Order; Dissent to Order by Chief Judge KOZINSKI; Opinion by Judge GOULD; Dissent by Judge FISHER.
ORDER
The opinion in the above-captioned matter filed on March 21, 2012, and published at 673 F.3d 1218, is amended as follows and is simultaneously filed with this order:
At slip opinion page 3268, line 2, add a footnote after <544 U.S. at 173, 125 S.Ct. 1497. >, stating: <Like the Supreme Court in Jackson, “[w]e do not rely on regulations extending Title IX’s protection beyond its statutory limits.” 544 U.S. at 178, 125 S.Ct. 1497. Our decision rests on “the statute itself,” not on regulations implementing Title IX. Id.; see also 34 C.F.R. §§ 100.7(e), 106.71.>.
Judges Gould and Paez have voted to deny the petition for panel rehearing and rehearing en banc. Judge Fisher has voted to grant the petition for panel rehearing and rehearing en banc. The full court has been advised of the petition for rehearing en banc. A judge requested a vote on whether to rehear the matter en banc, and the matter failed to receive a majority of the votes of the nonrecused active judges in favor of en banc consideration. Fed. R.App. P. 35. The petition for panel rehearing and rehearing en banc is denied.
No future petitions for rehearing or rehearing en banc will be entertained.
IT IS SO ORDERED.
Chief Judge KOZINSKI, with whom Judges O’SCANNLAIN, GRABER, FISHER, TALLMAN, BEA and M. SMITH join, dissenting from the order denying the petition for rehearing en banc:Bad facts make bad law. No facts make worse law. That’s what happened here *719when the panel majority allowed plaintiff Monica Emeldi to escape summary judgment even though she produced no evidence of causation, an element of her retaliation claim. In the place of evidence, the majority permits Emeldi to create a material issue of fact by speculation. This opinion undermines the pleading framework for Title IX and Title VII and erodes the well-established standards for summary judgment. Worse still, it jeopardizes academic freedom by making it far too easy for students to bring retaliation claims against their professors. Plaintiffs will now cite Emeldi in droves to fight off summary judgment: We may not have any evidence, but it’s enough under Emeldi. Defendants will go straight to trial — or their checkbooks — because summary judgment will be out of reach in the Ninth Circuit.
I
Monica Emeldi, a former Ph.D. candidate at the University of Oregon, had a falling out with her dissertation advisor. Emeldi v. TJniv. of Or., 698 F.3d 715, 721-22 (9th Cir.2012). Emeldi says that she complained to a university administrator about sex discrimination, the administrator relayed this complaint to Emeldi’s advisor and the advisor resigned as her dissertation chair in retaliation. Id. at 722, 725. Emeldi also asserts, again without evidence, that the advisor prevented Emeldi from finding a replacement, thus forcing her to withdraw. Id. at 723.
Under the established Title VII pleading framework, which the majority applies to this Title IX case, Emeldi must show a causal connection between her complaint and her advisor’s resignation. Surrell v. Cal. Water Serv. Co., 518 F.3d 1097, 1108 (9th Cir.2008). Emeldi says the administrator told the advisor about the discrimination complaint in a phone call between the two. Emeldi 698 F.3d at 722, 727. But Emeldi has no evidence that the administrator and the advisor discussed discrimination. To the contrary, the administrator stated under oath that she didn’t talk to the advisor about discrimination and that she couldn’t have because she never heard Emeldi make the complaint in the first place. Id. at 723, 727.
This case is not at the pleading stage. The parties have gone through discovery and Emeldi has come up with nothing to support her speculation that the discrimination complaint was discussed. All we’re left with is Emeldi’s claim, sourced to her own amended declaration, that the administrator said she “debriefed” the advisor about the conversation with Emeldi. Id. at 723, 727 n. 4, 729. Debriefing the advis- or is hardly an admission that they discussed discrimination. This is especially true in light of the fact that the administrator asked for and received Emeldi’s permission to call the advisor about Emeldi’s dissertation difficulties, id. at 735-36 & n. 3 (Fisher, J., dissenting), and in light of the fact that the administrator testified she’d never heard the discrimination complaint, id. at 723, 727 (majority opinion).
The majority finds the debriefing “evidence” sufficient to reverse the grant of summary judgment. It holds that “a jury reasonably could infer that [the administrator] passed Emeldi’s complaint on to [the advisor].” Id. at 727. This is a serious error that contravenes our own precedent, as the dissent notes: “[W]hen the non-moving party relies only on its own affidavits to oppose summary judgment, it cannot rely on conclusory allegations unsupported by factual data to create an issue of material fact.” Id. at 733-34 (Fisher, J., dissenting) (quoting Hansen v. United States, 7 F.3d 137, 138 (9th Cir.1993) (per curiam)). It’s also contrary to the teachings of the Supreme Court, by *720permitting Emeldi to plead her way out of summary judgment.
The Supreme Court has held that “mere pleadings themselves” can’t defeat summary judgment. Celotex Corp. v. Catrett, 477 U.S. 317, 324, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986). And even where some evidence is presented beyond the pleadings, that’s still not enough “[i]f the evidence is merely colorable, or is not significantly probative.” Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 249-50, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986) (internal citations omitted). The supposed admission about debriefing says nothing about whether discrimination was discussed, so it’s not even relevant evidence. But even if it were, it’s of vanishing probative value, far short of the threshold needed to stave off summary judgment.
