FREAKONOMICS: A Rogue Economist Explores The Hidden Side of Everything
by Steven Levitt & Stephen Dubner, William Morrow, 2005
I really wanted to like this book. There’s a lot about how economist Steven Levitt thinks that I resonate with, and the free association of how seeming unrelated things interrelate is certainly reminiscent of one of my favorite authors, James Burke and his best-selling and immensely entertaining book “Connections.”
But I didn’t like Freakonomics, for a variety of reasons, the first being that nowhere in the writing or editorial process did anyone bother to mention to the authors that modesty trumps egocentric writing. Between the introduction to the book and the chapter introductions, Levitt has more ego strokes (which is to say we’re trapped having to read about him) than any other living author I’ve encountered in thirty years of voracious reading.
Economists tend to be very focused on numbers. You could reasonably observe that all economists are “quants”, they quantify things, they believe facts if the numbers support them and disbelieve theories or “qualitative” information. Quants don’t like anything that’s not quantifiable. And that’s the fundamental flaw of this book: everything in life cannot, in fact, be reduced to measurable numbers and data fed into regression algorithms.
Rather surprisingly, though Levitt and Dubner criticize others about being unable to separate correlation and causal data, they too slip into this basic analytic error too. For example, “it turns out that obstetricians in areas with declining birth rates are much more likely to perform cesarean- section deliveries than obstetricians in growing areas — suggesting that, when business is tough, doctors try to ring up more expensive procedures.” Catch the logic error here? There are in fact many other reasons why the perceived risk of childbirth could increase in an area with a declining birth rate, including an aging of the population, a factor that can easily increase birth risk even as the birth rate declines.
What’s missing in the analyses is that there are random, chaotic, and apparently unrelated causes for behaviors. In addition to IQ, for example, healthy successful adults also can be measured on more qualitative scales like the so-called “emotional intelligence quotient” of EQ.
These are the fundamental ideas embodied in this book: incentives are the cornerstone of modern life, the conventional wisdom is often wrong, dramatic effects often have distant, even subtle, causes, “experts” – from criminologists to real estate agents, use their informational advantage to serve their own agenda, and the mantra of quants, knowing what to measure and how to measure it makes a complicated world much less so.
Do teachers help students cheat on tests? The answer is “no”, unless the test also affects the economic well being of the teacher (think standardized tests and their built-in incentive systems). Then the answer becomes “yes”, as is eloquently explained in Freakonomics. But, again, too much quant trips up the discussion, when it is confidently asserted that a dramatic one-year spike in test scores can be attributed to a good teacher, but when there’s a dramatic fall to follow, it’s probably cheating. Working with lots of teachers myself, I disagree. I have seen time and again how an inspirational teacher will encourage students to stretch, to get out of their comfort zones, and attain remarkable results. Once they leave that teacher and go back into the grist mill of modern education, their results plummet. Looked at analytically, is the good teacher ‘cheating’ by inspiring their students? Levitt seems to think so…
Freakonomics also discusses the rise in power, and later drop in power, of the Klu Klux Klan, observing that its power was “largely derived from the fact that it hoarded information” and that the KKK saw a precipitous drop in membership once its secrets were aired publicly. The story is fascinating and well-written, but the conclusion is dangerously false. Hate groups don’t gain power because of secret, hoarded information, they gain power by reinforcing the existing hates and biases of potential members. That is, people who were Klan members might have quit due to the troubling inability of the leadership to keep Klan information secret, but it didn’t change whether or not any given Klan member hated certain ethnic minorities, and it certainly didn’t affect whether these people would be willing to don a robe and try to intimidate or harass someone. It’s more a loss of brand luster than anything more significant, but that seems to pass Levitt by.
Another remarkable analytic error: the book analyzes the likelihood that people who profile themselves for online dating sites distort the truth, without ever acknowledging that there’s no reason to believe that the members of an online dating site are in fact representative of the population at large. That being concluded, it isn’t as damning to find from analysis that “more than four percent of online daters claimed to earn more than $200,000 year, whereas fewer than one percent of typical Internet users actually earn that much.”
