take more time to complete than the one additional hour specified in the CBA. Moreover, the District ignores Chakar’s evaluation’s reference to Donnelly’s long hours. Besides corroborating Donnelly’s claim that he did something before and after the required hours, Chakar’s statement makes clear that he, as Donnelly’s supervisor and a representative of the District, regarded Donnelly’s long hours not as preliminary work distinct from the principal activity *146for which Donnelly was employed, but as work that the District approved and encouraged as part of its educational mission. According to Chakar, Donnelly did not merely hang around the school until late in the day; rather “[h]e often stay[ed] late into the afternoon working with his kids to ensure their success.” “Working with the kids to ensure their success” is as good a definition of the “principal activity” of a secondary school teacher as we can imagine.
The cases the District cites for the proposition that, to survive summary judgment, the Portal-to-Portal Act requires more of a showing than Donnelly offers are inapposite. For example, in Kuebel v. Black & Decker, Inc., 643 F.3d 352 (2d Cir.2011), the plaintiff argued that his routine administrative tasks, done at home, rendered his commute time compensable work. We held that “[t]he general rule, however, is and always has been that the FLSA does not treat ordinary home-to-job-site travel as compensable,” id. at 360, and contrasted the plaintiffs losing argument with the long-established conclusion that “periods during which an employee is completely relieved from duty and which are long enough to enable him to use the time effectively for his own purposes are not hours worked,” id., quoting 29 C.F.R. § 785.16(a) (alterations omitted). The District here argues that Kuebel controls this case, and that since “there is no evidence that [Donnelly] was not free to leave the school and use his time after class instruction was over, or at the conclusion of seven hours and fifteen minutes of a day, freely and as he wished,” as a matter of law, Donnelly cannot show that his work was compensable under the FLSA. Appellee’s Br. at 23.
That argument goes too far. Donnelly alleges that he worked, but did so — as both the CBA and the regulations recognize will sometimes occur — during hours when he was not required to be at school. A reasonable jury could conclude that even if a teacher is permitted to go home after seven and a quarter hours, he or she is still required to prepare proper lessons and formulate and grade examinations, and that the District will not accept as an excuse for inadequate lesson plans and ungraded student papers that the teacher did all he or she could in the time required under the CBA. The law does not preclude teachers who grade papers or plan their lessons at home from ever counting that time for purposes of FMLA eligibility.
A jury reviewing the evidence in this record might well conclude that the evidence that Donnelly presents is insufficient to persuade it to find that he spent three or more hours beyond the CBA-maximum time engaged in activities integral to his employment. That factual inquiry is not ours to answer. The District has raised questions about the credibility and probative, force of Donnelly’s evidence that he worked enough additional compensable hours to qualify for FMLA leave. But these questions are, on this record, to be answered by the jury, which can assess Donnelly’s and Chakar’s testimony and credibility and determine how much time, if any at all, Donnelly worked beyond that specified in the CBA, and whether such time was compensable under the principles of the FLSA. See Hayes v. N.Y.C. Dep’t of Corr., 84 F.3d 614, 619 (2d Cir.1996) (holding that a district court “should not weigh evidence or assess the credibility of witnesses” in reviewing a motion for summary judgment). Summary judgment for the District may not, therefore, be justified on the ground of Donnelly’s ineligibility for FMLA leave.
II. Donnelly’s Qualifications for Tenure
The District next argues that the district court correctly ruled that Donnelly failed *147to adduce sufficient evidence that he was qualified for the tenured position that he was denied, on the authority of our decision in Zahorik. We disagree.
To establish a prima facia case of FMLA retaliation, a plaintiff must establish that (1) “he exercised rights protected under the FMLA,” (2) “he was qualified for his position,” (3) “he suffered an adverse employment action,” and (4) “the adverse employment action occurred under circumstances giving rise to an inference of retaliatory intent.” Potenza v. City of New York, 365 F.3d 165, 168 (2d Cir.2004). There is no dispute that Donnelly’s failure to be promoted to tenure was an adverse employment action and we have rejected above the only disputed issue as to whether Donnelly properly exercised FMLA rights. Thus, only the second and fourth factors remain in dispute.
