Writing a Methods Section Others Can Reproduce

Reviewers read your methods asking whether they could repeat the study from the page. First drafts often leave out details only the senior author still remembers.

Guide · 10 min read · Directive Publications Editorial Team

The standard your methods section is judged against

The ICMJE Recommendations, which many medical journals follow, put the test plainly. The methods section "should aim to be sufficiently detailed such that others with access to the data would be able to reproduce the results." For statistics, the bar is a knowledgeable reader with the original data who can judge whether your methods were appropriate and verify what you report.

That bar is easy to miss. A draft written by the person who ran the study tends to skip what felt obvious at the bench or on the ward: which version of the questionnaire, who delivered the intervention, what counted as a completed session, and which analysis was planned before the data came in.

This is why the methods section is a senior author's job to check, even when someone else drafts it. You are usually the person who remembers the protocol amendments and the reasons behind them.

One useful rule from ICMJE shapes the whole section: in general, the methods should contain only information that was available when the protocol or plan was written. Anything learned during the study, such as how many people enrolled or how many dropped out, belongs in the results.

Build the section in this order

Start from the reporting guideline for your design, not a blank page. ICMJE encourages journals to ask authors to follow these guidelines and names the EQUATOR Network as a good source for them.

Copy the guideline's methods items into your draft as working subheadings, then fill them in this order:

1. Pick the reporting guideline for your design. Its methods items become your skeleton.
2. Describe the design and setting. Name the study type and give the setting, locations and relevant dates; STROBE asks for periods of recruitment, exposure, follow-up and data collection.
3. Define the participants. Give eligibility criteria, the source population and how participants were selected. ICMJE also asks you, unless it is inappropriate, to report participants' sex and/or gender and to describe how you determined them. CONSORT 2025 adds, where relevant, eligibility criteria for sites and for the people delivering the interventions.
4. Describe interventions or exposures precisely. Enough detail to deliver them again (see the next section).
5. Specify every outcome. What was measured, how, and when.
6. Explain sample size and analysis. How the sample size was set, the method for each outcome, how missing data were handled, and the software used.
7. State ethics approval, consent and registration. Name the committee, describe consent, and give the registration details the journal asks for.

If an organisation was paid or contracted to help run the study, for example to collect or manage data, ICMJE asks you to say so in the methods. Senior authors are often the only ones on the byline who know about those contracts.

Describe interventions so someone else could deliver them

CONSORT 2025 item 13 asks for the intervention and comparator "with sufficient details to allow replication," plus where additional materials such as an intervention manual can be accessed. The TIDieR checklist exists to make that achievable. Published in The BMJ in 2014, it has 12 items, and although its emphasis is on trials, its authors intended it to apply across all evaluative study designs.

Use the items as questions to put to your own draft:

• Why: the rationale or theory behind the essential elements
• What: the materials used, where to access them, and each procedure
• Who provided it: each type of provider, their expertise and any specific training
• How and where: mode of delivery, individual or group, and the type of location
• When and how much: number of sessions, schedule, duration, intensity or dose
• Tailoring and modifications: planned personalisation, and any changes made during the study
• How well: how adherence or fidelity was assessed, and the extent to which the intervention was delivered as planned

The difference in practice, using an illustrative example rather than a real study:

Before: Participants in the intervention group received a supervised exercise programme.

After: Participants in the intervention group attended 12 supervised exercise sessions of 45 minutes, twice weekly for 6 weeks, in groups of up to 8. Sessions were delivered in hospital outpatient gyms by physiotherapists who completed one day of protocol training. Physiotherapists increased resistance when a participant completed all sets at the prescribed load. The session manual is in Supplementary File 2.

For drugs, ICMJE asks you to identify all drugs and chemicals precisely, including generic names, doses and routes of administration. For equipment, it asks for the manufacturer's name and address in parentheses; check the journal's house style for the exact format.

Define outcomes so nobody has to guess

"The primary outcome was blood pressure" leaves a reader unable to reproduce anything. CONSORT 2025 item 14 breaks a prespecified outcome into four parts, and the same discipline works for any design:

An illustrative rewrite: "The primary outcome was change in seated systolic blood pressure (mmHg) from baseline to 12 weeks, summarised as a mean in each group." Then say how it was measured, by whom, and with which device or validated instrument. STROBE item 8 adds that where there is more than one group, you should describe how comparable the assessment methods were.

