How to Create Figures for Research Publications
A figure should communicate one clear message about the data, such as a trend, a comparison, a distribution or a flow of participants. The chart type, labels, scale and caption should make that message clear and honest.
Choose the chart for the message
| Message | Suitable chart |
|---|---|
| Distribution of a numerical variable | Histogram, box plot, violin plot |
| Comparison of groups | Dot plots with intervals, box plots, bar chart for counts or proportions |
| Change over time | Line chart, with intervals |
| Relationship between two variables | Scatter plot |
| Survival | Kaplan–Meier curve with number at risk |
| Meta-analysis | Forest plot, funnel plot |
| Participant flow | Flow diagram (CONSORT, PRISMA) |
Good practice
- Label axes with units and use a sensible, honest scale (do not truncate bars).
- Show individual data points when numbers are small.
- Use colourblind-friendly palettes and distinguish groups without relying on colour only.
- Use fonts and line widths that remain legible when reduced to journal column width.
- Provide uncertainty (confidence intervals) where appropriate.
- Export at the resolution and format the journal requires.
Integrity
Do not selectively crop or adjust images in ways that alter their meaning. Describe any image processing; journals and COPE guidance treat inappropriate manipulation as a research integrity issue.
Illustrative example
Invented for teaching. A bar chart of mean blood pressure with truncated axis exaggerates a 3 mm Hg difference. Replacing it with a dot plot of individual values and the mean difference with its confidence interval shows the true scale of the effect.
Checking a figure before you submit it
Print the figure at the size it will appear in the journal and read the smallest text. Ask whether someone could understand the message in five seconds. Check that the numbers on the figure match the text and tables, that every group in the legend appears in the figure and that colours remain distinguishable in greyscale. Keep the source data and code so that the figure can be regenerated if a reviewer asks for changes.
Common pitfalls
- Chart junk and 3D effects.
- Missing axis labels.
- Colour schemes that cannot be read in greyscale.
Key takeaways
- One message per figure, honestly scaled and clearly labelled.
- Follow the journal's technical rules.
References and further reading
- International Committee of Medical Journal Editors (ICMJE). Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. Link
- Committee on Publication Ethics (COPE): guidance and flowcharts. Link
- Schulz KF, Altman DG, Moher D; CONSORT Group. CONSORT 2010 statement: updated guidelines for reporting parallel group randomised trials. BMJ. 2010;340:c332. doi:10.1136/bmj.c332
Related reading
This page is general educational information about research methods. It is not medical, legal or regulatory advice; follow your institution's ethics board and your target journal's instructions.