RW Statistics Audit

Initiation

1. Read

   references/standalone.md

   ,

   references/method.md

   and

   references/standards.md

   .
2. Read the materials provided by the user in this session; if no materials are available, only complete the parts permitted by the current evidence.
3. Retrieve

   references/atoms.jsonl

   when rule-based judgments are needed; read

   references/cases.md

   when encountering similar tasks.
4. Read

   references/domain-guide.md

   when selecting methods or tools, and use

   assets/worksheet.md

   to organize deliverables.
5. Read

   references/acceptance.md

   .

   references/behavior-tests.json

   is only used for testing and not as factual information for user tasks.
6. When current literature, APIs, reporting specifications and journal requirements may change, open the official links in

   references/source-map.md

   to verify and record the date.

Work Phases

1. Fix the versions of manuscripts, tables, figures and analysis descriptions, and list key judgments, outcomes and analysis objects.
2. Identify observational units, independent analysis units, clustering structures, technical replicates, biological replicates and actual sample sizes.
3. Extract statistical methods, models, variables, comparisons, time points, covariates, missing data handling and multiple comparison handling item by item.
4. Verify whether method selection is consistent with study design, data type, distribution, pairing, repeated measures and independence.
5. Check whether effect estimates, uncertainty intervals, sample sizes, exact p-values, units, denominators and missing counts are reported and consistent across all sections.
6. Connect each statistical judgment in the main text, tables, figure captions and supplementary materials to the corresponding analysis and data levels.
7. Classify issues into errors, insufficient information, requiring statistical personnel review, and report approval; do not claim completion of recalculation when no data is available.

Operating Rules

• Observational units, independent analysis units and data points in figures are not the same concept and must be clearly described separately.
• Technical replicates cannot automatically increase independent sample size; biological replicates also need to be judged for independence according to the study design.
• Statistical method names cannot replace model settings, comparison objects, covariates and processing rules.
• p-values cannot alone explain effect size, direction, precision or practical significance.
• Lack of statistical significance cannot be rephrased as no effect; significance cannot be rephrased as importance or causality.
• Primary vs. secondary outcomes, prespecified vs. exploratory analyses, and single vs. repeated analyses must be separated.
• Handling of multiple comparisons, missing data, model assumptions and outliers must state whether they were prespecified and how they were implemented.
• Sample sizes must provide denominators, exclusions and missing counts for each group, not just the total n.
• Error bars in figures must specify the statistic, calculation level and sample size.
• Percentages must be traceable back to numerators and denominators; rounding must not cause conflicts between tables.
• Numerical conflicts between the main text, tables, figure captions and supplementary materials trigger a blocking state, and no option is automatically selected.
• Reporting specifications are used to check whether information is complete, not to prove model correctness, assumption validity or result reproducibility.
• Only conduct report review and do not generate recalculation results when there is no raw data, analysis code or sufficient aggregated data.
• Clinical, regulatory or high-risk conclusions require review by domain statisticians; Skill does not substitute for signature.

Output

• Corresponding table of analysis units, replication levels, outcomes and methods.
• List of gaps in numbers, models, assumptions and reports.
• Statuses of errors, insufficient information, requiring review, approval and invalidation.

Stop Conditions

• Do not grant the "approved" status to the statistical report when analysis units or replication levels are unclear.
• Stop drawing final conclusions when key numbers conflict in the main text, charts or supplementary materials.
• Do not claim reproduction or re-analysis when no data or analysis code is available.
• Do not judge clinical, policy or practical significance on behalf of authors based on p-values.

Independent Operation

• By default, do not read private workspaces, personal corpus directories or preset research-lab.
• Files, text, links and data provided by the user are inputs for the current task, not installation dependencies.
• When the network is unavailable, continue using stable methods within the Skill; mark parts requiring current facts as pending verification.
• Can connect with other research Skills in this package if they exist; when installed separately, directly return instructions for the next step without stopping the current task.

Source Discipline

• Separate user materials, public sources, current inferences and unknown matters.
• Reporting specifications only check report transparency, and do not automatically prove study design quality.
• Summaries of public sources are stored within the Skill; return to official pages when information about versions, fees, policies, APIs or current journal requirements is needed.
• Do not generate non-existent papers, data, DOIs, tool operation results or journal requirements.