Freakonomics

The pressure to give this book rave reviews is enormous. Everyone seems to love it (the Freakonomics website will lead you to plenty of positive reviews), and Steven Levitt is an undeniably brilliant economist — my hat’s off to anyone who wins the John Bates Clark Medal. But this is not a brilliant book. And not just because the title is stupid. For all of the talk about the dazzling insights, the book is glib. It reads like a journalist describing the work of a scholar … which is exactly what it is.

The chapter entitled “What Makes a Perfect Parent?” will serve nicely to illustrate my complaint about the book, but the problems are not confined to this chapter. The authors begin the chapter by chiding parents who do not allow their children to play with friends whose homes have guns, while allowing the same children to play at homes with swimming pools. More children are killed in swimming pools than in gun accidents, so these parents are irrational. Fair enough, though criticizing people for being horrible at risk assessment is an old game and one not limited to parents.

The question the authors really want to answer, they claim, is this: “how much to parents really matter?” And they begin with this stage-setting thought:

Clearly, bad parenting matters a great deal. As the link between abortion and crime makes clear, unwanted children — who are disproportionately subject to neglect and abuse — have worse outcomes than children who were eagerly welcomed by their parents. But how much can those eager parents actually accomplish for their children’s sake?

If neglect and abuse leads to bad results, then attentive, caring parents must matter a great deal. How much can such parents accomplish? I would think the answer should be: a tremendous amount. But this isn’t the answer the authors are seeking. They want to find “the hidden side of everything,” and such an obvious conclusion is not interesting. So, just after telling us that experts of all kinds exaggerate their claims because “an expert whose argument reeks of restraint and nuance often doesn’t get much attention,” the author press on with the implausible claim that parenting really doesn’t matter all that much.

How do they support this claim? Here they find the sledding a bit rough. They begin with a provocatice comparison between two boys, one from a model white family in the Chicago suburbs and the other from a dysfunctional and abusive black family in Daytona Beach, Florida. That last fact suggests that these are real people, but we don’t find out until the epilogue that the black boy grew up to be Levitt’s co-author (Roland G. Fryer, Jr.) while the white boy grew up to be Ted Kaczynski, the Unabomber. How ironic that a book claiming to be about data rests one of its main claims on anecdote.

The authors frankly admit that “[c]ertain facets of a child’s outcome — personality, for instance, or creativity — are not easily measured by data.” Indeed, I would say that the most important outcomes that I hope for in my own children (integrity, honesty, charity, compassion, etc.) are not easily measured. What we can easily measure, however, is academic performance. Test scores. And the rest of the chapter is devoted to that.

A bit more than half of this chapter is devoted to a discussion of data gathered from the Early Childhood Longitudinal Study. The analysis in the book is based on a paper by Levitt and Freyer, an abstract of which appears here. The paper is about the gap in test scores between black children and white children in kindergarten and after two years of school. When discussing factors that seem to be correlated (or not) with test scores, the book states:

[A] mother who stays home from work until her child goes to kindergarten does not seem to provide any advantage. Obsessive parents might find this lack of correlation bothersome — what was the point of all those Mommy and Me classes? — but that is what the data tell us.

I don’t know a lot about Mommy & Me, but their website talks about making children feel “happy, healthy, and loved,” not about acing standardized tests. This is just a hunch, but I don’t think that most mothers who decide to stay home with their children think, “This will really give Johnny the edge on standardized tests.”

And why use the demeaning term “obsessive parents”? Sprinkled throughout this chapter, the term is introduced with this unhelpful definition: “Obsessive parents know who they are and are generally proud of the fact; non-obsessive parents also know who the obsessive parents are and tend to snicker at them.” Examples of things obsessive parents do: “trek to the local police station or firehouse” to have the car seat installed “just right”; take their children to museums even though such “culture cramming” does not improve test performance; and study parenting techniques. Now, these things may seem silly to someone who knows better — or someone whose only metric for success is a test score — but using a term like “obsessive parents” for well-intentioned acts like these is mean spirited.