Define harms too. CONSORT 2025 asks how harms were defined and assessed, including whether they were collected systematically or non-systematically.

Materials, software and code

Nobody can repeat a step if they cannot identify exactly what you used.

• Reagents and resources. Research Resource Identifiers (RRIDs) are ID numbers for key resources such as antibodies, cell lines, model organisms and software tools. The Resource Identification Initiative's draft instructions for authors show the format: supplier and catalogue number, then the identifier, as in (Supplier Cat# [number], RRID:AB_[number]). Check whether your target journal asks for them.
• Software. ICMJE asks you to specify the statistical software packages and versions used. "Analysed in R" is not enough; give the version and any key packages.
• Code. CONSORT 2025 asks trial reports to say where statistical code can be accessed. Deposit it with a persistent identifier and cite it, as set out in our guide to citing data, software and code.
• AI tools. If you used AI technology to conduct the study, ICMJE asks you to describe it in enough detail to replicate the approach, including the tool, version and prompts where applicable.

Statistics that someone with your data could check

Replace "appropriate statistical tests were used" with the method for each outcome. The CONSORT 2025 statistical methods items make a good checklist for any comparative study:

• The method used to compare groups for primary and secondary outcomes, including harms
• Who was included in each analysis, and in which group
• How missing data were handled
• Methods for subgroup and sensitivity analyses, distinguishing prespecified from post hoc

ICMJE also asks you to define statistical terms, abbreviations and most symbols, and to quantify findings with measures of uncertainty such as confidence intervals rather than relying solely on p-values. CONSORT 2025 also asks how the sample size was determined, including all the assumptions behind the calculation.

If the analysis changed after the study began, say so. CONSORT 2025 asks for important changes after the trial commenced, including outcomes or analyses that were not prespecified, with reasons. A reviewer who compares your paper with the registration or protocol can find the change anyway, and an unexplained one reads far worse.

Citing earlier methods and reusing your own wording

ICMJE gives a simple rule for deciding how much to write:

Two cautions. "As described previously" should lead to a source that actually contains the full method, not a paper that itself says "as described previously." And a citation does not cover changes: if you altered concentrations, timings or inclusion criteria, state the modification in your own paper.

Groups that run the same protocol across studies often want to reuse their own methods text. The Text Recycling Research Project's best practices for researchers treat consistent methods wording as legitimate where it aids accurate communication, provided it is accurate for the new study, does not infringe copyright or publisher policy, and is disclosed to editors at submission. Editorial policies differ, so check the journal's. Reused methods text is likely to appear in similarity screening; our guide to reading a similarity report explains how editors weigh that overlap.

Ethics, consent and registration statements

ICMJE says the methods section should include a statement that the research was approved by an independent local, regional or national review body, such as an ethics committee or institutional review board. When informed consent was obtained, the published article should say so. For animal studies, indicate whether institutional and national standards for the care and use of laboratory animals were followed.

A template to adapt:

If the committee waived consent, say that and state the grounds it accepted. For trials, CONSORT 2025 asks for the registry name, identifying number with URL, and date of registration. ICMJE recommends that journals publish the registration number at the end of the abstract, so check where your target journal wants it. Some journals collect ethics and registration details in a separate declarations section; follow their instructions.

The senior author's final read

Before submission, read the methods beside the protocol and registration record:

1. Every methods item in the relevant reporting guideline is addressed or marked not applicable
2. Each outcome has a variable, metric, aggregation method and time point
3. Intervention and comparator could be delivered from the text and linked materials
4. Every drug has a generic name, dose and route
5. Software and versions are named, and code location is stated
6. Prespecified and exploratory analyses are labelled
7. Changes from the protocol are reported with reasons
8. The approving committee is named and consent is described
9. Recruitment numbers, dropouts and other results have not crept into the methods

If a colleague outside your group could not answer "what exactly did they do?" for any paragraph, rewrite it before a reviewer asks.

When the methods are ready, check the target journal's own requirements. The Directive Publications For Authors page links to its author guidelines, and you can submit your manuscript once everything is in place.