On Appeal from the U.S. District Court, W.D. Pa. Judge Marilyn J. Horan, No. 2:20-cv-01952

Before: SHWARTZ, BIBAS, and PHIPPS, Circuit Judges Argued: Mar. 2, 2026; Filed: July 7, 2026

OPINION OF THE COURT

BIBAS, Circuit Judge . Academic debate is a full-contact sport. Debating ideas and questioning arguments can bruise people's egos. But like boxing, academic debate bars low blows. Academics may not harm one another by recklessly disregarding the truth. Administrators may not demote professors for calling out discrimination. And managers may not create or tolerate hostile work environments. A culture that cancels instead of counsels sacrifices persuasion on the altar of power. In doing so, it forgets that the remedy for disfavored speech is more speech, not coerced silence.

Page 2Dr. Norman Wang is an academic cardiologist. He worried that race-based affirmative action in medical schools and hospital residencies discriminates against some minorities in favor of others, could violate the law, and harms the very doctors whom it tries to help. Though most commentators opposed his views, some research supported it. After he published an article expressing those views, his bosses (also mostly doctors) allegedly denounced it as false and racist, fanned the flames on social media, demoted him at the hospital system, and got his article retracted.

Wang sued the university, hospital system, doctors, journal, and publisher for defamation and retaliation. The District Court dismissed some claims on the pleadings and granted summary judgment against Wang on the rest. That was partly wrong. True, Wang has not plausibly alleged state action. But he has plausibly alleged defamation against five defendants, and there are disputed facts about whether his employers retaliated against him for opposing their allegedly illegal discrimination. Thus, we will let many of Wang’s claims go forward.

Pittsburgh enjoys a public-private partnership to train doctors. The University of Pittsburgh (Pitt) is a public university with a medical school. It partners with a private hospital system, the University of Pittsburgh Medical Center (UPMC), and its subsidiary, University of Pittsburgh Physicians (which employs doctors in the UPMC hospital system). Medical Page 3students study at the university and join residents and fellows in some training at the hospital system.

Wang, like many of his colleagues, is both a professor at the medical school and a cardiologist at the hospital system. He has worked at both for almost two decades, serving patients and training residents and fellows. He no longer teaches regularly at the medical school. Nine years ago, he worked his way up to become director of a cardiology fellowship program at the hospital system. As program director, he helped select and train fellows.

Admission to medical schools, residencies, and fellowships is a competitive process based on grades, standardized-test scores, and “holistic” factors. App. 2642. Around 2018, Wang noticed that the American Heart Association, American College of Cardiology, and Accreditation Council for Graduate Medical Education (the hospital’s accreditor for residency and fellowship programs) were pushing another factor: race.

Wang was concerned. Though he is “very much for diversity,” he feared that these diversity programs “look[ed] overtly discriminatory and quota-based” and failed to “treat people as individuals.” App. 2467, 2469, 2490. So in March 2020, he critiqued these practices in the peer-reviewed Journal of the American Heart Association . His seventeen-page article included four tables, four figures, lots of data, and 108 endnotes full of scholarly sources. After explaining how these programs may violate the law and harm members of minority groups, he concluded: “Long-term academic solutions and excellence should not be sacrificed for short-term demographic optics.' App. 2642.

When Wang first published his article in March, it did not make a stir. All that changed in the summer of 2020. On July 29, a doctor emailed the president of the hospital system’s Physicians Services Division to flag and complain about Wang’s conclusions (not his data). The president looped in the medical school’s dean, who agreed, adding that the article was “pseudoscholarly … with handpicked data,” as well as “incredibly offensive and racist.” App. 2657, 2662.

At the same time, another doctor emailed Dr. Kathryn Berlacher, who pinged Dr. Samir Saba. Like Wang, Berlacher and Saba work at both the hospital system and the medical school. Saba is Wang’s superior in both workplaces; Berlacher is his colleague at the medical school but at the hospital gives guidance to him as director of the cardiology fellowship program, an umbrella that oversees Wang’s subspecialty program. The doctor who contacted Berlacher said she 'can't argue with facts” but objected to the “quotations he chose to use, the biased conclusions he made from the data, etc.” App. 2655. Berlacher agreed that, while “we can’t argue w[ith] facts,” “[i]t’s v[ery] clearly offensive and racist.' App. 2654. Saba concurred, expressing his “overwhelming feeling … that [Wang] cannot continue in [his leadership] role.” App. 2673. He looped in Dr. Mark Gladwin, chair of the medical school’s Department of Medicine and head of training at the hospital system, on the email chain. Gladwin lamented that, though Wang's article “was written before the George Floyd events, [the] timing of the print publication could not be worse.” App. 2667. Later, Saba replied that “[e]motions are very high, understandably.” Id. So on July 31, two days after getting those complaints, Page 5Berlacher and Saba met with Wang and fired him from his program directorship at the hospital system.

At that meeting, Wang raised his concerns about racial and ethnic preferences in “our [graduate-medical-education] contracts with the fellows” at the hospital system and said he “just wanted us to follow the law.” App. 961-63. Wang did not explicitly state that the hospital was acting illegally. At some point, though, Saba told him: 'you know what we're trying to do here, Norm,” App. 2490–91. Saba also said “that the views that [Wang had] expressed [we]re not in line with the views and policies that [we] were trying to create,” App. 2498.

Once Wang had been fired, the next question was whether to rebut Wang’s article or just smear it. Gladwin and the medical-school dean urged university and hospital-system leaders to “take[] the high road” by publishing an opposing paper identifying specific errors. App. 2657. They did not.

Instead, Berlacher, Saba, and other doctors launched a socialmedia campaign to discredit Wang's article. Two days after meeting with him, Berlacher tweeted that it “misinterprets data and misquotes people” and “is scientifically invalid and racist.” App. 145. Next, Berlacher authored a tweet on the hospital system’s Cardiology Department account (which she controlled) accusing Wang of using “misquotes, false interpretations, and racist thinking.' App. 145, 233. Saba reposted that tweet. That fusillade fanned “a firestorm on [T]witter, [F]acebook[,] and in all national cardiology circles.” App. 3013. Doctors called for Wang to be cut off from training to “#RetractRacists” and protect students. App. 2783. One fellow emailed Berlacher, describing an “uncomfortable” experience that he had had with anPage 6 “intense grilling” by Wang as evidence that he disliked minorities. App. 2701. Berlacher took this email as confirmation that it was “unsafe” for fellows to work for him and that he needed to be removed from his directorship and all educational roles “immediately.” App. 2700.

So they did. On August 4, two days after their tweets, Berlacher and Saba emailed Wang that “any educational environment in which you partake is inherently unsafe, increasing our learners' risk for undue bias and harm.' App. 2706. They banned him from any role teaching fellows, residents, or medical students. Berlacher had already tweeted to #medtwitter, reassuring other doctors that the students were safe. Over the next three days, the medical school’s dean and Gladwin sent hospital- and school-wide emails, denouncing Wang’s article as hurtful and offensive.

As a result, Wang suffered. He was uncomfortable going to work because long-time colleagues stopped talking to him. He cleaned out his office. Other doctors worried that he might kill himself, and the police even called to see if he was being harassed.

On August 3, Berlacher, Saba, the medical-school dean, and four other hospital-system doctors emailed the Journal of the American Heart Association ’s editor-in-chief, urging the journal to retract Wang’s article. They decried it as “deeply offensive and disturbing,' with 'several misquotes' and 'proven scientific falsehoods.” App. 2708. They attached to the email an eight-page 'point-by-point rebuttal' that they '[we]re Page 7planning to submit[] for publication.' Id. Starting July 31, hospital-system employees also sent back-channel messages to the American Heart Association, the journal’s proprietor.

Two days after Berlacher, Saba, and others emailed the editorin-chief, the Association and the journal’s editors retracted the article. The retraction notice said: “[T]he University of Pittsburgh Medical Center (UPMC)[] has notified the Editor-inChief that the article contains many misconceptions and misquotes and that together those inaccuracies, misstatements, and selective misreading of source materials strip the paper of its scientific validity.” App. 237. The Association also put out a press release that day. The next day, an updated notice added two alleged misquotations as a “sample of the misconceptions and misquotes identified in correspondence from UPMC.' App. 163. “Based on errors such as these, the editors” were retracting the article. App. 164. Later that month, an editor’s note on behalf of the journal, lead-authored by Dr. Marc Simon of the hospital system and medical school, accused “the retracted article … [of] misrepresent[ing] facts” and “evidence.” App. 195.

Yet the following months brewed a backlash to the backlash. Multiple institutions objected that Wang had been disciplined without due process for exercising his academic freedom. The university’s Tenure and Academic Freedom Committee asked the medical-school dean to reinstate Wang. The American Association of University Professors thought the same. The U.S. Department of Education started investigating whether the university’s “campaign of denunciation” violated Title VI and other laws. App. 3223. The medical-school dean said that 'there was a lot of consternation' over potentiallyPage 8 “limiting academic freedom” and “this was clearly a big issue for the faculty senate.” App. 2147. So in late October, under pressure, he told Wang that he had never lost any faculty status or privileges at the university and notified Berlacher, Saba, and Gladwin that they had no authority to bar Wang from university activities.

