Statistics & Reproducibility (sci-statistics)

When to trigger

• Results report P values but not effect sizes or n.
• "Three independent experiments" is claimed but replication is unclear.
• Multiple comparisons are run with no correction.
• A reviewer is likely to ask "were analyses pre-specified?" and there's no answer.

The reporting backbone (every quantitative claim)

Each claim needs: effect size + uncertainty + n + test + what n means.

• n stated, with the unit of replication (biological vs technical replicates; cells vs animals vs experiments).
• Effect size with 95% CI (preferred) or SD/SEM clearly labeled — not P alone.
• Exact P values (e.g., P = 0.013), not "P < 0.05", unless extremely small.
• Test named and justified (and its assumptions checked: normality, variance homogeneity, independence).
• Multiple comparisons corrected (Bonferroni/Holm/FDR) when many tests are run.