Correlation

If I could add one subject to the American high school curriculum, it would be statistics. In particular, too many people, including too many journalists, fall into the common statistical fallacy of confusing correlation with causation. An article will cover a study that shows a high correlation between two facts, but then discuss the study as if one therefore causes the other.

Examples are everywhere. A brief Google search revealed this example:

Researchers at the Aabo Akademi found that Finns who speak the language of their Nordic neighbors were up to 25 percent less likely to fall ill than those who do not.

The accompanying headline would probably read, “Does studying Swedish improve your health?” In fact, of course, Swedish has nothing to do with health. Instead, perhaps people who have time to learn Swedish work in less physically taxing jobs. Or maybe they live in cities where there is better medical care. Whatever the actual cause, it may be useful to know that speaking Swedish and good health are correlated, but we can’t jump to the conclusion that one causes the other.

The famous “flying spaghetti monster” letter pointed out that the global temperature has risen as the number of pirates has dropped. Does that mean that pirates prevent global warming?

In light of this pet peeve, I particularly enjoyed Friday’s xkcd comic: