The finding is written in the interview. The reversal just makes it official.

What the court held. Writing for the Seventh Circuit, then-Judge Amy Coney Barrett held that John Doe plausibly alleged Purdue's process was fundamentally unfair and plausibly alleged discrimination on the basis of sex. The court also held that withholding the investigative evidence from the accused was "itself sufficient to render the process fundamentally unfair."

The failure. The decision-makers credited the accuser without ever hearing from her. She did not appear before the panel and never submitted her own statement; her account reached the panel through a letter written by a campus advocacy office. Two of the three panel members "candidly stated that they had not read the investigative report." The panel refused to hear John's witnesses, including a roommate who was present at the alleged assault. In the court's words: "it is particularly concerning that Sermersheim and the committee concluded that Jane was the more credible witness, in fact, that she was credible at all, without ever speaking to her in person."

The prevention note. A credibility finding is an output of interviewing, not a substitute for it. A trained investigator interviews every party whose credibility the finding will rest on, documents the basis for each credibility assessment, tests the account against available witnesses instead of screening them out, and gives the respondent the evidence the decision will rely on.

What the court held. The Sixth Circuit held that when a public university's determination turns on credibility, "that hearing must include an opportunity for cross-examination," and that "if a public university has to choose between competing narratives to resolve a case, the university must give the accused student or his agent an opportunity to cross-examine the accuser and adverse witnesses in the presence of a neutral fact-finder."

The failure. After 23 witness interviews, the University of Michigan's investigator found the evidence in equipoise and recommended closing the case. A three-member appeals board reversed that recommendation "on a cold record," without interviewing anyone, and found the claimant more credible. The accused "never received an opportunity to cross-examine Roe or her witnesses, not before the investigator, and not before the Board." The board also discounted the respondent's witnesses because they were his fraternity brothers while applying no equivalent scrutiny to the claimant's sorority-sister witnesses. The court: "written statements cannot substitute for cross-examination."

The prevention note. Credibility cannot be reassessed from paper by people who interviewed no one. When a finding will turn on whose account is believed, the adjudicator must hear the witnesses, and the questioning of both sides' witnesses must be conducted with equal rigor. Asymmetric skepticism is the pattern courts flag as evidence of bias.

What the court held. The Third Circuit held that a private university's promise of "fair" and "equitable" treatment requires "at least a real, live, and adversarial hearing and the opportunity for the accused student or his or her representative to cross-examine witnesses, including his or her accusers," and that Doe plausibly alleged the university broke that promise. The court adopted the Purdue pleading standard for the Title IX claim.

The failure. The single-investigator model as implemented: one outside attorney interviewed the parties and witnesses and alone determined culpability, in a case that hinged on credibility. The student was expelled without any hearing. In the court's words: "Doe plausibly alleges that USciences deprived him of fairness because he never received a chance to cross-examine witnesses or any sort of real, live, and adversarial hearing."

The prevention note. A process in which the same person interviews, evaluates, and decides leaves every interview error invisible and uncorrectable. Even where a hearing model is used, the interviews feeding it must be conducted so that a later adversarial test does not expose what the investigator failed to ask.

What the court held. The Sixth Circuit held the student "amply stated a claim for sex discrimination," relying on "clear procedural irregularities," federal enforcement pressure, a 100 percent responsibility rate at hearings that academic year, and a decision the court called "arguably inexplicable" on its merits: "when the degree of doubt passes from 'articulable' to grave, the merits of the decision itself, as a matter of common sense, can support an inference of sex bias."

The failure. The hearing panel never addressed a direct contradiction between what the claimant told the investigator and what she said at the hearing on the central factual question, a contradiction the investigator himself flagged. The court called that silence "remarkable" "in a proceeding in which the credibility of accuser and accused were paramount." The incapacitation finding rested solely on the claimant's statement "I am not sober right now," with no apparent basis under the policy's own definition. The appeals officer did not engage with new impeachment evidence from the claimant's former best friend.

The prevention note. Inconsistencies between a party's interview account and later testimony are not landmines to step around; they are the record. A trained investigator documents them, resolves them through follow-up where possible, and ensures the finding addresses them expressly. A finding that ignores a flagged contradiction invites exactly this outcome.

What the court held. The Second Circuit held that Title IX discipline claims borrow the Title VII temporary presumption, so a complaint survives dismissal "if it pleads specific facts that support a minimal plausible inference of such discrimination." The court also observed: "When the evidence substantially favors one party's version of a disputed matter, but an evaluator forms a conclusion in favor of the other side (without an apparent reason based in the evidence), it is plausible to infer (although by no means necessarily correct) that the evaluator has been influenced by bias."

The failure (as alleged). The investigator did not seek out the witnesses the respondent identified: "She did not ask him about the witnesses, interview them, or follow up with them in any way." Her questioning of him was described as "akin to cross-examination calculated to elicit a confession," and he was not advised of rights and resources provided to the claimant. The panel declined to put his submitted questions to the claimant as "irrelevant." All of this occurred amid intense public criticism of the university's handling of assault complaints.

The prevention note. An investigation is defined by the witnesses it pursues, not just the ones it hears. Declining to interview identified witnesses, and interviewing one party as a subject while treating the other as settled, is the single most quotable investigative failure a plaintiff's lawyer can put in a complaint.

