Authorship Disputes: How to Prevent and Resolve Them

Authorship fights stall more papers than bad statistics, and almost all of them begin with a conversation that never happened.

Guide · 13 min read · Directive Publications Editorial Team

Prevent (the only cheap strategy)

Before writing ends, agree on:

1. Criteria for authorship
2. Who is on the list
3. Order and corresponding author
4. When the list will be revisited

Criteria for authorship. Name a standard instead of relying on shared instinct. In medicine that is usually the ICMJE criteria, which are cumulative rather than a menu: a substantial contribution to conception, design, acquisition, analysis or interpretation, and drafting or critically revising the manuscript, and approval of the version submitted, and agreement to be accountable for it. Other fields lean on their own conventions and on CRediT. Whichever you use, write your version in two sentences at the top of the project folder, and say explicitly what does not qualify on its own — funding the study, heading the unit, supplying a reagent or an existing dataset, one pass of language editing. What goes wrong without this is not that people are greedy. It is that everyone is applying a private threshold and nobody knows it. The collaborator who supplied the cohort believes supplying the cohort is authorship. The postdoc who ran the analysis believes only analysis is. Neither belief is unreasonable, neither was ever said out loud, and the two of them discover the gap in month eleven.

Who is on the list. Write down names, not categories. The people who get missed are predictable enough to check for deliberately: the technician who built the assay, the statistician consulted twice at the design stage, the collaborator at another institution, and the student who will have left the group before submission. In the same conversation, decide what happens to someone who joins later — the rule most groups settle on is that a new contributor is added when their work meets the agreed criteria, and that the decision is taken at the next revisit point rather than in the corridor. Then send the list. A list agreed in a room and never written down is a list four people remember four ways. Circulate it by email, ask each person to reply confirming their own line, and keep the replies. Silence is not agreement, and it is silence that later gets described as consent.

Order and corresponding author. Order carries meaning that differs across fields and even across subfields, so record the convention you are applying and not only the sequence: first author led the analysis and drafted the paper, last author supervised, middle authors in descending order of contribution, or whatever is actually true where you work. Equal-contribution notes belong on papers where the contribution really was equal; handed out to end an argument, they relocate the argument rather than settling it. Treat corresponding author as a job description, not a rank. That person's institutional address is printed on the paper, receives the editor's correspondence, signs the licence, handles proofs, and is where a query lands years afterwards. Say who it is, say that the address is institutional rather than personal, and say who inherits the role if that person leaves. Naming it early also takes it off the table as leverage, which is most of the reason to name it early.

When the list will be revisited. Set a moment, not an intention. Two named points do the work: when the first complete draft circulates, and again in the week before submission, when the list and the order are read out and each person confirms their own contribution line. The two events that break author lists are both foreseeable. Someone leaves the group and nobody wants to be the one who chases them. Or the paper comes back with major revisions and one person does three months of unplanned reanalysis that changes the result. Decide the rule for the second case now, while nobody knows who it will be: substantial new work during revision reopens the order.

The agreement itself is one page. Date it, put it in the shared project folder, email it to everyone named on it, and keep the replies. It is not a contract and nobody will enforce it anywhere. Its power is different — it is a dated record of what a group decided while nobody yet knew who would benefit, and that is precisely the thing no mediator can reconstruct afterwards. Write into it that it can be changed by agreement, and that changes are recorded the same way. An agreement people feel trapped by is one they will later argue was never really agreed.

Guides: Authorship Disputes · Who Counts as an Author?

If a dispute starts

1. Put contributions in writing (CRediT helps)
2. Use internal mediation (department / integrity office)
3. Tell the journal if the paper is already under review
4. Expect the institution to decide substance — journals pause; they rarely adjudicate credit

Before any of that, separate two questions that always arrive fused together: who meets the criteria, and who feels wronged. Only the first can be settled by a process. The second is often entirely justified, does not resolve at the same table, and treating the two as one question is why these things run for months. Also stop the group email thread. Every round with eight people copied in hardens positions and generates text that will be read back to its author later.

Put contributions in writing. Ask each person involved to write their own, in a fixed format so the statements can be laid side by side: one line per CRediT role — conceptualization, methodology, software, formal analysis, investigation, data curation, writing of the original draft, writing review and editing, supervision, funding acquisition — with something checkable attached to each claim. Which experiments, which script, which sections of the draft, roughly when. Then compare each statement against the criteria the group agreed, and against nothing else: not effort, not seniority, not who needs a first-author paper this year. What this step really does is convert "I was heavily involved throughout" into sentences that can be confirmed or corrected by someone who was there. Write yours even if you expect the comparison to go against you. A person whose account is specific and consistent is easy for an office to help; a person whose account is a feeling is not.

Use internal mediation. Take it to someone who is not a party to it. Usually that is the department head or the graduate school; where the head is one of the disputants, or where the dispute is between a supervisor and their own student, it is the research integrity office or an ombuds office. Ask three questions in your first message: what the process is, who makes the final decision, and roughly how long it takes. Bring the agreement if one exists, the contribution statements, and the dated artefacts behind them. Keep the submission about the paper — an account of which criteria each person meets is something an office can act on, while an account of how difficult a colleague has been is a complaint it will struggle to use. If you are a student, do not attend a meeting about credit alone with your supervisor. Ask the ombuds office or a second faculty member to sit in, and put your account in writing beforehand so the record does not depend on what anyone recalls afterwards.