D. The District Court rejects Wang’s defamation and retaliation claims

Within a week of the first complaint about his article, Wang was demoted, his article was retracted, and he went viral on Twitter for being a dishonest racist. So Wang sued a range of defendants on three theories. First, he claimed defamation by the university, the hospital system, its physician practice, Berlacher, Saba, Simon, the American Heart Association, and the journal’s publisher (Wiley Periodicals). But the District Court dismissed that claim. Even though most of the statements were defamatory statements of fact or opinions based on undisclosed facts, it reasoned, the Association’s and publisher’s statements were true and Wang (a limited-purpose public figure) had failed to plead that the rest of the defendants acted with actual malice.

Second, Wang brought various civil-rights retaliation claims against the university, the hospital system, and its physician practice, as well as Berlacher, Saba, and Gladwin. The District Court dismissed his Title VI claim against the hospital for failing to plead that a primary purpose of the federal funds was to fund employment, and the Title VI and 42 U.S.C. § 1981 claims against the university because any adverse action had come from the hospital system, not the university. Wang also Page 9brought claims under Title VII and the Pennsylvania Human Relations Act (PHRA). The District Court dismissed these claims against the university as to Wang’s statements in his article but not as to his statements in his July 31 meeting with Berlacher and Saba. Later, the court granted summary judgment against Wang on all remaining civil-rights claims, reasoning that Wang had not engaged in any protected activity at the July 31 meeting.

Third, Wang brought a First Amendment retaliation claim under 42 U.S.C. § 1983 against those same defendants. Applying Monell 's municipal-liability standard, the District Court dismissed the claims against the university as failing to allege action under a policy or by policymakers. See generally Monell v. Dep’t of Soc. Servs. , 436 U.S. 658 (1978). It also dismissed the claims against the hospital system and its physician practice, and later granted summary judgment to the individual defendants, reasoning that the hospital and the physician practice were not state actors, that Berlacher and Saba were not acting as employees of the university, and that Gladwin was not involved in Wang’s removal. Wang now appeals all these adverse rulings.

To make out a defamation claim, Wang must plausibly plead and prove that a defendant’s published communication was defamatory and sufficiently linked to him. 42 Pa. Cons. Stat. § 8343(a). We review the District Court's dismissal de novo, taking the complaint's factual allegations in the light most favorable to Wang. Fowler v. UPMC Shadyside , 578 F.3dPage 10 203, 206, 210–11 (3d Cir. 2009) (discussing Ashcroft v. Iqbal , 556 U.S. 662, 678 (2009)).

The District Court erred in dismissing Wang’s whole defamation claim. He has plausibly pleaded that the challenged statements were defamatory. Defendants either did not argue or cannot show that their statements were substantially true. Wang appears to be a limited-purpose public figure who must plead that the defendants spoke with actual malice, but he pleads enough to clear that hurdle as to Berlacher, Saba, the Association, the university, and the hospital system. So we will let much of Wang’s defamation claim proceed.

A statement can be defamatory if “it tends so to harm the reputation of another as to lower him in the estimation of the community or to deter third persons from associating or dealing with him.” Tucker v. Fischbein , 237 F.3d 275, 282 (3d Cir. 2001) (Alito, J.) (citation omitted). Statements of fact can be defamatory. So too can statements of opinion that “give[] rise to the inference that there are undisclosed facts that justify the forming of the opinion.” Meyers v. Certified Guar. Co. , 221 A.3d 662, 670 (Pa. Super. Ct. 2019) (quoting Restatement (Second) of Torts § 566 cmt. b (1977)). To see if a statement implies plausibly defamatory facts, we look at it in context. Graboff v. Colleran Firm , 744 F.3d 128, 136 (3d Cir. 2014).

The District Court rightly found that Berlacher’s and Saba’s tweets were opinions based on undisclosed facts, so they could amount to defamation. They launched a campaign to ostracize him, making his article go viral and generating a backlash.

Page 11Berlacher tweeted from her personal account: “@PittCardiology I’m looking at you. What do we stand for? What do you think of this OPINION piece that misinterprets data and misquotes people? @JAHA_AHA this is scientifically invalid and racist.' App. 145, 235. Next, the hospital-system's official Twitter account @PittCardiology tweeted that Wang’s article used “misquotes, false interpretations and racist thinking to defend a single person’s conclusion.” App. 145, 231. Saba then retweeted that on behalf of one of the hospital system’s institutes.

We cannot dismiss these statements as mere academic jousting. Of course, science is a process of revision, so “tentative scientific conclusions” based on disclosed data and methods are not actionable. Pacira Biosciences, Inc. v. Am. Soc’y of Anesthesiologists, Inc. , 63 F.4th 240, 247 (3d Cir. 2023). But these were not unfalsifiable “disagreements about the reliability of the methodology and data underlying the statements.' Id. Rather, these tweets accused Wang of misquoting and misreading data and sources—both of which are falsifiable claims. So they were more hit job than academic argument.

The Association made defamatory statements, too. It issued two retraction notices (through the publisher) plus a press release. The first notice alleged that the paper contained “many misconceptions and misquotes and that together those inaccuracies, misstatements, and selective misreading of source materials strip the paper of its scientific validity.” App. 143. It gave no examples. The companion press release called Wang’s article “a misrepresentation of the facts,” containing “deliberate misinformation [and] misrepresentation.' App. 144. Later that month, the journal's editorial board published an 'Editor's Page 12Note,” lead-authored by Simon, accusing Wang of “misrepresent[ing] facts” and “evidence.” App. 145.

The 'sting' of the tweets was that Wang's article was sloppy, riddled with errors, scientifically invalid, and motivated by racism. See Oweida v. Trib.-Rev. Publ’g Co. , 599 A.2d 230, 239 (Pa. Super. Ct. 1991); accord Schiavone Constr. Co. v. Time, Inc. , 847 F.2d 1069, 1083 (3d Cir. 1988) (applying analogous New Jersey law). Though “a simple accusation of racism” is not enough in Pennsylvania, these tweets went further and “could be construed to mean that [Wang] was acting in a racist manner” and that his racism led him to distort facts and evidence in connection with his work as an academic and a doctor. MacElree v. Phila. Newspapers, Inc. , 674 A.2d 1050, 1055 (Pa. 1996). In context, they implied that the article was full of errors driven by motivated reasoning. The Association’s notices, press release, and article likewise accuse him of spreading half-truths and lies. Wang plausibly alleges “the kind of harm which has grievously fractured his standing in the community of respectable society.' Tucker v. Phila. Daily News , 848 A.2d 113, 124 (Pa. 2004) (internal quotation marks omitted).

It would have been fine just to say that Wang’s article was wrong, harmful, even dangerous. Those are normal academic opinions. See Pacira , 63 F.4th at 245–47 (reasoning that a scientific article’s claim that a drug was “inferior” or “not superior' was an opinion, a tentative scientific conclusion that could not be defamatory). But these statements went much further, alleging misquotations, misstatements, and outright falsehoods. Wang’s academic integrity is central to his career as a professor, researcher, and doctor. And in Pennsylvania, “any Page 13language which unequivocally, maliciously and falsely imputes to an individual or corporation [a] want of integrity in the conduct of his or its business is actionable.” Cosgrove Studio & Camera Shop, Inc. v. Pane , 182 A.2d 751, 754 (Pa. 1962). When a journal retracts a paper for its “many” misstatements and inaccuracies, a reasonable reader infers that the journal found many falsehoods and errors because someone can objectively disprove the article's statements and claims. See Pacira , 63 F.4th at 246. So such statements can be actionable.

Defamation requires falsity. Defendants may raise truth as an affirmative defense and prove that their statements were true. 42 Pa. Cons. Stat. § 8343(b)(1). Courts may not dismiss on this ground, though, unless a statement’s truth is apparent on the complaint’s face. See Bethel v. Jendoco Constr. Corp. , 570 F.2d 1168, 1174 (3d Cir. 1978). Berlacher, Saba, and Simon forfeited the truth defense at this stage by not raising it below. The District Court found that the Association’s and publisher’s statements were true, but in doing so mistakenly looked to the second retraction notice, which defendants had attached as an exhibit to their motion to dismiss. But the complaint was ambiguous; though the language it quoted appeared in both the first and second retraction notices, the date it gave was that of the first retraction notice. Wang disputed that he was referring to the second retraction notice. Given that ambiguity, the District Court should not have relied on that exhibit, as it was “material beyond the complaint.” Dunn v. Castro , 621 F.3d 1196, 1204 n.6 (9th Cir. 2010).

Page 14Yet even if we read the complaint’s fuzzy language as incorporating the second retraction, making it fair game, the District Court erred by accepting truth as a defense. The defendants identified only two of Wang’s statements as even potentially misleading (and thus truthfully criticized as misquotations), yet neither bears the weight put on it.

The first of the “misquotes” unravels under scrutiny. App. 163. Wang’s article quoted a source stating that Ohio State University “simply made it a priority to rank [underrepresented in medicine] applicants more aggressively than in previous years, thus achieving success in matching them regardless of recruiting efforts, with the implication being that [the school] accepted less competitive applicants in an effort to increase diversity.” Id. (first alteration in original). That quotation was accurate. Yet the Association objected that it was “taken out of context and falsely implies that the Ohio State University (OSU) program ranked racial and ethnic groups underrepresented in medicines with lower academic credentials in order to achieve diversity.' Id. Even so, '[t]he full statement' identified by the Association still supports Wang’s reading: Underrepresented minorities, on average, scored slightly lower on the U.S. Medical Licensing Exam. App. 163. And though the source cautioned that the Licensing Exam scores are not perfect predictors of success and that the observed difference was not statistically significant, it admitted that the exam “scores are considered the most objective variable for comparison.' App. 205. Wang's quotation was fair, accurate, and supported his conclusion that medical-education programs were giving explicit racial preferences that arguably violated civil-rights law.