The authors conclude the chapter by listing eight factors that seem to be correlated with higher test scores and eight factors that are not. In looking at the 16 factors, the authors write: “To overgeneralize a bit, the first list describes things that parents are, the second list describes things that parents do.” (emphasis added) Let’s take a closer look. Here are the eight factors that are correlated with higher test scores, with my strikethrough on those that do not seem to fit their description:

* The child has highly educated parents
* The child’s parents have high socioeconomic status
* The child’s mother was thirty or older at the time of her first child’s birth
* The child had low birthweight
* The child’s parents speak English in the home
* The child is adopted
* The child’s parents are involved in the PTA
* The child has many books in his home

I think I am being quite generous here by counting education as something parents “are” rather than something parents “do” (and the same might be said for “high socioeconomic status,” which is often dependent on effort rather than status), but even giving them that one, it seems pretty clear that their “overgeneralization” isn’t much of a generalization at all, but a distortion. To be sure, the list of uncorrelated factors is all about things parents do, but that only tells us that many things we do are not likely to influence test scores. No big surprise there. The main point the authors are trying to make is that child-rearing “technique looks to be highly overrated.” Perhaps that is true, at least when the only thing you measure is test scores.

Cross-posted at Conglomerate.

33 comments for “Freakonomics

  1. I suppose that a blind spot the size of one of the lesser continents* is to be expected, just by looking at the title. It’s not hard to imagine that our esteemed authors have chosen to assess all value via an economic model — and what is economics if not metrics-obsessed? Unfortunately, this sounds like the kind of blind spot that not only prevents them from applying any non-metric value assessment, but prevents them from even realizing that there are values which their model can’t measure.

    *”Lesser” by the standard measurements of surface area, naturally. Not meaning to make a qualitative determination.

  2. main claims on anecdote which any economist with statistcal training tells you = incorrect.

    Well, you’ve saved me from buying the book.

    The authors made a serious mistake.

    xyz = matters that combine to make a baseline. Lack of xyz does result in dropping below the baseline. Therefore xyz don’t matter because it is -xyz that does.

    Interesting that “The child had low birthweight” correlates with higher test scores. It used to be a negative.

    But, seriously, it appears that mostly things that affect what parents do (parents with more education interact differently with children, buy more books, etc.), or that are things that they do, creates the result that the author is claiming it doesn’t. So that the entire chapter is written for effect, and to take digs at obsessive parents locked in the back room with the flash cards (some of the obsessive types do seem to be a bit of a pain) and to completely ignore happiness (and its effect on EQ).

    Thanks for the review.

  3. quote from the review linked to:

    One of his best-known, and in some quarters notorious, findings concerns America’s falling crime-rate during the 1990s. Towards the end of that decade, confounding the expectations of most analysts, the teenage murder rate fell by more than 50% in the space of five years; by 2000, the book notes, the overall murder rate was at its lowest for 35 years. Other kinds of crime fell too. Why? Some gave the credit to economic growth; others to gun control; still others to new methods of policing, or to greater reliance on imprisonment, or to increasing use of the death penalty, or to the ageing of the population.

    Mr Levitt goes carefully through these various explanations, checking them against the evidence. He finds that some of them do offer a partial explanation (more jail time, for instance), whereas others do not (greater use of the death penalty, new policing methods). But the most intriguing finding was that one of the most powerful explanations had not even been broached. That explanation was abortion.

    The reasoning is simple enough. In January 1973, the Supreme Court made abortion legal throughout the United States, where previously it had been available in only five states. In 1974, roughly 750,000 women had abortions in America; by 1980, the number was 1.6m (one abortion for every 2.3 live births). “What sort of woman was most likely to take advantage of Roe v Wade?â€? the book asks. “Very often she was unmarried or in her teens or poor, and sometimes all three…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…In the early 1990s, just as the first cohort of children born after Roe v Wade was hitting its late teen years—the years during which young men enter their criminal prime—the rate of crime began to fall.â€?

    The theory is the easy part, once you dare to articulate it. Testing it is quite another matter. But the book moves methodically and persuasively through the statistical evidence. It turns out, for instance, that crime started falling earlier in the states that legalised abortion before Roe v Wade; that the states with the highest abortion rates saw the biggest drops in crime (even controlling for other factors); that there was no link between abortion rates and crime before the late 1980s (when unborn criminals, as it were, first began to affect the figures); and that a similar association of crime and abortion has been found in other countries.

