Use this skill to make manuscript statistics transparent, reproducible, and appropriately bounded. It is a reporting and review skill, not a substitute for a statistician reanalysing raw data unless the user supplies the data and explicitly asks for computation.
Default stance
- Prioritize design transparency over decorative statistical language.
- Separate three questions: what was measured, what unit was analysed, and what inference was claimed.
- Treat the independent experimental unit as the default
n; do not silently treat cells, fields of view, repeated readings, spectra, model runs, or technical replicates as independent biological or experimental samples. - Prefer effect sizes, uncertainty intervals, sample sizes, and exact test definitions over significance-only phrasing.
- State missing information as
AUTHOR_INPUT_NEEDEDinstead of inventing sample sizes, tests, software, corrections, exclusion rules, randomization, or blinding. - If a journal-specific instruction, study-type guideline, or field standard conflicts with this skill, follow the more specific source and mark the source used.
Accepted inputs
The skill may receive:
- a Statistical analysis / Methods subsection
- Results paragraphs containing test statistics or p values
- figure panels, legends, captions, or source-data notes
- reviewer comments about statistics
- author notes in Chinese or English
- tables of reported comparisons
- raw or summary data, only when the user wants a concrete reanalysis or figure-statistics check
If the input is partial, run a bounded audit and state which parts cannot be assessed.
Workflow
- Classify the task. Decide whether the user wants audit, rewrite, draft, reviewer-response support, figure-statistics alignment, or data-backed reanalysis.
- Extract the design. Identify groups, treatments, time points, endpoints, blocking factors, repeated measures, randomization, blinding, exclusions, and missing-data handling.
- Define
nand replication. Separate independent experimental units, biological replicates, technical replicates, repeated measures, cells/fields/subsamples, simulations, and pooled observations. - Map claims to analyses, both ways. For each result claim, record the comparison/model, test family, assumptions, correction strategy, effect estimate, uncertainty, and exact p-value policy. Then map the other way: every test, statistic or interval that Statistical analysis declares must correspond to a result the paper reports. A declared-but-unused statistic (for example a paired t declared for every comparison while the text reports only mean differences and sign consistency across seeds) is removed from the Methods, not added to the Results.
- Check common failure modes. Use
references/common-failure-modes.mdwhen the text involves nested data, many comparisons, cell-level measurements, interaction claims, correlations, regression, outliers, small samples, or significance-only reasoning. - Check reporting completeness. Use
references/statistical-reporting.mdto verify that Methods and Results give enough information for readers and reviewers to understand the analysis. - Align figure statistics. Use
references/figure-statistics.mdwhen figure legends, panel labels, stars, error bars, box plots, violin plots, source data, or supplementary figure notes are involved. - Draft or revise. Produce ready-to-paste text whose claims stay within the supplied design and evidence. Do not upgrade statistical association into mechanism or causality. Report precision as numbers (interval, n), not as a caveat sentence; each reporting statement (unit and n, P-value policy, multiplicity, blinding, exclusions) appears once, in Statistical analysis. For descriptive designs without inferential tests, use the descriptive variant in
references/statistical-reporting.md. - Run final QA. Use
references/reviewer-checklist.mdbefore final delivery for severity labels, unresolved author questions, and reviewer-facing risk.
Output format
Unless the user asks for another format, return:
Statistics review scope
- Input reviewed:
- Boundary / missing materials:
- Study design readout:
- Independent unit and replication readout:
Major statistical issues
- [P0/P1/P2] Issue:
Evidence from supplied text:
Why it matters:
Fix:
Ready-to-paste revision
[Rewritten Statistical analysis / Results / figure legend text]
AUTHOR_INPUT_NEEDED
- [short factual questions only]
Reviewer-risk note
- What a statistical reviewer may still challenge:
The issue list and the reviewer-risk note are for the author. Neither goes into the manuscript; the
manuscript receives only the ready-to-paste revision, written without audit or defensive voice (see
anti-defensive-writing).
For a clean drafting request with enough information, skip the long issue list and return:
Draft Statistical analysis
[ready-to-paste text]
Reporting notes
- n definition:
- tests/models:
- multiple comparisons:
- software/version:
- unresolved fields:
Red lines
- Do not invent p values, sample sizes, degrees of freedom, confidence intervals, software versions, correction methods, preregistration, exclusion rules, or power calculations.
- Do not recommend a statistical test as final when the unit of analysis or design is unclear.
- Do not accept
n = number of cells/images/measurementsas independent replication without checking the experimental hierarchy. - Do not use “significant” as a synonym for important, large, causal, or biologically meaningful.
- Do not hide non-significant or weak results by rewriting them into stronger claims.
- Do not give medical, regulatory, or clinical-trial statistical advice beyond reporting checks unless the user provides the relevant protocol and asks for bounded manuscript wording.
Related files
| File | Open when |
|---|---|
| references/source-basis.md | You need the source hierarchy or want to justify why the skill emphasizes transparency, reproducibility, and design reporting |
| references/statistical-reporting.md | You are drafting or auditing Statistical analysis, Methods, Results, or Supplementary Methods text |
| references/common-failure-modes.md | You see nested measurements, many comparisons, interaction claims, correlation/regression, outliers, tiny samples, or overstrong p-value language |
| references/figure-statistics.md | You are checking figure legends, panel statistics, error bars, stars, box/violin plots, source-data notes, or graphical reporting |
| references/reviewer-checklist.md | You are finalizing an audit or preparing a reviewer-facing risk summary |
Source hierarchy
Use sources in this order:
- User-supplied manuscript, data, protocol, statistical analysis plan, reviewer comments, and journal instructions.
- Nature Portfolio reporting standards and reporting-summary requirements.
- Nature Methods / Nature Portfolio statistics guidance summarized in
references/source-basis.md. - Study-type reporting guidelines where relevant, for example CONSORT, STROBE, PRISMA, ARRIVE, or field-specific community standards.
- Conservative statistical reporting practice.
If the supplied material is insufficient for a defensible statistical recommendation, ask for the missing design facts or provide a bounded wording option rather than guessing.
Provenance: github.com/Yuan1z0825/nature-skills; original skill nature-statistics, license Apache-2.0. Imported into the local catalog; body preserved, only cross-references adapted.