Page 15The second allegation of misquotation is closer to the mark. Wang hypothesized that doctors from underrepresented minority groups do not improve medical access for underserved populations. He wrote that 'a 1995 California study … demonstrated [that] primary care physicians who were not board certified were 1.6 times more likely to work in rural underserved areas when compared with board-certified counterparts.” App. 420. That was not a quotation. But the substance of the paraphrase was off. The study actually found: “Non-board-certified family and general physicians were 1.6 times more likely than other non-board-certified primary care physicians to locate in rural underserved areas.” App. 210.

No party pressed any other misquotations, misstatements, or misrepresentations. Yet these two 'misquotes' were presented as “[a] sample of the misconceptions and misquotes,” falsely implying that there were many more. App. 163. There were not “many” other misquotations—just one modest error. App. 237. Defendants exaggerated greatly, and the District Court should not have considered documents beyond those in the complaint, so the truth defense cannot defeat Wang’s defamation claim at the dismissal stage.

Wang appears to be a limited-purpose public figure. He is not famous. But he 'voluntarily inject[ed] himself … into a particular public controversy and thereby bec[a]me[] a public Page 16figure for” discussing affirmative action in medicine. Gertz v. Robert Welch, Inc. , 418 U.S. 323, 351 (1974).

As a public figure, Wang must have plausibly pleaded that defendants spoke “with ‘actual malice’—that is, with knowledge that [each statement] was false or with reckless disregard of whether it was false or not.” N.Y. Times Co. v. Sullivan , 376 U.S. 254, 280 (1964). That does not require malice in the ordinary sense of ill will. Because defendants rarely admit lying or recklessness, “objective circumstantial evidence can suffice to demonstrate actual malice.” Schiavone , 847 F.2d at 1090. 1. Wang pleads Berlacher's and Saba's actual malice. As a threshold matter, Saba never raised his actual-malice argument before the District Court. But since the court ruled on that issue anyway, we address it and conclude that Wang properly pleaded enough objective circumstances against both Berlacher and Saba. Recall the complaint’s allegations, which we take as true, along with the documents integral to or explicitly relied on by the complaint. See In re Burlington Coat Factory Sec. Litig. , 114 F.3d 1410, 1426 (3d Cir. 1997). Wang published his peer-reviewed article in March 2020. Four months later, doctors alerted hospital-system and university staff to Wang’s article and views. Panicked, the doctors and staff 'criticized [Wang’s] conclusions.” App. 140. Berlacher and Saba quickly reacted to their criticisms by tweeting and retweeting socialmedia posts accusing him of misinterpreting data, misquoting sources, and racist thinking. App. 140, 141, 145. Yet Wang’s seventeen-page article was supported by 108 endnotes. And Wang alleges that the “citations and quotations in [his] article were clearly attributed, easily verified, and correct.” App. 150.

Page 17Mental states, including malice and intent, may be alleged generally, as long as the plaintiff pleads enough supporting facts to make inferring the mental state plausible. See Fed. R. Civ. P. 9(b); Biro v. Conde Nast , 807 F.3d 541, 544–45 (2d Cir. 2015). From Wang’s allegations, we could draw several plausible inferences. One is that Berlacher and Saba did not adequately investigate their claims that Wang had misquoted his sources. That inference supports dismissal, for a mere “failure to investigate' is not enough to show actual malice. McCafferty v. Newsweek Media Grp., Ltd. , 955 F.3d 352, 359 (3d Cir. 2020).

But it is just as plausible to infer that Berlacher and Saba knew, or at least recklessly disregarded the possibility, that their tweets were flat wrong. Several facts support that inference. First, the things that they said Wang got wrong were, as Wang alleges, easy to verify. Berlacher and Saba could easily have checked Wang's sources, all readily accessible and properly cited, and found that his citations and quotations were almost uniformly accurate. Second, the tweets themselves imply that Berlacher and Saba did in fact check Wang’s sources. As we explained above, some statements of opinion “give rise to the inference that there are undisclosed facts that justify the … opinion.” Meyers , 221 A.3d at 670 (cleaned up). Alleging that someone repeatedly “misquote[d]” his sources implies that the speaker checked them. App. 145.

These facts distinguish Wang's claims from most actualmalice cases, which usually turn on allegations that journalists did a shoddy job. See, e.g. , McCafferty , 955 F.3d at 359; Biro , 807 F.3d at 546. Wang does not just allege that Berlacher and Saba relied on bad sources (as in Biro ) or failed to interview Page 18people familiar with the subject (as in McCafferty ). Nor does he allege that Berlacher and Saba failed to “read every” relevant source on the question. Church of Scientology Int’l v. Daniels , 992 F.2d 1329, 1334 (4th Cir. 1993). Rather, Wang’s defamation claim fits squarely into the space left open by McCafferty . McCafferty had not alleged that the defendant “made up” the allegedly defamatory material or “published facts contrary to information it otherwise knew.” 955 F.3d at 360. Wang, by contrast, alleges facts—especially Berlacher’s and Saba’s own tweets—supporting an inference that they had checked Wang’s citations and quotations and found them (with one minor exception) substantially accurate. Nonetheless, they trashed his article on Twitter, telling the public that Wang had repeatedly misquoted his sources, knowing that was “contrary” to the information they had. Id. A complaint alleging that a speaker directly contradicted information that he had can survive a motion to dismiss. See id. ; Tholen v. Assist Am., Inc. , 970 F.3d 979, 984– 85 (8th Cir. 2020).

We could also infer that Berlacher and Saba intentionally avoided learning the truth. '[P]urposeful avoidance of the truth' shows actual malice. Harte-Hanks Commc'ns, Inc. v. Connaughton , 491 U.S. 657, 692 (1989). When a speaker chooses “not to listen” to an objective source that could verify a story, that choice supports finding actual malice. Id. at 684. Actual malice is even more apparent where facts show that a speaker or publisher decided-before investigating-to run with a particular narrative. See id.

Wang’s article was peer-reviewed. That means it “was vetted pursuant to the usual procedures for review of all articles in JAHA.” App. 143. That peer-review vetting process is designed Page 19to weed out the egregious errors that Berlacher and Saba accused Wang of making. And the allegations plausibly suggest that they “doubted the veracity of [their] story.” Tucker , 237 F.3d at 286–87. Given that, a cite check was the only thing that could have confirmed what they tweeted. If they failed to check the citations in the face of those doubts, then they “purposefully avoided further investigation with the intent to avoid the truth.” Michel v. NYP Holdings, Inc. , 816 F.3d 686, 703 (11th Cir. 2016).

Particularly given those concerns, the speed of Berlacher and Saba’s stampede strongly suggests a reckless rush. Within a couple of days of the first complaints, they demoted Wang. Within two days of that, they lambasted him on social media. No one seems to have paused to do a proper cite check; neither of the alleged errors was a misquotation, and one was not even wrong. Berlacher and Saba’s coordinated campaign of denunciation was thus a stark departure from normal academic practice. The allegations in the complaint suggest that the doctors were reacting to Twitter outrage, not reasoned scholarly inquiry backed by facts. This circumstantial evidence 'permit[s] the conclusion that the defendant[s] in fact entertained serious doubts as to the truth of [their] publication[s].” St. Amant v. Thompson , 390 U.S. 727, 731 (1968).

To be sure, each inference is plausible. And two of them suffice to let the defamation claims against Berlacher and Saba go forward. Because we must draw “all reasonable inferences” in Wang's favor at this stage, even if we thought he would struggle to “ultimately prevail on the merits,” we would let his claim proceed. Connelly v. Lane Const. Corp. , 809 F.3d 780, 790–91 (3d Cir. 2016).

1. 2. Page 20Wang fails to plead Simon' s actual malice. The case against Simon is inadequate. True, Simon, as lead author for the journal's editorial note, denounced Wang's article as 'a misrepresentation of evidence' with 'misrepresented facts.' App. 195. He “deeply regret[ted]” publishing it and was “grateful to the medical community for assisting in correcting this error.” Id. But he largely relied on the allegations of error that they had gotten from the hospital system, Wang’s employer. Plus, Simon (unlike Berlacher and Saba) never accused Wang of misquoting his sources-an accusation that would have been verifiable with a simple cite check. The most he said was that Wang had misrepresented the facts and evidence. Though that statement (an opinion premised on undisclosed facts) can still be defamatory, it is harder to disprove, making actual malice harder to allege. What is more, the complaint contains little suggesting that Simon rushed to retaliate against Wang without verifying his suspicions, at least as compared to Berlacher and Saba. So Wang has failed to plausibly allege that Simon wrote with actual malice.
2. 3. The allegations against the publisher are likewise inadequate . Alleging that a publisher of another’s statement acted with actual malice is hard. Wang must plausibly allege more than that it radically departed from its usual standards. McCafferty , 955 F.3d at 359-60. He does not. The publisher just passed on information from the Association, the hospital system, and its doctors, which it reasonably thought reliable. Again, mere failure to investigate, on its own, is not enough. Tucker , 237 F.3d at 286.
3. 4. Wang pleads the Association's actual malice . The closest call is about the Association. On the one hand, in places itPage 21 simply relayed the hospital’s and doctors’ statements, which it reasonably thought reliable. But the Association went further, and it did so hastily. Within two days of being asked to retract the article, it issued not only a retraction, but also a press release in its own name and voice. It “denounce[d] the views expressed in the article” as “contrary to our organization’s core values and historic commitment to promoting diversity and inclusion.' App. 166. It called the criticism for 'misrepresentation' “right.” Id. And it went much further, accusing him of “deliberate misinformation or misrepresentation'-intent that it could hardly know. App. 167. In its updated retraction statement, the Association did not merely pass along information as a publisher; it expressly accused Wang of lying.