  4. Gordon, thanks for this review.

    Does anyone have any theories on the low birth weight business? The only (really speculative) theory I can come up with would be: low birth weight is correlated (is it, though?) with multiple births, which are correlated with use of fertility drugs, which is correlated with higher age and income of parents (two other factors).

    And did it occur to anyone else that perhaps the reason handgun deaths are lower than swimming pool deaths is precisely because parents are doing a better job of keeping their children awar from the former?

    I’d been working on a post about the effect (if any) that data should have on our parenting choices. This book, I think, illustrates exactly the kind of thing that bothers me in the (ab)use of parenting data.

  5. Ethesis and Julie,

    Sorry, that was a bit confusing (though I think the book presents it this way, too). The list is factors that are correlated with test scores, and low birth weight and adoption are both correlated with lower test scores. The other six factors are correlated with higher test scores.

  6. It occurred to me last night that the three things parents “are” from my list really could be presented as three things parents “do.” I noted that about education and socioeconomic status, but it is also true of waiting until the age of 30 to have your first child. If you add all of those things to the others that parents can “do” to improve test scores — namely, speak English in the home, get involved in the PTA, and buy books for the home — then you have a list of six things that parents can do to affect test scores positively. Had the authors wanted that to be their conclusion, it seems like it would have been pretty easy.

  7. This sounds like the economic corollary to Judith Harris’s “The Nurture Assumption” of a few years back, which makes the argument from psychology that parents matter a lot less than peers do.

    On the working mother thing, it was my understanding that daycare kids actually did better on early standardized tests than at-home kids, but were also more aggressive.

  8. Interesting review. But I think there is a case to be made against over-parenting. Whether the book makes it or not appears to be debatable. There is something to be said for letting the back yard and empty field raise your kids. I’ve known plenty of parents who do all the wrong things. They fight in front of their kids, their children run around like ragamuffins, they don’t “childproof” their house, they don’t read books on parenting techniques or get them into preschool, they don’t hold FHE or wring their hands over the question of spanking. Yet if they LIKE their kids (love is a lot less relevant to parenting), the kids seem to turn out great. This evidence may be anecdotal, but when it comes to parenting, what else is there?

  9. If kids born to mothers over 30 have higher test scores, it is almost certainly not a direct effect of the mother’s age, but an ancillary effect of having the over-30 class disproportionately represented by intelligent women who would have had smart kids no matter that she waited until 30. Babies conceived by older women are more likely to have imperfections caused by deterioration of the ovum’s genetic material. (All of a woman’s eggs are the same age — she produces all of them before she’s born.)

  10. Gordon: I listened to book on CD and I don’t have it handy to check the accuracy of what I am about to say, so I may be out to lunch with this comment. However, I don’t think you are being entirely fair in your description of what parents are versus what they do. When the authors discuss what parents “do”, I thought they were referring to activities which they do as parents with their children. So, with respect to your example of owning books, they contrasted parents who have a lot of books in the house with reading to children. They argued that reading to your children is not positively correlated, but owning a lot of books is positively correlated. Although owning books is arguably something parents do, it is a reflection of their own education and personalities and is not a parenting activity such as reading to your children.

  11. Matt: Based on my recollection of the book, the authors would agree with you. Having your first child when you are more than 30 years old, tells us something about the mother, and probably both parents.

  12. It is worth pointing out that the chapter on parenting is only one chapter. Other issues discusssed in the book include questions such as why has the crime rate dropped significiantly, why do drug dealers still live with their mothers, how much do school teachers cheat to improve the scores of their students on standardized tests and how much cheating is there in sumo wrestling. The section on parenting and the section claiming that legalized abortion was a significant contributing factor to the reduced crime rate are probably the most controversial. Overall, I thought the book was quite interesting, but then again, I was an econ major.

  13. Gary, I thought the book was interesting, too, but I suspect Levitt’s papers would be more interesting. He doesn’t write about topics I normally read in economics, but the parts of the book I found objectionable read more like a journalist than a scholar.

    You have a good memory. They contrast having many books versus reading “nearly every day.” The former is correlated with test scores and the latter is not. Reading was one of the eight factors that was not correlated with test scores, and as I wrote above, “that only tells us that many things we do are not likely to influence test scores.” If that were all they tried to claim, it would have been interesting, but apparently not interesting enough because they felt the need to take this extra step of contrasting what we “do” with what we “are.”

