The CRediT author statement has moved from novelty to routine: a large majority of major publishers now ask for one at submission, and many will not advance a manuscript without it. Yet it is still frequently drafted in the last hour before submission, by one author, from memory. That is a missed opportunity, because a statement assembled carelessly does exactly what a contribution taxonomy is meant to prevent. This guide sets out a step-by-step method for producing one well. The authoritative how-to lives at how to write a CRediT author statement, and this article walks the same ground in practice.
First, know what CRediT is — and is not
Before assigning a single role, fix two facts in mind. CRediT is a controlled vocabulary of fourteen contributor roles, each with a canonical definition and a stable identifier, maintained as a NISO standard. It records what people did. It is not a definition of authorship, and it is not a scoring system. The decision about who qualifies as an author is made separately, under the ICMJE criteria in biomedical fields and equivalent norms elsewhere; CRediT supplements that decision and does not replace it. Conflating the two is the most common error in this area, and it is worth reading the full account of authorship and accountability alongside this guide.
The fourteen roles fall into four loose functional groups that make a useful checklist: planning and design (Conceptualization, Methodology, Software); research and analysis (Validation, Formal analysis, Investigation, Resources, Data curation); communication (Writing – original draft, Writing – review & editing, Visualization); and management (Supervision, Project administration, Funding acquisition). The canonical definitions are set out at the CRediT roles, and you should assign against those definitions rather than against your intuition about what a role name implies.
Step 1: list every contributor, then settle the author line
Start with people, not roles. Write down everyone who contributed to the work in any way, including those who may end up acknowledged rather than authored. Then apply your field’s authorship test to decide who belongs on the author line. In biomedical research that is the ICMJE four criteria: substantial contribution to conception or design or to acquisition, analysis or interpretation of data; drafting or critically revising the work; final approval of the version to be published; and accountability for the work. Most publishers apply CRediT only to named authors, so settle the author line first.
Step 2: assign roles to each named author against the canonical definitions
Take each author in turn and ask, for each of the fourteen roles, whether they genuinely performed that contribution as the definition describes it. Some pointers that prevent common mistakes:
- Investigation is performing the experiments or collecting the data — not the same as Data curation, which is annotating, cleaning, and maintaining the data for reuse.
- Methodology is designing or developing the method; Software is writing the code that implements it. In some fields these overlap, but assign both only where both genuinely happened.
- Writing – original draft is preparing the initial draft; Writing – review & editing is critical revision by members of the original research group. An author who only commented on a near-final draft did the latter, not the former.
- Funding acquisition, Resources, and Supervision are legitimate roles, but on their own they may or may not meet the authorship bar in your field — record the contribution honestly and let the authorship test, not the role, decide.
Be aware that even careful researchers given the same description sometimes disagree on which roles apply; the boundaries between adjacent roles are genuinely fuzzy. Treat the statement as an honest broad signal, not a precise measurement.
Step 3: add the degree-of-contribution qualifier where it helps
The standard supports an optional qualifier on each assignment — lead, equal, or supporting. It is not a percentage and it does not rank roles against one another; it distinguishes “I led this” from “I contributed to this.” Most published statements omit it because few publishers require it, but it is genuinely useful where several authors share a role: marking one author as lead on Writing – original draft and two as supporting conveys real information at almost no cost.
Step 4: confirm with every author
A contribution statement is a claim made on behalf of named people, so each named person should see and confirm their own roles before submission. This is not bureaucratic box-ticking. It is the step that catches the case where one author has, in good faith, attributed to themselves work that another person actually did — the failure mode that an honest taxonomy exists to surface. Circulate the draft statement; let each author correct their own line.
Step 5: format it for the journal
The conventional written form lists each author by name followed by their roles:
Zhang San: Conceptualization, Methodology, Software. Priya Patel: Data curation, Writing – original draft. Erin Wright: Visualization, Investigation. Adam Lloyd: Supervision, Software, Validation. Maria García-López: Writing – review & editing.
Many submission systems collect the same information through a structured form instead, which is better: a statement captured as structured metadata can propagate to Crossref and ORCID and be read by downstream systems, whereas a closing paragraph of prose cannot. Where the journal offers the structured route, use it. Where it only collects a narrative paragraph, write the paragraph above — but know that its value as machine-readable data is limited until the publisher’s plumbing catches up.
Common pitfalls to avoid
- Treating CRediT as the authorship test. It records contribution; it does not decide who qualifies as an author.
- Claiming roles for acknowledged work. If a medical writer drafted the text or a technician ran the experiments, do not absorb their roles into an author’s line.
- Over-assigning. Listing all fourteen roles for the senior author signals nothing. Assign only what was genuinely done.
- Leaving author order to CRediT. CRediT does not encode author order; that is a separate decision your field’s conventions govern.
Where shared vocabulary fits
CRediT itself is settled; the live problem is that its real-world implementation is uneven, with many venues collecting only narrative paragraphs rather than structured metadata. A shared, federated vocabulary that defines the roles consistently and points back to NISO for the standard is what lets a statement written for one system mean the same thing when read by another. Supplying that definitional layer is the role the CASRAI dictionary is designed to play; the contributor-roles vocabulary sits in the CRediT extensions domain.
Leave a Reply