Association vs Causation

This is perhaps the single most important concept in all of research methodology: just because two things are related does not mean one causes the other. Association (correlation) means two variables tend to change together. Ice cream sales and drowning deaths are correlated — both increase in summer. But eating ice cream does not cause drowning. The common cause is a confounder: hot weather increases both. Causation means one variable directly produces a change in another. Smoking causes lung cancer — not merely a correlation but a direct biological mechanism. Sir Austin Bradford Hill proposed nine criteria to evaluate whether an association is likely causal: (1) Strength of association — strong associations are more likely causal. (2) Consistency — the association is observed across different populations and settings. (3) Specificity — the exposure leads to a specific outcome. (4) Temporality — the cause precedes the effect (the only absolutely required criterion). (5) Biological gradient (dose-response) — more exposure leads to more effect. (6) Plausibility — a biologically plausible mechanism exists. (7) Coherence — the association is consistent with existing knowledge. (8) Experiment — removing the exposure reduces the outcome. (9) Analogy — similar exposures cause similar outcomes. These criteria are guidelines, not a checklist. No single criterion (except temporality) is required or sufficient. They are used as a framework for evaluating the totality of evidence. In clinical practice, confusing association with causation leads to incorrect treatments and policies. It is one of the most common errors in medical reasoning.