    My point is that “having a lot of books” is not fairly characterized as what parents are. You suggest a theory under which owning books may be a measurable proxy for what parents are: “Although owning books is arguably something parents do, it is a reflection of their own education and personalities and is not a parenting activity such as reading to your children.” They suggest the same theory: “most parents who buy a lot of children’s books tend to be smart and well educated to begin with.”

    Notice the leap of logic? You both infer that having a lot of books says something about the parents’ education and smarts, and further that these attributes are the reason for higher test scores. I have two problems with this: (1) if that is what they mean, then they are double counting, since “having highly educated parents” is already on the list and having parents with a high IQ was discussed outside of these lists as an obvious cause of higher test scores; and (2) it assumes a connection between having books and a well-accepted “cause” of higher test scores (parental education or IQ) without proving the connection. Perhaps having lots of books around entices children to read on their own. (They make fun of this possibility, but I think it is not only possible but likely.) The fact is that they don’t provide any evidence on the connection between having lots of books and higher test scores; they just assume an explanation that fits their theory of the world. All we really know is that owning a lot of books is correlated, and that looks to me like something parents do.

  14. One more thought. Matt’s argument and Gary’s argument suggest that it might be possible to scrunch almost all of the factors into two: the child has highly educated parents and the child’s parents have high socioeconomic status. Even if this is true and even if these are properly characterized as things that parents “are” rather than things that parents “do,” it is worth remembering that the only output measure here is test scores, so doing other things might still matter.

  15. Gordon: My recollection is that they were contrasting owning lots of books, not just with reading every day, but with reading to your children every day. I thought they were to trying to distinguish between certain parenting styles or activities, such as reading to your children, staying at home rather than working outside the home, sending them to “Head Start” programs (and others that I have forgotten) with other indicators of the parents’ personality or aptitudes. According to them, there seems to be little or no correlation between one’s parenting activities and a child’s academic performance. However, there is a correlation between a child’s academic performance and certain factors which they take to be indicators of the parents’ personality and aptitudes.

    I am not so sure they are really double counting in the sense you describe. At least, my interpretation of their arguments was that there is really one primary cause of the child’s performance, and that cause is a genetic one. I did not think that the factors which they described as correlated were thought to be independent causes, but rather different indicators that certain kinds of parents have high performing children, and that parenting styles such as reading to your children regularly are not correlated when controlled for these other factors. (But then again, this is a point that I was unclear about and I remember wishing as I heard this chapter on the CD that I could stop and cross examine the authors on this point because it was not entirely clear to me.)

    I agree with you that they did not provide any evidence on the connection between having lots of books and higher test scores. However, I thought that they were not talking about parents with lots of childrens books, but parents who had lots of books in general. I did not understand them to argue that the act of buying lots of books would result in your children performing better, but rather that parents who own lots of books are more likely to have children who perform well in the academic arena. Thus, I think I go back to my distinction (and, I think their’s) between doing something directly with or for your child, and doing things which are reflections of your own personality and which are taken to be proxies for the qualities that are measured on academic tests.

  16. One more thought on Gordon’s one more thought. I would agree wholeheartedly that they were talking about test scores and not about other, more important, aspects of a child’s character and personality. However, I must admit that I do often wonder how much of our character, such as our integrity, or our ability to feel and express empathy and charity, are heavily influenced by our genes or other biological causes. But that is a discussion for another day and another gazillion books.

  17. Maybe off the subject, but a question: If I mess up, if I hurt my child through my sins or shortcomings, then I truly repent, it’s as if the act never occurred, then any harm I caused my child is negated, and any harm my child causes another through my mistreatment is negated….is there an alternate place somewhere where our children are not screwd up? How does that work?

    I believe strongly in nature, because my son was so much like his father, who he never knew. The same crooked smile, slump of shoulders, voice, sense of humor, etc. I don’t have any answers to your question, Gary, but I have seen it first hand, the influence of genetics.

  18. “I’d like to see more discussion on the abortion-lowers-crime issue.”