Nor did the Association notify Wang of any errors or give him a chance to rebut those charges before publishing the accusations. That is questionable in the academy. Cf. Bd. of Regents v. Roth , 408 U.S. 564, 573 (1972) (explaining that an academic might be entitled to “notice and an opportunity to be heard” and “to refute the charge” when officials challenge his “reputation, honor, or integrity” (internal quotation marks omitted)). Although a departure from professional standards, standing alone, does not prove actual malice, it may contribute to that inference when coupled with inflammatory accusations and a deficient investigation. Thus, the Association’s allegedly false charges of intentional deception—made hastily, without a thorough review, and without giving Wang a chance to rebut them—are enough to plead actual malice. Cf. McCafferty , 955 F.3d at 359–60.

Once the District Court found that Wang had failed to state a defamation claim against the individual defendants, it reasoned that their employers could not be vicariously liable either. But with the defamation claim against two individuals back in the case, vicarious liability is back on the table.

Employers can be liable for defamatory statements that their employees make within the scope of their employment. To be within the scope of their employment, conduct must be the kind of action that the worker is hired to do, must happen substantially within the authorized work time and space, and must be at least partly motivated to serve the employer. Restatement (Second) of Agency § 228 (1958); Fitzgerald v. McCutcheon , 410 A.2d 1270, 1272 (Pa. Super. Ct. 1979).

Wang claims that Berlacher and Saba made their statements as part of their employment by the university or hospital system. He plausibly alleges that, as professors and leaders of the cardiology department, it was within the scope of their employment to “accuse[] [Wang] of professional incompetence, and academic dishonesty.” App. 149. But none of Wang’s factual allegations names the physician practice or statements made on its behalf, let alone plausibly alleges its actual malice or vicarious liability. So the District Court properly dismissed the defamation claim against the physician practice.

In sum, Wang plausibly alleged that Berlacher and Saba, speaking for the university and hospital system, defamed him, Page 23as did the Association. It is plausible that they either knew they were overclaiming the whole time, or at least buried their heads in the sand to avoid checking his citations and investigating their doubts. That is enough for now. The New York Times v. Sullivan actual-malice standard is supposed to be a shield for free expression, not a sword to silence disfavored speech.

Three different civil-rights laws converge to protect Wang’s calling out what he reasonably believed was illegal activity. At multiple stages, for multiple reasons, the District Court rejected this claim. But text and precedent support it.

No one may retaliate against an employee “because he has opposed any practice made an unlawful employment practice by” Title VII. 42 U.S.C. § 2000e-3(a). Section 1981 creates a cause of action for “retaliation against a person who has complained about a violation of another person’s contract-related 'right.' ' CBOCS W., Inc. v. Humphries , 553 U.S. 442, 445, 452 (2008); 42 U.S.C. § 1981(a). Pennsylvania also bars 'discriminat[ion] … against any individual because [he] has opposed any practice forbidden by' the PHRA. 43 Pa. Cons. Stat. § 955(d). Despite some differences, Title VII, the PHRA, and § 1981 all require the same three elements for a prima facie case of retaliation: protected activity by Wang, adverse employment action against him, and a causal connection between the two. Est. of Oliva ex rel. McHugh v. New Jersey , 604 F.3d 788, 798 & n.14 (3d Cir. 2010); Wilkerson v. New Media Tech. Charter Sch. Inc. , 522 F.3d 315, 318, 320 (3d Cir. 2008).

Once a plaintiff makes out a prima facie case, defendants may give legitimate, non-retaliatory reasons why they took Page 24their adverse actions. McDonnell Douglas Corp. v. Green , 411 U.S. 792, 802 (1973). The plaintiff may then show that those reasons are pretextual. Id. at 804. Reviewing de novo the District Court’s dismissal of Wang’s claims relating to his article and its grant of summary judgment on his claims relating to comments made during the July 31 meeting, we hold that these claims should go forward. Burtch v. Milberg Factors, Inc. , 662 F.3d 212, 220 (3d Cir. 2011); Moore v. City of Phila. , 461 F.3d 331, 340 (3d Cir. 2006), as amended (Sept. 13, 2006).

1. Wang opposed policies that he reasonably thought were unlawful. Wang criticized what he believed was the hospital's and university’s illegal practice of giving racial preferences. He made arguments like the ones that the Supreme Court later accepted in Students for Fair Admissions v. President & Fellows of Harvard College , 600 U.S. 181 (2023). That criticism was protected. Moore , 461 F.3d at 341. Yet the District Court disagreed because he had never called out the university or hospital system by name, and “advocacy, unconnected to employment practices,' is not enough. Curay-Cramer v. Ursuline Acad. , 450 F.3d 130, 136 (3d Cir. 2006). That was mistaken.

To start, it is not clear that an employee must oppose his own employer's illegal actions. All the statute seems to require is 'oppos[ing] any practice made … unlawful.' 42 U.S.C. § 2000e-3(a) (emphasis added). We have not passed on that issue, nodding toward it only in dictum. Curay-Cramer , 450 F.3d at 135. And we need not decide it here because, in context, Wang’s article plausibly opposed the university’s and hospital Page 25system’s admissions and hiring practices. He worked for those institutions. His article opposed “racial and ethnic preferences within graduate medical education training programs” industry-wide and nationwide. App. 415. He wrote that the accrediting bodies' diversity requirements, which the university and hospital system had to follow, may be illegal.

Discovery bore out Wang’s complaint. Gladwin read his article as criticizing affirmative-action practices, '[i]ncluding in graduate medical education” and the “selection of fellows in cardiology” at the hospital system and physician practice. App. 1339–40. In a school-wide email, Gladwin decried it as “antithetical to our values.' App. 2764. Because these comments “identif[ied] [Wang’s] employer and the practice—if not specifically, at least by context,” the District Court erred in finding that Wang’s claims in the article were not protected activity. Curay-Cramer , 450 F.3d at 135.

At the July 31 meeting, Wang testified that he raised his concerns about racial and ethnic preferences in “our [graduate medical education] contracts with the fellows” at the hospital system, and that he “just wanted us to follow the law.” App. 961–63. Saba told him, “you know what we’re trying to do here, Norm.” App. 2490–91. Saba also told him “that the views that [he had] expressed [we]re not in line with the views and policies that they were trying to create.” App. 2498. They understood his criticism perfectly and took it personally. Because there is a genuine dispute of material fact over whether Wang’s comments were protected conduct, the District Court erred in granting summary judgment on Wang's Title VII/PHRA/ § 1981 retaliation claim.

Page 262. The university and hospital system took non-trivial adverse actions against Wang. An adverse action is one that 'well might have dissuaded a reasonable worker from making or supporting a charge of discrimination.' Burlington N. & Santa Fe Ry. Co. v. White , 548 U.S. 53, 68 (2006) (cleaned up). As the District Court recognized, the demotion and ban on contacting fellows, residents, and students are enough for the hospital system. See id. at 68; Gilbert v. Des Moines Area Cmty. Coll ., 495 F.3d 906, 917 (8th Cir. 2007).

Wang also alleged that his bosses and colleagues at the university and hospital had retaliated against his speech by harassing him to “isolate[] [him] in his work environment in hopes that he would resign.” App. 104. Viral tweets called him racist. School-wide emails from leaders acting as both hospital and university employees denounced him and his views as “antithetical to our values and deeply hurtful to many of our URM faculty.' App. 2764.

His superiors at both the university and hospital were aware of the environment. Saba “predict[ed] … that the social environment here will become untenable for Norm [Wang] and this will lead to him leaving UPMC or to his reallocation to a peripheral site.” App. 597. The blowback was so intense that a colleague feared that Wang might kill himself, telling Berlacher: “One thing that crossed my mind … [i]s the issue of his overall suicide risk .… it must not be easy to get all this push back and come back to work.” App. 3051–52 (second elision in original). With all that Wang endured, “a reasonable jury might well conclude that this pattern of harassment might dissuade a reasonable worker from bringing or supporting a charge of discrimination.” Moore , 461 F.3d at 348.

Page 27The multi-front reputation-bashing that impugned Wang’s integrity as an academic is the sort of conduct that 'might [have] impair[ed]” Wang in current or future “employment situations,” and “well might have dissuaded a reasonable worker from making or supporting a charge of” retaliation. Nelson v. Upsala Coll ., 51 F.3d 383, 387 (3d Cir. 1995) (first two quotations); Burlington , 548 U.S. at 68 (third one) (internal quotation marks omitted). Advocating for a paper’s retraction for academic misconduct can harm Wang’s “reputation in the academic community” by “diminish[ing] his academic reputation, thus denying him opportunities” for future employment and publications. Wu v. Thomas , 863 F.2d 1543, 1545 (11th Cir. 1989) (upholding professors’ claims of Title VII retaliatory harassment). An academic’s reputation is his livelihood, so the harm to Wang’s reputation is non-trivial.

The retraction itself could likewise dissuade a reasonable academic from reporting discrimination. In academia, publications and citations are the coin of the realm. They affect salary, promotions, and job opportunities. It is “publish or perish,” so losing a publication is no small matter-not to mention the stigma of a retraction. Theidon v. Harvard Univ. , 948 F.3d 477, 483 n.4 (1st Cir. 2020).

