Pathform AI Browse packs

Research Figures

Research Figure Types: Choosing the Right One for Your Message

Most researchers have a default figure type they reach for, usually the one used in their first paper. That default is efficient until it meets a claim it cannot express, at which point the figure quietly misrepresents the argument and the text has to work harder than it should.

At a glance

Cluster
Research Figures
Reading time
5 min
Format
Concept walk-through
Sources
5 listed

Guide clusterResearch Figures

Direct answer

Choose a research figure type by matching it to the kind of claim you are making: flowcharts show process and decision order, path diagrams show causal relationships, schematics show structure and composition, timelines show sequence over time, and roadmaps show how stages of work produce evidence. The figure type should follow the claim, not the habit.

01Matching figure type to claim type

Figures carry different kinds of information, and the mismatch between a figure and its claim is a common source of confusion in papers. A process claim needs a diagram that shows order and decision, a relational claim needs one that shows direction and dependency, and a compositional claim needs one that shows structure and scale. When a figure is asked to do work it was not designed for, readers compensate by reading the caption more carefully, which is a sign the figure is underperforming.

The table below maps common claim types to figure types, with the question each figure answers for a reader.

Claim typeFigure typeQuestion it answers for the reader
A process with decisionsFlowchartIn what order do things happen, and where do paths split
A relationship between constructsPath or conceptual diagramWhat influences what, and in which direction
A structure or compositionSchematic or cutawayWhat are the parts, and how do they fit together
A sequence over timeTimeline or phased diagramWhat happens when, and what depends on what
A program of work producing evidenceRoadmap or logic diagramHow does the work build up to a defensible conclusion

02Flowcharts and when they help

Flowcharts are the most frequently used and most frequently overused research figure. They excel where a reader needs to see decisions and their consequences, such as participant flow in a trial or a screening procedure. They perform poorly when the underlying structure has no decisions, because the flowchart format implies branching that does not exist and readers start looking for missing branches.

If your process has no decision points, a simple linear diagram is honest and easier to read. If it has many, consider whether a table would serve better, since some procedures are clearer as a list of conditions with their outcomes than as a network of diamonds.

A flowchart with no decisions is usually a list pretending to be a process.

03Path and conceptual diagrams

Path diagrams show constructs and directional relationships, and they are the natural format for hypotheses about influence. Their strength is that they make the structure of an argument visible: what is expected to affect what, and which factors modify or carry those effects. Their weakness is that they invite over-specification, with so many arrows that the central argument disappears.

A useful discipline is to distinguish the primary path from the secondary effects by visual weight. The relationships central to your research questions should be drawn heaviest, with moderators and controls drawn lighter. The reader then knows immediately which arrows carry the argument and which are context.

Draw the relationships that answer your research questions with the most visual weight.
Distinguish moderators, which act on a relationship, from mediators, which sit on the path.
Group control variables rather than drawing each one as a separate box.
State in the caption what the arrows represent, since conventions vary between fields.

04Schematics, timelines and roadmaps

Schematics show structure, and their value depends on accurate proportion and clear labeling of parts. A schematic that misstates relative size or omits a component misleads readers even when the accompanying text is correct, because visual information is remembered more reliably than prose. Where proportions matter, use a scale and say so.

Timelines and roadmaps both organize work over time, and they are often confused. A timeline answers when things happen and is useful for planning and for reporting progress. A roadmap answers how the argument develops and what each stage establishes, which is why a good roadmap can be read without any dates at all. Choose based on the question the reader will be asking.

05Combining types without creating clutter

Complex projects often need more than one figure type, and the temptation is to combine them into a single composite. Composite figures usually fail because the reader has to hold two different mental models at once. It is better to use separate panels, each with its own caption and its own question, and to let the text connect them.

Where combining is unavoidable, keep one type dominant and the other reduced to an inset. A roadmap with a small timeline strip along the bottom can work, because the timeline is clearly subordinate. A roadmap with an embedded flowchart of equal visual weight usually cannot, because there is no obvious order in which to read the panels.

  1. Decide which figure type carries the primary claim.
  2. Build that figure first and write its caption as a standalone unit.
  3. Add the secondary type as an inset or a separate panel with its own caption.
  4. Remove the composite if a reader cannot say which panel to read first.

Worked example

Worked example: choosing a figure for a methods paper

A team wrote a methods paper describing a new approach to combining survey and administrative data. Their first figure was a flowchart, inherited from a previous paper, showing the sequence of processing steps as a chain of rectangles. Reviewers found it clear about order but silent about the design rationale.

The problem was one of claim type. The paper's central claim was not about a sequence of operations but about a relationship between two data sources and a correction strategy: administrative records were expected to correct systematic bias in self-reported measures, conditional on linkage quality. That is a relational claim, and the flowchart could not express it.

The team replaced the figure with a path diagram. Survey responses and administrative records appeared as parallel inputs, joined at a linkage stage whose quality was drawn as a moderator acting on the correction path. Measurement error in the self-report appeared as a mediator, since the correction operated by reducing it. The processing sequence from the original figure was retained as a small numbered list in the caption.

The revised figure made the rationale visible in a way the flowchart could not, and the reviewers' second report noted that the design intent was now clear from the figure alone. The team kept the original flowchart as a supplementary panel, where it served as a practical implementation guide rather than as the paper's conceptual anchor.

Common mistakes

Avoid thisDo this instead

Using a default figure type regardless of the claim

Identify the claim type first, then choose the figure type that answers the reader's question about it.

A flowchart with no decisions

Use a simple linear diagram or a numbered list when no real branching occurs.

Over-specified path diagrams

Give primary relationships the most visual weight and group controls so the central argument stands out.

Composite figures with two competing types

Split into panels with separate captions, keeping any secondary type clearly subordinate.

Frequently asked questions

How many figures should a research paper contain?

Enough to carry the claims that need visual support, and no more. Many journals set limits, and reviewers tend to comment on figures that could be described in a sentence. If a figure is only restating a number, move it to supplementary material.

Should figures be understandable without reading the text?

A reader should be able to state what a figure shows from the figure and its caption alone. The interpretation may need the text, but the content should not. This is why caption writing deserves real attention rather than a single sentence.

Can an AI tool recommend the right figure type?

It can suggest a type based on the kind of claim you describe, which is useful when you are unsure. The recommendation will be generic, and your field's conventions matter, so check what comparable papers use before committing to a format.

What if my figure type is unusual for my field?

Use it if it communicates better, and say so in the caption. A short note explaining what the diagram represents removes the novelty penalty. Fields change their conventions because individual authors make this argument well.

Sources and further reading

  1. ColorBrewer - colour schemes tested for print and screen legibilityCynthia Brewer, Pennsylvania State University
  2. WCAG 2.1 at a glance - contrast and non-text accessibility requirementsWorld Wide Web Consortium (W3C)
  3. Information for authors - artwork and submissionElsevier
  4. Author services - submission and figure guidanceWiley
  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