The most Emeldi can say about the phone call is that the administrator and the advisor discussed something about Emeldi’s conversation with the administrator. The Supreme Court warned against defeating summary judgment based on inferences drawn from such “ambiguous conduct”: “[Cjonduct that is as consistent with permissible competition as with illegal conspiracy does not, without more, support even an inference of conspiracy.” Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 597 n. 21, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986); see also Bell Atl. Corp. v. Twombly, 550 U.S. 544, 554, 127 S.Ct. 1955, 167 L.Ed.2d 929 (2007). The administrator’s phone call to the dissertation advisor is just such an example: It’s as consistent with a discussion about Emeldi’s dissertation as with any mention of discrimination. The majority is wrong to rely on it as evidence of causation.
The danger of the majority’s opinion should be obvious. If a plaintiff can escape summary judgment based on his own vague description of what someone else said during a conversation with a third party, defendants can never get summary judgment because the plaintiff will always have his own word to fall back on. This would thwart the Supreme Court’s directive that summary judgment be “regarded not as a disfavored procedural shortcut,” but as “an integral part of the Federal Rules” designed “to secure the just, speedy and inexpensive determination of every action.” Celotex, 477 U.S. at 327, 106 S.Ct. 2548 (internal quotation marks omitted). It would also cut against the grain of the Supreme Court’s recent opinions in Twombly and Iqbal, which required plaintiffs to provide more than “bare assertions” or a “ ‘formulaic recitation of the elements’ ” in pleading a claim. Ashcroft v. Iqbal, 556 U.S. 662, 681, 129 S.Ct. 1937, 173 L.Ed.2d 868 (2009) (quoting Twombly, 550 U.S. at 555, 127 S.Ct. 1955). Even at the motion to dismiss stage, plaintiffs must do something to “nudge[ ] their claims across the line from conceivable to plausible.” Twombly, 550 U.S. at 570, 127 S.Ct. 1955. Emeldi has made it all the way past summary judgment without doing even that.
II
The majority’s opinion would be bad enough if confined to the Title VII context. But this decision will impair the open exchange of ideas in our schools and universities if applied, as the panel majority does, in the Title IX context. Even accepting that the Title VII pleading standard applies to Title IX cases, no one claims the pleading standard should be lower for students suing professors in the Ivory Tower than for employees suing supervisors on the factory floor. The relationship between professor and Ph.D. student requires both parties to engage in candid, searing analysis of each other and each other’s ideas. Methodology, philosophy *721and personality often lead to intractable disputes and, when they do, the professor must be free to walk away without fear of a frivolous discrimination suit.
It’s not just the practicalities of academia that require this freedom. The First Amendment does, too. See Regents of Univ. of Cal. v. Bakke, 438 U.S. 265, 312, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978). In equating Title IX with Title VII, the panel overlooks the critical differences between academia and the outside world. It applies the law so loosely that one of the laxest interpretations of the pleading standard is now planted squarely in academia, just where the pleading standard should be highest. If this ill-considered precedent stands, professors will have to think twice before giving honest evaluations of their students for fear that disgruntled students may haul them into court. This is a loss for professors and students and for society, which depends on their creative ferment.
A great deal is at stake in the decision whether to allow a case like this to go to trial. In the Title VII context, subjecting employers to the expenses and risks of trial when the employee has presented nothing but unsubstantiated suspicions of discrimination imposes huge costs on businesses and makes them targets for holdup settlements. See, e.g., Wal-Mart Stores, Inc. v. Dukes, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011). The costs are even greater in the Title IX context, where the vagaries of litigation will chill academic freedom and intimidate institutions into granting degrees to undeserving candidates. Would any of us choose to go under the scalpel of a surgeon who “earned” his M.D. by bullying his medical school with unsubstantiated claims of unlawful discrimination? Emeldi is a very, very bad result, which bespeaks a major misapplication of long-standing legal principles to the sensitive area of academia. It invites all manner of frivolous suits while further diluting the authority of our schools and universities to maintain standards of academic excellence among students and faculty. I can only hope it will not be followed by other courts considering the issue.
GOULD, Circuit Judge:OPINION
In Jackson v. Birmingham Board, of Education, the Supreme Court held that retaliation by a federally funded educational institution against someone who complains of gender discrimination is actionable under Title IX. 544 U.S. 167, 171, 125 S.Ct. 1497, 161 L.Ed.2d 361 (2005). We must decide what a plaintiff must prove to prevail on a Title IX retaliation claim, and whether plaintiff Monica Emeldi adduced sufficient evidence of her claim to overcome summary judgment.
I
Monica Emeldi sued the University of Oregon, alleging that it prevented her from completing a Ph.D. program in retaliation for having complained of gender-based institutional bias in the University’s Ph.D. program, and gender discrimination by her faculty dissertation committee chair.
