Ri-freakin’-diculous: a review of, and warning about, Freakanomics

First Word

It’s difficult to write a complete, organized, whole review of a work that boasts of its own lack of focus: “Most books put forth a single theme, crisply expressed in a sentence or two, and then tell the entire story of that theme . . . . This book boasts of no such unifying theme” (14). Lacking nothing grandiose, they also add that they aim to explore the “hidden side” of everything (ibid.). So it must be the large scope of their aim that makes them seem aimless. They want to have no limits on what they can investigate (excepting the limit of 300 pages, or less, required to be a New York Times Best-seller). Levitt, the economics professor, and Dubner, Levitt’s personal praise patrol, have set out to write a neatly packed, sexy-titled, philosophical micro-epic, whose aim is either everything or nothing. In this reviewer’s opinion, their trajectory hit nothing first and the writers settled with what they got.
So even while not having their “unifying theme,” the writers did make some claim to get us readers grounded: “What this book is about is stripping a layer or two from the surface of modern life and seeing what is happening underneath” (12). Pause. Excellent idea. Now tell us how statistics are stripping layers away from surfaces. No answer, only the sound of crickets chirping amidst the graphs and tables. The writers want to be cutting-edge philosophical investigators, but fall into nightmarish scenarios like, for example, when they compare morality, which “represents the way that people would like the world to work,” with economics, which “represents how it actually does work” (13). Yikes! What a dichotomy! We’re off to a rocky start if they want to strip away layers to find something like truth underneath, because they’ve already tossed out one of the most fundamental layers in human society. (Don’t worry, they bring morality back later when they talk about incentives, and they have a new, more convenient definition for it.) By the end of the book this reviewer was unsure about how many layers had been stripped away or how many layers of pseudo-scientific stucco had been plastered on to his brain. Assumption after assumption, generalization after generalization, and it’s not quite clear how all their sets of data could possibly be revealing anything without covering more things over. This review will address some details a little farther below. But first, the introduction and underlying framework of the book should be examined.

Introduction and Framework

The writers assert that they have five presuppositions:

1. “incentives are the cornerstone of modern life”;
2. “the conventional wisdom is often wrong”;
3. “dramatic effects often have distant, even subtle causes”;
4. “‘experts’ . . . use their informational advantage to serve their own agenda”; and
5. “knowing what to measure and how to measure it makes a complicated world much less so” (13-14).

(Regarding the first presupposition: incentives (internal or external, financial or social) are inseparable from a personal sense, or lack of a sense, of morality—what is right, what is good, or at least what is beneficial, what is pleasurable, etc. Regarding the second: this reviewer assumes they mean “wrong” amorally; also, attacking conventional wisdom is easy and in so doing writers and scientists create infinite new conventional wisdoms by doing it. Regarding the third: fine. Regarding the fourth: beware that experts also wrote Freakanomics. Regarding the fifth: this presupposition makes sense, but in light of the preceding one it is worrying that we readers have no clue what or how to measure, and so we are invited to utterly trust the writers’ hidden, expert knowledge.)
Their list isn’t a bad foundation for a book, but they could have added a few more to make sense of the remainder of the book:

1. generalizations can be considered the same thing as knowledge;
2. predictive data can be bent back and applied to the past to describe what would have happened; and
   
3. the authority of statistics is unquestionable, or in Dubner’s words, “Steven Levitt may not fully believe in himself, but he does believe this: teachers and criminals and real-estate agents may lie, and politicians, and even CIA analysts. But numbers don’t” (17).

With the hasty introduction out of the way, they begin to look at various mini-themes, each of which makes a neat chapter that relates two usually inconceivably related things. This is the merit of the book. The writers have great creativity. This is, no doubt, why many people really like this book. Go check the five-star reviews on Amazon.com and see for yourself; people who rate the book highly are dazzled and wowed by their insights, and some reviewers are even so pleased despite not agreeing with the conclusions. This reviewer, however, is not one of them, and two stars out of five is all that could be mustered in support of Freakanomics.
Their initial promise was delightful: "We will ask a lot of questions, some frivolous and some about life-and-death issues” (12). They weren’t lying. Questions about abortion are not separated by many pages from ones about baby names. But the range of questions from shocking to silly does not necessarily make any of them the right questions. Again this reviewer had some philosophical hopes coming into this book, but was very disappointed and even angry throughout the reading.

Exhibit One

Chapter 5, tagged with the question What Makes a Perfect Parent? is a great example. Levitt and Dubner put forth eight things that do and eight other things that do not correlate to public school standardized test scores. The list isn’t quite random. Using a magical tool called “regression analysis,” which remains a mystery for us non-experts, even if we do try to fill in the gap with Wikipedia, they found the following correlates (with a “+” or “-” to show positive or negative correlation):

1. the parents are highly educated (+);
2. the parents have high socioeconomic status (+);
3. the mother was 30 years old or more when she had her first child (+);
4. the child had low birth weight (-);
5. the parents speak English in the home (+);
6. the child is adopted (-);
7. the parents are involved in a PTA (+);
8. and the family has many books in the home (+).

