It seems that at least once a week, we hear some breathless announcement that X activity has been found to be linked with Y behavior. Correlation seems to be the social science tool du jour, and the media eats it up.
Why? Because we have all sorts of data, and computers that can sort that data into unlimited comparisons, so it’s easy for someone to just start feeding in numbers and seeing what pops out.
The thing about correlation, though, is that by itself it’s meaningless. It’s a possible indication of an area for further study, but without context it’s useless for making policy. And it certainly doesn’t prove causation.
What’s fun and instructive is to explore some of the strange correlations that exist out there.
For example, did you know that people who have to look at the keyboard to type are much more likely to prefer thin crust pizza over deep dish as compared to the rest of the population?
Why? Who knows. You might speculate that Chicago (which tends to prefer deep dish) has better typing classes in the schools than New York (which tends to prefer thin crust) and that difference affected the overall population. But without further study, you really have no knowledge other than a bizarrely interesting statistic. (Disclosure: I’m from Chicago, can type without looking at the keyboard, and prefer deep dish pizza.)
Studies will also show you that people who don’t like camping are linked with a lesser ability to burp at will, as well as the fact that loving non-fiction books is linked to a likelihood of pigging out when you’re upset, and also that people who prefer soft serve ice cream don’t like roller coasters as much as the rest of us. (10 Crazy Correlations Between Unrelated Things – Business Insider)
So remember when you hear that a new study has conclusively shown that A is linked with B, that only means that someone put some numbers in a computer and looked for any kind of variance they could find, and in this case, 12% of people who did A, also did B, commpared to only 7% of the rest of the population doing B. Those numbers tell you absolutely nothing about why anyone did B.
This is a lesson that reporters should take more time to learn themselves and pass on to their readers.