Correlation vs. Causation: The Most Common Reasoning Error
Confusing correlation with causation is arguably the single most common error in everyday reasoning, media reporting, and even scientific interpretation. Two things occurring together does not mean one causes the other. Understanding why requires examining the many ways correlations can arise without direct causation.
What Correlation Actually Means
A correlation exists when two variables tend to change together. A positive correlation means they rise and fall together; a negative correlation means one rises as the other falls. Correlation can be measured statistically (the correlation coefficient ranges from -1 to 1), but the key point is that correlation is a purely mathematical relationship.
Ice cream sales and drowning deaths are positively correlated. Does ice cream cause drowning? Of course not. Both are caused by a third variable: hot weather. When the weather is hot, people buy more ice cream and also swim more, leading to more drowning incidents. This is the classic illustration of why correlation does not imply causation.
Yet every day, news headlines trumpet correlations as if they were causal findings: 'Study links coffee to longer life,' 'Research connects video games to violence.' The word 'links' is doing heavy lifting, implying causation without actually establishing it.
Why Correlations Occur Without Causation
There are several reasons two variables can be correlated without one causing the other. The most common is confounding: a third variable causes both. Hot weather confounds the ice cream-drowning correlation. Poverty confounds many correlations between social factors and health outcomes.
Reverse causation is another explanation. If we observe that people who exercise more are healthier, it might be that health enables exercise rather than exercise causing health. Sick people may exercise less because they are already unwell.
Coincidence and data mining also produce spurious correlations. If you test enough pairs of variables, some will correlate by pure chance. The divorce rate in Maine correlates with per capita margarine consumption -- this is meaningless noise in the data, not a real relationship.
How to Establish Causation
Establishing causation requires more than correlation. The gold standard is the randomized controlled experiment, where subjects are randomly assigned to treatment and control groups. Randomization ensures that confounding variables are equally distributed between groups, so any difference in outcomes can be attributed to the treatment.
When experiments are not possible (you cannot randomly assign people to smoke for 30 years), researchers use criteria like the Bradford Hill criteria: strength of association, consistency across studies, specificity, temporal precedence (cause before effect), dose-response relationship, biological plausibility, and coherence with known facts.
In everyday reasoning and debate, establishing causation requires showing a plausible mechanism (how would A cause B?), temporal order (A precedes B), and ruling out major alternative explanations. Simply pointing to a correlation is never sufficient.
Exploiting the Confusion in Debate
Debaters routinely exploit the correlation-causation confusion. A politician might point to a correlation between their policy and a positive outcome, implying causation. The effective counter is to identify alternative explanations: Was the trend already occurring? Are there confounding factors? Could the causation run in the opposite direction?
Conversely, when your opponent dismisses a causal relationship by saying 'that is just correlation,' recognize that correlation is necessary for causation even if it is not sufficient. A complete absence of correlation is evidence against causation. The appropriate response is to provide additional evidence beyond the correlation -- mechanism, temporal ordering, control for confounders -- to build the case for causation.
- •Correlation means two variables change together; it does not mean one causes the other.
- •Confounding variables, reverse causation, and coincidence all produce non-causal correlations.
- •Establishing causation requires controlled experiments or careful application of causal criteria.
- •In debate, always ask: Is there a confound? Could causation run the other way? Is this coincidence?
- •Correlation is necessary for causation but never sufficient on its own.