Tell the journal if the paper is under review. Someone should email the editorial office within days, with the manuscript number and title in the subject line, in a few sentences: an authorship question is unresolved, you are asking for the manuscript to be placed on hold, the matter is with a named institutional office, and you will update the editor by a stated date. Do not argue the merits to the editor and do not attach the correspondence. The reason to do this early is what happens otherwise — the excluded person writes to the journal first, and the editor learns of the dispute from someone the authors never mentioned. Two things not to do while a dispute is open: submit or resubmit anywhere, and change the author list in the submission system to make the problem look settled.

Expect the institution to decide substance. Journals follow COPE-aligned processes, and those processes are procedural. An editor will pause a paper, will require written agreement from every author — including anyone being added or removed — before a list changes, and may decline to move until the institution reports. What an editor will not do is weigh whose contribution mattered more; they were not in your lab and have no way to find out. So the outcome turns on the record you build internally, which is one more argument for the contribution statements. Plan on weeks to months, and plan on the manuscript sitting still throughout.

A worked example. A student builds an assay, runs the first experiments, and leaves the group. Fourteen months later the group has reworked the protocol, collected fresh data, and submitted without her, on the reasoning that none of her data is in the paper. Nobody contacted her. Work the question rather than the feeling: does the paper still rest on something she contributed — the design, the method as it appears in the methods section, an early figure, an interpretation that shaped the study? If it does, and she meets the criteria once she has read the draft and approved it, she is an author, and writing to her now is far cheaper than a correction later. If her data was genuinely replaced and what survives is groundwork, acknowledgment may be the right answer — and that is still a conversation you have with her, not a conclusion you reach on her behalf. The part no integrity office will overlook is identical in both branches: nobody asked her.

What damages labs

• Gift authorship as currency
• Silent removal of a student contributor
• Corresponding-author power games

Gift authorship as currency. The unit head added to everything the unit produces. Two colleagues who put each other on papers. The person who controls an instrument, added by custom. It feels harmless because nobody was deprived; the list only got longer. Two things are lost anyway. Authorship is accountability as well as credit, so you have attached a name to work that person cannot defend, and if the paper is ever questioned, the questions reach them too. And you have spent your criteria. A group that trades names cannot later invoke criteria to settle a disagreement, because everyone in the room can name the exception. Trainees learn the working rule immediately: names are currency, and currency belongs to whoever has the power to spend it.

Silent removal of a student contributor. The person removed is almost always the one with the least power and the most riding on it, and they usually find out when the paper appears. Be clear about what this is. It is not a disagreement about criteria; it is a decision taken about someone in their absence, and that is how an office will read it however defensible the underlying judgement was. It is also mostly unavailable in practice: journals ask for written confirmation from every author, including the person being removed, before a list changes after submission, and after publication the change becomes a public correction notice attached to the paper. The alternative costs an afternoon — send the draft, ask for revisions and approval, and if no reply comes, keep the dated record of having asked.

Corresponding-author power games. The recognisable versions: holding the submission account and declining to show anyone the record, submitting a version the group has not seen, editing the author list in the portal, forwarding the editor's correspondence selectively or not at all. Fix it structurally rather than personally. The final submission file and the completed author list circulate to everyone before submission; any author may ask to see the submission record; the role uses an institutional address, so the account cannot walk out of the building with a person. If you already have a corresponding author who will not share the file, ask once in writing, then take that written record to the same neutral person you would use for any other authorship dispute.

If you think you are being left off

Ask early, ask directly, and ask in writing. One short message to whoever is leading the paper: your understanding of the current author list, the contributions you believe qualify you with one line per role, and a request to talk this week. No accusation, no history, nobody copied in. Send it well before submission if there is any way to do so, because every part of this is cheaper before the file is uploaded.

Keep your own record from the beginning of a project, not from the moment you start worrying. Dated notebook pages, commits, analysis scripts, the email in which you were asked to do the work, the drafts you returned with comments. You are not building a case against anybody. You are making it possible to answer the question a year later, when nobody can reconstruct from memory who did what.

If the answer is no, or if no answer comes, escalate to one neutral person rather than to the group. A single message to the graduate school, the integrity office or an ombuds reads as a request for help. The same message sent to everyone reads as an accusation, and it narrows what the people copied in are then able to do.

Apply the criteria to yourself with the same honesty you want from everyone else. Running samples someone else designed, or being present in the group while the work happened, is not authorship on its own. If your contribution belongs in the acknowledgments, ask for the acknowledgment, and then ask for a defined role on the next paper — in that order, and in the same conversation.

Bottom line

Everything expensive about authorship happens because a decision was postponed. The four questions at the top of this post take a group half an hour, and that half hour exists at exactly one point: while the work is still running and nobody yet knows how the paper will land. Postponed, the same four questions cost you an integrity office, a paused manuscript, and a working relationship that does not fully recover. If you are reading this because a dispute has already started, the order is fixed — contributions in writing, one neutral person, the journal told if the paper is under review, and patience while the institution decides the substance.

Authorship is credit and accountability. Write the agreement early. More at Directive Academy.