From Data to Defensible Results: A Reporting Checklist for Researchers
Align questions, variables, methods and interpretation while making uncertainty visible.
Start before running the analysis
Translate each research question into an estimand, comparison or mathematical object. Define outcomes, predictors, units, exclusions and transformations before exploring results. Separate confirmatory analysis from exploratory work.
Choose methods because their assumptions match the data-generating process and question, not because they are familiar. Document missingness, dependence, multiplicity and measurement limitations.
Report more than a p-value
Provide effect sizes or substantive magnitudes, uncertainty intervals, denominators and enough descriptive information to interpret the analysis. Explain model specification and diagnostics. Statistical significance does not establish importance, causality or replication.
Use tables and figures to reveal structure rather than duplicate prose. Label units, samples and uncertainty clearly.
Make the result reproducible
Preserve data provenance, code, software versions, decision logs and an analysis-ready data dictionary. Share materials when ethical and lawful, or explain restrictions and access procedures.
In the discussion, distinguish what the analysis estimates from what the broader theory suggests. State sensitivity analyses and limitations that could change the conclusion.