Multiplicity

TipExecutive summary

This page outlines how I approach multiplicity control in confirmatory analyses to support valid inference, clear decision logic, and review-ready reporting.

Why it matters

When multiple hypotheses or endpoints are tested, the risk of false-positive findings increases. A clear multiplicity strategy helps ensure conclusions are:

  • Statistically valid
  • Prespecified
  • Traceable to SAP/TLF logic
  • Clearly interpretable for reviewers

What I focus on

  • Defining hypothesis families
  • Clarifying testing order / hierarchy
  • Distinguishing confirmatory vs descriptive results
  • Aligning multiplicity logic across:
    • protocol/SAP language
    • TLF shells
    • output footnotes and interpretation

Practical contribution areas

  • Multiplicity planning support during study design/SAP development
  • Testing strategy traceability in outputs
  • Review-friendly labeling of inferential vs nominal results
  • Consistency checks across tables and figures

Common pitfalls I help prevent

  • Ambiguous testing hierarchy
  • Mixing confirmatory and exploratory conclusions
  • Inconsistent SAP vs output interpretation language
  • Nominal p-values presented as confirmatory evidence