Michael Hiltzik

See some hilarious charts showing that correlation is not causation

Think correlation implies causation? This website will make you think again

The most powerful weapon that debaters wield against the unwary is causation: marijuana use leads to heroin addiction, pornography to rape, video games to mass murder, high consumption of margarine to divorces in Maine.

Whoops. The first three of these are common juxtapositions of parallel trend lines, and the fourth is from the website Spurious Correlations, the work of a Harvard Law School student named Tyler Vigen. (hat tip to Business Insider.)

Vigen's site aims to underscore the common warning that correlation does not prove causation by providing charts of absurd correlations. As a dividend, the site allows you to make your own!

As we write, some of the more outstanding correlations on the site include per capita consumption of cheese (U.S.) and number of people who died by becoming entangled in their bedsheets (correlation: 0.947), and people who drowned after falling out of a fishing boat with the marriage rate in Kentucky (0.952). Vigen seems to have a fascination with death rates, but let it go.

Underlying his effort is a serious issue. Real policy discussions are often infected by spurious implications of causality, offered disingenuously. One of the more notorious was offered a few years ago by Arthur Laffer, the conservative economist also responsible for the bogus state competitiveness index we reported on last week. In 2010, Laffer purported to demonstrate that unemployment benefits cause unemployment.

Laffer's device was a chart matching unemployment benefits against the unemployment rate. As it happened, Laffer's own graphic showed that the rise in the unemployment rate followed the rise in unemployment rather than preceded it, but the timing was so close that he tried to slip the correlation-means-causation argument past the goalie. 

The idea that correlation does not imply causation is so widely understood in the abstract that it has almost become a cliche. But specious correlations, often drawn from academic papers, are mainstays of attention-grabbing headlines the world over. We've found at least one website devoted to outing spurious correlations that make it into the press.

But Tyler Vigen's work may do more to undermine this threadbare debating tool than any number of earnest debunkings. Unless, that is, you really believe that the age of Miss America is closely related to murders by steam, hot vapors and hot objects.   

 

Copyright © 2016, Los Angeles Times
Loading
61°