Emeldi was a Ph.D. student in the University of Oregon’s College of Education, in its Department of Special Education. Her advisor and dissertation committee chair, Edward Kame’enui, took a sabbatical starting in the fall of 2005. Emeldi asked Robert Horner, another professor, to replace Kame’enui as her dissertation chair. Horner agreed. During the time of Emeldi’s work with Horner, Emeldi and *722other Ph.D. candidates complained to Mike Bullís, Dean of the College of Education, about lack of adequate support for female Ph.D. candidates. In May 2007, Emeldi produced a memo summarizing a meeting between Bullís and several graduate students. That memo lists, as one of fifteen topics discussed, the students’ concern about the Department’s lack of female role models. The memo says:
Students request that qualified Women be hired into tenured faculty positions [emphasis]. Students attempted and were unable to identify a current female appointment to a tenured faculty position. Students need to experience empowered female role models successfully working within an academic context [emphasis]. Doctoral students request that the college model a balance of gender appointments that reflect the proportion of student gender population ratios.
While the University maintains that no one other than Bullís knew of the memo, Emeldi’s position was that she was told that all Department faculty received copies; that it was “common knowledge in the College of Education” that she was dissatisfied with the Department’s level of support for women; that Horner, her dissertation chair, was treating her less favorably than his male graduate students and did not give her the same support and attention that he gave male candidates; that Horner often ignored her and did not make eye contact with her; that, when Emeldi attended Horner’s group meetings with his graduate students, either she was not on the agenda, or no substantial or meaningful work of hers was discussed; and that Horner’s male students had opportunities that were not available to his female students, such as access to more and better resources, including more office space and better technology for collecting data.
Whatever their teacher-student relationship at first, Emeldi’s relationship with Horner as Ph.D. advisor soured. The reasons for this development are unclear. The University vigorously disputes that Horner treated his male students more favorably than his female students, and its position is that Emeldi’s relationship with Horner deteriorated because Emeldi “refused to listen to Dr. Horner regarding the necessary changes to produce a dissertation that would be a focused piece of scholarship.”1 Emeldi attributes the worsening relationship to Horner’s gender animus.
In October 2007, Emeldi met with University administrators Annie Bentz and Marian Friestad to discuss her worsening relationship with Horner. Emeldi says that she complained to Friestad about the Department’s “institutional bias in favor of male doctoral candidates, and a relative lack of support and role models for female candidates.” To illustrate her experience of this “institutional bias,” she said that she “identified the chair of [her] dissertation committee, Dr. Rob Horner, as being distant and relatively inaccessible to me.” According to Emeldi, Friestad then alerted Horner that Emeldi had accused him of discriminating against her. While Friestad does not dispute that she spoke with Horner, her version of the eonversa*723tion Emeldi described is markedly different. Friestad, who is an administrator and professor, says that Emeldi never alleged discrimination in their discussion about Horner and that Friestad and Horner discussed only Emeldi’s dissertation, not an allegation of discrimination. However, Emeldi in her amended declaration explicitly said that Friestad told Emeldi that Friestad had “debriefed” Horner on the conversation Friestad had with Emeldi. Within a few weeks, Horner, by email, resigned as Emeldi’s dissertation chair. According to Emeldi, Horner then told other Department faculty members that Emeldi should not be granted a Ph. D., and should instead be directed into the Ed.D. program, which Emeldi says is a less prestigious degree. The University denies that this occurred.
Emeldi sought a new dissertation chair, but did not find one. According to Emeldi, she asked fifteen faculty members in her Department, some of whom said that they were too busy and some of whom said that they were not qualified to supervise her research. The University doesn’t dispute that she inquired of fifteen faculty members, but criticizes Emeldi’s efforts to obtain a new dissertation chair as inadequate, arguing that she did not try to recruit two faculty members who were qualified and available, including her former advisor Kame’enui. While seeking a new dissertation chair, Emeldi also pursued the University’s internal grievance procedure, which, she says, contributed to her inability to find a willing faculty member. Unable to complete her Ph.D. without a dissertation chair, Emeldi abandoned her pursuit of the Ph.D. degree, thus effectively withdrawing from the University.
Emeldi then filed this lawsuit in Oregon state court. The University timely removed the action to federal court, but mistakenly said in its notice of removal that Emeldi’s suit was filed in Linn and Multnomah Counties, when in fact the suit was filed in Lane County. The University then filed an amended notice of removal correcting these errors, but the amendment was filed after the 30-day removal deadline had expired. Emeldi sought remand on the basis that the defective notice of removal was fatal to federal jurisdiction, but the district court rejected this argument.
After a period of discovery, the University moved for summary judgment, which Emeldi opposed. The district court granted summary judgment for the University on the alternative grounds that Emeldi did not engage in protected activity and that she adduced no evidence showing that the University’s adverse actions were causally related to her protected activity. Emeldi v. Univ. of Or., No. 08-6346, 2010 WL 2330190, at *2-5 (D.Or. June 4, 2010). Emeldi timely appealed.
II
We have jurisdiction pursuant to 28 U.S.C. § 1291. We review a grant of summary judgment de novo. Oliver v. Keller, 289 F.3d 623, 626 (9th Cir.2002). “Summary judgment is warranted when ‘there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.’ ” Wash. Mut. Inc. v. United States, 636 F.3d 1207, 1216 (9th Cir.2011) (quoting Fed.R.Civ.P. 56(a)).