A. Tenure Review under Zahorik
Ordinarily, a plaintiff-employee challenging an adverse employment action as discriminatory or retaliatory need not do much to establish his qualification for the position he holds or seeks. As we have held in the related context of employment discrimination, “the qualification necessary to shift the burden to defendant for an explanation of the adverse job action is minimal; plaintiff must show only that he possesses the basic skills necessary for performance of the job.” Slattery v. Swiss Reinsurance Am. Corp., 248 F.3d 87, 92 (2d Cir.2001) (internal quotation marks and brackets omitted). Moreover, “the qualification prong must not be interpreted in such a way as to shift into the plaintiffs prima facie case an obligation to anticipate and disprove the employer’s proffer of a legitimate, non-[retaliatory] basis for its decision.” Gregory v. Daly, 243 F.3d 687, 696 (2d Cir.2001). It is unusual for a plaintiff to fail to meet this standard.
The district court held the plaintiff here, however, to a much higher standard. The district court concluded that, in order to show that “he was qualified for his position,” Potenza, 365 F.3d at 168, Donnelly must meet the exacting standard we have applied in the context of allegations of discriminatory denial of tenure to university professors. Donnelly, 2011 WL 1899713 at *1, citing Zahorik, 729 F.2d at 93-94.
Because of Zahorik’s centrality to this case, we review it in some detail. The four plaintiffs in Zahorik — assistant professors at Cornell in the Departments of Psychology, Sociology, Russian Literature, and Community Service Education, alleged that “each was the victim of discriminatory treatment on the basis of sex in Cornell’s decisions to deny them tenure and that Cornell’s tenure criteria and procedures have an illegal disparate impact within the meaning of Griggs v. Duke Power Co., 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971).” 729 F.2d at 88. Although we acknowledged that “[tjenure decisions are not exempt under Title VII,” id. at 93, we held that, in order to survive summary judgment, a university professor seeking to challenge a tenure denial must show that “some significant portion of the departmental faculty, referrants, or other scholars in the particular field hold a favorable view on the question” of the scholar-plaintiffs tenure candidacy. Id. at 93-94. In crafting this standard, the Zahorik court expressed its and other courts’ concerns that “[wjhere the tenure file contains the conflicting views of specialized scholars, triers of fact cannot hope to master the academic field sufficiently to review the merits of such views and resolve the differences of scholarly opinion.” Id. at 93, citing Lieberman v. Gant, 630 F.2d 60, 66 (2d Cir.1980). In other words, because the academic expertise needed to adjudge *148a scholar worthy of tenure is so technical, courts and juries will in many cases have great difficulty determining whether a tenure denial was based on legitimate or discriminatory considerations.
Zahorik identified five specific factors that make the review of university tenure denials by judges or juries particularly difficult. First, “tenure contracts entail commitments both as to length of time and collegial relationships which are unusual.” Id. at 92. Tenured university professorships are thus unusual not simply because of the duration of the commitment involved, but also because “collegial or professional relationships” among tenured scholars cannot “be eased by transfers among departments. Professors of English ... remain in that department for life and cannot be transferred to the History Department.” Id. Second, “academic tenure decisions are often noncompetitive,” as “a denial of tenure to one person does not necessarily lead to tenure for another.” Id. Third, “university tenure decisions are usually highly decentralized,” in that the “decision at the departmental level is of enormous importance both because of the department’s stake in the matter and its superior familiarity with the field and with the candidate.” Id. Fourth, “the number of factors considered in tenure decisions is quite extensive. The particular needs of the department for specialties, the number of tenure[d] positions available, and the desired mix of well[-]known scholars and up-and-coming faculty all must be taken into account.” Id. And fifth, “tenure decisions are a source of unusually great disagreement. Because the stakes are high, the number of relevant variables is great and there is no common unit of measure by which to judge scholarship, the dispersion of strongly held views is greater in the case of tenure decisions than with employment decisions generally.” Id. at 93.
We have no occasion to question Zahorik’s correctness in the context for which it was designed. Whatever the merits of that analysis in the university context, however, neither we nor any other appellate court we have identified has ever applied the standards Zahorik pronounced to teachers denied tenure in elementary or secondary schools. We decline to do so now.
