Biostatistician authorship is one of the most consistently disputed applications of the ICMJE authorship criteria. The four criteria (substantial intellectual contribution, drafting or critical revision, final approval, and accountability) were written to apply to any contributor role, but they don’t map cleanly onto how statistical collaboration actually happens in a research unit. That gap is why a growing number of universities and cancer centers — not journals, and not ICMJE itself — have published their own statistician-specific authorship guidance. This page covers why the ambiguity is structural, what several real institutions have done about it, and how to apply the general criteria to a statistical collaborator’s actual contribution.
Why statistician authorship is a recurring hard case
ICMJE’s four criteria were not written with any single discipline or role in mind, which is exactly why they’re difficult to apply consistently to statistical collaboration. Three structural features of how biostatisticians work make the case genuinely ambiguous, rather than just contested:
- The contribution model varies by engagement type. A department of biostatistics or a core facility may support a project anywhere from a one-off consultation (running a pre-specified test on a dataset someone else designed) to full co-investigator involvement (developing the analysis plan, contributing to study design, interpreting results, and writing the statistical methods section). ICMJE’s criteria apply identically to both scenarios in principle, but only the second clearly clears the bar in practice.
- Statistical contribution is often paid, fee-for-service work. Many biostatistics cores and consulting units bill for their time the same way a lab charges for reagents or sequencing. ICMJE and every institutional policy reviewed for this page are explicit that payment for a service does not by itself disqualify someone from authorship, and it does not substitute for authorship when the underlying contribution qualifies — but the presence of a fee-for-service relationship is exactly what makes some research teams default to an acknowledgment out of habit, without actually evaluating whether the four criteria were met.
- “Just running the numbers” undersells what substantial contribution can look like. A statistician who only executes a pre-specified analysis plan designed entirely by others, with no role in interpretation or manuscript drafting, has a real case for acknowledgment rather than authorship. But a statistician who develops the analysis plan, chooses the modeling approach, interprets what the results mean substantively, and drafts or critically reviews the methods and results sections has met the same substantial-contribution threshold as any other co-investigator — the work is intellectual, not clerical, even when it’s also billed.
The result is a large middle ground between “clearly authorship” and “clearly acknowledgment,” and ICMJE’s own recommendations don’t resolve it — the ICMJE roles-and-responsibilities page lists general examples of qualifying and non-qualifying contributions but does not address statisticians, methodologists, or other specialized technical collaborators as a distinct category. That silence is precisely what has pushed individual institutions to write their own supplementary guidance.
How real institutions have filled the gap
Several universities and academic health centers have published statistician-specific authorship policies that apply ICMJE’s four criteria to the specific engagement patterns biostatistics units see day to day. These are institutional supplements to ICMJE, not replacements for it — every one of them frames itself as ICMJE-derived rather than as an alternative standard.
Rutgers Cancer Institute of New Jersey
Rutgers Cancer Institute’s Biostatistics Authorship Policy states that collaborating statisticians qualify for authorship on manuscripts and abstracts when the ICMJE criteria are met, and explicitly instructs that MS- and PhD-level statisticians should not be listed under acknowledgments when those criteria have been satisfied — they should be listed as authors instead. The policy also addresses byline placement in practice: it notes it is customary for a faculty statistician to appear as second author when they have co-led the overall study design, analysis, conduct, and manuscript preparation, giving research teams a concrete placement convention to apply once qualification is established.
Virginia Commonwealth University (VCU) Department of Biostatistics
VCU’s Department of Biostatistics published a proposed departmental guideline on authorship that walks directly from the ICMJE criteria (substantial contribution to conception/design or data analysis/interpretation; drafting or critically revising the manuscript; final approval) to how a statistician’s specific involvement — developing the analysis plan, choosing methodology, interpreting output — should be evaluated against them. The guideline exists specifically because department faculty saw the same disputes recur project after project and wanted a documented reference point rather than relitigating the question each time.
Yale School of Public Health
Yale’s guidance, published by the Yale Center for Analytical Sciences (ysph.yale.edu/ycas/working/authorship), emphasizes agreeing on authorship criteria early in the collaboration rather than after the manuscript is drafted, and states that MS- and PhD-level biostatisticians should be listed as authors when they meet the scientific-contribution criteria, in addition to receiving appropriate remuneration for their services — making explicit that payment and authorship are not mutually exclusive or substitutable.
Northwestern University Quantitative Data Sciences Core (QDSC)
Northwestern’s Quantitative Data Sciences Core states in its published FAQ that it endorses the ICMJE criteria and treats authorship as generally expected whenever a statistician provides substantive input on study design or analysis. This guidance traces back to Authorship Guidelines approved by the Research Affairs Committee of the Northwestern Faculty Senate in April 2017, giving the core’s practice an actual university-governance basis rather than just departmental custom.