III
We start with the statutory premise that Title IX of the Education Amendments of 1972 bars gender-based discrimination by federally funded educational institutions. It provides, “No person in the United States shall, on the basis of sex, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any ed*724ucation program or activity receiving Federal financial assistance.... ” 20 U.S.C. § 1681(a). In Jackson, the Supreme Court held that “[r]etaliation against a person because that person has complained of sex discrimination” is a form of gender-based discrimination actionable under Title IX. 544 U.S. at 173, 125 S.Ct. 1497.2
Until now, we have not had occasion to say what a plaintiff must prove to prevail on a retaliation claim under Title IX. We join our sister circuits in applying the familiar framework used to decide retaliation claims under Title VII.3 In this framework, a plaintiff who lacks direct evidence of retaliation must first make out a prima facie case of retaliation by showing (a) that he or she was engaged in protected activity, (b) that he or she suffered an adverse action, and (c) that there was a causal link between the two. Brown v. City of Tucson, 336 F.3d 1181, 1192 (9th Cir.2003). We have emphasized that to make out a prima facie case, a plaintiff need only make a minimal threshold showing of retaliation. As we have explained, “ ‘The requisite degree of proof necessary to establish a prima facie case for Title VII claims on summary judgment is minimal and does not even need to rise to the level of a preponderance of the evidence.’ ” Davis v. Team Elec. Co., 520 F.3d 1080, 1089 (9th Cir.2008) (quoting Wallis v. J.R. Simplot Co., 26 F.3d 885, 889 (9th Cir.1994) (ellipses omitted)); see also Humphries v. CBOCS W., Inc., 474 F.3d 387, 405 (7th Cir.2007) (noting that “[ejstablishing a prima facie case” is “not ... an onerous requirement”).
Once a plaintiff has made the threshold prima facie showing, the defendant must articulate a legitimate, non-retaliatory reason for the challenged action. Davis, 520 F.3d at 1089. If the defendant does so, the plaintiff must then “show that the reason is pretextual either directly by persuading the court that a discriminatory reason more likely motivated the employer or indirectly by showing that the employer’s proffered explanation is unworthy of credence.” Id. (internal quotation marks and citation omitted).
We stress three reasons for adopting the Title VII framework for Title IX retaliation claims. First, the legislative history of Title IX “strongly suggests that Congress meant for similar substantive standards to apply under Title IX as had been developed under Title VII.” Lipsett v. Univ. of P.R., 864 F.2d 881, 897 (1st Cir.1988). The House Report provides:
One of the single most important pieces of legislation which has prompted the cause of equal employment opportunity is Title VII of the Civil Rights Act of 1964.... Title VII, however, specifically excludes educational institutions from its terms. [Title IX] would remove that exemption and bring those in education under the equal employment provision.
H.R.Rep. No. 92-554, at 46 (1972), reprinted in 1972 U.S.C.C.A.N. 2462, 2512.
Second, we have found the Title VII framework useful in assessing claims of discrimination and retaliation outside the *725Title VII context, even where its application is not mandatory. See, e.g., Diaz v. Eagle Produce Ltd. P’ship, 521 F.3d 1201, 1207 (9th Cir.2008) (applying the Title VII framework to a claim under the Age Discrimination in Employment Act); Keyset v. Sacramento City Unified Sch. Dist., 265 F.3d 741, 754 (9th Cir.2001) (applying the Title VII framework to an equal protection claim); Gay v. Waiters’ & Dairy Lunch-men’s Union, Local No. 30, 694 F.2d 531, 538 (9th Cir.1982) (applying the Title VII framework to a claim brought under 42 U.S.C. § 1981).
Third, the Supreme Court has often “looked to its Title VII interpretations of discrimination in illuminating Title IX.” Olmstead v. L.C. ex rel. Zimring, 527 U.S. 581, 616 n. 1, 119 S.Ct. 2176, 144 L.Ed.2d 540 (1999). Following this approach, we hold that the Title VII framework generally governs Title IX retaliation claims.
IV
A
The first requirement of a prima facie case of retaliation is that the plaintiff engaged in protected activity. Viewing the evidence presented at summary judgment in Emeldi’s favor, we hold that Emeldi’s complaints to Bullis and Friestad about gender-based institutional bias, and to Friestad about Horner’s unequal treatment of female graduate students, were protected activity under Title IX.
As an initial matter, we have no doubt that Title IX empowers a woman student to complain, without fear of retaliation, that the educational establishment treats women unequally. Jackson, 544 U.S. at 174, 125 S.Ct. 1497; see also generally Lucy M. Stark, Exposing Hostile Environments for Female Graduate Students in Academic Science Laboratories, 31 Harv. J.L. & Gender 101 (2008). Emeldi’s complaint to Friestad that there was institutional bias against women in the Ph.D. program and that her dissertation chair, Horner, was treating his male graduate students more favorably than his female graduate students, is thus unmistakably a protected activity under Title IX. The protected status of her alleged statements holds whether or not she ultimately would be able to prove her contentions about discrimination. See Moyo v. Gomez, 40 F.3d 982, 984 (9th Cir.1994).
