Mediation for Authorship Disputes

When an authorship disagreement has moved past a first informal conversation but hasn’t reached formal institutional adjudication, most documented institutional dispute-resolution models call for one specific intermediate step: mediation. Mediation is the stage where a neutral third party helps the disputing researchers work out their own resolution, rather than having a decision imposed on them. This guide focuses specifically on that one mechanism — who mediates, how a mediation session is actually structured, what stays confidential and what doesn’t, and what a realistic outcome looks like. For the full escalation ladder an authorship dispute typically follows, from informal discussion through formal adjudication and appeal, see CASRAI’s guide to How Authorship Disputes Get Resolved: The Institutional Process. For how to prevent a dispute from reaching this point in the first place, see Authorship Agreements: Why Research Teams Should Sign One Before Writing Begins.

What mediation is, and isn’t

Mediation is facilitation, not decision-making. A mediator’s job is to help the parties talk to each other productively, surface what the disagreement actually turns on — often mismatched expectations about who did what, rather than a factual question an outsider could simply settle by reading lab notebooks — and support the parties toward a resolution they both genuinely agree to. A mediator has no authority to impose an outcome. That’s the specific feature that separates mediation from formal adjudication, where a designated official (an institution’s research integrity lead or another senior, conflict-free “Responsible Person,” in the language of the UK Research Integrity Office’s UKRIO Model Authorship Dispute Procedure) reviews evidence and issues a binding decision. UKRIO’s own model describes the mediator role in exactly those terms: a “neutral, non-decision-making Mediator” whose function is “to facilitate the conversation, not to decide it.” CASRAI covers that formal-adjudication end of the process, including realistic timelines and the frequent absence of a built-in appeal route, in the institutional-process guide linked above.

Mediation is also distinct from arbitration, a related but less common alternative-dispute-resolution model in which a neutral third party is empowered to hand down a binding ruling, generally only if the parties agreed in advance, in writing, to be bound by it. Almost none of the publicly documented institutional authorship-dispute models use binding arbitration; mediation (non-binding, consent-based) and formal institutional adjudication (binding, evidence-based) are the two mechanisms that actually appear in practice.

When institutions turn to mediation

Across the institutional models that publish their process in enough detail to compare, mediation gets invoked at a fairly consistent trigger point: direct discussion between the parties has stalled without resolution, or there’s a real or perceived power imbalance — commonly a student–supervisor or postdoc–PI relationship — that makes an unmediated conversation unlikely to be fair to the less senior party. UKRIO’s model, for instance, recommends escalating from direct discussion to mediation specifically when either of those conditions applies, and treats roughly 30 working days as the point at which an unresolved mediation should move on to formal adjudication rather than continue indefinitely. NIH’s Office of Intramural Research takes a parallel approach in its own authorship dispute resolution process: after direct dialogue between the parties, researchers can bring the disagreement to NIH’s Office of the Ombudsman for what NIH’s own policy describes explicitly as “confidential, informal mediation.”

The case for building a defined mediation stage into an institutional process isn’t hypothetical. A 2024 literature review on coercive authorship published in the Journal of Korean Medical Science identified a formal institutional process for authorship disputes, alongside ombudsperson support specifically for junior researchers, as among the concrete mechanisms institutions can put in place to address the underlying power dynamics that produce these disputes in the first place — mediation, run by a genuinely neutral party, is the practical form that first recommendation usually takes.

Selecting a mediator

Institutional models source mediators a few different ways: through HR, from a trusted senior colleague with no stake in the outcome, through a standing organizational ombuds office, or, less commonly for an internal authorship dispute specifically, an external professional mediator. Whichever route an institution uses, the same basic qualifications recur: no stake in the outcome, no supervisory or evaluative relationship over either party (a mediator who will later write one party’s performance review is not a neutral choice), and enough standing to be credible to both sides.

The clearest published benchmark for what “neutral” should mean in practice comes from the field most closely associated with this kind of internal, non-adversarial resolution work: organizational ombuds practice. The International Ombudsman Association (IOA) defines four core standards for that role — independence (structurally and functionally unaligned with any other function in the organization), neutrality/impartiality (unaligned with any party, while considering all parties’ legitimate interests), confidentiality (holding communications in strict confidence absent explicit permission to disclose), and informality (not participating in any formal investigation or decision-making process). Those standards technically govern a standing ombuds office rather than every ad hoc mediator drawn from HR or a senior colleague, but the four principles work as a practical checklist regardless of who’s asked to take on the role: is this person actually independent of both parties, genuinely neutral, bound to keep what’s said confidential, and free of any formal investigative or decision-making authority over the outcome. A mediator who fails any one of those four tests isn’t functioning as a mediator in the sense institutional models intend.

