Wednesday, October 5, 2011

Freakonomics by Steven D. Levitt and Stephen J. Dubner

Freakonomics: A Rogue Economist Explores the Hidden Side of Everything (Freakonomics, #1)Freakonomics: A Rogue Economist Explores the Hidden Side of Everything by Steven D. Levitt and Stephen J. Dubner
My rating: 4 of 5 stars

I think I read this book years ago, but (re?)reading it today, I'm not sure I actually did read it before. It deals a lot with the causes of crime (with regards to abortion, bagels, crack dealing, and guns) cheating (in education, real estate, and um, sumo wrestling), and the prevalence of racism.

The thing that I like about this book is that it relies on data for it arguments on cause and effect in public policy with little to no regard for whether the results happen to align with liberal or conservative ideologies. For years I've argued that our political parties are a mish-mosh of beliefs that don't necessarily fit together in any logical way.

There are a number of things I don't like about this book:

1) It explains some general statistics and regression analysis only towards the end of the book, even though a lot of the data referenced in the book depends on an understanding of this. When they finally explain correlation, even where their isn't a relationship of causation, the authors offer a number of guesses about why the two factors are correlated. Can we do better than guesses? Also these correlation guesses also seem to bias the reader.

2) For those of us who do have some understanding of statistics and studies, the authors don't provide enough information about the studies and data to let us decide for ourselves how accurate the data, or how rigorous the studies are. This despite the fact that the authors talk at length about inaccurate causes from other sources.

3) Because the book jumps around between topics with "no unifying theme" as the authors claim, there is no real deep examination of any issue that allows the reader to form a fully informed thought on an issue. Certainly I'm more interested in the causes of crime now, but I don't particularly have a handle on all the available information or applications of the information. For example, on the issue of guns, the writers present a number of interesting points. Gun ownership in Switzerland is high and they don't have as high a crime rate so "guns don't cause crime." Okay, but do guns cause more death? Why isn't there a comparison with Japan which has no guns, but still has crime- and compare how many crimes result in death? Data divorced of real context is pretty useless. Unfortunately the book jumps around too much to provide adequate context and data for any of the issues it discusses.

4) I understand that morality is not the job of economists but at the same time some application of some moral system is necessary when examining these issues. A number of time the writers joke that a sufficiently high disincentive would cure the problem. For example, if you want to cure bagel theft, you could consistently apply the death penalty. (I'm not sure if this was an actual example in the book but it's close.) Of course the writers are joking! Sort of. They're not joking that would work but they concede that it's out of bounds for most politicians. This leaves the reader unmoored in the very real world of their statistics of cause and effect. It's not enough to say that "some Americans are uncomfortable with number of citizens" in jail, but to apply a real analysis to applying the solutions of imprisonment instead of increasing education, opportunity, and quality of life. Instead the writers seem to dash those off by negating the effects of an improved economy. Is it possible that the gains of a good economy are not affecting all equally for example? Likewise, the emphasis the writers place on abortion is not in any way balanced by the same exploration of increased opportunities, contraception, education, or even incentives to put babies up for adoption for disadvantaged populations.

I get it, data is king. I'm a believer. But failing to paint a complete portrait of the issues results in the partial data. This partial data can be just as misleading as the absence of data. By focusing on one problem at a time and considering all the related issues and data for thinking about the problem holistically the writers might help clarify thought rather than just further confusing it.

Still 4 stars.

View all my reviews

No comments:

Post a Comment

Related Posts Plugin for WordPress, Blogger...