Other academic biostatistics units with published statistician-authorship guidance
The same pattern — an ICMJE-derived, statistician-specific policy published by the collaborating unit itself — also appears at Vanderbilt Biostatistics, NYU School of Global Public Health’s Biostatistical Collaboration and Consultation Core, Ohio State’s Department of Biostatistics, and the University of Utah’s Study Design and Biostatistics Center, among others. The consistency across independently written policies is itself informative: none of these institutions is inventing a new authorship standard for statisticians. Each is applying the same four ICMJE criteria to the same recurring engagement patterns and arriving at compatible guidance, which is a reasonable signal that the underlying interpretation is sound rather than institution-specific.
Practical decision criteria: authorship, acknowledgment, or a listed role short of authorship
Applying the pattern above to a specific collaboration, ask the following in order. This does not replace a real conversation between the research team and the statistician — every institutional policy reviewed here recommends agreeing on expectations at the start of the collaboration, not at manuscript submission — but it structures that conversation around the same criteria the published policies use.
- Did the statistician contribute to conception/design, or to data analysis/interpretation, beyond mechanically executing someone else’s fully pre-specified plan? Developing the analysis plan, choosing the statistical approach, or substantively interpreting what the results mean scientifically all count. Running a single pre-agreed test with no interpretive input is the clearest case for acknowledgment rather than authorship.
- Did they draft or critically revise any part of the manuscript for intellectual content? Writing or substantively editing the statistical methods and results sections satisfies ICMJE’s second criterion. A statistician who only supplies output (tables, p-values) without writing or reviewing how it’s described has not met this criterion yet — but can, if invited to review the draft before submission, which several of the policies above explicitly recommend doing as a matter of course.
- Will they review and approve the final submitted version, and take responsibility for the statistical integrity of the work? This is often the easiest criterion to satisfy in practice and the one most often skipped procedurally — simply routing the near-final manuscript to the statistician for sign-off closes this gap.
- If the honest answer to (1) or (2) is no, is a fee-for-service payment being treated as a substitute for evaluating these criteria? Every policy surveyed above is explicit that it isn’t supposed to be. Being paid for statistical work neither confers authorship on its own nor disqualifies someone who has, separately, met the substantive criteria.
- If authorship criteria clearly aren’t met, is CRediT more useful than a bare acknowledgment line? A statistician who ran pre-specified analyses without design or interpretive input, and didn’t contribute to the writing, is accurately captured by Formal Analysis and, where applicable, Methodology or Software CRediT roles in a contributor-role statement, even where a full ICMJE byline isn’t warranted. This is more informative to a reader than an unattributed “we thank the biostatistics core” line, and several of the institutional policies above implicitly point toward exactly this outcome by distinguishing “acknowledgment” from a documented, role-specific credit.
The decision point that trips teams up most often in practice is treating the statistician’s involvement as fixed at the start of the project. Contribution level frequently changes as a study proceeds — a consultant brought on for one analysis ends up co-developing the interpretation, or a co-investigator statistician’s role narrows after a design change. Every policy surveyed here recommends revisiting authorship expectations as the collaboration evolves, not only once at the outset.
How this differs from the general ICMJE criteria
Nothing here is a fifth criterion or a discipline-specific exception to ICMJE — see ICMJE Authorship Criteria: The 4 Requirements for the general framework this page applies. What the institutional policies above add is not a new rule but a worked interpretation of the existing rule for a specific, recurring pattern of contribution: paid, service-structured statistical collaboration that can range from purely mechanical to fully intellectual. The same reasoning extends to other specialized technical collaborators (bioinformaticians, data scientists, core-facility staff) who face a structurally similar ambiguity, though institutional guidance specifically addressing those roles is less consistently published than it is for biostatistics.
Frequently asked questions
Does being paid for statistical consulting disqualify a biostatistician from authorship?
No. Every institutional policy examined here states plainly that a fee-for-service or consulting-fee arrangement does not, by itself, disqualify someone from authorship when the ICMJE substantive-contribution, drafting, approval, and accountability criteria are otherwise met. Payment and authorship are evaluated independently.
Can a statistician be listed as an author without contributing to the writing?
Not under ICMJE’s criteria as written — drafting or critically reviewing the manuscript for intellectual content is one of the four required criteria, and all four must be met. In practice this is usually satisfied by having the statistician write or substantively review the statistical methods and results sections, which most of the institutional policies above explicitly recommend building into the workflow.
What should happen when a statistician’s contribution doesn’t meet the bar for authorship?
Acknowledge the contribution specifically (naming the person and their role, e.g., “statistical analysis performed by”) rather than a generic thanks, and consider documenting it with a CRediT role such as Formal Analysis where a contributor-role statement is used. See ICMJE Authorship Criteria: The 4 Requirements for what generally distinguishes an acknowledgment-level contribution from authorship.
Who decides whether a statistician qualifies for authorship — the journal, the institution, or the research team?
ICMJE sets the criteria journals expect to be applied, but ICMJE does not adjudicate individual authorship disputes and does not publish role-specific guidance for statisticians. That’s why the institutional policies covered on this page exist: they’re written by the department, core facility, or biostatistics unit itself to give its own statisticians and collaborating research teams a shared, ICMJE-consistent reference point, agreed early in the collaboration rather than argued over at submission.