The following are supposedly non-factors in a child’s likelihood at getting better test scores:

1. the child’s family is intact;
2. the family recently moved into a better neighborhood;
3. the mother didn’t work between birth and kindergarten;
4. the child attended Head Start;
5. the parents regularly take the child to museums;
6. the child frequently watches television;
7. and the parents read to their child every day.

(What would Claiborne Barksdale say to that last item on the non-factor list?)
Here’s the problem: they beg the questions about the basis for their magical regression analyses. They lead the reader to assume he has learned something because a bunch of variables have been correlated with test scores. But what are test scores? It is “conventional wisdom” that standardized test scores have their own range of error. A crack in the foundation of these statistics could make the house of cards fall. Instead of bringing this up, they begin to make claims on the Nature versus Nurture question, in favor of nature—some admitted generalization about what the parents are versus what the parents do. This question cannot begin to be answered in such a way! They’ll quickly fall into an infinite regression of causes: being causes doing, which leads to being, which causes doing, which . . . What? Has the reader learned anything really? In fact, the writers themselves subtly admit that the reader has not gained much insight from these correlations and non-factors. With regard adopted children, who statistically have lower test scores, they bring up another “study”:

Sacerdote found [that] the parents weren’t powerless forever. By the time the adopted children became adults, they had veered sharply from the destiny that IQ alone might have predicted. Compared to children who were not put up for adoption, the adoptees were far more likely to attend college, to have a well-paid job, and to wait until they were out of their teens before getting married. It was the influence of the adoptive parents, Sacerdote concluded, that made the difference. 174

This is Levitt and Dubner admitting that their foundation among standardized test scores and “intelligence quotients” is shaky and possibly meaningless. On the one hand, students’ test scores are more likely to be higher because of genetics and things occurring before birth (Nature > Nurture; we should probably weed out the bad ones now and start the eugenics race). On the other hand, the very same negative correlate to test scores, I mean adoption, is a positive correlate to college attendance and a well-paid job. The reader must navigate an intricate maze of glitter and jazz to find that these two writers are using a lot of words and numbers to say very little, and even this little bit is spuriously got.

Exhibit Two

Let’s take a look at another philosophical blunder from the chapter 4, which correlates higher abortion rates to lower crime:

What sort of woman was likely to take advantage of Roe v. Wade? Very often she was unmarried and in her teens or poor, and sometimes all three. What sort of future might her child have had? One study has shown that the typical child who went unborn in the earlier years of legalized abortion would have been 50 percent more likely than average to live in poverty; he would have also been 60 percent more likely to grow up with one parent. These two factors—child poverty and a single-parent household—are among the strongest predictors that a child with have a criminal future.
. . .
In other words, the very factors that drove millions of American women to have an abortion also seemed to predict that their children, had they been born, would have led unhappy and possibly criminal lives. 138-139

This reviewer will let C. S. Lewis answer, from chapter 5 of The Problem of Pain: “But it must always be remembered that when we talk of what might have happened, of contingencies outside the whole actuality, we do not really know what we are talking about.” This doesn’t speak against using sets of data as predictive tools, but it points out the inanity of bending a generalized prediction back on the past. If it is difficult enough to find sets of data of the present conditions and to find all the variables of today that will affect tomorrow, then how much more difficult would it be to find all those things from the 1970s and ’80s that are clouded by time from the detective’s eye? And furthermore, the quote from Lewis cries out about the irresponsibility of justifying an action, whether intellectually or morally, on such shaky ground as the future optative grammar affords.

Exhibit Three

This is the reviewer’s last complaint. The book is filled with praise for Levitt, who remains like a Jedi Master with magical abilities that cannot be communicated to the common rabble. It seems as if Dubner has been following this fairly successful professor around like a puppy dog, waiting for crumbs to fall from on high; and now the journalist is generous enough to share with the rest of us vermin if we buy the book. But the praise in Freakanomics is stale. And, despite the creativity in the connections, the basic material isn’t really very fresh. At the beginning they present it like it’s hot off the press:

Now, as the crime-drop experts (the former crime doomsayers) spun their theories to the media, how many times did they cite legalized abortion as a cause?
Zero. 6

Dramatic, right? It makes it seem as if they, or at least Levitt, received oracles from God about some of these data. But really, 130 pages later, we find out that it’s old news that we must have been snoozing through because no one wrote a New York Times Best-seller about it:

Whether or not one feels strongly about abortion, it remains a singularly charged issue. Anthony V. Bouza, a former top political official in both the Bronx and Minneapolis, discovered this when he ran for Minnesota governor in 1994. A few years earlier, Bouza had written a book in which he called abortion “arguably the only effective crime-prevention device adopted in this nation since the late 1960s.” 142

Now we’re well informed.

The Summary

What does this book do? It groups things into genera, begs all the right questions, praises itself, and leaves the reader either dazzled with pretty colors and flashy titles or angrily skeptical. The writers made it clear that morality was not in the picture (too idealistic), that numbers don’t lie (they mean “lie” in an amoral sense), and that economics, or at least statistical analysis, is utterly incapable of responding to anomalies. They confirm this last point themselves with the final comparison between Ted Kaczynski and Roland G. Fryer Jr., both of whom break the trends discovered by, or the molds made by, Levitt and Dubner. It comes as a bitter blow to their statement that “economics represents the actual world” (206). Obviously the generalized is not the same as the actualized.