    Okay, here’s my two cents. Abortion most likely does lower the crime rate, but so what? Pro-choicers who cite this connection act as if it bolsters their case, which is pretty sick. Are we to lower crime through eugenics and infanticide (which is what I think it amounts to)? Sounds like fascism to me.

  19. Ok, lots of points. Here are a few clarifications:

    * Yes, Gary, reading to children everyday. I think you have it exactly right. They want to make the point that “parenting technique” is irrelevant, though I am not sure where “parenting technique” leaves off and parenting personality and aptitudes begin.

    * Children’s books or other books? Most of the chapter they refer to books, but in the part I quoted, they refer to children’s books. I’m not sure it matters too much for their argument.

    * On genetics: early in the chapter, they assert that “a long line of studies … [has] already concluded that genes alone are responsible for perhaps 50 percent of a child’s personality and abilities.” They want to know “what accounts for the other half?”

    * As to whether parenting techniques matter beyond test scores, my best guess is that is matters somewhat, but that genetics and peers and other environmental factors matter a lot, too. The authors mostly punt on this.

    * The abortion-reduces-crime argument is pretty straightforward, though problematic in many ways. Here goes: the women who most benefitted from legalized abortion were poor women, both because the price of abortions was was reduced and access increased. The children of these women would have been the most likely to commit crimes, not only because they are poor but because they are “unwanted.” They write: “Legalized abortion led to less unwantedness; unwantedness leads to high crime; legalized abortion led to less crime.”

    Two points: (1) this might be true, but the argument is so sketchy in the book that it is hard for me to evaluate (and I am not interested enough in the point to read the paper); (2) the authors go through a thought experiment at the end of the chapter and conclude, “the trade-off between higher abortion and lower crime is, by an economist’s reckoning, terribly inefficient.”

    Nevertheless, this apolitical economist, who allegedly speaks only the truth as revealed by the data, concludes: “What the link between abortion and crime does say is this: when the government gives a woman the opportunity to make her own decision about abortion, she generally does a good job of figuring out if she is in a good position to raise the baby well. If she decides she can’t, she often chooses the abortion.” It is hard to know where to begin with reasoning like that, though I am sure that others will have some idea. Let me make the modest point: you simply cannot get to the conclusion that a woman generally does a “good job of figuring out if she is in a good position to raise the baby well” from aggregate data on abortion and crime. You may believe they authors are right and they may be right, but this is not someone who is reasoning from data.

  20. I wonder whether the main stream media’s very quick and strong approval of this book is fueled because it comes down on the side of “abortion is a good thing.” Abortion is one of the great divisive issues of our time, and there is no doubt on which side of this issue we find the main stream media.

    I have not read the book, so I cannot really comment on it. But positive correlation is not always definitive.

  21. I think it is evident that Mormon parents think that everything they do will have such a profound influence on how their kids “turn out.” However, as parents we can only do so much. Our kids don’t come to us as a blank sheet of paper for us to write on and create a perfect masterpeice.
    OUr children may have our genes (usually) but they also have their own spirit. And we may try to set up their environment to the extent we have control over it, but they also come with agency.
    When we look at our precious babies sent to us, they are angels and we love them so much. We have a huge list of hopes and dreams for them. We want them to be happy, succesful, faithful, etc.
    Will we love them the same if they fail to live up to any of those expectations? And if they fail, will we blame ourselves for our failure or will we blame them for not learning what they taught?

  22. Levitt covers up a mountain of evidence raising severe empirical doubts about his abortion-cut-crime theory. For example, the first generation born after the legalization of abortion had, by far, the highest murder and serious violent crime rates in recent history. My introduction to this important subject is here:

    http://www.amconmag.com/2005_05_09/feature.html

    For further data and analysis, see here:

    http://www.iSteve.com/abortion.htm

    As for the parenting advice, Levitt is basically following Judith Rich Harris’ 1998 “The Nurture Assumption,” which argued that genes and peers are more important than what parents do. I took partial exception to Harris’ view in book review here:

    http://www.isteve.com/nurture.htm

    where I show a number of important areas where family nurture is dominant.

  23. Addendum: The first generation born after legalization had by far the highest _juvenile_ (up through age 17) violent crime.

