Literacy teaches the skills for evaluating evidence directly: distinguishing correlation from cause, reading confidence intervals, converting relative risk into absolute terms, spotting undeclared conflicts, and recognizing when a sample is too small. Examples come from real published papers. For anyone wanting to check a health claim independently.
Practical tools for judging evidence: relative versus absolute risk, confidence intervals, control groups, disclosures, and wording that hides weak data.
A p-value is the probability of data at least as extreme as the observed data if the null hypothesis were true — not the probability that the hypothesis is true, and not the size of the effect.
Peer review is a check by outside experts on a study's methods and reasoning before publication — a filter that improves papers but cannot verify data, replicate results, or catch deliberate fraud.
A 2025 analysis in PNAS finds reviewers often disagree with each other — and shows which specific fixes actually help.
A five-step checklist — source, study, design, size, conflicts — separates checkable findings from marketing wearing science's clothes.
Writing down the plan before the data arrive — outcomes, sample size, analysis — forecloses the quiet ways results get shaped afterward.
Exaggeration often enters the pipeline before any journalist writes a word — in the university's own announcement.
Science is cumulative by design: single findings are provisional by default, and the checks that confirm or retire them take time.
Fake peer review leaves fingerprints: promises of days-fast review, unlisted editors, and fees that surface only after acceptance.
The range around a study's headline number — not the number itself — shows how much the result could plausibly move.
The same finding can be reported as a 50 percent increase or as a few extra cases per thousand — both true, only one honest without the other.