In response, the university (though not the hospital system) says Wang cannot impute its employees’ actions to it. But that confuses § 1983's state-action inquiry with Title VII's. Under Title VII, we routinely impute employees’ harassment to their employers when senior employees know about it. See Huston v. Procter & Gamble Paper Prods. Corp. , 568 F.3d 100, 107 (3d Cir. 2009). And even with dual employers, under Title VII we do not limit liability to the employer who had more control.

Page 28Faush v. Tuesday Morning, Inc. , 808 F.3d 208, 215 (3d Cir. 2015). Wang raised a genuine dispute about whether Pitt “management knew or should have known about the harassment, but' instead of stopping it, fanned the flames. Jensen v. Potter , 435 F.3d 444, 453 (3d Cir. 2006) (Alito, J.), overruled in part and on other grounds by Burlington , 548 U.S. at 60, 64. At the very least, his superiors at the university (Saba and Gladwin) knew of the retaliation and “failed to undertake prompt corrective measures within [their] control.” EEOC v. Glob. Horizons, Inc. , 915 F.3d 631, 641 (9th Cir. 2019) (quoting EEOC guidance). Because employees “sufficiently senior in [the university’s] governing hierarchy, or otherwise in a position of administrative responsibility' over Wang, knew of his harassment and publicly accused him of academic misconduct, there is a genuine dispute of material fact over whether the university is responsible for the adverse actions. Huston , 568 F.3d at 107.

1. 3. There is enough for causation. As the District Court rightly found, there are genuine disputes about causation and possible legitimate, nondiscriminatory reasons for the adverse actions. Title VII (and thus the PHRA and § 1981) requires only but-for causation. Univ. of Tex. Sw. Med. Ctr. v. Nassar , 570 U.S. 338, 360 (2013). That is easily met here, as Wang’s opposition to race-based discrimination led to everything that happened to him in the summer and fall of 2020. Thus, Wang’s Title VII/PHRA claims against the hospital and University proceed. Because Wang's § 1981 claim against Pitt was dismissed before discovery even though it was sufficiently pleaded, we will vacate the District Court’s dismissal of that claim and remand for further proceedings.

Page 29B. The Title VI claim survives against the hospital, but not the university

Title VI outlaws discrimination (and by extension, retaliation) by recipients of federal funds. 42 U.S.C. § 2000d; 28 C.F.R. § 42.107(e) (protecting complainants from retaliation); 34 C.F.R. § 100.7(e) (same). Title VI retaliation claims are governed by the same framework as Title VII, PHRA, and § 1981 claims. See Woodson v. Scott Paper Co. , 109 F.3d 913, 920 (3d Cir. 1997); Hankins v. Temple Univ. (Health Scis. Ctr.) , 829 F.2d 437, 441 (3d Cir. 1987). The university and hospital system both take federal money. To state a claim under Title VI, though, Wang must also plead that “a primary objective of the Federal financial assistance is to provide employment.' 42 U.S.C. § 2000d-3.

The District Court dismissed the Title VI claim against the university for lack of retaliation; the reasoning above applies equally to revive this part of the claim. And it dismissed the Title VI claim against the hospital system for failing to allege the required primary purpose. Yet Wang did allege that the hospital system “receives federal funds to employ residents and fellows in its residency, G[raduate] M[edical] E[ducation], and fellowship programs.” App. 137. Medical residency straddles the line between education and employment. Cf. Doe v. Mercy Cath. Med. Ctr. , 850 F.3d 545, 556, 559 (3d Cir. 2017) (finding it plausible that residency programs are education and residents are employees). At this stage, all Wang need do is allege that the hospital system gets federal money to fund residents. He barely clears that bar.

Page 30But his allegations against the university come up short. He alleges only that it “receives federal funds that, among other things, assist it in providing a medical education for students at' the school. App. 137 ¶ 6. That statement does not allege a primary purpose of funding employment. 42 U.S.C. § 2000d-3. The District Court dismissed Wang’s Title VI claim on other grounds, so Wang will have an opportunity to amend.

Wang also sued the university, hospital system, physicians practice, Berlacher, Saba, and Gladwin under 42 U.S.C. § 1983 for violating his First Amendment rights. The District Court dismissed the claims against the institutional defendants for lack of state action and later granted summary judgment to Berlacher and Saba for the same reason and to Gladwin because he was not involved in demoting Wang. Even if Wang could make out a First Amendment retaliation claim, those rulings were right.

Because the First Amendment does not restrict private parties, only 'state actors' can be liable. Leshko v. Servis , 423 F.3d 337, 339 (3d Cir. 2005). People and business count as state actors when the state encourages or joins in the activity, or is “integrally related” to the actor, so that the “source fairly can be said to be the state.” Id. at 340. What is more, § 1983 has a tighter standard for vicarious liability than Title VII. Title VII follows agency law, often holding employers vicariously liable and imputing supervisors’ actions to them. Moody v. Atl. City Bd. of Educ. , 870 F.3d 206, 215-16 (3d Cir. 2017). Yet for his § 1983 claim, Wang must show that 'the state has exercised Page 31control over the particular conduct that gave rise to the plaintiff's alleged constitutional deprivation.' Kach v. Hose , 589 F.3d 626, 649 (3d Cir. 2009).

Wang has not made that showing. Though affiliated with the public university, the hospital system and physician practice are private corporations. They, not the university, are responsible for their employees' actions. See id. at 646. Their academic affiliation alone is not enough to make them liable; “[a]ction taken by private entities with the mere approval or acquiescence of the State is not state action.” Am. Mfrs. Mut. Ins. Co. v. Sullivan , 526 U.S. 40, 52 (1999).

Wang’s allegations trying to link the hospital system’s and physician practice’s actions to the university are conclusory. True, the people who disciplined him and retaliated against him were employed by both the hospital system and university. Yet dual employment is not enough for state action. See Borrell v. Bloomsburg Univ. , 870 F.3d 154, 160 (3d Cir. 2017). The key question is whether each employee “was wearing his [hospitalsystem] hat or his [university] hat when he decided to” take adverse action. Id. Berlacher, Saba, and Gladwin were all wearing their hospital-system hats. In dismissing Wang from his hospital-system directorship and restricting his access to hospital-system residents and fellows, Berlacher and Saba acted as hospital-system employees. They testified that the decision to remove Wang as director and bar him from contacting residents and fellows was made without consultation with anyone from the university. Berlacher's and Saba's tweets, as well as the tweet from the hospital system's account, neither stated the university’s views nor were made on its behalf. See Lindke v. Freed , 601 U.S. 187, 191 (2024) (treating social-media posts Page 32as state action only when the official who posted had actual authority to speak for the state and purported to do so on social media). And Berlacher did not even have supervisory authority over Wang at the university.

Saba’s purported bar on Wang’s interacting with medical students unaffiliated with the hospital system is a closer case. True, university officials knew about the supposed bar yet did not lift it for two months. Even so, that is not enough for state action. To start, it was effectively a restriction in name only, as Wang never taught classes at the medical school, mostly teaching residents and fellows at the hospital. And Saba never purported to bar him on behalf of the university. Three days later, Berlacher and Saba sent another email from Saba’s hospitalemail address, telling Wang: '[W]e can no longer have you serve in any medical education role in the institution, specifically pertaining to the general cardiology and electrophysiology fellows, while also including medical students and residents.” App. 861. In context, they were wearing their hospitalsystem hats. Saba had no authority to take away his university privileges. He did not consult with the medical-school dean. The dean later reversed Saba's decision, telling Berlacher, Saba, and Gladwin that 'personnel decisions regarding Dr. Wang at [the hospital] are not applicable to the University.” App. 912.

Nor is Gladwin responsible for that supposed bar. Though he emailed the medical school that “[w]e have taken immediate action and removed [Wang] from [his] leadership position,” he said nothing about interacting with medical students and did not in fact remove Wang from any positions. App. 2764. He, Berlacher, and Saba all testified without contradiction that the Page 33decision was made without Gladwin. Gladwin’s after-the-fact defense of Berlacher’s and Saba’s actions is not enough to create state action, as he “played no affirmative part in depriving” Wang of his rights. Rizzo v. Goode , 423 U.S. 362, 377 (1976).

That leaves the public university itself. We have called Pitt an “instrumentalit[y] of the state, both in name and in fact.” Krynicky v. Univ. of Pittsburgh , 742 F.2d 94, 103 (3d Cir. 1984). Thus, we have previously held, “the University of Pittsburgh [is a] state actor[] when [it] make[s] employment decisions.” Molthan v. Temple Univ. of Commw. Sys. of Higher Educ. , 778 F.2d 955, 961 (3d Cir. 1985). Yet the District Court and the parties treat it as a “municipal subdivision” subject to Monell liability. Frazer v. Temple Univ. , 25 F. Supp. 3d 598, 607 (E.D. Pa. 2014); Monell , 436 U.S. at 691.

