1. Start with the exact question
We define the population, exposure or intervention, comparator, outcome, and timeframe. “Does cannabis help?” is usually too broad to answer responsibly.
2. Find the primary record
We use papers, trial registrations, regulator documents, statutes, filings, and official statements. Other reporting can reveal a lead but does not replace the underlying record.
3. Match confidence to design
Systematic reviews and controlled human trials can support different claims than observational, animal, or laboratory research. We report sample size, comparator, product, major limitations, and conflicts when relevant.
4. Seek contradiction
We search for null findings, adverse outcomes, retractions, corrections, and competing interpretations. A single study rarely settles a broad question.
5. Date the conclusion
Medical and legal pages carry review dates. New evidence can strengthen, narrow, or reverse a conclusion.