How a mediation session is typically structured

Institutional documentation rarely spells out session mechanics in detail, but the shape follows standard alternative-dispute-resolution practice: an intake step where the mediator sets ground rules and confirms who will attend and what’s on the table; a joint session where each party describes their view of the dispute without interruption; private caucusing, if needed, where the mediator meets separately with each side when a joint conversation alone isn’t productive; and, if the parties reach agreement, a written summary of what was agreed — a revised author order, an updated CRediT contribution statement, or an agreed description of a specific role — that both parties confirm in writing. That written record matters beyond just closing out the dispute: if a similar disagreement resurfaces later, or if the same project produces a second dispute, a documented mediated agreement is exactly the kind of contemporaneous record a later formal adjudicator would weigh heavily, for the same reason an authorship agreement drafted at project outset is weighed heavily — it was created before there was any incentive to shade it in one party’s favor.

What mediation does not produce is a ruling. If the parties reach agreement, that agreement is what moves forward. If they don’t, nothing is imposed at this stage — the dispute proceeds to whatever formal adjudication step the institution’s process defines next.

Confidentiality: what’s actually protected, and what isn’t

General alternative-dispute-resolution practice treats mediation conversations as confidential and not carried forward as evidence if the dispute proceeds to formal adjudication — the underlying logic is that people need to be able to speak candidly about a genuinely difficult interpersonal situation without every statement becoming a fact to be litigated later. NIH’s model reflects this directly, describing its Ombudsman route in its own policy language as confidential.

That said, this is a general practice, not a guaranteed legal privilege, and institutions vary in how formally they’ve written it down. Before a mediation session starts, it’s worth asking directly, and getting a direct answer to: can the mediator tell the eventual formal decision-maker that mediation happened and didn’t resolve the dispute, without describing what either party actually said? Can any documents produced specifically for the mediation (draft proposals, position statements) be introduced as evidence later if the case goes to formal adjudication? A well-run process answers both of those clearly, in writing, before mediation begins — don’t assume a blanket confidentiality guarantee that a specific institution hasn’t actually committed to in its own policy.

What mediation can actually resolve

Realistic outcomes fall into three categories: full resolution, where the parties agree on a revised author list, order, or contribution statement; partial resolution, where some points are settled and the remainder is formally referred onward; or impasse, where mediation doesn’t produce agreement and the dispute escalates to formal adjudication. Under UKRIO’s published model, roughly 30 working days from the first mediated meeting is the general benchmark for how long an institution allows this stage to run before moving the case on — a reasonable proxy for how long mediation is generally given to work elsewhere too, even where an institution hasn’t published its own explicit figure. In most institutional models, a mediated outcome isn’t binding in the way a formal adjudication decision is — it’s an agreement the parties consented to, not a ruling imposed on them, which is precisely the point of using mediation in the first place.

When mediation is the wrong tool

Mediation assumes a good-faith disagreement about credit between people who are willing, even if reluctantly, to talk. It isn’t the right mechanism in a few specific situations: where the underlying issue looks like research misconduct rather than an honest disagreement about contribution — fabricated or falsified contribution claims, for instance — institutional models redirect straight to the relevant misconduct process instead; where the power imbalance between the parties is severe enough (documented harassment, retaliation risk, an active grievance) that a facilitated conversation, however neutrally run, isn’t a safe or fair setting — that belongs in a formal HR or grievance channel, not mediation; and where a hard external deadline (a paper on the verge of submission or already accepted) doesn’t leave room for even an expedited multi-week process — in that situation, documenting the dispute clearly and using a provisional author listing with a note that authorship remains under discussion is generally the safer choice over forcing a rushed decision either way.

Frequently asked questions

Is mediation for an authorship dispute legally binding?

No, not by default. In the institutional models that document this explicitly, mediation produces an agreement the parties consent to, not a ruling imposed on them. If mediation doesn’t produce agreement, nothing is decided at that stage — the dispute moves on to formal adjudication, which does produce a binding decision.

Who can request mediation for an authorship dispute?

Typically either party to the dispute, or the person who raised the original concern, can request it — usually as the next step once direct, informal discussion between the parties hasn’t resolved the disagreement, or where a power imbalance makes direct discussion alone unlikely to be fair.

Is what I say during mediation confidential?

Generally, yes, as a matter of standard practice — and some institutional models, like NIH’s Office of the Ombudsman route, describe this explicitly in their own policy language. But it isn’t a guaranteed legal privilege by default, and institutions vary in how formally they’ve documented it. Ask directly, before mediation starts, exactly what a mediator can and can’t carry forward to a later formal decision-maker.

What happens if mediation doesn’t resolve the dispute?

The dispute proceeds to the institution’s formal adjudication step, where a designated official reviews the evidence — contribution records, correspondence, any existing authorship agreement — and issues a binding decision. See CASRAI’s guide to how authorship disputes get resolved for what that stage involves and how long it realistically takes.

Can a journal require the authors to go through mediation before it will proceed?

No. Journal editors, per both ICMJE’s and COPE’s guidance, don’t adjudicate or arbitrate authorship disputes themselves — they can pause processing and refer an unresolved dispute back to the authors’ institution, but the choice of mediation, adjudication, or any other institutional mechanism is the institution’s process to run, not the journal’s.

Related CASRAI resources

Topic cluster: CRediT & Authorship Attribution