Figure Legend Auto-generation
Use this page to understand how the current product can draft figure-legend text when the necessary sample-size, preprocessing, and reporting metadata are present, and why the result still needs human review.
What this page generates or verifies
In the current product, this page refers to a legend draft helper. It does not mean that every chart family receives a fully finished legend with no user editing.
The current draft can help assemble legend text around:
- analyzed sample size and exclusions
- group-wise attrition summaries when available
- preprocessing disclosures
- outlier-handling notes
- a prompt to include inferential details in the final legend when the figure is claim-bearing
- a major caveat or safety note carried from the current snapshot state
Inputs
The legend draft is built from the same general figure-safety and reporting context that feeds export review.
- sample attrition and analysis-unit summaries
- preprocessing and outlier-handling snapshots
- reporting completeness state
- batch, missingness, denominator, survival, or other caveat summaries when present
This means support is conditional. The content can differ by figure type, by available metadata, and by how complete the current snapshot bundle is.
If key metadata is absent or only partially resolved, the legend helper can fall back to placeholders, omit some disclosure language, or leave inferential details for manual completion rather than guessing.
Outputs
The current implementation supports a draft legend string, not a universal final legend contract.
- a
Draft / Needs reviewlegend-oriented text bundle shown next to the Methods draft - placeholders when essential details are still missing
- inferential reminders to include exact test, correction, effect size, CI, and sample size in the final legend
- disclosure-oriented text that can help the user finish a submission-ready legend manually
A typical output might read like a structured draft rather than a polished final caption, for example: "Draft / Needs review: Data are shown for analyzed samples after preprocessing; exclusions and attrition should be reported here; include exact test, correction, effect size, CI, and n before submission."
For Group Comparison figures, this is most relevant for pages such as Box Plot, Violin Plot, Strip Plot, and Group Comparison Mean and SEM.
Blocking or caution logic
- The current behavior is snapshot-driven and not strongly chart-family-specific in a way that supports a blanket public claim.
- Inferential details such as effect size, confidence interval, and sample size are included conditionally, depending on available metadata and disclosure state.
- A figure legend draft should still be reviewed against Effect Size, CI, and N Reporting before submission.
- This page should not claim fully automatic legend generation or that every required disclosure is always present.
How to use it
- Use the legend draft as a first assembly pass.
- Check whether the chart and current metadata actually support the disclosure you want to keep.
- Review and edit the final wording before manuscript or journal submission.
- Treat the draft as figure-specific starter text, not as proof that journal-specific wording, reporting rules, or all required disclosures have already been satisfied.