Correlation and Causation
Tyler Vigen · 2015
"Two variables moving together over time is never, by itself, evidence that one causes the other — with a large enough dataset, data dredging alone will always turn up strong-looking correlations between things that have nothing to do with each other."
Bachelor's degrees awarded in psychology track almost perfectly with the number of groundskeepers in Utah, year over year. Nobody thinks psychology graduates are causing a groundskeeper boom. That's the entire lesson, made visible.
Tyler Vigen's Spurious Correlations project compares thousands of real datasets against each other and publishes the ones that happen to line up — margarine consumption tracking the divorce rate in Maine, or the number of judges in Indiana tracking viewership of a sitcom. These aren't fabricated; the numbers are real and the statistical correlation is real. What's missing is any causal mechanism connecting them. The project exists specifically to make an abstract warning ('correlation isn't causation') impossible to forget, by showing just how easily a strong correlation appears between variables that obviously can't be causing each other.
According to the Spurious Correlations project, what is 'data dredging' and why does it produce misleading results?
Read more about the topic
The explanation above is written with AI assistance. These are the originals — go to them to check it.
- Spurious CorrelationsTyler Vigen (tylervigen.com)
Deductive vs Inductive Reasoning
"Deduction moves from general premises to a certain conclusion (if the premises are true, the conclusion must be true); induction moves from specific observations to a probable — never certain — general conclusion, and confusing the two (treating an inductive leap as if it had deductive certainty) is a common source of bad reasoning."