Tuesday, April 27, 2010


For such a popular and acclaimed book I expected something more substantial though a cynic might argue why did I think it appealed to so many people? This really isn't any primer on economics or statistics but is instead illustrations on question solving. For instance, why do so many crack dealers still live with their parents? Answer is that they're so low in the orgnization that they actually make little money but the authors Levitt and Dubner actually go through the data to explain this and show how it works that way.

Towards the end they criticize experts who come up with definite answers just to get publicity but I can't tell if Levitt and Dubner expect us to understand that they're doing exactly the same thing or if in fact they really are unaware. Look at their discussion of why crime rates have been falling for a couple of decades. They talk about the "diverse army of experts" who came up with a "phalanx of hypotheses to explain the drop in crime". They then list the most common, state that only three have any evidence and then that "one of the greatest measurable causes" isn't on the list and was never mentioned in any newspapers. That unmentioned reason is the legalization of abortion which resulted in fewer unwanted children and therefore fewer adults inclined towards crime.

Now the thing is that Levitt and Dubner are no different than any of these other experts. They criticize what they see as wrong, they have a main cause that they propose and they have their own data/evidence to support it. I'm not saying they're wrong because their reason does make sense but making sense and being true are not the same thing. If legalized abortion as a cause is measurable and so obvious then doesn't it raise a red flag that nobody has discussed this before? Is Levitt (the actual economist of the duo) really so insightful that he found something everybody missed? (And let's just assume that the no mentions in newspapers does actually mean experts weren't discussing this - there is a large gap between mass media and academics/policy experts.) Doesn't it make more sense that Levitt found something that's already been discussed and dismissed? Or that he has the wrong data? Again, I don't know. Freakonomics seems plausible but like so many social issues this is very complex and it's impossible to completely separate all the causes - in fact I would think it's possible that there is no actual cause at all, at least in the sense that they mean.

They also sometimes use the wrong data which doesn't quite create more confidence. They discuss (p149) the parents of an eight-year-old who they allow to go to a house with a swimming pool but not to one with a gun. Levitt & Dubner claim that this is both wrong and irrational because the statistics show that the liklihood of dying at a pool is greater than that of dying from guns. Now I'm just going to assume that their statistics are true - the source in the footnotes is one of Levitt's own articles in the Sun-Times. But what their statistics are measuring are numbers across the entire country and in aggregate - 1 death under age 10 per 11K pools, 1 similar death per 1 million guns (which he translates to a total 550 pool deaths and 175 gun deaths per year). What really is needed are statistics concerning deaths in households that actually have guns. In other words, how many children actually die from playing in a household with guns rather than a number based on total deaths and total guns. Also it could only help his case but Levitt doesn't distinguish between intentional and unintentional gun deaths.

But when you expand this the parents decision seems more reasonable. Most "near-misses" at a swimming pool create mainly a frightening event (one statistic I found is permanent damage happens to less than 20% of children hospitalized but of course most wouldn't need hospitalization) but with guns there is a wide variety of permanent injuries that could result. So instead of focusing on just deaths Levitt needs to pull in statistics on injuries, disfigurements and other permanent damage.

This example is actually a fairly small one in the book and to some degree doesn't entirely matter. What they're really doing is raising a question and then exploring possible ways that data might be utilized and examined to help provide answers. That process is the real core of the book even though I suppose most people are actually just taking away that real estate agents sell their own houses for more than they get for customers, probably remembering a reason or two why this should be so and likely forgetting how Levitt figured it out. Maybe this is one reason I found the short newspaper articles included in the paperback edition more interesting than most of the original book.