What it shows
A causal diagram of a confounder, using the familiar example that ice cream sales and drownings rise together. Hot weather (Z) causes both: more people buy ice cream (X), and more people go swimming, so drownings (Y) rise.
There is no causal arrow between ice cream sales and drownings. They are joined only by a dotted line for the correlation that appears in the data.
How to read it
Solid arrows mean “causes”; follow them out from hot weather.
The dotted line between X and Y marks a correlation only, with no causal link, as the key states.
The variables
- Z, the confounder: hot weather.
- X: ice cream sales.
- Y: drownings.
- Z → X arrow: “More people buy ice cream”.
- Z → Y arrow: “More people go swimming”, the route by which hot weather raises drownings.
- X and Y: “Correlated but neither causes the other”.
Ways to use it in teaching
Students propose a confounder for a correlation reported in the news, then draw its diagram.
Introduce controlling for Z: what would the X–Y correlation look like on days of the same temperature?
In a research design session, it sets up the difference between observational studies and experiments.
For how studies name and control their variables, see our guide to research variables.
Sources
- Confounding. (n.d.). In Wikipedia. Retrieved October 18, 2026, from https://en.wikipedia.org/wiki/Confounding
- Correlation does not imply causation. (n.d.). In Wikipedia. Retrieved October 18, 2026, from https://en.wikipedia.org/wiki/Correlation_does_not_imply_causation
- Spurious relationship. (n.d.). In Wikipedia. Retrieved October 18, 2026, from https://en.wikipedia.org/wiki/Spurious_relationship