Although both university and high-school teachers may be awarded “tenured” positions that provide long-term employment and protections against arbitrary dismissal, the two contexts have very little in common. While high school teachers and scholars at universities and colleges are both subject to individual performance evaluations at a predetermined stage of their careers that can lead either to the significant reward of protected long-term employment or to adverse employment events, subjecting those two processes to the same analysis in the face of allegedly unlawful denial of promotion does not logically follow.
The distinction between the two situations is apparent from an examination of the five factors cited in Zahorik. First, while it is true that high school and university tenure contracts both afford long-term job security to those teachers or scholars who enter into such contracts, high school tenure contracts are not impossible to break for the very reasons that the District now cites as necessary to prevent Donnelly from receiving tenure. The record in this very ease reveals that a tenured teacher in the District was dismissed for excessive absences, his tenure notwithstanding. See also N.Y. Educ. Law § 3012 (explaining that tenured public school teachers are removable for cause, which includes inter alia “conduct unbecoming a teacher[,] ... inefficiency, incompetency, physical or mental disability, or *149neglect of duty”). It is thus not clear that the “tenure” offered to teachers in the District offers materially more job protection than is common to unionized public sector employees outside the educational context.
Moreover, transfers of the kind that the Zahorik court noted are impossible in universities, in which tenure is awarded in a particular department, can and do occur in the context of secondary schools. High school teachers of one subject do not necessarily “remain in that department for life.” Zahorik, 729 F.2d at 92. High school math teachers, for example, can sometimes be “transferred to the History Department.” Id.; see also Valtchev v. City of N.Y., 400 Fed.Appx. 586, 590 (2d Cir.2010) (discussing teacher who began service teaching math and was transferred to teaching history and English). And as the closing of the WMA and Donnelly’s subsequent transfer indicate, teachers in a school district may even be transferred from one school to another. Thus, the inflexibility imposed on administrators by tenured positions, and the importance of collegial relationships in small departments, are not necessarily factors in high school hiring.
The second factor noted in Zahorik, the non-competitive nature of tenure decisions, may apply in some degree in Donnelly’s situation. Probationary teachers like Donnelly are evaluated on their own merits, and are promoted to tenure if they meet the requirements set forth by statute, regardless of the fates of other similarly-situated probationary teachers being considered for tenure at the same time. See N.Y. Educ. Law § 3012(2). In this regard, however, we think that some language in Zahorik overstated the relevance of non-competition even in the context of university tenure decisions. In many universities, the number of tenured positions in a department is strictly limited, whether by resource constraints or by the number of tenured positions assigned to a particular department. Thus, university tenure decisions often are in effect competitive, with assistant professors not only vying for promotion with each other, but on occasion even competing for a particular tenured position against lateral applicants from other institutions, including already established scholars holding tenured positions. This fact — as the Zahorik opinion recognizes in connection with its fourth factor— considerably complicates university tenure decisions. Whether a non-tenured university professor is “qualified” for tenure may well not be a question that can be answered in the abstract, by reference to some fixed minimum criteria of quality or quantity of scholarly output, but one that varies over time depending on shifting pools of available talent. There is no evidence of similar complications in Donnelly’s case. By statute, if he performed satisfactorily during his probationary period, he was qualified for promotion to a tenured position.
The third factor — the deference given to a university department’s own recommendation regarding a candidate’s tenure — is completely irrelevant in this case. In Zahorik, each of the plaintiffs was subject to a vote of tenured faculty in her respective department. Zahorik, 729 F.2d at 89-91. This “decentralized” decision-making, combined with the small number of tenured appointments in each department, makes statistical evaluations across a university’s departments or over time relatively uninformative. The District’s own description of its tenure review process includes no such element. Indeed, by statute, only one person’s tenure recommendation ultimately matters: the superintendent alone is tasked to report a recommendation to the Board of Education of each district. N.Y. Educ. Law § 3012(2).