The main reason I was asked to review Freakonomics, however, was because I’m an outspoken advocate of attachment parenting (see my parenting blog to learn my biases) and half of Freakonomics wrestles – unsuccessfully – with the question of What Makes a Perfect Parent?
In a classic case of using what I call “Limbaugh Logic”, Levitt, a self-avowed expert (which is reinforced time and again in this book) says that experts “don’t so much argue the various sides of an issue as plant his flag firmly on one side.” Which is, of course, exactly what Levitt does throughout this book too, though his flag has the motto “numbers are facts” emblazoned upon it.
To understand how Levitt approaches parenthood, imagine Mr. Spock from the popular TV show “Star Trek” was raising a child. He’d carefully analyze all the facts associated with a behavior, draw the conclusion most supported by the “facts”, and be baffled why others aren’t following the same approach. The discussion of the statistical facts surrounding fear of flying and the safety of flying versus driving have just this tone, with Levitt asking almost plaintively why people are afraid to fly when driving is, statistically, equally dangerous.
Are airbags dangerous to small children? Levitt notes that “fewer than five children a year have been killed by airbags since their introduction” without ever acknowledging that the warnings that parents keep their children away from airbags have a clear and significant effect on this yardstick. The question is how many kids would be killed by airbags if there were no warnings, but that’s not examined.
There’s a big flaw in the reasoning presented that becomes clear in the parenting section too: conclusions are drawn from the analysis of data based on various studies, but just like medical research, there’s no discussion of what criteria were used to evaluate the base research. Was that study on children growing up to be criminals funded by a liberal think-tank or a conservative pro-imprisonment organization? Was the data upon which so much is based regarding the efficacy of parenting techniques based on research funded by a group with a particular view they sought to promote? If you don’t think that this is a critical part of evaluating modern research, you’re asleep at the analytic wheel.
In the discussion of test results (which Levitt correlates to various positive or negative parenting techniques) a startling lack is any discussion of cultural or ethnic bias in the test itself. For example, one of the factors that correlates strongly with high test scores is “The children’s parents speak English in the home”. But that’s only true if the test is given in English. So concluding or even highlighting that parents who speak English correlate strongly with children who have higher test scores is fundamentally flawed.
Falling deep into the well of quant thinking, Levitt also draws the correlation from the data that whether a family is intact doesn’t seem to matter (in terms of being a good parent. Remember, the question isn’t “what approach to parenting produces children who test well”) and that parents splitting up “has little impact on a child’s personality” or “academic abilities.” But what of emotional intelligence? What of the experience of childhood? Hard to quantify, sure, but that’s exactly my point. In a world that’s purely quantitative, major measures are ignored, and Freakonomics is a powerful example of this error.
It should be no surprise that the authors conclude that nurture is not of much importance in parenting, and that it’s nature that’s more important. As Levitt says “this is not to say that parents don’t matter… [but] most of the things that matter were decided long ago – who you are, whom you married, what kind of life you lead.” This is a dangerous line of thinking, and the very fact that modern parents are striving to create a maximally nurturing environment for their children is a tremendous boon for our generation and future generations. It’s easy to forget the dramatic changes in how children are reared and limit the discussion to that which we have data for, but self-enlightenment always brings about positive change in society, whether we can quantify it or not.
The book ends oddly, too, by turning the analytic technique on its head for the ability to make what is ostensibly an ironic observation about how nurture doesn’t guarantee outcome: the poor black child profiled in the book grew up to be a Harvard economist, and the wealthy white child became the Unibomber. But so what? It’s a trivial matter to isolate a single anomolous case from any dataset and suggest it proves a postulate. It’s just very, very weak science.
Freakonomics is a very interesting book to read. Skip the introduction and ask someone to rip out the chapter introductions, then write across the cover “The World Is Not Quantifiable” and you’ll have a fascinating read, without being lulled into the false confidence of the quant. Or just highlight the sentence in the introduction where the authors explain their ultimate bias: “Morality, it could be argued, represents the way that people like the world to work – whereas economics represents how it actually does work.”