Reproducibility · Beginner
Template 2: Study protocol
Complete before collecting or examining the relevant outcomes. Amend with a dated record when plans change. This worksheet does not replace institutional forms or the reporting standard for your study type.
3 min read · Oct 26, 2026
Key points
- Adapt the prompts to the actual study.
- Complete every statement from verified records.
- Follow the target journal and institutional requirements.
Contents
Complete before collecting or examining the relevant outcomes. Amend with a dated record when plans change. This worksheet does not replace institutional forms or the reporting standard for your study type.
Complete before collecting or examining the relevant outcomes. Amend with a dated record when plans change. This worksheet does not replace institutional forms or the reporting standard for your study type.
Purpose and design
Title, version, date and contributors: [Details]
Question and rationale: [Question; supporting evidence]
Design and setting: [Experimental, observational, computational, qualitative or mixed; where and when]
Target population or system: [Who or what the conclusion concerns]
Primary objective and outcome: [Definition, unit, timing and aggregation]
Secondary outcomes: [Definitions and priority]
Sampling and procedures
Eligibility: [Inclusion and exclusion criteria, with reasons]
Recruitment or sampling: [Source and selection method]
Independent unit: [Unit of assignment or sampling relevant to the inference]
Repeated measurements and clusters: [What belongs within each unit]
Sample size rationale: [Power, precision, informational adequacy, complete population or feasibility]
Inputs to that rationale: [Effect or precision target, variability, assumptions, losses and calculation method]
Conditions and controls: [What each comparison isolates]
Allocation and concealment: [Method and who can see assignments]
Masking: [Who is blinded at each stage; explain unavoidable knowledge]
Procedure: [Chronological steps with materials, settings and timing]
Quality criteria: [Calibration, acceptance criteria and handling of failures]
Analysis and records
Primary analysis: [Estimand, model, contrast and uncertainty method]
Missing data: [Expected causes and handling strategy]
Exclusions: [Rules, decision timing and logging]
Multiplicity: [Comparison families and error control objective]
Sensitivity analyses: [Reasonable alternatives and the uncertainty each addresses]
Stopping rule: [Fixed target or justified sequential rule]
Data management: [Raw records, identifiers, access, backups and retention]
Reproducibility outputs: [Dictionary, scripts, README and shareable materials]
Ethics and permissions
Responsible institution or review body: [Name and contact route]
Review status: [Pending, approved or documented determination; actual reference if available]
Consent and publication permissions: [Applicable process]
Safeguards and escalation: [Risks, responsible person and reporting process]
Data restrictions: [Confidentiality, licences and permitted sharing]
Registration or archive: [Actual identifier and date, or planned destination clearly labelled]
Approval to begin
Outstanding requirements: [Unresolved permissions, training or resources]
Review by collaborators: [Who checked what, and when]
Amendments: Record changes in Template 16 with the reason and whether outcomes had been examined.
Tutorial: make another person's decisions visible
Use Article 4 to develop the protocol before collecting the outcome. A protocol should tell a second researcher how to respond to predictable problems. Specify what happens when a measurement fails, a participant withdraws or a device produces an impossible value. Record the rationale as well as the rule.
Hypothetical engineering example: Two sensor designs will be evaluated under the same sequence of reference conditions. The protocol identifies each independently assembled device, calibration procedure, environmental range, test order and summary error metric. Repeating a reading on one device is useful for repeatability, but the protocol must not count those readings as independently manufactured devices. A device that fails before the final condition remains in the failure record rather than silently disappearing from the comparison.
Worked amendment: “Version 2 adds a reference check after each test session because drift was observed during preliminary calibration. No comparison outcomes had been examined when this decision was made.” This communicates the reason, timing and consequence of the change. If outcomes had already been examined, record that fact instead.
Completion check: Walk through one ordinary case and one failure case using only this document. Any step that requires an unwritten decision should become an explicit instruction or an honestly recorded unresolved question.