We need not decide whether that was right. Even if it was, on the face of Wang’s complaint, the university is not plausibly liable under § 1983. That provision does not allow vicarious liability. Parkell v. Danberg , 833 F.3d 313, 330 (3d Cir. 2016). To hold the university liable, Wang must show either that its policy or custom directly caused his constitutional harm or that it knowingly acquiesced in constitutional violations. Id. ; Monell , 436 U.S. at 694. He alleges neither. On policy or custom, Wang does not argue that the university’s policies harmed him, but implies just the opposite—that the employees violated the university’s policies by infringing on his academic freedom. Plus, the crucial decisions were made by Berlacher and Saba; Gladwin and the medical-school dean just relayed and commended others’ actions. And though Wang argues that the either Page 34the university “implemented” or the dean “ratified” Berlacher and Saba’s bar on interacting with medical-school students, he never says how. App. 324–25; City of St. Louis v. Praprotnik , 485 U.S. 112, 124 (1988). Although the medical-school dean knew of the bar, it apparently applied only at the hospital-system and was decided by Wang's superiors there. Neither Gladwin nor the dean controlled Wang’s hospital-system employment, and Wang did not interact with medical students at the university. In short, Wang’s conclusory allegations do not plausibly allege the university’s “participation in or actual knowledge of and acquiescence in the wrongful conduct.' Chavarriaga v. N.J. Dep’t of Corr. , 806 F.3d 210, 222 (3d Cir. 2015). So the District Court properly rejected Wang's First Amendment retaliation claim.

The First Amendment is part “of a profound national commitment to the principle that debate on public issues should be uninhibited, robust, and wide-open.' N.Y. Times v. Sullivan , 376 U.S. at 270. And civil-rights protections “serve as a spur or catalyst to cause employers to self-examine and to selfevaluate their employment practices and to endeavor to eliminate, so far as possible, the last vestiges of discrimination.' McKennon v. Nashville Banner Publ’g Co. , 513 U.S. 352, 358 (1995) (cleaned up).

Taking Wang’s allegations as true, Berlacher, Saba, the Association, the university, and the hospital system did not just exercise their own academic freedom; they defamed and retaliated against Wang for calling out what he reasonably believed was their own discrimination. His defamation claim may proceed Page 35against those five, as may his civil-rights retaliation claims against all appellees. And the District Court should grant him leave to amend his Title VI claim against the university. But his First Amendment claim fails for lack of state action. We will thus AFFIRM IN PART, VACATE IN PART, and REMAND to the District Court for further proceedings consistent with this opinion.

Page 36Counsel for Appellant

Shawn M. Rodgers GOLDSTEIN LAW PARTNERS

John R. Renner [Argued] Michael E. Rosman Michelle A. Scott CENTER FOR INDIVIDUAL RIGHTS

Counsel for Appellee Wiley Periodicals Inc.

Elizabeth Seidlin-Bernstein Michael L. Berry [Argued] BALLARD SPAHR

Counsel for Appellee American Heart Ass’n, Inc.

Howard A. Chajson Steven L. Ettinger [Argued] John C. Conti Jeffrey J. Wetzel DICKIE MCCAMEY & CHILCOTE

Counsel for Appellee Marc Simon

Bryon M. Chowka Robert J. Ridge [Argued] CLARK HILL

Counsel for Appellee University of Pittsburgh

Jeremy D. Engle MARCUS & SHAPIRA

Jeremy W. Brinster Brian Boynton [Argued] Bruce M. Berman Molly M. Jennings WILMER CUTLER PICKERING HALE & DORR

Counsel for Appellees University of Pittsburgh Medical Center, University of Pittsburgh Physicians, Samir Saba, Mark Gladwin, and Kathryn Berlacher

Emilie R. Hammerstein Taylor Brailey Page 37Terrence H. Murphy Theodore A. Schroeder [Argued] LITTLER MENDELSONPage 38 SHWARTZ, Circuit Judge concurring in part, dissenting in part.

I agree with the disposition my colleagues reached on all claims except Wang’s defamation claims. I conclude the defamation claims fail for two reasons. First, the challenged statements are non-actionable opinions of academic research and scientific methodology based on disclosed facts. Second, even if those statements were actionable, Wang is a limited purpose public figure, but his complaint does not allege facts from which to infer that the challenged statements were made with actual malice. Accordingly, I would affirm the District Court’s order dismissing Wang’s defamation claims.

Wang is a cardiologist affiliated with the University of Pittsburgh Medical Center ('UPMC') and the University of Page 39Pittsburgh School of Medicine ('Pitt').² At UPMC, Wang serves as a doctor and previously directed the fellowship program in clinical cardiac electrophysiology. At Pitt, Wang is a faculty member.

Wang wrote an article entitled 'Diversity, Inclusion, and Equity: Evolution of Race and Ethnicity Considerations for the Cardiology Workforce in the United States of America from 1969 to 2019” (the “Article”). App. 139 (FAC ¶ 20), 17894. He submitted the Article to the Journal of the American Heart Association ('JAHA'). JAHA is published by the American Heart Association (“AHA”) and Wiley Periodicals, Inc.³ ('Wiley'). In March 2020, AHA and Wiley published the Article in JAHA after the 'normal review and vetting process.” App. 139-40 (FAC ¶ 21).

The Article “trac[ed] the history of the use of race and ethnicity as factors in determining admission into medical schools, residency programs, and fellowships.' App. 139 (FAC ¶ 20). It “asserted that the medical profession had not been successful in reaching its goals of increasing the percentages of underrepresented races and ethnicities in the medical profession generally, and cardiology in particular.' App. 139 (FAC ¶ 20). The Article argued that, “to achieve those goals,' 'programs . . . applied different standards to applications by members of underrepresented races and ethnicities and raised questions about the legality, Page 40effectiveness, and wisdom of doing so.” App. 139 (FAC ¶ 20). The Article contained 108 endnotes with embedded hyperlinks.

In late July 2020, “various individuals on social media” criticized the Article, App. 140 (FAC ¶ 22), including Dr. Samir Saba and Dr. Katheryn Berlacher, who are both affiliated with UPMC and Pitt. On August 2, 2020, Saba retweeted a Twitter post about the Article and wrote the Article 'uses misquotes, false interpretations, and racist thinking to defend a single person’s conclusion[s]. We are outraged that @JAHA_AHA published this shameful and infuriating piece.” App. 145 (FAC ¶ 43) (quoting App. 233). Later that day, Berlacher retweeted a Twitter post about the Article and added that it 'misinterprets data and misquotes people' and is 'scientifically invalid and racist.' App. 145 (FAC ¶ 44) (quoting App. 231, 235).⁴

On August 5, 2020, AHA and Wiley retracted the Article “[d]espite the fact that [it] was vetted pursuant to the usual procedures for review of all articles in JAHA.” App. 143 (FAC ¶ 39). The Article remains available on JAHA’s website but contains a label noting its retracted status. In connection with the retraction, Wang contends that AHA and Wiley made three defamatory statements against him. First, AHA and Page 41Wiley issued a statement on August 5 that accompanied the Article’s retraction (the “Retraction Statement”). It stated, in relevant part:

The author’s institution, the University of Pittsburgh Medical Center (UPMC), has notified the Editor-in-Chief that the article contains many misconceptions and misquotes and that together those inaccuracies, misstatements, and selective misreading of source materials strip the paper of its scientific validity. . . . The Editor-in-Chief deeply regrets publishing the article and offers his apologies.

App. 143 (FAC ¶ 39), 237. The Retraction Statement did not provide examples of misconceptions or misquotes from the Article but noted JAHA would 'be publishing a detailed rebuttal.” App. 237. The next day, AHA and Wiley posted an updated Retraction Statement (the “Updated Retraction Page 42Statement')⁵ with virtually the same paragraph 6 as well as two specific “misconceptions and misquotes identified in correspondence from UPMC' to serve as a 'sample' of the Article’s errors.⁷ App. 163-64.

Page 43Second, AHA issued a statement (the “AHA Statement”), providing, in relevant part:

The [Article] has rightfully drawn criticism for its misrepresentations and conclusions. As an organization focused on the relentless pursuit of longer, healthier lives for everyone everywhere, the [AHA] denounces the views expressed in the article and regrets its role in enabling those views to be promoted. Those views are a misrepresentation of the facts and are contrary to our organization’s core values and historic commitment to promoting diversity and inclusion in medicine and science. . . . While [JAHA] and the other AHA scientific journals are editorially independent of the Association,

Page 44we take our responsibility to ensure factual accuracy seriously. The independent editors of JAHA and the [AHA] are reviewing the journal’s peer-review and publication processes to ensure future submissions containing deliberate misinformation or misrepresentation are never published. The journal can and will do better.

App. 166-67; see also App. 144 (FAC ¶ 39).

Third, JAHA published another article entitled “A Path Forward,” which “stated that [Wang’s] institution had requested the article be retracted ‘[based on] specific scientific errors as well as misleading and incomplete quotations.’” App. 144 (FAC ¶ 39) (quoting App. 170-71).

Separately, Wang also identified as defamatory an article published by JAHA and authored by Dr. Marc Simon, a UPMC and Pitt colleague. This piece, entitled 'Equity, Diversity, and Inclusiveness in Cardiovascular Healthcare and Medicine,” (the “Simon Article”), App. 145 (FAC ¶¶ 45), 19596, claimed that the Article “misrepresented facts” and “evidence” “[in its efforts] to argue against affirmative action in the field of cardiology,” App. 145 (FAC ¶ 45) (quoting App. 195). The Simon Article’s sources were set forth in endnotes with hyperlinks.

Wang filed defamation claims against Defendants, alleging that: (1) Saba and Berlacher defamed him in their August 2 tweets, (2) Simon defamed him in the Simon Article, (3) UPMC and Pitt defamed him via their employees' statements, (4) AHA defamed him in the Retraction Statement, the AHA Statement, and A Path Forward, and (5) Wiley Page 45defamed him in the Retraction Statement and A Path Forward.

