When I was in graduate school, one of my professors, the legendary memory expert Gordon Bower, told a story of a paper he once wrote. He had conducted a study of word memory, but found that the results were muddled by an incidental social aspect of the experiment. The social effect intrigued him so that he wrote up the results and submitted them to a journal of social psychology, outside of his field. The editor of the journal responded, "Well Gordon, nice work. You've successfully replicated one of the best-known results in all of social psychology."
Why am I telling you this? Well, it came to mind as I was reading the baby-naming chapter of renowned economist Steven Levitt's new book Freakonomics, recently excerpted in Slate. Levitt (with journalist Stephen Dubner) set out to use "the best analytical tools that economics can offer" to uncover the "hidden side of everything," promising to reveal surprising and counterintuitive aspects of the world around us. His manifesto, from the introduction:
It is well and good to opine or theorize about a subject, as humankind is wont to do, but when moral posturing is replaced by an honest assessment of the data, the result is often a new, surprising insight.Analytical rigor, astonishing revelations. Does he deliver?
I first started to worry when Levitt stated as fact the well-worn urban legends about babies named OrangeJello, LemonJello, and Shithead. A 10-second Google search would have been sufficient to throw those into question, didn't he bother? As a matter of fact, he did. In the endnotes, he cheerfully admits that his source for Shithead "might have been misinformed, of course, or even outright lying." And as for the 'Jellos:
Although these names have the whiff of urgan legend about them -- they are, in fact, discussed on a variety of websites that dispel (or pass along) urban legends -- the authors learned of the existence of OrangeJello and LemonJello from Doug McAdam, a sociologist at Stanford University, who swears he met the twin boys in a grocery store.Oh, you heard it from a guy who swears he saw them once in a grocery store! Phew, thank goodness for "the best analytical tools that economics can offer."
Hereon, we proceed with caution.
In discussing the origins of name trends, Levitt's primary thesis is that fashions which originate with the upper classes gradually trickle down the economic ladder. This, naturally, is no revelation -- in fashion-based industries like apparel, it's an explicit, institutionalized process. (After all, we call the elite "trendsetters" because they set the trends.) The revelation is meant to come in the form of predictive power. Levitt uses data about California parents' economic status and name choices to propose a list of names that, "unlikely as it seems," are candidates to become "mainstream names" ten years from now. Names like Emma, Isabel, and Grace. But wait a second, aren't those already mainstream names?
Here's a graph of the popularity of Levitt's suggested future girl's names, using the data available at the time he wrote the book:
In fact, of his 24 predictions for "unlikely" names that could possibly hit the mainstream in a decade, 7 were already top-100 names, including 2 of the top 15 (Emma and Grace). Looking boldly out into the future, he predicted the present. Oops. So much for revelations.
What was the economics-based methodology for those questionable choices? Apparently, the author looked at a list of of names favored by rich Californians and chose the ones he personally found attractive. So much for rigor.
All of which brings us back to the Gordon Bower story. When anyone, even a brilliant scholar, walks into a whole new domain of knowledge, he's at a disadvantage: he doesn't know what is and isn't known. Thus it can be hard to tell the revelatory from the obvious. Any mom of a preschooler could have told Levitt that Emma wasn't a very clever prediction. It was news to him, though, so he didn't bother to dig deeper -- even to learn that Emma was already the #2 name in America. Which is a shame, because he was sitting on an absolute treasure trove of data which doubtless does have secrets to reveal.
The fashion analysis was only half of the baby-name presentation in Freakonomics. The other, much less silly half, concerns the real-world impact of race-specific names, which I'll talk about next time.