tl;dr

As I noted on my Twitter feed, I have recently finished reading Freakonomics by Steven Levitt and Stephen J. Dubner. For those who aren’t familiar with it, it’s an attempt to apply the mindset and tools of the economist to areas of study where they would not normally be applied, such as figuring out if Sumo wrestlers in Japan throw matches sometimes, or why crack dealers live with the mothers if drug dealing is such a lucrative industry.

I loved the book, but despite my attempts to pickle it with alcohol the PhD wielding bit of my brain did twitch a number of times during the book. I’d prefer to read some of the more startling conclusions in peer-reviewed journal papers, I think, rather than a pop culture novel that cannot hope to present much of the raw data the conclusions are drawn from. After all, extraordinary conclusions require extraordinary proof, and the entire raison d’être of the book is to present startling conclusions.

Some parts of the work seemed very solid to me. The aforementioned Sumo wrestling investigation is an outstanding piece of investigation (you can read more about this here, although note the bottom of that article contains a partial rebuttal to the conclusions in Freakonomics from an expert on Sumo). However, one bit that didn’t sit right with me was the question posed by this scenario:

Consider the parents of an eight-year-old girl named, say, Molly. Her two best friends, Amy and Imani, each live nearby. Molly’s parents know that Amy’s parents keep a gun in their house, so they have forbidden Molly to play there. Instead, Molly spends a lot of time at Imani’s house, which has a swimming pool in the backyard. Molly’s parents feel good about having made such a smart choice to protect their daughter.

But according to the data, their choice isn’t smart at all. In a given year, there is one drowning of a child for every 11,000 residential pools in the United States. (In a country with 6 million pools, this means that roughly 550 children under the age of ten drown each year.) Meanwhile, there is 1 child killed by a gun for every 1 million-plus guns. (In a country with an estimated 200 million guns, this means that roughly 175 children each year die from guns.) The likelihood of death by pool (1 in 11,000) versus death by gun (1 in 1 million-plus) isn’t even close: Molly is roughly 100 times more likely to die in a swimming accident at Imani’s house than in gunplay at Amy’s.

My problem with this is that it seems to ignore the fact that if anyone has a pool, they probably only have one; whereas if anyone has a gun, they probably have more than one. Thus, by extrapolating from number-of-guns-in-country to households, an important multiplier — the average guns per household — has been overlooked. The real difference in the odds (it seems to me) must be much less than the “roughly 100 times more likely” quoted here. However, as I doubt the average number of guns owned in any given gun-owning household is surely less than 100, the conclusions of the book still hold — but by a narrower margin than this would suggest.

These commenters on the official Freakonomics blog have a different criticism of the same conclusions, namely:

The question in the book about which is more dangerous.. swimming pools or guns… is complete rubbish. Children are not given the same access to guns are they are to swimming pools! Are you saying that if children were allowed play near and with guns to the same level they play near and with swimming pool there would be less deaths from guns?

I’m not sure I agree with this. It seems to me the 1-in-1,000,000 statistic already takes into account the child’s (hopefully small) access to the guns. It does go to show, though, just how tricky this sort of analysis can be.

Admittedly I think such nit-picking at the statistics shouldn’t be taken as a criticism of the book itself. The fundamental premise of the book is to present, in a series of worked examples, the results of using economics tools such as regression analysis on real-world problems they would not normally be applied to, and interpreting the results in a purely scientific manner even when there are contentious moral issues involved (such as the question, did crime fall in the USA throughout the 1990s because of the legalisation of abortion following Roe vs Wade in 1973?).

This is not to say that the moral dimension is not important, of course; it’s just that Levitt and Dubner conciously exclude it from their analysis. This aspect reminds me a little of Derren Brown’s book Tricks of the Mind, which discusses things like cold reading techniques used by fraudulent mystics. Both books have a similar undercurrent of an appeal for more rationalism and logic in people’s worldviews.

So, anyway, yes. A very recommended read.

Science