    When Levitt concocted his theory back in 1999, he forgot to look at narrow breakdowns of the crime rate by age, so he simply assumed that crime was falling first among those born after legalization. When I pointed out in our Slate.com debate that it first fell among those born before legalization, his name was already tied to this theory, so he’s just bulled onward.

  24. Thanks for this review, Gordon. And Steve Sailer’s comments (including links to his reviews and articles) really helped me put the positive media hype about Freakonomics into perspective, and clarify the problems with the abortion-cut-crime theory.

    JKS (#24) added some good points. The “blank paper” analogy has always seemed incomplete to me, because it seems to leave out heredity, and also (as JKS mentioned) the fact that our spirits existed as individual entities long before they entered our physical bodies.

    Perhaps spirits are assigned to a particular parentage because what they get in terms of heredity and environment (including parenting skills) will be sufficient to allow them to progress and reveal their true character (on earth and in the spirit world) before the final judgment. I think that longer time frame also allows for all the positive adjustments that annegb (#14) refers to in terms of the effects of the Atonement.

    I think anyone who has children sees enough evidence in their lives to feel that what they as parents do, matters. In the case of children with learning disabilities, what a parent does (personally, or through obtaining professional help) can definitely affect test scores. And as been mentioned above, test scores are probably not the most important things that parents influence.

  25. Gordon:

    Did Levitt’s book cite any studies that deal with qualitative changes in violent crime since the late 70s, or did he only address quantitative changes?

    Quantity is only one side of the crime coin, and I have a suspicion–not verified; I haven’t researched this–that violent crime has become more perverse, more gross since the 70s. Anyone know? If so, it adds an additional dimension to the debate that would need to be adequately addressed before someone could assert that abortion has improved society from a criminal standpoint.

    If my suspicion is correct, is it possible that–giving Levitt the point–while abortion may have reduced aggregate criminal numbers, it has also–by devaluing life–given rise to a new level of callousness in violent crimes, particularly against women and children?

  26. When the authors decided to premise the book on the assumption that higher standardized test scores is a good thing, they are only following what most of American society seems to already believe.

    Our entire education system is being subordinated to standardized test performance. It’s not as bad as Japan yet, but it is certainly heading there. This obsession with performance tests, however, is simply a symptom of a larger problem:

    Our society assumes that the only thing education is good for is making money. If something doesn’t translate into higher salaries, it is quickly dismissed.

    In the Book of Mormon, we call this “priestcraft.”

  27. Richard T: My recollection is that Levitt was discussing reasons for the decline in the crime rate in general and that he noted that there was a decrease, not an increase, in violent crime during the 90’s.

    Seth: The book is not premised on the assumption that higher standardized test scores is a good thing. Most of the book has nothing whatsoever to do with standardized tests. The chapter on the correlation between academic performance and other factors is simply his attempt to identify those factors which are correlated and those which are not. It really has nothing to do with the utility of standardized tests. There is, however, another chapter in the book which analyzes data showing evidence of significant cheating by the teachers who are required to administer standardized tests to their students.

  28. Mr. Sailer,
    Thanks for those links. Your article on on nurture was very sensible.

    Seth Rogers,
    Amen.

  29. There was a book published about 10 years ago titled “What Money Can’t Buy’. Dr. Meyers, a sociologist at Chicago University, ran some regression equations on longitudinal data to produce the familiar correlation of wealth and performance on standardized tests.
    Then she controlled for wealth, looked at some other variables, and ran it again. The strongest variables were the ratio of time spent watching vs time spent reading to the child (less TV is better), and the consistancy of discipline within the home (observed during face-to-face interviews in the home). Both of these had stronger effects than wealth alone.
    I have not read the book, so I don’t know if they addressed Dr. Meyer’s study. If they didn’t, I’d consider them guilty of cherry picking the research.

  30. I liked your review of Freakonomics. However, I read the book and liked it very much. Obviously a book like Freakonomics should not be viewed as gospel. Every experiment and attempt at analysis by any social scientist will probably be flawed. You can always stand on the sidelines and take potshots.

    I liked the book because it is accessible to the layman and quite thought provoking. I would definitely recommend it to anyone. Sure it would be nice if Leavitt had just written it himself, but nerdy economists are impossible to understand. I am sure a book solely by Leavitt would have put us all to sleep, so I am glad he had a reporter dumb it down.

Comments are closed.