In assessing a defamation claim, a “court must determine: (1) whether the defendants have harmed the plaintiff’s reputation within the meaning of state law; and (2) if so, whether the First Amendment nevertheless precludes recovery.” Marcone v. Penthouse Int’l Mag. for Men, 754 F.2d 1072, 1077 (3d Cir. 1985) (citations and quotations omitted).

A

To succeed on a defamation claim under Pennsylvania law, a plaintiff must show, among other things, “[t]he defamatory character of the communication” at issue. 42 Pa.

Page 46Cons. Stat. § 8343(a)(1).⁹ 'A communication is defamatory if it tends so to harm the reputation of another as to lower him in the estimation of the community or to deter third persons from associating or dealing with him.” Birl v. Philadelphia Elec. Co., 167 A.2d 472, 475 (Pa. 1960) (citations omitted); Graboff v. Colleran Firm, 744 F.3d 128, 136 (3d Cir. 2014). Statements of opinion, as opposed to statements of fact, may be viewed as having such an effect if they are based on undisclosed facts. Remick v. Manfredy, 238 F.3d 248, 261 (3d Cir. 2001); Parano v. O’Connor, 641 A.2d 607, 609 (Pa. Super. Ct. 1994) (holding that “expressions of . . . opinion” were “not actionable unless they imply undisclosed, false and defamatory facts”).

In the course of academic debate, criticisms and characterizations of scientific methodology and scholarly analysis are better understood as statements of opinion rather than fact. See ONY, Inc. v. Cornerstone Therapeutics, Inc., 720 F.3d 490, 497 (2d Cir. 2013) (“[W]hile statements about contested and contestable scientific hypotheses constitute assertions about the world that are in principle matters of verifiable ‘fact,’ for purposes of the First Amendment and the laws relating to fair competition and defamation, they are more closely akin to matters of opinion, and are so understood by the relevant scientific communities.”); Pacira BioSciences, Inc. v.

Page 47Am. Soc’y of Anesthesiologists, Inc., 63 F.4th 240, 246 (3d Cir. 2023) (determining that “tentative scientific conclusions” were opinions in libel claim); Saad v. Am. Diabetes Ass’n, 123 F. Supp. 3d 175, 178 (D. Mass. 2015) (holding that medical journal's expression of concern about 'the reliability of the data” in professor’s articles was protected opinion).

The Majority holds that accusations of “misquoting and misreading data and sources” are “falsifiable,” thereby rendering them 'more hit job than academic argument.' Majority at 11. Setting aside the Majority’s characterization of these statements as a “hit job,” one academic accusing another of 'misreading' data is not actionable because it reflects an opinion about the other’s work based on disclosed facts. See Pacira, 63 F.4th at 247 (“[M]ere disputes about the reliability of a scientific study’s disclosed methodology cannot create an actionable falsehood . . . as such disputes do not address whether the statements themselves are verifiable.”); ONY, 720 F.3d at 497 (declining to recognize a cause of action where “plaintiff allege[d] that the inferences drawn from . . . data were the wrong ones”).

That is what happened here. First, AHA and Wiley criticized the Article for 'contain[ing] many misconceptions and misquotes and that together those inaccuracies, misstatements, and selective misreading of source materials strip the paper of its scientific validity.' App. 143, 163-65, 237.¹⁰ Likewise, the Simon Article stated that the Article “misrepresented facts in its effort[] to argue against affirmative Page 48action in the field of cardiology” and included the “misrepresentation of evidence.” App. 145, 195-96.

In making these claims, each author explicitly referred to the Article-the subject of their criticism-and discussed the bases for their opinions, namely the sources listed in the Article’s endnotes. Indeed, the Updated Retraction Statement even identified specific errors as a 'sample of [Wang's] misconceptions and misquotes.” App. 163.

Thus, JAHA's readers-medical professionals and academics equipped to evaluate the challenged statements and the sources on which they were based—had the opportunity and information to assess for themselves the disputed conclusions. See Pacira, 63 F.4th at 249 (determining statements were based on disclosed facts where “[academic] journal’s readers were provided the basis for the statements, have the expertise to assess their merits based on the disclosed data and methodology, and thus are equipped to evaluate the opinions the authors reached”); Cheng v. Neumann, 51 F.4th 438, 447 (1st Cir. 2022) (concluding an opinion was based on disclosed facts and thus “not actionable because [it] provide[d] a link to the source material, which, in the context of the [controversy], enable[d] readers to draw their own conclusions ‘based on facts accessible to everyone’” (citations omitted)).

The Majority insists “[i]t would have been just fine to say that Wang’s article was wrong, harmful, or even dangerous” because “[t]hose are normal academic opinions.” Majority at 12. That is exactly what the speakers did here. In any event, whether “normal” or not, courts should be cautious when weighing in on academic debates because doing so chills scientific progress and the rigorous discourse required to bring Page 49it about. See Joseph v. Springer Nature, No. 20 civ. 4672 (JPC), 2021 WL 1372952, at *7 (S.D.N.Y. Apr. 12, 2021) (declining to consider a retraction statement as actionable where the underlying dispute concerned 'a critique of a scientific work” in a contested area of research “that the Court has no business solving”), aff’d sub nom. Joseph v. Springer Nature Am. Inc., No. 21-959-cv, 2021 WL 6105369 (2d Cir. Dec. 21, 2021) (summary order). As the Court of Appeals for the Seventh Circuit held, 'scientific controversies must be settled by the methods of science rather than by the methods of litigation. . . . More papers, more discussion, better data, and more satisfactory models-not larger awards of damagesmark the path toward superior understanding of the world around us.” Underwager v. Salter, 22 F.3d 730, 736 (7th Cir. 1994); see also Pacira, 63 F.4th at 247-48 (holding that a defendant’s characterization of academic work as employing a 'flawed method' was not actionable because 'data and methodology may be the basis of future scholarly debate, but they do not form the basis for trade libel under New Jersey law” as concluding 'otherwise would risk 'chilling' the natural development of scientific research and discourse”).¹¹

Page 50Second, Saba and Berlacher's tweets are also nonactionable opinions about an academic work based on disclosed facts because they explicitly refer to the Article they critiqued and the reasons for their views. See Veno v. Meredith, 515 A.2d 571, 573-75 (Pa. Super. Ct. 1986) (concluding that an article constituted non-actionable opinion because it “merely stated an opinion” on another article and the defendant author did not 'imply that he was privy to some additional facts, beyond [the article] which supported his opinion”). Furthermore, although their characterization of the Article as “racist” may be inflammatory, it is not actionable under Pennsylvania law. See McCafferty v. Newsweek Media Grp., Ltd., 955 F.3d 352, 358 (3d Cir. 2020) (“In Pennsylvania, ‘a simple accusation of racism’ is not enough . . . Rather, the accusation must imply more, by for instance suggesting that the accused has personally broken the law to ‘act[ ] in a racist manner.’” (second alteration in original) (quoting MacElree v. Philadelphia Newspapers, Inc., 674 A.2d 1050, 1055 (Pa. 1996)); 12 see also McAndrew v. Scranton Republican Publ'g Page 51Co., 72 A.2d 780, 783 (Pa. 1950) (“It would doubtless be very annoying for a man to be charged with bigotry of any kind, yet it has been held that for a man to be charged with such bigotry is not defamation.”); Rybas v. Wapner, 457 A.2d 108, 110 (Pa. Super. Ct. 1983) ('A publication which charges that an individual is actuated by unpleasant or undesirable prejudice may offend his sensitivities, but is not thereby libelous.”).

Because Defendant’s statements are nonactionable opinions, I would affirm the order dismissing Wang’s defamation claims. 13Page 52 ethnicity for employment reveals a lack of knowledge regarding legal permissibility and fellow status.” App. 414. The Retraction Statement highlighted that Wang omitted the sentence immediately following the quoted excerpt, which provided:

Although the measurement of an applicant's achievement reflected in his or her application and, by extrapolation, the applicant's potential for being a successful fellow, cannot be fairly quantified, USMLE score comparison shows no significant difference between URM and nonURM fellows matched to the OSU program.

App. 163. In other words, the Article suggested that the original work stood for the proposition that less qualified candidates were accepted into the OSU program in the name of diversity, but the original work stated that there was no significant difference in test scores (i.e., qualification) between candidates from different racial groups.

Second, in a section of the Article relating to the qualification of URM, Wang suggested another researcher's hypothesis that 'working in an underserved area may be attributable to inability to secure a job in other areas because of low professional qualifications” was bolstered by:

a 1995 California study that demonstrated primary care physicians who were not board certified were 1.6 times more likely to work inPage 53 App. 164. In other words, the original source compared rural employment rates for non-board-certified family and general physicians for non-board-certified primary care physicians, but Wang described it as comparing rural employment rates between board-certified and non-board-certified primary care physicians. While Wang did not include quotation marks, he (incorrectly) described the results of a study and, at the very least, misrepresented what it demonstrated.

Third, the Retraction Statement’s assertion of “many” errors does not change the truth-as-a-defense analysis. The use of the word “many” could change the meaning of a statement such that it is defamatory-for example, one alleged act of misconduct may be less damaging than fifty allegations, depending on the severity of that single act. However, 'defamation case law affords leeway in the reporting of numbers if the discrepancy does not alter the gist of what is being reported and does not produce a different effect in the mind of the reader.” Croce v. N.Y. Times Co., 345 F. Supp.

