Just finished reading Freakonomics, the much-discussed popularization of applied statistics by Steven D. Levitt and Stephen J. Dubner, economist and writer respectively. Although Levitt won the Nobel Prize for Economics, this best-seller really is about “applied statistics” rather than economics. It’s also undeniably accessible to any moderately educated reader. It takes much less time to read than you would expect any book on either statistics or economics to take. It would be a better book if the reader could be expected to spend more time on it, as it feels dumbed down in its current state.
On the plus side, almost all of the case studies are interesting, ranging from the impact of radily available abortions to a cheating scandal by Chicago teachers (yes, teachers) to the minimal effect that parents can have on how their kids turn out. The mathematical arguments are clear and easy to follow. The political and social conclusions are clear and well-argued, although certainly not beyond question. And even though the book is about statistics more than economics, there is an economic theme that runs through the entire book as a common thread: the importance of incentives.
A concluding section on the impact of first names is particularly interesting, addressing not only the possible bias caused by certain names but also the history and future of the distribution of names.
There’s a lot of other intriguing material, mostly social and political rather than economic or mathematical — how the media and the politicians have spread incorrect information about crime rates, for instance. Questions of causality vs. correlation are raised in many places, though not always satisfactorily addressed. And therein lies the minus side of this book: Levitt and Dubner combine an incomplete discussion of causality with the superficiality that I mentioned in the first paragraph.
On balance, Freakonomics, is worth reading, and definitely worth thinking about, but it could have been a much stronger book.