*150This difference is of critical importance in deciding whether to extend the Zahorik standard to the high school context. The standard we ultimately applied asked whether anyone in the tenure process thought that the candidate was worthy of promotion. It is reasonable, in a case in which the university’s ultimate decision may be made by a committee, in which outside experts are typically consulted for their opinions about the significance and quality of a candidate’s writing, and in which all the tenured members of the candidate’s department — his or her putative peers — vote on the candidate’s qualifications, to conclude that a candidate who lacks any support at all for the position from any of the multiple experts whose opinion has been sought is not qualified. Here, by contrast, the record reflects that only two people, Chakar and Washington, evaluated the quality of Donnelly’s teaching in determining his qualifications for tenure, and that their negative recommendation determined the conclusions of Muhammad, and thus of the District. To ask that Donnelly, in order to proceed with his case, identify a dissenting vote in order to show that he was qualified for promotion when he had been evaluated for the promotion by only two supervisors would impose a burden that, as a practical matter, is far more stringent than that imposed in Zahorik.
The fourth factor — the extensive “number of factors considered in tenure decisions,” Zahorik, 729 F.2d at 92 — is also not present before us. The only factor that the District has identified in determining Donnelly’s eligibility for tenure is, generically, the “excellence” expected of a teacher in his third year in the District, which Chakar concluded Donnelly lacked. The other factors the Zahorik court identified — the department’s need “for specialties, the number of tenure positions available, and the desired mix of welI[-]known scholars and up-and-coming faculty” — are all missing from this case. Id. Unlike university teachers, whose scholarly writings will often be determinative in tenure decisions, and which will certainly be heavily considered along with teaching, collegiality, and the effective execution of administrative responsibilities, high school teachers are not expected to produce writing, and are evaluated solely in light of classroom performance.
We recognize the vital importance of effective teaching of elementary and secondary school students. We also recognize that evaluations of teachers’ effectiveness will often be subjective, and that reasonable people may differ about the quality of a teacher’s work. But these factors do not distinguish teaching from many other professional contexts in which courts must assess the qualifications of employees. Here, the fact that performance must be evaluated in only one arena distinguishes the promotion of high school teachers from the award of tenure in universities, where teaching is only one, and in many cases not the most important, factor in determining promotion.
Finally, while it is true that teaching is an art, not a science, and that, as with many professions, evaluations of performance may differ, there is no indication in this record or in common experience that in the high school context “tenure decisions are a source of unusually great disagreement,” or that “the dispersion of strongly held views is greater in the case of tenure decisions than with employment decisions generally.” Id. at 93. In Zahorik we concluded that such disagreement is characteristic of the university situation, and attributed that disagreement to the facts that “the stakes are high, the number of relevant variables is great and there is no common unit of measure by which to judge scholarship.” Id. Of those facts, *151only the significance of the stakes is present here. As noted above, the “number of relevant variables” is limited in the present situation, and the lack of a “common unit of measure by which to judge scholarship,” is irrelevant in high school tenure decisions. Id. Crucial to our concern in Zahorik was our belief that “triers of fact cannot hope to master the academic field sufficiently to review the merits of [the conflicting views of specialized scholars] and resolve the differences of scholarly opinion.” Id. Almost no one in a jury pool is likely to have experience with the sort of arcane scholarly issues that, for better or worse, commonly decide university tenure battles. By contrast, nearly all jurors will have direct experience, as students, with effective and ineffective elementary and secondary school teachers, and many will have had additional experience as parents or other guardians judging teachers’ effectiveness.
We therefore conclude that Zahorik is applicable only in the context of challenges to university tenure denials, and only in those eases where the Zahorik factors are independently relevant. High school tenure decisions are fundamentally a different type of employment decision, motivated by different goals and resulting in a different employment contract. As such, the requirement, in Zahorik, that a tenure candidate must show that “some significant portion of the departmental faculty, refer-rants, or other scholars in the particular field hold a favorable view” of the applicant, id. at 94, is inapplicable to high school tenure contexts.6
B. Remaining FMLA Retaliation Factors
The conclusion that Donnelly need not make the additional showing required by Zahorik does not end our inquiry. The FMLA-retaliation plaintiff must still establish inter alia that “he was qualified for his position,” and that “the adverse employment action occurred under circumstances giving rise to an inference of retaliatory intent.” Potenza, 365 F.3d at 168.
