Core Idea
Definition
Correlation means two variables are associated in their movement or occurrence, while causation means a change in one variable contributes to a change in the other.
In Plain English
Just because two things happen together does not mean one is causing the other.
How It Works
Patterns in data are tempting because they feel like explanations. But correlation can arise for many reasons: one variable may cause the other, the other may cause the first, both may be caused by a third factor, or the pattern may be coincidental. The model works by slowing the leap from pattern to mechanism. It asks what pathway would make this relationship causal, what alternative explanations exist, and what evidence could distinguish between them. It is central to science, policy, analytics, and everyday judgment because humans are natural pattern-seekers but not naturally disciplined causal thinkers.
When to Use
- •When interpreting data, trends, or research claims
- •When diagnosing why a change happened after an intervention
- •When evaluating advice based on anecdote or weak statistics
- •When testing product, marketing, or operational hypotheses
- •When building arguments about what actually drives an outcome
Examples
Everyday
You notice that on days you drink more coffee you feel more tired. Coffee may not be causing tiredness; tiredness may be causing coffee consumption.
Professional
A company sees that high-performing teams use a certain tool. The tool may help performance, or high-performing teams may simply adopt better tools more often.
Extreme Case
A public policy is credited for an improvement that actually came from unrelated demographic or economic changes happening at the same time.
Common Mistakes
- •Treating temporal sequence as proof of causation
- •Ignoring confounders that influence both variables
- •Assuming a visible pattern must have a simple direct cause
- •Rejecting all correlation as meaningless instead of seeing it as a clue
Limits & Failure Modes
- •Not every causal claim can be proven cleanly in messy real-world systems
- •Even causal relationships can be conditional or context-dependent
- •Strong correlation can still be useful for prediction even without causal clarity
- •Demanding perfect causal proof can create paralysis in practical decision-making
How to Practice
confounder search
For any apparent relationship, ask what third factor could plausibly be influencing both variables.
causal pathway test
Describe the exact mechanism by which one variable would produce the other and look for evidence that the mechanism is present.
prediction vs explanation
Separate whether a pattern is useful for forecasting from whether it truly explains what is happening.
Related Cognitive Biases
illusory correlation
People perceive relationships that are weak, nonexistent, or misunderstood.
post hoc fallacy
People assume that because one event followed another, the first caused the second.
confirmation bias
People highlight patterns that support the story they already want to believe.
Related Mental Models
Related Skills
Advanced Notes
Historical Origin
The distinction is foundational in statistics, epidemiology, economics, and philosophy of science.
Philosophical Context
It connects to theories of causal inference, counterfactual reasoning, and the challenge of learning mechanism from observation.
Further Reading
- The Book of Why by Judea Pearl and Dana Mackenzie
- Causal Inference in Statistics by Judea Pearl, Madelyn Glymour, and Nicholas P. Jewell
- Calling Bullshit by Carl T. Bergstrom and Jevin D. West