Emeldi says that she complained to Friestad about the Department’s “institutional bias in favor of male doctoral candidates, and a relative lack of support and role models for female candidates.” Illustrating her experience of this “institutional bias” in speaking with Friestad, she says that she “identified the chair of [her] dissertation committee, Dr. Rob Horner, as being distant and relatively inaccessible to me.”
It is a protected activity to “protestf ] or other wise oppose[ ] unlawful ... discrimination.” Moyo, 40 F.3d at 984; see also Bigge v. Albertsons, Inc., 894 F.2d 1497, 1501 (11th Cir.1990). In the Title IX context, “speak[ing] out against sex discrimination”—precisely what Emeldi says that she did—is protected activity. Jackson, 544 U.S. at 178, 125 S.Ct. 1497. Accordingly, we hold that Emeldi has alleged facts that, if true, demonstrate that she engaged in an activity protected by Title IX.
B
The second requirement of a prima facie case of retaliation is that the plaintiff suffered an adverse action. Viewing the evidence presented at the summary judgment stage in Emeldi’s favor, we hold that Horner’s resignation constitutes an adverse action.
*726In the Title VII context, the Supreme Court has said that the adverse action element is present when “a reasonable [person] would have found the challenged action materially adverse, which in this context means it well might have dissuaded a reasonable [person] from making or supporting a charge of discrimination.” Burlington N. & Santa Fe Ry. Co. v. White, 548 U.S. 53, 68, 126 S.Ct. 2405, 165 L.Ed.2d 345 (2006) (internal quotation marks and citations omitted). We have held the adverse action requirement was satisfied, for example, when an employee was forced to use a grievance procedure to get overtime work assignments that were routinely awarded to others, Fonseca v. Sysco Food Servs. of Ariz., Inc., 374 F.3d 840, 848 (9th Cir.2004), when an employee was assigned more hazardous work than her co-workers, Davis, 520 F.3d at 1089-90, and when an employee was laterally transferred or received undeserved poor performance ratings, Yartzoff v. Thomas, 809 F.2d 1371, 1376 (9th Cir.1987).
We will not establish a different rule on adverse action for Title IX than for Title VII. Women students should not be deterred from advancing pleas that they be treated as favorably as male students. A student cannot complete the University’s Ph.D. program without a faculty dissertation chair, and the loss of a chair is an adverse action.
This sort of adverse action bears analogy to the concept of constructive discharge, in which a retaliating employer creates working conditions so “ ‘extraordinary and egregious [as] to overcome the normal motivation of a competent, diligent, and reasonable employee to remain on the job.’” Poland v. Chertoff, 494 F.3d 1174, 1184 (9th Cir.2007) (quoting Brooks v. City of San Mateo, 229 F.3d 917, 930 (9th Cir.2000)). Here, although the University did not formally dismiss Emeldi from the Ph.D. program, as a practical matter, it rendered her unable to complete the degree. A reasonable person in Emeldi’s position — someone who had been abandoned by her dissertation chair and who was unable, despite diligent efforts, to secure a replacement chair — could justifiably feel unable to complete the Ph.D. program. A reasonable person would find these events “materially adverse” insofar as they “might have dissuaded” such person from complaining of discrimination in the Department. Burlington N. & Santa Fe Ry. Co., 548 U.S. at 68, 126 S.Ct. 2405. We therefore conclude that Horner’s resignation was an adverse action.
C
The third requirement of a prima facie case of retaliation is a causal link between the protected activity and adverse action. “At the prima facie stage of a retaliation case, ‘the causal link element is construed broadly so that a plaintiff merely has to prove that the protected activity and the negative ... action are not completely unrelated.’ ” Poland, 494 F.3d at 1180 n. 2 (quoting Pennington v. City of Huntsville, 261 F.3d 1262, 1266 (11th Cir.2001)) (alteration omitted). Emeldi has met this standard. From the record, we conclude that Emeldi has produced evidence from which a rational fact-finder could find a causal link between Emeldi’s complaints of gender discrimination in the Department and the adverse actions identified above.
First, the proximity in time between Emeldi’s complaint to Friestad about Horner and Horner’s resignation as her dissertation chair is strong circumstantial evidence of causation. See Cornwell v. Electra Cent. Credit Union, 439 F.3d 1018, 1035 (9th Cir.2006) (“[C]ausation sufficient to establish the third element of the prima facie case may be inferred from ... the *727proximity in time between the protected action and the allegedly retaliatory employment decision.” (quoting Yartzoff, 809 F.2d at 1376)).4
Second, Emeldi has articulated a theory of how Horner found out about her complaints: Friestad relayed them to him. Emeldi alleges that she complained of Horner’s gender bias — among other things — at her October 2007 meeting with Friestad. Friestad admits that she relayed Emeldi’s complaints to Horner, but denies that Emeldi raised concerns about discrimination at the meeting. Friestad also insists that she did not inform Horner of any allegations of discrimination. Nonetheless, a reasonable jury, crediting Emeldi’s recollection that she complained specifically to Friestad about Horner’s favoring of male Ph.D. candidates, could find a causal link between Friestad’s conversation with Horner and his resignation from the dissertation chair post. Stated another way, a jury reasonably could infer that Friestad passed Emeldi’s complaint on to Horner, and that Horner’s resignation not long thereafter5 as Emeldi’s dissertation chair was a response to Emeldi’s complaint.
