Pathform AI Browse packs

Research Methodology

How to Draw a Methodology Flowchart That Matches Your Design

A methodology flowchart is a process diagram, which makes it easy to draw and easy to get wrong. The common failure is a flowchart that describes the order of activities rather than the flow of evidence, which leaves a reader unable to see where bias could enter or where units are lost.

Guide clusterResearch Methodology

Direct answer

Draw a methodology flowchart by following the path of a single unit of study, whether that is a participant, a sample or a case, from entry to inclusion in the analysis. Show every decision point, every exclusion and its reason, the measurements taken at each stage, and the analysis applied at the end. Label each step with what it produces.

Quick facts

  • ClusterResearch Methodology
  • Reading time6 min
  • FormatStep-by-step method
  • Sources5 listed

Follow the unit of study, not the calendar

The most reliable way to build a methodology flowchart is to pick one unit of study and trace it. For a clinical or survey study, that unit is a participant. For a laboratory study, it is a sample or specimen. For a qualitative study, it is a case, a site or an interview. Follow that unit from the moment it enters the study until it either contributes to an analysis or leaves the study.

Tracing a unit forces the flowchart to include exclusions, which is the part most drafts leave out. Where a unit leaves, the diagram shows the reason and the number. That single change turns a diagram of activities into a diagram of the evidence base, which is what a methods reviewer actually reads.

If no unit ever leaves your study, you have either an unusually clean design or an incomplete diagram.

The elements a methodology flowchart should contain

Flowcharts in methodology sections use a small vocabulary consistently. Entry and exit points are rounded. Processes are rectangles. Decisions are diamonds with labeled branches. Data stores, such as datasets or biobanks, are drawn differently from processes. Using shapes consistently means a reader can follow the diagram without reading the caption, which matters when the figure appears in a paper alongside several others.

Beyond shapes, the diagram needs three kinds of labels: what each step produces, the reason for each exclusion, and the number of units passing each point. Numbers are what make the diagram auditable. A reader who can see that two hundred units entered and one hundred and sixty reached the analysis knows immediately what the effective sample is and where the losses occurred.

01

Set the entry point and define precisely what qualifies a unit for inclusion.

02

Draw each measurement or procedure as a process box and label the variable or artifact it produces.

03

Add a decision diamond wherever a unit may leave the study, with the reason on the exit branch.

04

Record the number of units at every transition, including the number excluded at each decision.

05

Close with the analysis stage, naming the statistical or analytic method applied to the final set.

06

Check that the numbers balance at every junction before publishing the figure.

Making decisions and branches explicit

Decision points are the most informative part of a methodology flowchart and the part writers most often hide. Where a study applies eligibility criteria at multiple points, or switches analysis depending on a data property such as a distributional assumption, the diagram should show the branch and its trigger. Hiding the branch does not remove it; it only removes the reader's ability to evaluate it.

Branches should be mutually exclusive and labeled with the condition that routes a unit one way or the other. If a branch has no label, the reader cannot tell whether the routing was pre-specified or decided after seeing the data, which is a distinction that affects how much weight the results carry.

  • Label every branch with the condition that determines the route.
  • Keep decision criteria pre-specified rather than described as they arose during analysis.
  • Show destination for units that leave, including follow-up or referral where relevant.
  • Avoid loops unless the design genuinely includes iteration, and label the iteration criterion if it does.

Where controls, blinding and randomization fit

Design features that protect against bias should appear in the diagram at the point where they operate, not in a separate box. Randomization is a step that allocates units to conditions. Blinding is a condition that applies during measurement. A control condition is a parallel path that runs alongside the intervention path. Placing them correctly lets a reader see the design's protections at a glance.

Timing is equally important. If outcomes are measured at several points, the diagram should show them on a horizontal axis or as clearly labeled stages, because the difference between a concurrent and a delayed measurement changes what the study can conclude. Ambiguity about timing is a common reason reviewers ask for the study to be clarified before acceptance.

