In Praise of Statistical Thinking

In an essay in the Chronicle of Higher education, Eric G. Wilson (an English professor at Wake Forest) writes "In Praise of Melancholy."

Melancholy, by the way, is just a fancy word for depression. The word itself derives from the ancient Greek meaning "black bile," because it was once thought that depression was caused by an excess of black bile in the body.

Anyway, the author of the essay mentions that recent polls show that 85% of Americans rate themselves as happy. He also claims that this is a disturbing trend because melancholy has been the inspiration for much of the world's great art, music, and literature. Yet we are content to "annihilate" it through positive psychology, psychotherapy, and the use of antidepressants like Prozac.

Although there are a lot of interesting things about this argument, there is a fundamental problem that prevents it from ever getting off the ground ... and it's something that I emphasize over and over in Psych 144. In the abstract it's this: evaluating a claim about a statistical relationship requires a comparison of one variable across levels of the other. In the concrete: the author claims that people are happier than they used to be, but he only presents data on how happy they are now. Who's to say they weren't just as happy in the "olden days?" If they were, then the whole idea that we are in the process of "annihilating" melancholy falls flat.

I'm not an expert on happiness (or what psychologists are more likely to refer to as "subjective well being"), but I'm pretty sure there are data out there that show whether or not there has been a change in the happiness levels of Americans across time. I'm also pretty sure that those data show that there really hasn't been much change over time--certainly not changes on the order of what the Wilson suggests in his essay.

But this is a blog, not a journal article. So you'll have to look up those data yourself.