Third, Emeldi offered evidence that Horner exhibited gender-based animus in other contexts. See Coghlan v. Am. Seafoods Co. LLC, 413 F.3d 1090, 1095 n. 6 (9th Cir.2005) (stating that “evidence establish[ing] the employer’s animus toward the class to which the plaintiff belongs” is relevant to proving causation). Specifically, Emeldi said that Horner gave more attention and support to male students and that he ignored her and did not make eye contact with her. She contended that, when she attended Horner’s graduate student group meetings, she was “not on the agenda, or when [she was] on the agenda, that no substantial/meaningful work [of hers was] discussed.” She gave specific examples of Horner’s male students being given opportunities that were not available to his female students. For example, Horner allegedly gave one male student access to more office space and better technology for collecting data than similar female students.
As the above discussion reveals, there is ample circumstantial evidence to establish causation. Emeldi also points to other evidence in the record that would support a jury inference of causation: (1) that Horner resigned as Emeldi’s dissertation chair *728without designating or providing assistance in securing a replacement chair is circumstantial evidence of retaliatory intent; (2) that Horner praised Emeldi on the progress of her dissertation, could, together with other evidence, support the inference that his stated reasons for resigning as her dissertation chair were pretextual;6 and (3) that Emeldi could not secure a replacement dissertation chair, despite asking fifteen faculty members, is circumstantial evidence that Horner poisoned his colleagues against her.
These items together provide a sufficient basis for a jury to find that Emeldi’s protected activity brought about Horner’s resignation.
The dissent argues that Emeldi’s position is based on impermissible speculation, citing Cafasso v. Gen. Dynamics C4 Sys., Inc., 637 F.3d 1047, 1061 (9th Cir.2011). But this case is nothing like Cafasso. There, we rejected Cafasso’s claim that her employer eliminated her department and terminated her employment in retaliation for her inquiries about suspected fraud. Cafasso, 637 F.3d at 1060-61. The employer’s position was that the action was part of a corporate reorganization unrelated to Cafasso’s inquiries; Christopher Marzilli, the official who terminated Cafasso, testified that he did not know about her inquiries when he made the reorganization decision; and “Cafasso admitted in deposition that she had no reason to disbelieve Marzilli’s account.” Id. at 1060. Cafasso nevertheless argued for a “cat’s paw” theory of liability, see generally Poland v. Chertoff, 494 F.3d 1174, 1182-83 (9th Cir.2007), which we rejected as speculative. Cafasso, 637 F.3d at 1061. Cafasso would have had to establish “that one of Marzilli’s subordinates, in response to Cafasso’s protected activity, set in motion Marzilli’s decision to eliminate Cafasso’s department and job, and that the subordinate influenced or was involved in the decision or decisionmaking process.” Id. (internal quotation marks, citation, and alteration omitted). Because Cafasso did not “set forth non-speculative evidence of specific facts” that this chain of events in fact occurred, we concluded that to find liability would require “undue speculation.” Id.
We do not disagree with the principle that mere speculation cannot raise an issue of fact. But here, by contrast, Emeldi proffered non-speculative evidence supporting reasonable inferences of causation. Her declaration states that she complained to Friestad about gender discrimination in the Department and, at this stage, her assertions must be accepted as true. The dissent reaches a contrary conclusion only by disregarding tried and true principles governing summary judgment. The dissent first asserts that Emeldi’s complaint to Friestad that Horner was “distant and relatively inaccessible” is not a claim of gender bias. However, the correct approach is to consider Emeldi’s complaint in its context. See T.W. Elec. Serv., Inc. v. Pac. Elec. Contractors Ass’n, 809 F.2d 626, 631 (9th Cir.1987) (“[T]he court’s ultimate inquiry is to determine whether the ‘specific facts’ set forth by the non-moving party, coupled with undisputed background or contextual facts, are such that a rational or reasonable jury might return a verdict in its favor based on that evidence.” (citing Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 257, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986))). Here, where the complaint that Horner was “distant and relatively inaccessible” immediately followed Emeldi’s complaint of institutional *729bias, a jury could reasonably infer that she was giving an example of the institutional bias that led to inadequate support for women Ph.D. candidates, and indeed the normal reading of her “distant and relatively inaccessible” criticism of Horner in context is that he was relatively inaccessible by contrast to his accessibility for male Ph.D. candidates.7
The dissent further argues that, even if “gender discrimination was discussed” between Friestad and Emeldi, it is only speculative to infer that Friestad relayed Emeldi’s complaints of discrimination to Horner. Again, the dissent reaches this conclusion only by ignoring the general rules governing summary judgment. As noted above, as the nonmoving party Emeldi was to be believed and reasonable inferences given her. In her amended declaration, she explicitly states that Friestad told her that Horner was “debriefed” on their discussion. If we assume Emeldi’s statements are true, a reasonable inference arises that Friestad “debriefed” Horner about Emeldi’s complaints of gender discrimination. These facts are sufficient to state a prima facie case.
D
Because Emeldi established a prima facie case of retaliation, we inquire whether the University has stated a legitimate, non-retaliatory reason for the challenged action, and if so, whether Emeldi has shown that the reason is pretextual. See Davis, 520 F.3d at 1089.
