05/15/2014 10:49 am ET | Updated May 15, 2014

Cheeky 'Spurious Correlations' Site Shows Why We Shouldn't Be Too Quick To Trust Graphs

You've probably heard that "correlation does not equal causation." And sometimes, correlation doesn't mean much at all.

If you need any convincing of that, have a look at Spurious Correlations. This funny-but-wise website, created by Harvard law student Tyler Vigen, shows a series of absurd graphs like these:

spurious correaltions

marriage rate kentucky

Presumably, eating sour cream doesn't cause motorcycle deaths. And it's a good bet that marriages in Kentucky have no effect on the risk of falling out of a fishing boat and drowning. More to Vigen's apparent point, just because one thing follows the same trend as another doesn't necessarily indicate that the things are connected in a meaningful way.

Vigen's critique of correlations doesn't mean, however, that he finds them devoid of meaning.

"It's up to us as rational thinkers to decide if there is actually a connection between the data or if it's merely a coincidence," he says in a YouTube video. "Correlations are a great starting place, but it's up to us to be critical of the correlations we see."

Dr. Steven Strogatz, a mathematics professor at Cornell University, seems to agree.

"On their own, correlations don't tell us much," Strogatz told The Huffington Post in an email. "If you have a huge enough collection of charts like these, you can cherry-pick them to find lots of meaningless correlations. The much more impressive thing is to be able to predict that two such graphs will continue to be tightly correlated in the future. If you can do that, you might be onto something."


Amazing Food Science Experiments