Donnelly has made an adequate showing in both respects to warrant a jury trial. First, as we have noted above, the ordinary standard for showing qualification is not exacting: the “qualification necessary to shift the burden to defendant for an explanation of the adverse job action is minimal; plaintiff must show only that he possesses the basic skills necessary for performance of the job.” Slattery, 248 F.3d at 92 (internal quotation marks omitted). Thus, in the context of secondary-school tenure denials, when evaluating the plaintiff-teacher’s prima facie case, we do not determine whether the teacher was entitled to tenure; we simply assess whether the plaintiff has demonstrated that he held the basic qualifications to be eligible for promotion. See, e.g., Helmes v. S. Colonie Cent. Sch. Dist, 564 F.Supp.2d 137, 147-48 (N.D.N.Y.2008).
*152Donnelly has made that showing. He held the necessary educational and licensing credentials to serve as a teacher, and he had worked as a teacher for the necessary period of time to be considered for promotion. The record indicates nothing egregious about his performance suggesting that he was manifestly unsuitable for promotion. Of the nine evaluations he received during his probationary period, Donnelly received the highest evaluation possible in six: the remaining three all occurred after the medical leave. To be sure, these evaluations included recommendations for improvement, as was standard for such evaluations, even for tenured teachers. Donnelly’s file also included two questionable interactions with students that might have led the District to question his suitability as a tenured teacher. Neither the interactions with students nor the recommendations for improvement, however, establish as a matter of law that Donnelly was not qualified for his position. His basic credentials, and his record of strong performance reviews which only began to change after his medical leave, are sufficient for him to meet the qualifications standard.
Second, Donnelly has also established that “the adverse employment action occurred under circumstances giving rise to an inference of retaliatory intent.” Potenza, 365 F.3d at 168. While we have never previously construed this requirement in the context of FMLA retaliation, our analysis of wrongful retaliation in other contexts—e.g., Title VII retaliation, see Mack v. Otis Elevator Co., 326 F.3d 116, 129 (2d Cir.2003)—suggests that such an inference can be established when there is a basis for a jury to conclude that “a causal connection [exists] between the plaintiffs protected activity and the adverse action taken by the employer.” Id. (internal quotation marks omitted); see also Hicks v. Baines, 593 F.3d 159, 164 (2d Cir.2010).
Donnelly has amply established a basis for such a causal inference. The “very close” temporal proximity required in other employment cases is present here. See Clark County Sch. Dist. v. Breeden, 532 U.S. 268, 273, 121 S.Ct. 1508, 149 L.Ed.2d 509 (2001) (internal quotation marks omitted). Before Donnelly’s medical leave, his teaching evaluations were extremely positive; after this leave, the evaluations deteriorated. Moreover, Donnelly presents more evidence than mere temporal proximity. The negative evaluations he received expressly penalize Donnelly for his excessive absences, including those taken—assuming his eligibility—pursuant to the FMLA. That the District manifestly penalized Donnelly for absences that a jury could find were protected by the FMLA provides a sufficient basis to send the question of the District’s retaliatory intent to the jury to reach a final determination.
We have never definitively applied the burden-shifting framework from the context of employment discrimination to FMLA retaliation, and we need not establish that application here, as Donnelly’s case must proceed to trial regardless of whether such a pretrial shift in burden occurs. Donnelly has presented direct evidence that his arguably FMLA-protected leave was held against him in the tenure process. That evidence clearly suffices to meet the “minimal” burden of showing a prima facie case.7
*153In reaching this conclusion, we emphasize that we neither express nor imply any view as to the ultimate merits of Donnelly’s claim. We hold only that he has presented sufficient evidence to permit a jury to decide whether the District unlawfully denied Donnelly tenure because he took a protected FMLA leave, or whether the District lawfully deemed Donnelly deficient in his performance and therefore unqualified for the position of tenured teacher.
CONCLUSION
For the foregoing reasons, we reverse the district court’s judgment and remand the case for trial regarding (1) Donnelly’s FMLA eligibility and (2) whether the District unlawfully denied him tenure in retaliation for exercising his rights under the FMLA.