The Internet Blowhard’s Favorite Phrase
Why do people love to say that correlation does not imply causation?
Karl Pearson, English mathematician and eugenicist, in 1912
Photo by Wikimedia Commons.
Depressed people send more email. They spend more time on Gchat. Researchers at the Missouri University of Science and Technology recently assessed some college students for signs of melancholia then tracked their behavior online. "We identified several features of Internet usage that correlated with depression," they said. Sad people use IM and file-share. They play video games. They surf the Web in their own, sad way.
Not everyone found the news believable. "Facepalm. Correlation doesn't imply causation," wrote one unhappy Internet user. "That's pretty much how I read this too… correlation is NOT causation," agreed a Huffington Post superuser, seemingly distraught. "I was surprised not to find a discussion of correlation vs. causation," cried someone at Hacker News. "Correlation does not mean causation," a reader moaned at Slashdot. "There are so many variables here that it isn't funny."
And thus a deeper correlation was revealed, a link more telling than any that the Missouri team had shown. I mean the affinity between the online commenter and his favorite phrase—the statistical cliché that closes threads and ends debates, the freshman platitude turned final shutdown. "Repeat after me," a poster types into his window, and then he sighs, and then he types out his sigh, s-i-g-h, into the comment for good measure. Does he have to write it on the blackboard? Correlation does not imply causation. Your hype is busted. Your study debunked. End of conversation. Thank you and good night.
The correlation phrase has become so common and so irritating that a minor backlash has now ensued against the rhetoric if not the concept. No, correlation does not imply causation, but it sure as hell provides a hint. Does email make a man depressed? Does sadness make a man send email? Or is something else again to blame for both? A correlation can't tell one from the other; in that sense it's inadequate. Still, if it can frame the question, then our observation sets us down the path toward thinking through the workings of reality, so we might learn new ways to tweak them. It helps us go from seeing things to changing them.
So how did a stats-class admonition become so misused and so widespread? What made this simple caveat—a warning not to fall too hard for correlation coefficients—into a coup de grace for second-rate debates? A survey shows the slogan to be a computer-age phenomenon, one that spread through print culture starting in the 1960s and then redoubled its frequency with the advent of the Internet. The graph below plots three common versions of the phrase going back to 1880 as they turn up in Google Books. It's that right-most rise that interests me—the explosion of correlations that don't imply causation in the 1990s and 2000s. Beware of spurious correlations, I know! But it is tempting to say the warning spread in the squall of data on the Web, as a means of warding off the cheap associations that ride a stormy sea of numbers. If now we're quick to say that correlation is not causation, it's because the correlations are all around us.
Let's go back a little further, though, to the origins of the phrase itself. Those first, modest peaks of "correlation is not causation" show up in print in the 1890s—a date that happens to coincide with the discovery of correlation itself. That's when the British statistician Karl Pearson introduced a powerful idea in math: that a relationship between two variables could be characterized according to its strength and expressed in numbers. Francis Galton had futzed around with correlations some years before, and a French naval officer named Auguste Bravais sketched out some relevant equations. But it was Pearson who gave the correlation its modern form and mathematics. He defined its role in science.
Philosophers had spent centuries, by that point, on the question of how the mere association of events might reveal their causal links and what it means to say that one thing can ever cause another. The ambiguity of correlations was well-known. Victorian logician Alexander Bain wasn't breaking new ground in 1870 when he warned his readers of the "fallacy of causation," whereby we might assume that, say, "the healthy effect of residence at a medicinal spa is attributed exclusively to the operation of the waters," as opposed to being caused by "the whole circumstances and situation." The confusion between correlation and causation, he said (not quite using the famous phrase), "prevails in all the complicated sciences, as Politics and Medicine."
With the arrival of Pearson's coefficients and the transformation of statistics, that "fallacy" became more central to debate. Should scientists even bother with a slippery concept like causation, which can't truly be measured in the lab and doesn't have a proper definition? Maybe not. Pearson’s work suggested that causation might be irrelevant to science and that it could in certain ways be indistinguishable from perfect correlation. "The higher the correlation, the more certainly we can predict from one member what the value of the associated member will be," he wrote in one of his major works, The Grammar of Science. "This is the transition of correlation into causation."