Page 54Even if any of Defendants’ statements were actionable under state law, Wang’s claims are still prohibited under the First Amendment’s public figure doctrine. Like my colleagues, I conclude that Wang is a limited purpose public figure and failed to allege actual malice as to Defendants Wiley

Thus, if the alleged defamatory statements are viewed as factual rather than opinions, then truth as a defense would apply to all of Wang's defamation claims. See Budhun v. Reading Hosp. & Med. Ctr., 765 F.3d 245, 259 (3d Cir. 2014) (“While ordinarily a party may not raise affirmative defenses at the motion to dismiss stage, it may do so if the defense is apparent on the face of the complaint.' (citation omitted)).

Page 55and Simon, 14 but I part company with their conclusion that the pleadings are sufficient to establish AHA, Saba, and Berlacher’s allegedly defamatory statements were made with actual malice.

As a limited purpose public figure, Wang had to plead that the allegedly defamatory statements were made with 'actual malice.' Marcone, 754 F.2d at 1081-82. ''Actual malice' is a term of art that does not connote ill will or improper motivation.” McCafferty, 955 F.3d at 359; Masson v. New Yorker Mag., Inc., 501 U.S. 496, 510 (1991) (“Actual malice under the New York Times standard should not be confused with the concept of malice as an evil intent or a motive arising from spite or ill will.”). Instead, “[t]he actual malice standard requires that the defendant make the defamatory remarks with ‘knowledge that [the statement] was false or with reckless disregard of whether it was false or not.’” McDowell v. Paiewonsky, 769 F.2d 942, 951 (3d Cir. 1985) (alteration in original) (quoting N.Y. Times Co. v. Sullivan,Page 56 376 U.S. 254, 280 (1964)). “[R]eckless disregard for the truth means that the defendant ‘in fact entertained serious doubts as to the truth' of the statement,' id. (quoting St. Amant v. Thompson , 390 U.S. 727, 731 (1968)), “or that the defendant had a ‘subjective awareness of probable falsity,’” id. (quoting Gertz v. Robert Welch, Inc., 418 U.S. 323, 335 n.6 (1974)). This 'standard is a subjective one, based on the defendant’s actual state of mind,” though a court may rely on “objective circumstantial evidence” to determine actual malice. Schiavone Constr. Co. v. Time, Inc., 847 F.2d 1069, 1089-90 (3d Cir. 1988); see also Kendall v. Daily News Publ’g Co., 716 F.3d 82, 93 (3d Cir. 2013) (“The subjective nature of this inquiry requires that there be some evidence showing, directly or circumstantially, that the defendants themselves understood the potential defamatory meaning of their statement.”).

Actual malice exists where a 'publisher's allegations are so inherently improbable that only a reckless man would have put them in circulation,' or 'where there are obvious reasons to doubt the veracity of the informant or the accuracy of his reports.” St. Amant, 390 U.S. at 732; see also Schiavone, 847 F.2d at 1090 ('Where the defendant finds internal inconsistencies or apparently reliable information that contradicts its libelous assertions, but nevertheless publishes those statements anyway, the New York Times actual malice test can be met.”). For example, in McCafferty we reasoned that actual malice could be established where a publisher “made up [a commentator’s] statements, used them to accuse him of breaking the law when it knew he was innocent, or published facts contrary to information it otherwise knew.' 955 F.3d at 360. In Tucker v. Fischbein, we recognized actual malice where an attorney publicly mischaracterized a filed Page 57complaint and only later attempted to explain his false statements by claiming that he had failed to read the complaint before speaking on it. 237 F.3d 275, 285 (3d Cir. 2001). In Schiavone, we concluded the defendant acted with actual malice where reliable, independent sources contradicted the veracity of a defamatory claim, and the publisher deliberately omitted 'exculpatory' information and 'add[ed] editorial comment to draw attention” from readers. 847 F.2d at 109192.

Absent circumstances that indicate a publisher was or should have been on notice of the falsity of their statements, we have generally declined to conclude that defamatory statements were made with actual malice. See, e.g., Marcone, 754 F.2d at 1091 (concluding no actual malice was shown where false statements were not based on facts so “inherently improbable that [publisher] should have been put on notice of the probability of falseness”).

Applying this framework to Wang’s factual allegations, and assuming that Defendants' statements are false and otherwise defamatory, Wang's allegations do not show that any of the statements were made with knowledge or reckless disregard of their falsity. See Marcone, 754 F.2d at 1091 ('Although the lack of truthfulness of a statement is a prerequisite to liability, mere falsity, without more, is generally not sufficient to establish actual malice.'). Because I agree Page 58with the Majority that Wang failed to establish actual malice as to Wiley and Simon, I will not discuss those claims.¹⁵

As to AHA, the Majority holds that Wang alleged actual malice by relying on: (1) assertions in the AHA Statement, and (2) AHA's failure to give Wang an opportunity to respond. Majority at 20-21. Neither establishes actual malice.

First, the AHA Statement-'denounc[ing]' Wang's views and claiming he engaged in “deliberate misinformation or misrepresentation,' App. 166-67-was strongly worded, but not sufficient to show actual malice. Passionate opposition to a speaker’s views does not constitute actual malice, even if motivated by 'spite or ill will.' Masson, 501 U.S. at 510. Moreover, AHA's claim that Wang relied on 'deliberate' falsehoods does not show that it acted with knowledge of falsity or reckless disregard for truth. Rather, it is reasonable for AHA to have surmised that if Wang misrepresented sources Page 59in making his argument, he did so “deliberate[ly]” to bolster his conclusions.¹⁶

Second, that AHA did not to provide Wang with notice of the retraction or an opportunity to rebut the claims against him does not constitute actual malice. Indeed, we have previously determined no actual malice existed where a publisher failed to offer the subject of a news report the opportunity to respond. McCafferty, 955 F.3d at 359. Even if Wang had alleged that giving an author notice and an opportunity to respond was standard practice in the industry (which he did not), “an extreme departure from professional Page 60standards, without more, will not support a finding of actual malice.” Tucker, 237 F.3d at 286 (citing Harte-Hanks Commc'ns, Inc. v. Connaughton, 491 U.S. 657, 665 (1989)).¹⁷

Wang also failed to plead actual malice as to Saba and Berlacher. Wang contends that actual malice was shown by: (1) the Article's easily verifiable assertions, (2) Saba and Berlacher's strong opposition to Wang's position on racebased admissions, and (3) their orchestrated “attack campaign” against Wang, App. 143 (FAC ¶ 38). Even if true, none constitute actual malice.

First, Saba and Berlacher’s alleged failure to review all of the sources in Wang’s article does not demonstrate actual malice. See Marcone, 754 F.2d at 1089 (“Failure to investigate, without more, does not demonstrate actual malice.”); see also Harte-Hanks, 491 U.S. at 688 (“[F]ailure to investigate before publishing, even when a reasonably prudent person would have done so, is not sufficient to establish reckless disregard.”). “[T]he purposeful avoidance of the truth is” distinguishable from a mere failure to investigate, HarteHanks, 491 U.S. at 692, but this case is not one in which the publisher of the statement failed to interview a key witness or necessarily did not consult other objective evidence. Id. at 69293. Furthermore, the complaint does not plead facts to suggest Page 61Saba and Berlacher purposefully avoided learning whether their claims were false.

The Majority insists—without citation—that Berlacher and Saba must have checked all of Wang's sources and determined they were “substantially accurate,” but still decided to “trash[]” the Article. Majority at 18. This inference is not supported by the complaint. There is no allegation that Saba and Berlacher checked each source. Rather, the allegations show that Saba and Berlacher claimed some sources Wang quoted or relied on were not accurately conveyed and their mere failure to fully investigate does not constitute actual malice. Marcone, 754 F.2d at 1089.

Second, Wang argues that he has pleaded actual malice by pointing at Saba and Berlacher’s opposition to his views and their characterization of the Article as 'scientifically invalid and racist.” App. 145 (FAC ¶ 44). That does not show they made their statements with knowledge or reckless disregard of their falsity. See Masson, 501 U.S. at 510 ('Actual malice under the New York Times standard should not be confused with the concept of malice as an evil intent or a motive arising from spite or ill will.'). Although it 'cannot be said that evidence concerning motive or care never bears any relation to the actual malice inquiry,” Harte-Hanks, 491 U.S. at 668, the vitriol of a challenged statement alone does not show that it was made with actual malice for defamation purposes. In a similar vein, the Majority adds its view that Saba and Berlacher's criticism was 'a stark departure from normal academic practice.” Majority at 19. Even if this were true, as explained above, 'an extreme departure from professional standards, without more, will not support a finding of actual Page 62malice.' Tucker, 237 F.3d at 286 (citing Harte-Hanks, 491 U.S. at 665).

Third, even if Saba and Berlacher attempted to induce others to criticize Wang's work, that conduct would not establish actual malice because it does not suggest they knew or reasonably should have known their claims were false. See McCafferty, 955 F.3d at 359 (“‘Actual malice’ is a term of art that does not connote ill will or improper motivation.”).

Because Wang’s complaint does not establish that Saba and Berlacher acted with actual malice, the public figure doctrine precludes recovery even if Wang’s defamation claims against them were otherwise actionable.¹⁸

I agree with the Majority that our First Amendment standard “is supposed to be a shield for free expression, not a sword to silence disfavored speech,” Majority at 23, and that is the very reason why I would dismiss these defamation claims. Thus, I respectfully dissent from Part II of the Majority's opinion.