Design featureWhere it appears in the diagram
Inclusion criteriaAt the entry point, before any measurement
RandomizationAs a step between baseline and intervention, with the comparison arm as a parallel path
BlindingAs a condition attached to the measurement step
Follow-up measurementAs a later stage on the timeline, with its interval stated
Analysis set definitionAt the final junction where exclusions are completed

Checking the flowchart against your text

Once drawn, the flowchart becomes a test of your methods section. Every arrow in the diagram should correspond to a sentence in the text, and every procedure described in the text should appear somewhere in the diagram. Where the two disagree, one of them is wrong, and finding that before a reviewer does is the entire benefit of drawing the figure early.

Pay particular attention to the numbers. The count at entry, the count excluded at each stage, and the count analyzed must reconcile exactly with what the text reports. Reconciliation failures are among the most common technical objections raised in review, and they are almost always present in the diagram before they are noticed in the prose.

Worked example

Worked example: a cohort study flowchart

A team studies whether a structured rehabilitation program improves mobility six months after joint replacement. The flowchart begins with an entry box defining eligibility: adults scheduled for a first unilateral replacement at one of three participating hospitals, able to provide consent, and resident within the hospital catchment for follow-up.

A decision diamond then removes units that do not meet criteria or decline, with a separate count for each reason. Allocation follows as a process box, generating two parallel paths, one for the structured program and one for usual care. Baseline measurement sits before allocation, which the diagram makes visually obvious by placing it to the left of the split.

Two follow-up stages appear on the right, at three and six months, each attached to a measurement step and a note that outcome assessors were blinded to allocation. A later diamond removes participants who missed both follow-up measurements, with the reason recorded as loss to follow-up, and the remaining units feed into the analysis box, which names the adjusted comparison against the baseline score.

The numbers reconcile at every junction: the entry count equals the sum of those removed and those allocated, and the analysed count equals the allocated count minus the losses in each arm. The team also annotated the pre-specified analysis set in a footnote, which pre-empted the question of whether participants who switched arms were included.

Common mistakes

A flowchart of activities rather than unitsTrace one participant, sample or case from entry to analysis so exclusions and losses become visible.
Unlabeled decision branchesState the condition that routes each unit at every diamond, so the reader can see whether the criterion was pre-specified.
Numbers that do not reconcileAdd counts at every transition and confirm that entry equals allocated plus excluded at each stage.
Design safeguards in a separate boxPlace randomization, blinding and controls at the stage where they operate, not in an isolated note.

Frequently asked questions

Should a methodology flowchart include the statistical tests?

Name the analytic approach in the final stage rather than listing every test in the flow. The diagram should show what data reaches the analysis and in what form. Detailed model specifications belong in the text, where they can be fully explained.

How detailed should a methods figure be for a journal submission?

Detailed enough that a reader can audit the effective sample size and see where bias protections apply, and simple enough to read at column width. Many journals recommend a participant flow diagram for intervention studies, so check the specific author guidelines first.

Can an AI tool produce a methodology flowchart I can submit?

It can lay out the steps you have already specified, which is useful when making a clean version of a sketch. It cannot verify your counts, and it will not know your field's reporting conventions. Treat the output as a draft to check line by line against your protocol.

What if my study has no exclusions at all?

Then say so explicitly and show the counts anyway. A diagram with no losses is credible only when the numbers support it. Reviewers treat missing counts as an omission rather than as evidence of a perfectly executed study.

Sources and further reading

  1. PRISMA - reporting guidance for systematic reviewsPRISMA Executive
  2. CONSORT - reporting guidance for randomised trialsCONSORT Group
  3. STROBE - reporting guidance for observational studiesSTROBE Initiative
  4. ARRIVE - reporting guidance for animal researchARRIVE Guidelines
  5. JATS - Journal Article Tag Suite, the standard article structureNational Center for Biotechnology Information, U.S. National Library of Medicine

These are the organisations whose published guidance the recommendations on this page follow. Listing them does not imply that they endorse Pathform AI, and this page is not affiliated with them.

Put this guide into practice

Get the editable layouts this guide describes, ready to retype and submit.

Browse template packs