Thursday, June 30, 2016

Read This, and You’ll Remember Why Correlation Doesn’t Equal Causation

Thank you, Roy Germano and Christopher Jon Sprigman, for explaining this much-loved but little-understood axiom:

Correlation is not causation. We hear this phrase all the time, but what does it actually mean?

“A famous example of the perils of using a correlation to suggest a causal relationship is the association between ice cream sales and drowning deaths. Like clockwork every June, ice cream sales go up, and so too do the number of deaths by drowning. Without a deeper understanding of how the world works, we might be tempted to use this correlation to tell a story about how eating ice cream causes people to drown — especially if we’re a company that makes snow cones and thinks that warning people off of ice cream will boost sales of our competing product.

“Now, most people are nowhere near gullible enough to fall for the argument that ice cream causes drowning — even if it’s made in a paper full of scientific-sounding metrics and graphs that show ‘strong to very strong’ correlations. Most people would realize that the correlation between ice cream sales and drowning is spurious. The only link between the two variables is that they are both independently related to the weather. When it’s hot, people look for ways to cool off. They go out for ice cream. They also go swimming. Unfortunately when more people go swimming, drowning deaths increase.