The University says that Horner resigned for a proper reason, that is, because Emeldi did not follow his research advice. Further, University administrators did not provide a dissertation chair because, the University says, the faculty members who Emeldi solicited were unwilling to take Emeldi as a student for legitimate reasons, such as being unavailable or unqualified to advise her research. If credited by the jury, the University states legitimate, non-retaliatory reasons for Horner’s resignation.
But Emeldi has presented evidence from which a reasonable jury could conclude that the University’s account is pretextual. For substantially the same reasons we concluded that Emeldi proffered sufficient evidence of causation, we likewise conclude that Emeldi’s evidence is sufficient to show pretext. See Farrell v. Planters Lifesavers Co., 206 F.3d 271, 286 (3d Cir.2000) (explaining that the causation and pretext inquiries are often overlapping). The proximity in time between Emeldi’s complaints of unequal treatment and Horner’s resignation as Emeldi’s dissertation chair; Friestad’s admission that she relayed Emeldi’s complaints to Horner; Horner’s resignation without providing assistance in securing a replacement chair; other evidence of Horner’s gender-based animus; Horner’s praise for Emeldi; and Emeldi’s inability to secure a replacement dissertation chair, all considered together, could lead a reasonable jury to conclude that Emeldi’s complaints of unequal treatment, and not Horner’s dissatisfaction with her research, motivated Horner’s resignation.8
*730Because a reasonable jury could conclude from the evidence presented at summary judgment that Horner’s resignation was gender-based retaliation, the district court erred in granting summary judgment.
E
We pause to elaborate on the sufficiency of evidence that Emeldi presented in response to the University’s motion for summary judgment. When deciding whether an asserted evidentiary dispute is genuine, we inquire whether a jury could reasonably find in the nonmovant’s favor from the evidence presented. See Anderson, 477 U.S. at 251-52, 106 S.Ct. 2505 (stating that summary judgment requires determination of “whether the evidence presents a sufficient disagreement to require submission to a jury or whether it is so one-sided that one party must prevail as a matter of law”).
We cannot say that a reasonable jury would be compelled to reject liability. We are mindful that the University has offered evidence that would support a verdict in its favor. For starters, the testimony of Horner and Friestad contradicts Emeldi’s account, and emails corroborate the University’s version of events. Making matters worse for Emeldi, her own account at times may appear to be inconsistent. In deposition, she testified that she “would be speculating” if she said why she believed Horner’s resignation as her dissertation chair was gender-based retaliation.9 Further, the record does not disclose why, *731despite unsuccessfully soliciting fifteen faculty members, Emeldi overlooked two professors who, the University says, were qualified and available to replace Horner as Emeldi’s dissertation chair. All of this is to say that the University may have a convincing case at trial. However, that the University has presented strong evidence in its defense does not undermine our conclusion that there is a genuine dispute of factual issues that requires resolution by a jury.
V
Emeldi also challenges federal subject-matter jurisdiction over her case. The basis of her jurisdictional challenge is that the University’s original notice of removal mistakenly said that the action was filed in Linn and Multnomah Counties, when in fact it was filed in Lane County. This error was corrected by an amended notice of removal, but Emeldi protests that the amendment was untimely.
“[T]he propriety of removal is determined solely on the basis of the pleadings filed in state court.” Williams v. Costco Wholesale Corp., 471 F.3d 975, 976 (9th Cir.2006) (per curiam). Where, as here, the state court pleadings establish federal jurisdiction, an obvious factual error in the notice of removal is not fatal to jurisdiction. But even if the notice’s mistaken listing of the county from which the case was removed were fatal to jurisdiction, the University’s amendment would cure the defect. See Pochiro v. Prudential Ins. Co. of Am., 827 F.2d 1246, 1248 (9th Cir.1987) (holding that insufficient verification of a notice of removal was not fatal to jurisdiction because the technical error was later corrected by amended notice).
VI
We reverse the district court’s grant of summary judgment on Emeldi’s state law claim for the same reasons as Emeldi’s Title IX claim.10 Also, we reverse the district court’s award of costs because the University is no longer the prevailing party under Federal Rule of Civil Procedure 54(d). See Cascade Health Sol. v. Peace-Health, 515 F.3d 883, 917 (9th Cir.2008).
REVERSED and REMANDED.
. The summary judgment record contains evidence of email communication between Horner and Emeldi in mid-2007. In July 2007, Emeldi submitted to Horner a "Dissertation Prospectus” that laid out her research plans. In September 2007, Horner provided feedback on Emeldi’s proposal. Horner’s feedback staled that Emeldi had proposed a "tremendously interesting project,” and had "done brilliantly in [her] efforts,” but also expressed concern that "the reader struggles to find the details that can be examined within a dissertation.”
. Like the Supreme Court in Jackson, "[w]e do not rely on regulations extending Title IX's protection beyond its statutory limits.” 544 U.S. at 178, 125 S.Ct. 1497. Our decision rests on “the statute itself,” not on regulations implementing Title IX. Id.-, see also 34 C.F.R. §§ 100.7(e), 106.71.
