For authors
Statistics & Data Analysis
Statistical, analytical and computational methods must match the scientific question, be reported transparently and permit independent evaluation or reproduction where feasible.
Required elements
- Define outcomes, endpoints, variables and analysis populations before reporting results.
- Describe sample-size or precision rationale where applicable.
- Report statistical tests, model assumptions, effect estimates, uncertainty intervals and multiplicity handling as appropriate.
- Describe missing-data handling, outlier rules and sensitivity analyses.
- For experimental optimization, report design-of-experiments factors, responses and model diagnostics.
- For pharmacokinetic, bioequivalence or dose-response analyses, report model structure and acceptance criteria.
- For machine-learning studies, separate training, validation and test procedures and report performance with uncertainty.
- For meta-analysis, justify effect measures, model choice, heterogeneity assessment and sensitivity analyses.
- State software, version and relevant packages; provide code or model files when feasible.
Method follows question
PSHP does not prescribe one statistical model for all studies. Authors must justify analytical choices in relation to design, data structure, estimand, assumptions and intended inference.