What the court held. In a case that was exclusively one account against another, where the claimant did not appear at the hearing and the finding rested entirely on her statements to investigators, the Sixth Circuit held due process required some opportunity for circumscribed cross-examination: written questions through the panel, with remote appearance sufficing. The holding is narrow: the panel "must provide a means for the ARC panel to evaluate an alleged victim's credibility, not for the accused to physically confront his accuser."

The failure. The claimant never appeared; the respondent was not told in advance she would be absent; the panel decided credibility without her. In the court's words: "the panel resolved this 'problem of credibility' without assessing Roe's credibility. In fact, it decided plaintiff's fate without seeing or hearing from Roe at all. That is disturbing and, in this case, a denial of due process." The timeline compounded it: a month to interview the claimant, four more months to notify the respondent, four more to convene the hearing. And the principle the court stated reaches every interview in the file: "Few procedures safeguard accuracy better than adversarial questioning."

The prevention note. If the finding will rest on an account, the fact-finder needs a way to test that account. Investigators who know the hearing may proceed without a party must build the interview record accordingly: complete, specific, internally tested, and documented well enough to be examined in the party's absence.

What the court held. Where a student faces severe discipline and witness credibility is central, fundamental fairness requires a mechanism for cross-examination, direct or indirect, before a neutral adjudicator who independently finds facts. And that adjudicator "cannot be a single individual with the divided and inconsistent roles occupied by the Title IX investigator in the USC system." The court put the structural problem plainly: "a right of 'cross-examination' implemented by a single individual acting as investigator, prosecutor, fact finder and sentencer, is incompatible with adversarial questioning designed to uncover the truth. It is simply an extension of the investigation and prosecution itself."

The failure. The investigator had "unfettered discretion to chart the course and scope of her investigation and to determine credibility, and exercised that discretion in questionable ways." She rejected the respondent's fabrication-motive theory "almost immediately" despite corroborating leads, "did not follow up with presumably identifiable and available witnesses," and failed to verify a checkable fact with the athletic department before making an unequivocal pronouncement about it. The appeal layer could not realistically disturb her credibility findings.

The prevention note. Investigative discretion is where cases are won and lost. Leads that could corroborate the disfavored account get pursued, checkable facts get checked, and the file documents both. When a court can list the follow-ups an investigator declined to make, the finding is already gone.

What the court held. "Westmont's investigation and adjudication of Jane's accusation was fatally flawed. Westmont did not provide John with a fair hearing; indeed, it did not comply with its own policies and procedures." Where the outcome turns on credibility, each adjudicator must hear from critical witnesses in some form before assessing credibility, and "an adjudicatory body cannot base its credibility determinations on information in its possession that is not made available to the accused."

The failure. The panel did not hear testimony from three critical witnesses, yet relied on their prior statements to corroborate the claimant and impeach the respondent, selecting among non-testifying witnesses' statements by what the court called a "seemingly arbitrary method." The panel withheld the investigator's interview notes and the detailed hearing notes, giving the respondent only oral summaries that were "significantly less detailed." The investigator also sat as an adjudicator, and the panel appeared simply to adopt her credibility determinations.

The prevention note. Interview notes are part of the record the finding must survive on, not private work product for the panel. If a statement will be used to corroborate or impeach, the party it is used against sees it, and the witness who gave it is heard in some form. Anything less converts the interview file into grounds for reversal.

What the court held. In the court's own words: "We hold that where, as here, John was facing potentially severe consequences and the Committee's decision against him turned on believing Jane, the Committee's procedures should have included an opportunity for the Committee to assess Jane's credibility by her appearing at the hearing in person or by videoconference or similar technology, and by the Committee's asking her appropriate questions proposed by John or the Committee itself. That opportunity did not exist here." The court did not require direct cross-examination by the accused.

The failure. The claimant never appeared before the review committee in any form; only the respondent appeared and spoke. The committee credited her written account over his live account, in a case with no third-party witnesses to the encounter, without ever observing her or putting a single question to her.

The prevention note. A written statement is an interview no one conducted. When a finding will rest on believing one account over another, the process must give the fact-finder some live means of testing the credited account, and the investigator's interviews must anticipate that test rather than assume the paper record will carry the finding alone.

What the court held. Judge Saylor held that a private university's discipline is measured against its contractual promises and "basic fairness," and that Doe plausibly alleged Brandeis's single-examiner process denied him both. The opinion contains the most-quoted judicial warning about the single-investigator model: "The dangers of combining in a single individual the power to investigate, prosecute, and convict, with little effective power of review, are obvious. No matter how well-intentioned, such a person may have preconceptions and biases, may make mistakes, and may reach premature conclusions."

The failure (as alleged). An essentially inquisitorial proceeding: one special examiner investigating, prosecuting, and adjudicating; no opportunity to confront the accuser or witnesses; no notice of the specific factual bases of the charges until after the responsibility finding. The court also supplied the line every investigator should carry into the first interview: "Whether someone is a 'victim' is a conclusion to be reached at the end of a fair process, not an assumption to be made at the beginning."

The prevention note. Role labels decide interview technique. An investigator who begins by classifying one party as the victim has already chosen which account gets tested and which gets accepted. Equal-rigor interviewing, the same standards, the same follow-up discipline, the same documentation, applied to every party, is what keeps the process on the right side of this opinion.