nature-statistics

Nature Statistics is a manuscript statistical-reporting skill that audits, revises, or drafts the statistics content of papers aimed at Nature and other high-impact journals. It solves a recurring cause of reviewer pushback and retraction risk: statistical reporting that is opaque, overstates precision, or conflates measurement units with inference. It is explicitly a reporting and review skill, not a substitute for a statistician reanalyzing raw data unless the user supplies the data and asks for computation.

Its default stance prioritizes design transparency over decorative statistical language, separates what was measured from what unit was analyzed and what inference was claimed, treats the independent experimental unit as the default n, and prefers effect sizes and uncertainty intervals over significance-only phrasing. It follows a structured workflow: classify the task (audit, rewrite, draft, reviewer-response, figure alignment, or reanalysis), extract the design, define n and replication level, map each claim to its analysis, then check common failure modes (pseudoreplication, uncorrected multiple comparisons, wrong interaction inference, analysis-unit mismatch, small-sample overclaim, and correlation-as-causation) using bundled references. It aligns figure statistics (error bars, stars, n per panel), checks reporting completeness, and marks unknowns as AUTHOR_INPUT_NEEDED rather than inventing sample sizes, tests, or corrections. Output is severity-labeled issues plus ready-to-paste revised text. It handles English and Chinese drafts.

Target users are researchers, corresponding authors, and lab members preparing high-impact submissions or responding to reviewer statistics comments, especially those needing publication-ready Statistical analysis text and a conservative audit of replication and inference claims.