Describe research labour without turning it into a score
A byline compresses many forms of work into a sequence. CRediT supplements it with a controlled list of roles. Readers can more readily discover who handled methodology, software, data or writing. Important work below an authorship threshold can also be described appropriately where the publication permits contributor information.
A role is neither a prize nor ownership of a project component. Several people may share a role, and one person may hold several. Allocation depends on tangible activity. Degree or leadership can be described if a journal supports that detail, but must not be inferred from the label itself.
Read fourteen roles across four parts of research
Planning: Conceptualization develops ideas and aims; Methodology designs methods; Funding acquisition secures support; Project administration coordinates delivery; Supervision provides scholarly oversight. Funding and senior oversight alone do not automatically create authorship.
Materials and conduct: Resources supplies materials, samples or access; Investigation performs data collection or experiments; Data curation organises, documents and preserves data. These may overlap, yet their specific outputs make them distinguishable.
Technical analysis: Software creates or implements code; Formal analysis applies statistical, mathematical or computational techniques; Validation checks reproducibility or results; Visualization prepares scholarly displays. Merely operating standard software does not necessarily amount to the Software role.
Communication: Writing – original draft produces the first substantial prose; Writing – review & editing performs intellectual review or revision. Routine proofreading should not be inflated into critical scholarly editing.
Assign roles through an evidence-aware conversation
- Ask participants independently to list activities and outputs.
- Read the official definitions rather than relying on short labels.
- Resolve duplicate and overlooked work against project records.
- Add a brief verb-and-object statement to every allocation.
- Decide authorship, order and acknowledgements separately.
- Review the matrix after analysis, drafting and revision.
A self-reported role begins discussion; it is not automatically decisive. At the same time, less visible labour such as curation and validation must not disappear because it features rarely in presentations. A facilitator and confidential route can reduce the effect of hierarchy.
Use records proportionately. Version history, analysis logs and data dictionaries can refresh memory, but a missing digital trace does not necessarily mean no contribution. Discuss work with the person concerned rather than auditing them silently.
Avoid points, titles and planned-work errors
Do not count roles. Conceptualization is not inherently more valuable than Data curation, and five minor labels need not outweigh two central roles. Do not invent intensity levels unless the journal supports them. A role must not make someone accountable for material they could not inspect.
Separate expected from completed work. Planning matrices may contain intentions; the published statement describes what happened. Automatically assigning roles from job titles is unreliable. “Professor”, “programmer” and “research assistant” are positions, not CRediT categories.
Tools do not become authors. A responsible person records, checks and discloses technological assistance under relevant policy. Human contractors are acknowledged or considered for authorship according to actual scholarly work, consent and the journal's rule. Transparency need not expose confidential participant data.
Keep the contributor statement consistent
Before submission, authors confirm their own description and inspect the whole statement for obvious omissions. Journal form, manuscript and repository should use consistent roles. Preserve the more detailed internal matrix if the platform only accepts labels.
The authorship criteria guide answers eligibility. The order decision record addresses rank. More resources sit in the publication ethics hub. The single foundation route is the guide to documenting review evidence.
Run the final matrix in both directions. Starting with each person, check that consequential work is represented; starting with each role, check that all material contributors appear. Compare the result with methods, data and acknowledgements sections. Resolve contradictions before they spread across publication formats.
Create a separate matrix for every paper from a project. Shared dataset work may recur, while analysis, figures and writing differ. Do not copy an earlier statement without review. After publication, inspect the visible contributor record and ask the editor to correct missing roles, switched names or truncated metadata.