. See, e.g., Papelino v. Albany Coll. of Pharm. of Union Univ., 633 F.3d 81, 91-92 (2d Cir.2011) (applying the Title VII framework to a Title IX retaliation claim); Frazier v. Fairhaven Sch. Comm., 276 F.3d 52, 67 (1st Cir.2002) (same); Clinger v. N.M. Highlands Univ., Bd. of Regents, 215 F.3d 1162, 1168 (10th Cir.2000) (same).
. The dissent suggests that Cornwell is irrelevant because, it urges, we there rejected a causal connection. But the idea that a causal connection can be shown by proximity in time between protected activity and adverse action is the well-established rule followed in many cases. E.g., Dawson v. Entek Int'l, 630 F.3d 928, 937 (9th Cir.2011); Bell v. Clackamas Cnty., 341 F.3d 858, 865 (9th Cir.2003); Yartzoff, 809 F.2d at 1376; Miller v. Fairchild Indus., Inc., 797 F.2d 727, 731 (9th Cir.1986). The application of that rule in Cornwell does not negate its application here. In Cornwell, we concluded that Cornwell presented "no evidence raising an inference” that his demotion was caused by his complaint because there was no evidence that the person who demoted him knew about Cornwell’s complaint before his demotion. 439 F.3d at 1035. We also concluded that the "gap” of nearly eight months between Cornwell’s complaint and his termination was "too great to support an inference” that his termination was caused by his complaint. Id. Here, by contrast, Friestad "debriefed” Horner about the content of her discussion with Emeldi, thus "raising an inference” that Horner knew about Emeldi’s complaint and resigned as a result. See id. Also in contrast to Comwell, there was only a short time of a few weeks between Emeldi’s discussion with Friestad and Horner’s resignation without replacement.
. Emeldi's conversation with Friestad took place on or about October 19, 2007. Horner resigned on November 19, 2007. In Homer’s November 20, 2007 email to administrator Mike Bullís, Homer stated that Friestad had contacted him a few weeks beforehand.
. See Wells v. Colo. Dep’t of Transp., 325 F.3d 1205, 1218 (10th Cir.2003) (stating that evidence of pretext is also probative of causation). We note that the email in which Horner praised Emeldi’s dissertation work also contains criticism of her scholarship.
. Emeldi's declaration asserted, “I described one possible cause of that problem as an institutional bias in favor of male doctoral candidates, and a relative lack of support and role models for female candidates. I mentioned the content issues in the [May 2007] Student Advisory Board Memo and my concern about gender inequity of the faculty. I identified the chair of my dissertation committee, Dr. Rob Horner, as being distant and relatively inaccessible to me.”
. The dissent parts company with the majority by concluding that there is no genuine dispute of material fact as to whether the adverse actions suffered by Emeldi were caus*730ally related to her complaints of institutional gender bias or as to whether there was pretext. As we have noted, the dissent’s conclusions ignore traditional rules for applying Rule 56.
Specifically, the dissent complains that Emeldi did not provide other evidence supporting her assertions. An example concerns Emeldi’s complaint that at Horner's graduate student group meetings Emeldi was not on the agenda or if on it her meaningful work was not discussed. These statements are not speculative but based on Emeldi’s personal knowledge and would be admissible at trial. Emeldi had direct percipient knowledge of what happened at the graduate student group meetings she attended. The dissent argues there are no minutes in the record so one cannot verify their substance, and that “there is no proffered testimony of other students or faculty members to give credence to Emeldi’s perceptions that Horner was slighting her (and presumably other women students).’’ But her declaration that she “was publicly and chronically ignored in research team meetings by Rob Horner” generates a genuine dispute of material fact. The dissent's insistence on corroborating testimony of others inserts into the law governing summary judgments a precondition that has never been recognized. See SEC v. Phan, 500 F.3d 895, 910 (9th Cir.2007) (holding that district court erred in disregarding declarations as "uncorroborated and self-serving”); see also 10A Charles Alan Wright & Arthur R. Miller, Federal Practice & Procedure § 2727 (3d ed. 2011) ("[F]acts asserted by the party opposing the motion [for summary judgment], if supported by affidavits or other evidentiary material, are regarded as true.”). Like much of the dissent, this point goes to the weight of Emeldi’s evidence, not to its admissibility and sufficiency to withstand summary judgment.
The dissent characterizes as "speculative” Emeldi’s difficulties gaining a replacement chair of her dissertation committee. But it is not speculative for Emeldi to say that she asked fifteen faculty members who declined for various reasons. A reasonable jury could infer that she was blackballed as a troublemaker because of her claims of institutional gender bias in the Ph.D. program.
. While the University characterizes Emeldi's answer as an admission that she does not know whether Horner’s resignation was motivated by retaliatory animus, we note that Emeldi clearly says, elsewhere in her deposition, that she believes Horner’s conduct was gender-based retaliation. Emeldi might explain to a jury’s satisfaction her anomalous deposition comment in a way that would be consistent with the University's liability. Further, in light of the other evidence that we have noted, it would be incorrect to view Emeldi’s word choice as conclusive against her.
. We need not decide whether, in the context of gender-based retaliation, the state law cause of action is coextensive with Title IX. Because the district court granted summary judgment on Emeldi's state law claim for the same erroneous reasons as her Title IX claim, we need not decide the state statute’s coverage.